USF Libraries

The origin and evolution of the Easter Seamount Chain

MISSING IMAGE

Material Information

Title:
The origin and evolution of the Easter Seamount Chain
Physical Description:
x, 265 leaves : ill. (some col.) ; 29 cm.
Language:
English
Creator:
Liu, Zhengrong
Publisher:
University of South Florida
Place of Publication:
Tampa, Florida
Publication Date:

Subjects

Subjects / Keywords:
Easter Seamount Chain   ( lcsh )
Dissertations, Academic -- Marine Science -- Doctoral -- USF   ( fts )

Notes

General Note:
Includes vita. Thesis (Ph. D.)--University of South Florida, 1996. Includes bibliographical references (leaves 212-223).

Record Information

Source Institution:
University of South Florida
Holding Location:
University of South Florida
Rights Management:
All applicable rights reserved by the source institution and holding location.
Resource Identifier:
aleph - 023484029
oclc - 37046546
usfldc doi - F51-00197
usfldc handle - f51.197
System ID:
SFS0040015:00001


This item is only available as the following downloads:


Full Text

PAGE 1

THE ORIGIN AND EVOLUTION OF THE EASTER SEAMOUNT CHAIN by A dissertation submitted in partial fulfillment of the r e quirements for the degree of Doctor of Philosophy Department of Marine Science University of South F l orida December 1996 Major Professor: David F. Naar, Ph D

PAGE 2

Graduate School University of South Florida Tampa, Florida CERTIFICATE OF APPROVAL Ph.D. Dissertation This is to certify that the Ph.D. Dissertation of ZHENGRONG LIU with a major in Marine Science has been approved by the Examining Committee on August 20, 1996 as satisfactory for the dissertation requirement for the Doctor of Philosophy degree Examining Committee: Major Professor: David F. Naar, Ph.D Member: John S. Compton, Ph.D. Member: Marc J. Defant, Ph D Member: Albert C Hine, Ph.D. Member: Sarah E. Kruse, Ph.D Member: Sarah F Tebbens, Ph.D.

PAGE 3

ACKNOWLEDGMENTS I would like to express my sincere thanks to Dr. David F. Naar my major advisor for his guidance and support during my graduat e study at the Department of Marine Science, University of South Florida I am grateful to him for providing luxurious scientific facilities and all the related geophysical data in my study area especially for creating the chance for me to pursue a Ph D degree in an easygoing but productive environment. Thanks to my c ommittee m e mb e rs Dr. J Compton, Dr. M. Defant Dr. A. Hine Dr. S. Kruse, and Dr. S. Tebbens for sharing their time and providing me access to their lab facilities, and for their valuable comments on my dissertation Special thanks to Dr. Edward Van Vleet, our international student advisor, for his advice He has always been generous and helpful. I appreciate comm e nts and suggestions from Dr. T Atwater Dr. M -H. Cormier, Dr. D Forsyth Dr. R. Hagen Dr. K Macdonald Dr. G Neumann, Dr. P Ihinger, Dr. D Sche irer, Dr. J.-G Schilling and Dr. D Turcotte. I thank Dr. R. Hey Dr. F. Martinez, Dr. R Duncan, Dr R. Hagen, and Dr. S Tebbens for providing me with the related geophysical data used in this dissertation. Some programs were provided by Dr. M.-H Cormier Dr. D. Forsyth, Dr. R. Hagen, Dr. K. Macdonald, Dr. S. Kruse Dr. S. Miller, Dr. G. Neumann, Dr. V Paskevich, Dr. D. Scheirer, Dr. W Smith Dr C Weiland, Dr. P. Wessel, and Dr. D Wil s on for data processing Many people have provided various assistance and shared their expertise during this study. I would like to extend my thanks to D. Myhre for maintaining our compute r faciliti es, and M. Kuykendall, Y Rappaport and X. Liu for revi e wing the early drafts of this dissertation Further acknowledgments are listed in th e individual manuscripts Fund ing is f rom the NSF USF, and the American Chemical Society (24362-02).

PAGE 4

TABLE OF CONTENTS LIST OFT ABLES iv LIST OF FIGURES v ABSTRACT viii ThiTRODUCTION 1 CHAPTER 2. SIDE-SCAN PROCESSING OF GLORI-B AND SEABEAM 2000 4 ABSTRACT 4 nnRODUCTION 4 GLORI-B AND SEABEAM 2000 SYSTEM DESCRIPTIONS 7 DATAANDFORMATS 8 IMAGE PROCESSING PACKAGE 10 IMAGE DATA PROCESSING 10 PROCESSING SEABEAM IMAGE DATA 12 Converting Image Data to CDF Files 13 Shading the Image 14 Removing Striping Noises 14 Smoothing Heading 16 Flipping the Swaths 16 Converting from CDF to IMG Format 16 Mosaicking the Images into a Single Grid 17 PROCESSINGGLORI-B IMAGEDATA 20 Preparing Navigation Data 21 Converting Image Data to CDF Format 21 Incorporating Navigation Information into the Image Data Files 22 Computing the True Ground Ranges 22 Shading the Image 23 Removing Striping Noises 23 Correcting the Aspect-Ratio 23 Registering the Image into Grids 24 Mosaicking the IMG Files into a Single Grid 25 MERGING TilE IMAGE DATA 25 Calculating Histograms 26 Merging the Images 26 SUMMARY ANDFUfUREDIRECTIONS 28 CHAPTER 3 SWATH BATHYMETRY PROCESSING OF GLORI -BAND SEABEAM 2000 31 ABSTRACT 31 1

PAGE 5

INTRODUCTION GLORI-B AND SEABEAM 2000 SYSTEM DESCRIPTIONS DATA AND FORMATS BATHYMETRY PROCESSING PACKAGE LINE FILTER THE DATA PROCESSING PROCEDURE SEABEAM 2000 BATHYMETRY DATA PROCESSING Correcting CrossTrack Bias Converting Data to GRD Format FlliNG THE TOPOGRAPHY DATA OF THE ISLANDS INTO THE SEABEAM 2000 GRID GLORI-B BATHYMEfRY DATA PROCESSING Procedure of GLORI-B Data Processing Preparing Navigation Data Converting the Data into RAS Format Reducing Random Noises and Filling Small Holes Cutting the Dropped Edges Removing CrossTrack Biases Reducing AlongTrack Biases Removing Duplicated Pings Converting the Data from RAS to GRD Format Trimming the Data Clipping Speckles Calibrating GLORI-B Data Using the SeaBeam Grid Merging GLORI-B and SeaBeam Grids Interpolating the Merged Data to Fill the Data Gaps Reducing the Along-Track Bias Further Restoring the SeaBeam 2000 Data FlliNG THE REST OF THE GRID ASSESSMENT OF PROCESSED BATHYMETRY SUMMARY CHAPTER 4 FORMATION OF THE EASTER SEAMOUNT CHAIN AND IMPLICATIONS FOR DEEP EARTH STRUCTURE ABSTRACT INTRODUCTION Review of Previous Work Hot spot Other models Age measurements Tectonic Setting DATA AND OBSERVATIONS ALONG THE ESC En-echelon Volcanic Ridges En-echelon Ridges Elsewhere Failed Propagators Fracture Zone Seamounts Volcanism and Lineations Seamount Morphology Progression of Side-scan Intensity Seamount Age Progression HOT BUBBLE MODEL 11 31 34 35 39 40 42 45 45 47 48 49 50 53 54 55 60 60 61 61 61 62 63 63 64 64 65 66 66 67 70 72 72 73 76 76 78 79 80 84 84 90 92 93 94 95 99 102 104 110

PAGE 6

Constraints for the Model Descriptions of the Model Applications of the Model Implications of the Model CONCLUSIONS CHAPTER 5. EFFECTIVE ELASTIC THICKNESS OF THE LITHOSPHERE ALONG THE EASTER SEAMOUNT CHAIN ABSTRACf INrRODUCTION METHOD The Forward Method The Admittance Method The Coherence Method RESULTS Along the ESC and near the San Felix Island In the Easter-Salas y Gomez Region DISCUSSION CONCLUSIONS CHAPTER 6 EVOLUTION OF THE SOUTHEAST PACIFIC AND THE EASTER SEAMOUNT CHAIN ABSTRACf INrRODUCilON DATA INrERPREr A TION VOLCANIC AGE PROGRESSION ALONG TilE ESC RECONSTRUCTION OF TilE TEcrONIC HISTORY Chron 10 Chron 7 Chron 6b Chron 6 Chron 5b Chron 5ab Chron 5a Chron 5 Chron 3 Present DISCUSSION CONCLUSIONS REFERENCES APPENDICES VITA APPENDIX 2.1. IMAGE PROCESSING P ACK.AGE APPENDIX 2.2. IMAGE PROCESSING SCRIPT APPENDIX 3.1. BATHYMETRY PROCESSING PACKAGE APPENDIX 3 2 BATHYMETRY PROCESSING SCRIPT APPENDIX 4.1. IMPLICATION FOR DEEP EARTII STRUCIURE APPENDIX 5 .1. SPECTRAL MODELS 111 110 113 118 122 124 126 126 127 131 131 137 140 141 141 146 158 162 163 163 164 170 193 197 201 201 201 202 202 203 203 204 204 205 205 210 212 224 225 229 238 243 252 257 End Page

PAGE 7

LIST OF TABLES Table 4 .1. Information about individual ridges Table 4 2 Relative migration rate between plate and plume Table 5.1. Default parameters Table 5.2. Elastic thickness estimated along the ESC Table 6 .1. Volcanic age of dredge s amples IV 87 119 134 158 196

PAGE 8

LIST OF FIGURES Figure 1.1. Tectonic location Figure 2.1. Location map showing the shiptracks of the Gloria Expediti o n Figure 2 2 Flow chart of the image data processing using the IP system Figure 2 3. Shading profiles calculated from SeaBeam 2000 and GLORI-B image data Figure 2.4. Side-scan images showing the results of the GLORI-B and SeaBeam 2000 image processing Figure 2 5 Histograms for GLORI-B and SeaBeam 2000 image data Figure 2.6 Three dimensional view of the data in the test area Figure 3.1. Tectonic location and s hiptracks of the Gloria Expedition Figure 3.2 Bathymetry data in the test area Figure 3.3. Flow chart of the bathymetry data processing Figure 3.4. Flow chart of the SeaB e am 2000 bathymetry data processing Figure 3 5 Flow chart of the GLORI B bathymetry data processing Figure 3.6. Average residual cross track profiles Figure 3. 7 Bathymetry maps in th e t est region showing the procedure of the data processing Figure 3 8 Assessment of the processed bathym e try data Figure 4.1. Tectonic setting of the ESC and major features in the southeastern Pacific Figure 4 2 Location map showing th e shiptracks o f Legs 5, 6 and 7 of the Gloria Exp e dition Figure 4.3. Comparison of differ e nt data sets in the Easter Salas y Gomez Region v 2 6 11 15 19 27 30 33 37 44 46 51 56 58 69 75 82 86

PAGE 9

Figure 4.4. Interpretation maps 89 Figure 4.5. Data near Easter Island and Salas y Gomez Island 98 Figure 4 6 The topographic profiles of some islands and seamounts in the ESC 101 Figure 4.7. Progression of the side-scan intensity along the ESC 103 Figure 4.8. Volcano ages along the ESC 106 Figure 4.9. Volcano ages along two of the major en-echelon ridges 109 Figure 4.10 A model of the deep structure of the Earth's dynamic system 112 Figure 4.11. Geometric models of hot bubble ridges on the seafloor 117 Figure 5.1 Tectonic location and estimation of the effective elastic thickness along the ESC 129 Figure 5 2 Example for forward method 133 Figure 5.3 Comparison of modeled and observed gravity 136 Figure 5.4 Theoretical patterns of an admittance model 139 Figure 5.5. Estimation of effective elastic thickness along the ESC using forward method 143 Figure 5.6. Estimation of effective elastic thickness along the ESC using admittance method 145 Figure 5.7. Comparison admittance and coherence methods 148 Figure 5.8. Estimation of effective elastic thickn ess along the Easter and Salas ridges using forward method 151 Figure 5.9. Estimation of effective elastic thickness along the Easter and Salas ridges using admittance m e thod 154 Figure 5.10. Non-iso s tatic compensation 157 Figure 5 .11. Comparison with studies elsewhere 160 Figure 6.1. Predicted seafloor bathymetry and age in the southeastern Pacific basin 166 Figure 6.2 Magnetic data and identified isochrons 174 Figure 6 3. GLORIA data interpretation for the Easter-Salas y Gomez region 177 Figure 6.4. A pseudofault imaged in Gloria data 181 VI

PAGE 10

Figure 6.5. Seafloor lineations near failed western propagator of the Mendoza paleo-microplate 183 Figure 6 .6. Southern portion of the failed eastern rift of the Mendoza paleo-microplate 186 Figure 6.7. Gloria transit near the Nazca-ESC intersection and the Nazca fracture zone 188 Figure 6.8. Gloria data near the San Felix Island 190 Figure 6.9 Interpreted tectonic features and magnetic isochrons 192 Figure 6.10 Seamount and seafloor ages along the ESC 195 Figure 6.11. Schematic reconstructions of the southern portion of the Nazca plate and its corresponding part on the Pacific plate 200 Vll

PAGE 11

THE ORIGIN AND EVOLUTION OF THE EASTER SEAMOUNT CHAIN by ZHENGRONG LIU An Abstract Of a dissertation submitted in partial fulftllment of the requirements for the degree of Doctor of Philosophy Department of Marine Science University of South Florida December 1996 Major Professor : David F. Naar, Ph.D. Vlll

PAGE 12

This dissertation addresses the origin and evolution of the Easter Seamount Chain (ESC) a major bathymetric feature in the southeastern Pacific, by processing and interpreting GLORI-B and SeaBeam 2000 side scan and swath bathymetry data, and modeling gravity and magnetic anomaly data. These data were used to test existing hypotheses such as the Hot Spot model, Hot Line model, Leaky Fracture Zone model, and Diffuse Extension model. None of the previous hypotheses explains all the observations. A modified Hot Spot model is proposed for the formation of the ESC after analyses of the available geophysical and geochemical data. The tectonic reconstructions of the ESC and the Pacific-Nazca plates since 30 Ma are constrained by side-scan, swath bathymetry, radiometric age, gravity, and magnetic data. Gravity modeling shows that the effective elastic thickness is close to 3 km along most of the ESC, -9 km on the older seafloor southeast of the Nazca fracture zone and -13 km near the San Felix Island. Bathymetry side-scan and radiometric data show a general age progression along the ESC, as predicted by a modified Hot Spot model and suggest that the San Felix Island is not part of the ESC Reconstructions suggest that the volcanic chains to either side of the East Pacific Rise were formed by separate hot spots, as previously proposed by some investigators. A microplate was formed as the Mendoza rift propagated northwards during chrons 5c (16.4 Ma) through 5a (12.3 Ma). The cessation of seafloor spreading along this propagator formed the Mendoza failed rift. A new Euler pole of Nazca hotspot motion, calculat e d using the trends of the ESC, the Galapagos chain, and the Juan Fernandez chain is located at 85.9"N, 171.4 "E. Side-scan image and swath bathymetry data near the western end of the ESC indicate that the ESC is composed of several volcanic ridges in a dextral en-echelon pattern The trends of the individual ridges are about 10 oblique to the overall trend of lX

PAGE 13

the ESC but close to the direction of the Pacific Nazca plate motion. Side-scan and radiometric dates along two major ridges show an age progression A Hot Bubble model is proposed to explain the general en-echelon pattern of volcanic ridges The model suggests that a weak unsteady plume may rise as a series of mantle blobs or bubbles which are sheared, while rising, into elongated ellipsoids and act as sources for the enechelon ridges Admittance gravity modeling shows slightly (1 0-15 % ) non isostatic compensation of the young seamounts at the far western end of the ESC suggesting dynamic support for bathymetric loads may be ass ociated with the nearby Easter hot spot Abstract Approved: Major Professor: David F. Naar, Ph D Professor, Department of Marine Science Date Approved : X

PAGE 14

CHAPTER 1 INTRODUCTION The Easter Seamount Chain (ESC), a major bathymetric and volcanic feature, runs about 3000 krn east-west in the southeastern Pacific (Figure 1.1). The origin of this feature is unknown. Several hypotheses have been proposed to explain the origin of the ESC, including the Hot Spot model (Hagen et al., 1990; Handschumacher et al. 1981; Hey et al., 1995; Morgan, 1971; Morgan, 1972 ; Okal and Cazenave 1985; Pilger and Handschumacher, 1981; Schilling et al., 1985a ; Wilson, 1963a; Wilson, 1963b; Wilson, 1973), the "leaky fracture zone" model (Clark and Dymond 1977 ; Epp, 1984), the "hot line" model (Bonatti and Harrison 1976; Bonatti et al., 1977), and the "diffuse extension" model (Sandwell et al., 1995; Winterer and Sandwell, 1987). Side scan image, swath bathymetry, gravity, and magnetic data collected during Legs 5, 6, and 7 of the Gloria Expedition aboard the RN Melville in early 1993 conducted along the ESC (Hey et al. 1995 ; Naar et al. 1993a; Naar et al., 1993b; Poreda et al. 1993a) are used in this study Interpretations are also constrained by the predicted bathymetry (Liu and Naar, 1996b; Smith and Sandwell, 1994) from global gravity anomaly calculated from recently declassified Geosat altimetry data (Smith and Sandwell 1995a ; Smith and Sandwell, 1995b) and ETOP0-5 data (National Geophysical Data Center, 1988). Previous magnetic data from National Geophysical Data Center are also used in the magnetic interpretation. The observations from these data are used to re-examine the previous hypotheses about the ESC and to reconstruct the tectonic history for the ESC and the southeastern Pacific New 40Ar -39Ar age data are calculated from dredge samples 1

PAGE 15

N 'Tl .... ()Q c:: @ ....... ....... t"'"' 0 () l:>l ..... .... 0 ::s 3 l:>l '? ti 0 ..... ..... (1) 0.. .... ::s (1) Vl e; (1) 3 l:>l ()Q ::s (1) ..... cs .... Vl 0 () g-0 ::s Vl ,........ Vl (1) (1) n ::r l:>l '"0 ..... (1) ..., 0\ "-' Location o active h o t s p o t x f a ilin g h o t spo t C Crou g h h ots p ot E East e r h o t s p o t J Juan F e rnandez S San F e li x hot s p o t i \ \ 1 0 9 8 7 \\ \ \ \ \ ------ISO' W 145' W 1 40' W 135' W 1 30' W 125' W 120' W tO' S IS' S 20 S 4_1" 25 S 6361> Sa Sab !ih 6.1 Sh 6a xld 5<: 6 35' S Mic:ruplatc tos-w too w

PAGE 16

collected during Leg 7 of the Gloria Expedition by Dr. R. Duncan and from the RN Sonne 80 survey (O'Connor et al., 1995). This dissertation is organized as five manuscripts with the authors listed in the Vita. Chapters 2 and 3 present the side-scan image and swath bathymetry processing, respectively. The image and bathymetry data collected with both GLORI-B and SeaBeam 2000 systems were processed and merged using the two software packages, Image Processing and Bathymetry Processing packages which we developed. Chapter 4 presents the interpretation of the side-scan image and bathymetry data along the western end of the chain. A modified hot spot model is proposed to explain the en-echelon volcanic ridges near the western end of the ESC. A new Nazca-hotspot Euler pole is calculated at 85.9 N, 171.4E using the trends of the ESC, the Galapagos chain and the Juan Fernandez chain. In chapter 5 the effective elastic thickness of the lithosphere is estimated along the ESC. The results suggest that a hot spot model provides the best explanation for the formation of the ESC, that the San Felix Island is not part of the ESC, and that some young seamounts near the western end of the ESC may not be isostatically compensated. In chapter 6, the evolution of the ESC and the tectonic history of the southeastern Pacific since 30 Ma are reconstructed based on the identified tectonic features and isochrons. The Mendoza paleo-microplate is proposed to have formed as the Mendoza rift propagated northward between chrons 5c (16.4 Ma) and 5a (12.3 Ma). The proposal that the Crough volcanic chain formed from a hot spot different from the one that formed the ESC is supported by the tectonic reconstructions 3

PAGE 17

ABSTRACT CHAPTER 2 SIDE-SCAN PROCESSING OF GLORI-B AND SEABEAM 2000 A software package has been developed to process GLORI-B and SeaBeam 2000 side-scan image data collected during Legs 5 and 6 of the Gloria Expedition in the southeast Pacific. We describe procedures to process the side-scan image data and to remove artifacts using the software package, which we call the IP (Image Processing) package The IP package is composed of 45 individual programs written in ANSI C. On line help is provided in each program Various data structures are used by the different programs i n the IP package and programs to display i mages on screen or generate maps f rom IMG files are included It also has interfaces to GMT (Generic Mapping Tool s) and WHIPS (Woods Hole Image Processing System) allowing incorporation of some GMT and WHIPS programs INTRODUCTION This paper discusse s the techniqu es and methods used to process side -scan image data collected using both GLORI-B and SeaB e am 2000 systems during Legs 5 and 6 of 4

PAGE 18

Figure 2.1. Location map. A: Map of the Southeast Pacific, showing the Pacific plate, the Nazca plate, and the Easter Microplate. The background image is the ETOPO 5 data with darker colors indicating the deep areas The large box indicates the survey region of the Gloria Expedition The small box indicates the test area that will be used to illustrate the data processing. B: The ship tracks of the expedition in the densely surveyed area C: The processed image. East of llO" W SeaBeam 2000 image data are shown along the center of the shiptracks; GLORI-B image data are used along the outer edges of the swaths. West of llO" W, only GLORI-B image data are used.

PAGE 20

the Gloria Expedition (Hey et al., 1995; Naar et al. 1993a; Naar et al., 1993b) along the Easter Seamount Chain in the southeast Pacific during 1993 (Figure 2 1). Swaths of bathymetry data were also acquired with both systems at the same time, which is discussed elsewhere (Liu and Naar, 1996d). Image data collected by the GLORI-B become noisy near the nadir; SeaBeam image data complements the GLORI-B data and provides quality data in this area (Mitchell, 1991; Searle, 1992). Various radiometric and geometric distortions and system artifacts are present in the raw image data (Mitchell, 1991; Mitchell and Somers, 1989; Searle et al., 1990). A software package, th e IP (Image Processing) package, has been developed for processing these image data sets (Liu et al., 1994). The initial task of image data processing is to remove distortions and artifacts from the individual SeaBeam and GLORI-B image data sets Afterwards, the individual data sets are mosaicked or merged into a grid and processed further. Finally, any remaining biases, distortions, and noises in the data must be corrected, removed, or reduced during the processing before the data can be used for interpretations and analyses. GLORI-B AND SEABEAM 2000 SYSTEM DESCRIPTIONS The GLORIA (Geological LOng Range Inclined ASDIC--Allied Submarine Devices Investigation Committee) system was designed to collect acoustic side-scan image data of the seafloor in deep oceans (Caress and Chayes, 1993; Goff and Kleinrock, 1991 ; Kleinrock 1992; Kleinrock et al., 1992; Rushy, 1970 ; Somerset al., 1978; Tyee, 1987; Vogt and Tucholke 1986) The system has recently been upgraded to collect bathymetry data as well and is now called GLORI B (Somers and Hugget, 1993) The GLORI-B is built in a fish vehicle, which is towed -300 m behind a ship at a depth of -50 rn. The instrument, consisting of two parallel rows of long transducer arrays on each side, 7

PAGE 21

transmits a 2-second-long 6.5 kHz FM pulse every 20-40 seconds (30 seconds during the Gloria Expedition) within a narrow beam (2 7 "). The returning signal is correlated (to compress the pulse) and is recorded in digital form for subsequent processing The intensity and travel time of the return signal are used to calculate imagery of the seafloor. The full image swath is 45-km wide. At 8 knots towing speed, the data have an along track resolution of-125m and a cross-track resolution of -45 m The SeaBeam 2000 is a hull-mounted multi narrow beam system designed toward production of bathymetry maps rather than seafloor acoustic images (Caress and Chayes, 1993; Goff and Kleinrock 1991; Kl e inrock, 1992 ; Kleinrock et al., 1992; Tyee, 1987 ; Vogt and Tucholke, 1986) However, seafloor acoustic intensity is recorded enabling acoustic imaging of the seafloor. The system transmits a 2 7 -wide beam at 7 msec-long pulse, which gives a range resolution of about 5 m for imagery and a cros s -track resolution of about 14m for the outer beams in 5-km water depth. The outer beams extend 60" to either side of vertical. The raw imag e data includ e both acou s tic intensity and beam angle, which has 12-bit resolution stored in 16 bit words, and a sub s et of 4 bit gray scale data, 1000 pixels per ping (D. W. Caress, personal communication, 1995). At a depth of 3 km, the system records image swaths of -10km wide DATA AND FORMATS The large box in Figure 2.1A (115 W/103 W 29 S/25 S) is the densely surveyed region where we have mosaicked the collected data (Figure 2.1B) and merg e d them into a grid (Figure 2.1 C). The small box shows the test area for which we will illustrate the procedure of th e image processing using the IP package There are no image gaps between adjacent -45 km wide swaths except for very shallow areas b e cause the distance between 8

PAGE 22

shiptracks is -30 km. The noisy image near the center of GLORI-B image swaths are replaced by high resolution SeaBeam image data that overlap 10-km of the GLORI-B image swaths long the shiptracks Data collected using various systems are stored or processed in files of different formats. Two major formats of binary files are used in the data processing and/or storage and described below: CDF and IMG formats. The former is a type of netCDF (network Common Data Format) (Unidata Program Center 1991). A header in each netCDF file provides information about the data. CDF Most of the GLORI-B and SeaBeam image data are converted and processed in files of this format to correct for radiometric and geometric distortions This is the same format used by WHIPS (Woods Hole Image Processing System) (Paskevich, 1992). In this format, image data are stored in two dimensional arrays Each row of the array is a ping of the image data across swath A swath of image data is represented by a number of consecutive pings along a shiptrack. Each ping is composed of a given number of sampl es of intensity values r e flected from seafloor. The header in each CDF file provid e s information such as time heading, altitude, location, the number of values for each ping, etc IMG GLORI B and SeaBeam image data ar e mo saicked merged, and stored in thi s format. This format is much like the GMT (Generic Mapping Tools) GRD format (Smith and Wessel, 1990; Wessel and Smith 1991) except for the size of the storage variable. In a GRD file, each cell is stored as a floating point variable, usually 4-byte An unsigned character is used to stor e the value of an IMG cell because the pixel values of image data range from 0 to 255, which can be represented by a single byte However, the value 0 is reserved for cell s without data. 9

PAGE 23

IMAGE PROCESSING PACKAGE The IP (Image Processing) package is composed of 45 programs written in ANSI C (Appendix 2.1). It is designed as an open package and has been installed on SUN/UNIX and SGIIIRIX platforms New programs can be added to it in order to handle more complex problems for image processing Some WIDPS programs (Paskevich, 1992) are also included in the package and are used in the data processing for image corrections Each of the IP programs can be run separately as a regular UNIX command or executed sequentially in a UNIX shell script to produce the desired result. Many of the programs have multiple functions and, therefore, can handle a set of similar tasks depending on the context of the command line Many of the programs use the GMT style syntax for setting command line parameters. An on-line help is provided in each of the programs to help users with the program usage Many of the programs were so named that provide indications of their input/output file formats. For ex ample img11Ws will tak e IMG files. Most of the IP programs are u sed for converting, mosaicking, merging and displaying the data sets Program grd2img converts the image data between the GRD and IMG formats It provides a linkage between the IP and GMT packages. The IP package also includes programs for displaying image data on screen or producing maps. Though the package is developed for GLORIB and SeaBeam 2000 image processing it can also be used for processing side-scan o r acoustic image data from other systems. IMAGE DATA PROCESSING Figure 2.2 shows the procedure of the image data processing. The UNIX script in Appendix 2.2 provides an example of the imag e processing u sing the IP package. Each 10

PAGE 24

Image Data Processing --r-==-==-==:....;;;;-1 I Raw GLORI-B Imag e Data I I Navi g ation Data 1 I I 1 (Convert the data to WHIPS fonnat) ( Convert into DXF Fonnat ) 1 I I 1 I Data in CDF files of WHIPS Format ( Convert into WHIPS format ) 1 Processing GLORI-B Image Legen d . I NaVJganon Data 10 WHlPS _ ____ 1 r (Merge navigation int o the data ( Cal c ulate h eadi n g from navigation data ) 11 I SeaBeam image data I ( Smooth headin g II ---l SJlC(:d log s I I Convert to WHIPS f orma t ) ( Correc t slan t range ) [r-S b adi n g Pro _fi_ tl-,e II Calculate s hadin g profile CDF files i n WHIPS f o rmat I ...-----'------....... .... ,./ I I c Shade the data with a profile r II I Shading Profile II Shade th e data with a profil e ) I Data after shading ___ _..V""'c;___ ....... (Filte r with box car (IX71j) (Filter with box car ( 9X71j) I H igh frequen cy co nponentl I Low frequency conponent I ( Comb i ne the two conponents ) I (Restor e headin g./ ::::: I ( Corre c t v elocity ) l Ve l ocity co rrected data f-( Smoo th data with a 2X2 box car) I (Restore h eading I Data after s hading I ___ ... _v/;;___, _____ ............. __ (Filte r with box car (I X 7 1]) (Filte r with box car (9X7 IV I High frequency con p o nent] I Low frequency conpo nent l ( Combin e the two conponents ) I Restore h eading) I ( Smooth h ea ding ) r{ Da t a aft e r heading s m oo thed J ( Aip the s wath ) _l b eading ) I (Convert int o liz format) Convert int o liz format) 1 II (Regis t e r t o grid in IMG format) II (Regis ter t o grid in IMG format) I I (Mosaic into a grid file in IMG forma!) J l (Mosaic int o a grid fil e in IMG forma!) -----------: I G L ORI-B data fil e in IMG f orma t Find over l a p area SeaBeam data fil e in IMG format I I I 't' I I ( Mask th e data A mask o f over l aped re g i o n M as k the da ta) 1 I I I GLORI-B data in maks I I SeaBeam data i n maks I 1 I I I ( Calcu.late hi s t ogram) Merging the Two ( Calculate hi s t ogram ) : 1 Image Data Sets 1 I : IIi ] GLORI-B hi s t ogram I SeaBeam ,.. hi stogram l I I ( A dju s t pixel value ofGLORI-B data) (Adjust pixel value of S eaBeam data)! I 't' (Merge the two dat a s e ts treating the) --;: Y : I I adjus ted GLORI B data data w i th higher pri o rity .r-1. adjusted SeaBeam data I L----------_I Figure 2.2. F l ow c hart of the ima ge data processing u sing th e IP software package Included in the das h ed lin es s ho w the individual parts of processing GLORI B image data processing SeaBeam 2000 image data, and merging the two data sets. 11

PAGE 25

following section corresponds to a portion of the script The GLORI-B and SeaBeam 2000 image data are processed separately to correct various distortions in the data. Then they are merged into a single grid of IMG file. Several radiometric and geometric distortions can be readily identified in image data (Chavez, 1986 ; Cobra, 1990 ; Hagen et al. 1996 ; Miller et al., 1991; Scheirer and Macdonald, 1993; Scheirer et al., 1996). These include : random and electronic noise such as speckle noise and striping ; problems related to the design and sampling characteristics of sonar systems such as power drop-off and multiple returns ; aspect-ratio distortion because of variations on ship speed; and heading offset because the heading is calculated from the shiptrack. These distortions are removed in the next two sections prior to data merging and mosaicking. PROCESSING SEABEAM IMAGE DATA The procedure for the SeaBeam image processing is shown inside the dashed line box on th e upper right porti o n of Figure 2 2 SeaB e am data are converted to CDF format processed in this format to remove and reduce radiometric and geometric distortions, and then converted to IMG format in order to merg e th e m with GLORI-B image data. The processing steps include : (1) format conversions: converting from SeaBeam files to CDF file s, calculating geographic coordinates, registering image data into IMG fil e s, (2) geometric corrections: aspect ratio corrections, 12

PAGE 26

* smoothing of heading, (3) radiometric corrections: shading correction, striping noise removal, (4) miscellaneous: creating a shading profile, flipping swaths, mosaicking the image. Converting Image Data to CDF Files In order to process SeaBeam image data in a format that maintains the lateral geometry of the data points in a swath, the data are converted to CDF format. Four-bit gray scale SeaBeam image data are preprocessed and stored in a format similar to the SeaBeam bathymetry data (Liu and Naar 1996d; Scheirer and Macdonald, 1993 ; Scheirer et al 1996) The program sbss2cdf (modified from Rick Hagen s program (Hagen et al. 1996)) is used to convert the image data to CDF ftles The SeaBeam system collects image data at various cross-track resolutions depending on the water depth. Therefore, it is likely that a single image file may include several segments of different resolutions. To restore a 1:1 aspect ratio when using velocity or gss_vel, data in a given image file must be of a constant resolution. Sbss2cdfidentifies the resolution for each segment in a image data file and splits into several ftles, each of which has a constant resolution The program converts the data into CDF format before saving them The resolutions of the CDF data files are stored in a separate log file. 13

PAGE 27

Shading the Image There is a systematic radiometric drop-off across the SeaBeam image swaths. A shading correction is applied to the data to normalize the image across track with respect to signal attenuation and the power drop-off characteristic of the system so that intensity values of the image data in the near and far range can be more accurately compared. This correction attempts to restore the value of a pixel to the value that would be expected in the absence of radiometric distortion. A shading profile is created from each CDF me using the program shade3gen This program calculates an average pixel value for each sample position and divides it by the average value of the entire image or a specified normalization value to produce a profile. From these profiles an average profile (Figure 2.3A) is computed using sumprof and awk. The average profile is stored in an ASCII file, shade3.dat, in the working directory, which is used by shade3 for applying the shading corrections to each of the data flies. Removing Striping Noises Striping occurs in the cross-track direction. During data acquisition some pings of sonar data have lower values than adjacent pings because of vehicle instability or noises in the water column. A combination of high-and low-pass spatial filters are applied to the data to remove striping noise. The method used to remove the striping noise is based on generating two separate images from the input data using filter. One image represents the high-frequency components present in the image, except for the noise frequency; the second image represents the low-frequency components without the noise The results of the two spatial filters are then added using wtcombo to create a new image that is very 14

PAGE 28

1.6 1.6 I A: SeaBeam Shading Profile I 1.4 1.4 i: 1.2 1.2 0 u u to:: t.:: ...... '-0 0 0 0 u u 1.0 1.0 tl) tl) c c 0 0 i: i: 0.8 0 8 0.6 0.6 -500 -400 -300 -200 -100 0 100 200 300 400 500 s tbd <--beam number across track -->port 1.6 1.6 1.4 1.2 u to:: '-0 0 u 1.0 tl) c 0 i: 0.8 0.6 I B: GLORI-B Shading Profile I 1.4 1.2 "() c '-0 0 u 1.0 0 8 0.6 tl) c 0 c -500 -400 -300 200 -100 0 100 200 300 400 500 stbd <-beam number acros s track--> port Figure 2.3. Shading profiles calculated from SeaBeam 2000 image data (A) and from GLORI-B image data (B). 15

PAGE 29

similar to the original one, but with less noise The filter shapes used to remove the striping noise from the SeaBeam images are a 1 ping by 71 samples for the high-pass filter, and a 9 pings by 71 samples for the low-pass filter. Smoothing Headings The SeaBeam is a hull-mounted system. The ship headings should be accurately recorded. However, frequent and sharp changes of the heading directions will introduce some data gaps in the cross-track direction, especially when the data are registered into a high resolution grid. A slightly smoothed headings will reduce the gaps. The heading is smoothed by averaging 5 pings using the program avg_heading. Flipping the Swaths The data points within each swath are registered from right to left in the SeaBeam files The program flip is applied to reorder data for each ping of an image file. The header distorted by flip is restored by restorehdr. Converting from CDF to IMG Format Geographical coordinates are computed using gss_vel for data points in image for registering the data onto a grid. The center beam location from the header provides a reference for other samples in each ping. The geographical offset from the center beam for 16

PAGE 30

each sample location is calculated from the heading of the ping and the distance from the center beam using simple triangulation From the offsets and center beam locations the geographical location for each datum is calculated The results are saved into a flle as a series of triplets (longitudes latitudes, and intensities). Before this computation the program also makes the aspect ratio correction by duplicating a number of pings between two consecutive pings so that the aspect-ratio of the cells is 1:1. The resolution information required for this program is provided from the log flle created by sbss2cdf The triplets are registered into IMG flles using .xyz2img. The grid size is .001 by 001 degrees. If there is more than one intensity value in a cell, the average intensity is calculated and used in that cell The cells with no data are set to 0. Mosaicking the Images into a Single Grid An IMG flle is obtained from each CDF flle in the previous step Each of the image flles is a small segment of the image swath covering a small portion of the surveyed area along the shiptracks These IMG files ar e finally mosaicked into a single grid using imgmos The average value is applied to cells if there are overlapped data points. At this point, we have finished proces s ing the SeaBeam image data. The SeaBeam image data (stored in the IMG file, SeaBeam.img) will be merged with GLORI-B image data later Figure 2.4C shows an example of SeaBeam image data in the test area. 17

PAGE 31

Figure 2.4 Image maps showing the results of the GLORI-B and SeaBeam 2000 image processing. A: GLORI-B image data The overlapped outer swaths are mosaicked by calculating the averages for the pairs of overlapped cells. B : GLORI-B image data. The overlapped outer swaths are mosaicked by selecting the larger value for the pairs of the overlapped cells. C: The SeaBeam 2000 data after processing D : M e rged GLORI-B and SeaBeam 2000 image data without matching their histograms. E : The final processed image data. The GLORI-B and SeaBeam 2000 data are merged after their pixel values have been adjusted so that their histograms will have similar patterns

PAGE 32

0 25 50 75 100 125 150 175 200 225 250 Intensity 19

PAGE 33

PROCESSING GLORI-B IMAGE DATA The GLORI-B side-scan image data have various systematic distortion s inherent in the sonar data which should be removed before any meaningful information can be extracted (Chavez, 1986; Cobra, 1990; Miller et al., 1991; Paskevich, 1992). This section illustrates the use of the IP package to process the GLORI B image data The GLORI-B image data are converted to CDF format for the subsequent processing A series of corrections are applied to each of the files to make the radiometric and geom e tric corrections including slant ranging s hading filtering, velocity correcting, box car smoothing, heading smoothing. Afterwards the data are converted and mosaicked into an IMG file The general flow of the data processing i s illustrated in the dash e d line box on the top left portion of Figure 2.2. Below is a list of the basic processes ( 1) format conversions : converting imag e data from RA WPS format to CDF format, calculating geographic coordinates, registering image data into IMG files, (2) geometric corrections: water column offset rem o val, slant t o-ground range projection aspect ratio corrections, smoothing of heading, (3) radiometric correction : shading correction, striping noise removal blocky effect removal, 20

PAGE 34

(4) miscellaneous processes: preparing navigation information, merging navigation data into image data ftles, calculating headings from navigation information, computing a shading profile, mosaick:ing the image. Preparing Navigation Data The ship navigation data must be incorporated into the image data for processing because the GLORI-B data do not include any navigation information. A navigation file is downloaded onto a ship-board computer daily and is available for at-sea processing. The navigation data are converted to DXF format using sbnav2dxf and merged into a single flle. This file is converted to the WHIPS format using dxj2whips, which will be merged into the image data later for the subsequent processing Converting Image Data to CDF Format The raw GLORI-B image data is stored in binary files and must be converted into CDF format for geometric and radiometric corrections. The program rawps2whips is used to convert a raw data file to a CDF file A raw file usually consists 6 hours of data, 737280 bytes (=720 ping x 1024 bytes). If a file contains less than this amount, the number of complete pings in the file must be determined and supplied to the program. 21

PAGE 35

Incorporating Navigation Information into the Image Data Files There is no navigation information in the above converted CDF files. The navigation data are merged into the image data using mrgnav. A corresponding segment of the navigation data is located and added into the header record of each CDF file. However, there is still no heading information in the navigation data just merged The track orientation is computed from the ship's navigation data using sshead, and is used as an approximation for the GLORI-B vehicle's heading. If there is no current and the ship travels in a straight line, there is no difference between the heading of the GLORI-B fish and the track orientation of the ship. However, ships seldom follow straight lines during surveying due to waves, wind, current, or the ship's navigation system instabilities. The GLORI-B fish follows a few hundred meters behind the ship, and thus has a much smoother track and heading. Therefore, a smoothed track orientation would more closely represent the true heading of the fish The headings are smoothed by averaging 41 pings using avg_heading. This smoothing process also minimizes/eliminates data gaps and holes in the gridded image (compare to the less smooth track orientation) Computing the True Ground Ranges The water column offsets are removed and the slant-to-ground-ranges are calculated based on the fish altitude and the image resolution. There is a period of time when the acoustic wave is traveling through the water column and not returning signals from the seafloor because the GLORI-B transducers start recording immediately after an acoustic pulse. This period appears in the image as a dark column on both sides of nadir. This 22

PAGE 36

distortion translates into a spatial displacement to the first bottom return The water column offsets are corrected using the fish's altitude. Data collected by the GLORI-B system are recorded as slant range distance, or travel time, with respect to the position of the sonar system. To measure the horizontal distance of seafloor features from the nadir, slant range must be converted to true horizontal, or ground range, distance. Assuming that the seafloor is flat, the slant-range is corrected to restore the proper cross-track geometry without introducing significant distortion using slant. Shading the Image Like the SeaBeam image data, there is a systematic radiometric drop-off across the shiptrack in the GLORI-B image. An average shading profile is calculated using all the GLORIB image data (Figure 2.3B). This profile is applied to each of the image files to make the shading correction. Removing Striping Noises The striping noises are removed using a similar method as described for the SeaBeam image data. Correcting the Aspect-Ratio During a survey the ship's velocity could change between 8.3 and 18.5 kmlhr (or 4.5 10 kts), which cause the pixel resolution in the along track direction to vary from 23

PAGE 37

approximately 70 to 150m for a 30-second pulse-repetition rate. In contrast, the pixel resolution in the cross-track direction is usually 50 m A program velocity is applied to regulate the aspect ratio and create a new image with a ratio of 1: 1. The distortion is removed by stretching a group of pings referred to as segments. Stretching is actually accomplished by duplicating individual pings The number of pings to be duplicated is determined by the aspect ratio calculated from the navigation data and the cross-track resolution After a ping is duplicated several times, a small area of image block is formed due to the constant pixel values. And there may be a step between two consecutive blocks. These effects must be reduced without over smoothing and blurring the image A box car filter of 2 by 2 cells is used to reduce the effect using lowpass2b2. The header record is destroyed by lowpass2b2 and must be restored with restorehdr. Registering the Image into Grids The geographical locations for the data points in each CDF files are calculated and saved into ASCII files as a series triplets of longitudes, latitudes and depths. These triples are then registered into a grid in an IMG flle as done for SeaBeam image data files. We limit the swath's width to be no more than 32 km during this conversion because the distance between tracks is -30 km. It is not necessary to keep data beyond 16 km from the nadir because they are overlapped by more accurate data from the adjacent swaths The grid size of these IMG files is again .001 by 001 degree 24

PAGE 38

Mosaicking the IMG Files into a Single Grid As we did for the SeaBeam data, the gridded image files are mosaicked into a single grid. In this case the GLORI-B swaths overlap more often side by side, and thus we attempt to utilize the better data in the overlapped areas during the mosaicking. There are three methods to choose the pixel values for the overlapped cells : (1) calculate an average for each pair of the overlapped cells; (2) select the smaller value of the pair; or (3) choose the larger value of the pair. The outer beams of GLORIA image data usually become dark due to the absence of return signals. There may be significant intensity contrasts between the outer beams and the inner beams of the adjacent swath. The first two methods would could not minimize these contrasts, and, therefore, likely produce two significant intensity contrast lines on both sides of the overlapped portion of swaths Figure 2.4A shows the mosaicked image using the first technique. Two parallel track lines can be identified between two swaths. Selecting the larger values for the overlapped areas during the mosaicking provides the best results (Figure 2.4B). The program imgmos is used again to mosaic the IMG flies with the option mode set to select the larger pixel value for overlapped cells. MERGING THE IMAGE DATA The two mosaicked grids GLORI-B and SeaBeam image data, are finally merged by replacing the inner portion of the GLORI-B swaths with the high quality SeaBeam 2000 image data along the corresponding shiptracks There is a systematic difference between GLORI-B and SeaBeam images. If the two sets of image data are merged without any modification, clear borders could be easily identified because of the intensity contrast 25

PAGE 39

between the two data sets (Figure 2.4D). To avoid this problem, we adjust the pixel values for one or both of the data sets so as to match their histograms before merging the two images. We stretch both of their pixel values to the full range of 1-255 because most of the pixel values of the images are compressed in the range between 20 and 120. Therefore, we match their histograms and extend their intensity range. Calculating Histograms Because the GLORI-B and SeaBeam images cover different areas only data in the overlapped areas should be used for comparison However, because the GLORI-B image near the nadir is so noisy, these cells should not be used, either. Therefore, only a few cells toward the outer beams of the SeaBeam swaths are used to compute the histograms. The two cells on the outer edges of the SeaBeam swaths are cut because they may be unreliable. The next five cells on the edges are isolated from the other cells (Figure 2 .5A) using the programs imgshrink and imgmath The program imghistgram is used to calculate the histogram from these cells (Figure 2 5B) The corresponding set of cells from GLORI B image are used to calculate a histogram for the GWRI-B image (Figure 2 5B ). Merging the Images Figure 2.5B shows that the profil es of the two histograms do not especially near the peaks. To merge the images we need to adjust the pixel values in the images so that their new histograms have similar proflles A program imgs c ale is designed to stretch or compress the range of the pixel values. Any two digital values divide the range of pixel 26

PAGE 40

C) '0 -26 7 -26.8 -26 9 A: Overlapped Pixels -26 7 -26.8 C) -27.0 -27 0 ; >. u c:: -27.1 -27 1 -27.2 -27 2 -27.3 -27 3 -107 5 -107.4 -107.3 -107.2 -107.1 107.0 -106.9 -106. 8 -106.7 -106 6 -106.5 longitude 0.020 t -;:=============::::;-'-----'-----'----'----'---'---r 0.020 I B: Histogram of Pixel Values I 0 018 0 016 0.014 0 012 I, II I I I I 0 I .. I I I 0 I I I I I I I I : / O I I O.Q18 0 .016 0.014 0 .012 >. u c:: I J \ : ', / I f .. -;: g 0 010 0" SeaBeam GLORIB 0 .010 g : : ._, "",,, r : "" l : : .... I I J \ I I I "' I I I 1 I I I 1 : : : I I I 0 008 0.006 : : : '\.t. I I y \ 0.004 I 0 I 0 I 1 0 I I I 0 0 : \ : ; , 0 002 I : . ...... ,\. -... 0 008 0.006 0 .004 0 .002 "'..... 0.000 .1-L ---.-.......c::.:::.-_-.---..-----.-----r----..,....--......-=::;=:=::;=;::::====;:===r=:Jl. 0 000 0 20 40 60 80 100 120 140 160 180 200 220 240 pixel value Figure 2.5. A: Pixels selected for calculating histograms in B. B: Histograms from GLORI-B image data (solid line) and from SeaBeam 2000 image data (dashed line). 27 0" C)
PAGE 41

values 0-255 into three segments By moving up or down one of the digital values and linearly changing the pixel values in the two segments on both sides we can stretch one segment and compress the other On the other hand, stretching or compressing the pixel values linearly in the three segments, we can match this pair of digital values to another given pair This method is used by imgscale to rearrange the pixel values of image data, therefore to change the histograms of the image By applying the program several times with different pairs of digital values, we can rearrange the histogram of an image to a given shape. Imgscale is used once for GLORI-B and SeaBeam image data, respectively, because the patterns of the histograms in our data are very simple A horizontal line at frequency 0 001 in Figure 2.5B intersects the histograms at pixel values 12, 161 for the SeaBeam profile, and 41.5, 155 for the GLORI B profile. Then imgscale is used to rearrange the pixel value of the images so that both pairs of values will match a new pair of pixel values, 20 and 220 Therefore, the pixel values in both of the images are stretched in the range between 20 and 220 with new histograms that are closely matched. Mter the above adjustment, the program imgmos is used again to merge the data by replacing the overlapped GLORI-B cells with SeaBeam data (Figure 2.4E). At this point, we have acquired a mosaicked grid with cell size 001 by 001 degrees in an IMG file that contains both GLORI-B and SeaBeam image data. We have SeaBeam image data for Leg 6 only There are no SeaBeam image data used to the west of 11 ow in the mosaicked region (Figure 2.1C). SUMMARY AND FUTURE DIRECTIONS Using the IP package, we are able to remove the distortions and reduce the system artifacts in both GLORI-B and S e aBeam image data The image data collected with the two 28

PAGE 42

different systems are fmally scaled to the same intensity range and mosaicked into an image in a regular grid for the Eas ter-Salas y Gomez region (104/115 W, 25/29 S) with resolution of about 100 by 100m. The major features identified from th e processed image data are reliable for data interpretation The data processing package IP is compatible with BP (Bathymetry Processing) package (Liu and Naar, 1996d), WHIPS, GMT, and MB (MultiBeam) system (Caress and Chayes 1993) The software IP is freely available by anonymous ftp at sunset.marine.usf.edu (contact zjl@marine.usf.edu or naar@marine usf.edu with questions). Our image processing did not consider the influence of topography on the sonar system Figure 2.6A is a three-dimensional vi e w of the seafloor in the test area from the processed GLORI-B and SeaBeam bathymetry data (Liu and Naar, 1996d) The intensity in Figure 2 6B is the processed image data. Th e dark peak on the seamount in the center of the figure is not real Rath e r the GLORI-B sys tem does not imag e b e yond a certain angle Therefore shallow seamounts at th e sides of the GLORI B swaths app e ar dark, but should appear brighter. Topographic information could be used to enhanc e the image in the future (Bird et al. 1996; Mitchell 1991), and fill in such types of obvious imagery gaps 2 9

PAGE 43

A : Topography of the Test Area B : Sonar Image Affected by Topog r aphy Figure 2 6. Three-dimensional view of the data in the test area A : Topography of the seafloor from processed GLORI-B and SeaBeam 2000 bathymetry data. B : The merged image data, showing the effects of seafloor topography on the GLORI-B imaging system which in this case forms an image gap (dark area ) within an area of e x pected strong returns (bright area) 30

PAGE 44

ABSTRACT CHAPTER 3 SWATH BATHYMETRY PROCESSING OF GLORI-B AND SEABEAM 2000 A software package has been developed to process GLORI-B and SeaBeam 2000 swath bathymetry data collected during Legs 5 6 and 7 of the Gloria Expedition in the southeast Pacific GLORI-B bathymetry is calculated using an interferometry method with a modified GLORIA towfish that now has parallel rows of transducers on both sides We describe procedures to process the bathymetry data and remove or minimize artifact<; using this software package, which we call the BP (Bathymetry Processing) package The package is composed of 37 individual programs written in ANSI C On-line help is provided with each program The BP package uses data structures compatible with the GMT (Generic Mapping Tools) package, allowing incorporation of some GMT and other compatible programs. INTRODUCTION This paper discusses the techniques and methods of processing swath bathymetry data collected with both GLORI-B and SeaBeam 2000 systems, during Legs 5, 6, and 7 of 31

PAGE 45

Figure 3.1. Location map A: Map of Southeast Pacific, showing the Pacific plate, Nazca plate, and Easter Microplate. The background image is the ETOPO 5 data with darker region indicating the deep areas Large box is the surveyed region of the Gloria Expedition The small box indicates the test area that will be used to illustrate the data processing B: The shiptracks of the expedition in the densely surveyed area, the large box. C : The processed bathymetry data in the large box. In the densely surveyed area, the data include the SeaBeam 2000 swaths centered along the shiptracks, GLORI-B data to the both s ides of SeaBeam 2000 swaths, the interpolated data between GLORI-B swaths. Out of the survey area there are SeaMARC II data in the area to the west of 109 w and ETOPO 5 data to the east of 109 w.

PAGE 46

longitude -120 -115 -110 -105 -100 -95 -90 -85 -80 -75 -10 -15 -20 4.> '0 -25 -30 -35 small box: location of the test area image : bathymetry from ETOPO 5 data -40 longitude -115 -114 -113 -112 -111 -110 -109 -108 -107 -106 -105 -104 -25 .. .................. ... ______ Cl.l '0 -26 -27 -28 -2 9 B: Ship tracks in Easter-Salas y Gomez Region -25 Cl.l '0 -26 -27 -28 -29 -5000 -4500 -4000 -3500 -3000 -2500 -2000 -1500 1000 -500 33 0 -70 -10 -15 -20 4.> '0 -25 -30 -35 -40 -103 -25 -26 4.> '0 -27 ..... .!!! -28 -29 -25 -26 4.> "0 -27 ..... .!!! -28 -29 500

PAGE 47

Gloria Expedition (Hey et al., 1995; Naar et aL, 1993a; Naar et al., 1993b) along the Easter Seamount Chain, southeast Pacific during 1993 (Figure 3 .1). The capability of collecting swath bathymetry was recently added to the GLORI-B (Somers and Hugget, 1993). However, there are many system artifacts in the data. A software package, which we call the BP (Bathymetry Processing) package, was developed to process the data (Liu et al., 1994). The BP package is used to process each of the data sets individually to remove or minimize the system artifacts first. Then the data are mosaicked and merged into a grid, and processed together to further reduce the artifacts. To accomplish these objectives, we must process data from different sonar systems and merge files of different formats. In addition, biases, distortions, and noises in the data sets must be corrected, removed, or reduced during the processing before the data can be used for interpretation or analyses GLORI-B AND SEABEAM 2000 SYSTEM DESCRIPTIONS The GLORIA (Geological LOng Range Inclined ASDIC) system was originally designed to collect acoustic side-scan image data of the seafloor in deep oceans (Caress and Chayes, 1993; Goff and Kleinrock, 1991; Kleinrock, 1992; Kleinrock et al., 1992; Mitchell, 1991 ; Mitchell and Somers, 1989 ; Rushy 1970; Searle 1992; Searle et al., 1990; Searle et al 1989 ; Somerset aL, 1978; Somers and Hugget, 1993; Tyee, 1987; Vogt and Tucholke, 1986) The new capability of acquiring swath bathymetry was recently added to the system based on the interferometry method used by SeaMARC ll; and the system is now called GLORI-B (Somers and Hugget, 1993). The GLORI-B is built in a fish vehicle, which is towed -300 m behind a ship at a depth of -50 m The instrument, consisting of two parallel rows of long transducer arrays on each side, transmits a 2 second-long 6.5 kHz FM pulse every 20-40 seconds (30 seconds during the Gloria 34

PAGE 48

Expedition) within a 2.7" narrow beam. The returning sig nal is correlated (to compress the pulse) and is recorded in digital form for s ub sequent processing The phase differences of the return signals to the two transducer arrays are used to compute beam angles, which are used together with travel time to calculate the depth to the seafloor that reflects the signals. The full swath of bathymetry data is -24-km wide. At 8 knots towing speed, the data have an along-track resolution of -125 m, a cross-track resolution of -45 m, and a vertical resolution of -50 m. SeaBeam 2000 is a hull-mounted system developed out of previous work on narrow-beam echo sounders and as such, is designed toward production of bathymetry maps rather than seafloor acoustic images common to traditional side-looking sonars (Caress and Chayes, 1993; Goff and Kleinrock, 1991; Kleinrock, 1992; Kleinrock et al., 1992; Tyee, 1987; Vogt and Tucholke 1986). The system collects acoustic side-scan data as well, however, the processing of side-scan image data is discussed elsewhere (Liu and Naar, 1996c). SeaBeam 2000 uses a 7 msec long transmit pulse. At a water depth of 5 km, the 2.7 beam width gives along-track and cross-track bathymetry resolution from 233 m beneath the ship to 264 m in the outer beams. The SeaBeam 2000 bathymetry has 1 m vertical resolution. The roll and pitch of the ship are very precisely removed because an error of0.1 is detectable in the data. SeaB e am's outer beams extend to 60 either side of vertical At a water depth of 3 km, the system produces swaths of -10 km wide DATA AND FORMATS Most of the bathymetry data were collected during Legs 5, 6, and 7 of the Gloria Expedition. However the GLORI B system was not used in Leg _7 Figure 3.1 shows the location of the shiptracks in the survey area. The large box (115 W/103 W, 29 S/2YS) is 35

PAGE 49

Figure 3.2. Bathymetry data in the test area A : The SeaBeam 2000 data befor e processing; B: The GLORI-B data before processing; and C: The processed data including both GLORI B and SeaBeam 2000 data, and the interpolated data in the gaps between the swaths

PAGE 50

01) '0 -26 7 2 6 8 26.9 e -27.o ..... 01) '0 -27 1 -27.2 -27.3 26 7 -26 8 -26 9 e -27.o ] 01) '0 -27.1 -27.2 -27.3 -26.7 -26 8 -26 9 e -27.o -27 1 -27 2 2 7.3 depth(m) 0 -500 -1000 -1500 -2000 -2500 -3500 -4000 -4500 -5000 -107 5 -107.4 -107 3 -107 2 -107.1 -107 0 -106 9 -106.8 -106 7 106.6 -106 5 longitude 37

PAGE 51

the Easter-Salas y Gomez region for wh i ch w e have mosaick e d and m e rged th e data int o a grid (Figure 3 1C). The small box shows the test area for which we will illustrate the data processing using the BP package. The swath of GLORI-B bathymetry data extends -12 km t o each side of the shiptrack, and is overlapped by a higher resolution SeaBeam swath that extends -5 km t o each side of the shiptrack (Figures 3 2A and 2B). The SeaB e am 2000 data ar e not only used to calibrate the GLORI-B data but also to replace the l o wer res o lution GLORI-B data in the overlapped area. GLORI B data can be processed without the SeaBeam 2000 data However, the overall accuracy of the processed data i s significantly increased by incorporating the SeaBeam 2000 data Data gaps are presen t between adjacent GLORI B swaths because the distance between the adjacent shiptracks is great e r than the bathymetry swath width (but not the side-scan swaths which are 45 km wide ) ( Figur e 3 2B ) These gaps are filled by interpolating the merged GLORI-B and SeaBeam 2000 data Figure 3 2C illustrates the final result of the processed bathym e try data in the test area. The topography data of the islands were digitized from publish e d contour maps because the sonar systems cannot a c quire topography data above sea level (Hagen et al., 1990 ). These data were incorporated into the grid before the interpolation In areas outside th e surveyed region SeaMARC II data (provid e d by R Hey), or ETOPO 5 data ( t o th e east of 109 W ) are used. Various data are stored or processed in files of differ e nt formats. There are two major formats of binary files used in the data processing and/or storage: GRD and RAS. The first one is a type of netCDF (network Common Data F o rmat ) ( Unidata Program Center, 1991) A header in each netCDF file provides information about the data GRD Most of the data are converted mosaicked, processed merged and finally, stored in this format. This is also the same format used by mo s t of the GMT (Generic 38

PAGE 52

Mapping Tools) (Smith and Wessel, 1990; Wessel and Smith, 1991) programs. A GRD file contains a regular grid of data points such as bathymetry A special value NaN is used for cells that do not have data. The header of a GRD file provides information such as dimension of the grid, the size of the cells, and range of the values, etc. RAS This format is used in the early stage of GLORI-B bathymetry data processing A RAS file consists of a number of consecutive pings representing bathymetry data along a swath Each ping has 605 elements of floating point values The first 601 elements contain bathymetry values for a ping (300 bins on each side plus the center bin) and the last 4 elements contain values for the center beam s longitude, latitude, heading, and average depth The number of pings for a file is unlimited. BATHYMETRY PROCESSING PACKAGE The BP software package is composed of 37 programs written in ANSI C (Appendix 3.1 ). The package has been installed and operated on SUN/UNIX and SGIIIRIX platforms Some GMT programs can be incorporated in the data processing because the BP package uses data structures compatible with GMT package. The BP package is an open system. New programs can be incorporated in order to deal with more complex problems Each of the programs can be run separately as a regular UNIX command or executed sequentially in a UNIX shell script to produce the desired result. Many of the programs provide multiple functions to handle a set of similar tasks, depending on the setting of the command line. Most of the programs in the BP package use the GMT style syntax for setting command line parameters. An online help is provided in each of the programs to help users for the program usage The names of the 39

PAGE 53

programs also provide an indication of what type of data structures the program will work with For example, grdpatch will take GRD files. Although the package is designed for processing swath bathymetry data from GLORI-B and SeaBeam 2000, some programs may be used to process other data in GRD format. Therefore, the BP package is an additional tool for GMT users. LINE FILTER There are along-track biases in the data collected with most multibeam sonar systems, especially in GLORI-B swath bathymetry data. Along-track b i ases are system artifacts characterized by narrow highs and lows parallel to the shiptracks These biases are not consistent along the shiptracks and, therefore, cannot be removed with a cross-track profile. The biases may be due to system drift or noise in the water A line filter is developed, which effectively minimizes or removes the biases The line filter is a combination of two median filters, one with a long boxcar and the other with a square boxcar Considering bathymetry data in a regular grid, the first filter calculates a median for each cell in a narrow boxcar centered at the cell The long axis of the boxcar should orient parallel to the along-track biases The second filter calculates a median for each in a square boxcar centered at the cell. The first filter obtains the general trend of the features in shiptrack directions while the second filter acquires the regional depth of the seafloor If there is no bias, the two trends should be close (to the median depth). The differences of the two medians are used to subtract the original data to remove the biases. The length of the long axis of the first filter boxcar should be compatible with, and usually is the same as, the length of the side of the second filter boxcar. 40

PAGE 54

The line filter is sensitive to long narrow features It removes only those features with their long axis parallel to and longer than the boxcar of the first filter, but much narrower than the boxcar of the second filter. The line filter is not sensitive to features in other orientations or shapes In order to identify reliable trends, the boxcars of the filters should be large. However, they are limited by two factors : (1) the larger the boxcar the longer time it will take for the computer to sort the numbers in the box to fmd the median; (2) the along-track biases are local features that cannot be removed if the boxcar is longer than the length of the biases We recommend that the dimension of the long boxcar be 1/3 or 1/2 of the length of the biases. For a grid of cell size .003 by 003 degrees a 30 by 30 km boxcar consist s of 10,000 numbers It takes significant CPU (Center Processing Unit) time (on a SUN SP ARC 20) to sort these numbers to find a median for each cell in the second filter. In practice, a mean filter, instead of a median filter, is often used in the place of the second filter Thus, the second filter computes an average value for each cell within the boxcar The computing time is reduced further when a better algorithm is implemented in the mean filter During calculation of the average value for each cell, the boxcar moves from one cell to the next. A summation of the values of all the cells within the box has to be calculated. Because there are many cells overlapping in two consequent locations of the boxcar, the change in the summation is due to the removal of a column on one side of the boxcar and addition of another column on the other side. Therefore, we need only to adjust the summation accordingly, instead of adding all the numbers in the box every time. This algorithm is implemented in the program grdmean. By using a mean filter, we gain computing speed, but lose accuracy. Many times the mean filter cannot find the true regional depth of the seafloor For example, if there is a positive feature of a single cell size with a height of 9 m standing on a flat seafloor of 0 m height, for a boxcar of 10 by 10 cells, the mean height of the seafloor is 0.09 m instead of 41

PAGE 55

0 m, the median height This may cause two side-effects The filter slightly smoothes the normal features on seafloor with dimension less than the boxcars; and the biases cannot be removed cleanly (there is a little residual) Fortunately, these side-effects are comparatively much smaller than biases and noises in the data THE DATA PROCESSING PROCEDURE The general flow of the bathymetry data processing is shown in Figure 3.3. The more detailed GLORI-B and SeaBeam 2000 data processing are illustrated in Figures 3 4 and 5, respectively This procedure is exemplified using the BP programs in a UNIX script (Appendix 3 2). Each following section corresponds to a portion of the script. Generally, the data processing involves several data sets. Each set is processed separately until a GRD file is obtained, then the grids are merged into a single grid. SeaBeam 2000 bathymetry data are processed first to obtain a grid The topography data of the islands (above sealevel) are gridded into individual grids, then pasted onto the SeaBeam grid to patch the data holes at islands. This merged grid is our best data set, which is used as a calibration standard for GLORI-B data. GLORI-B data are processed separately to remove noise and minimize biases. The SeaBeam grid is used to replace the overlapped GLORI-B data near the center track of the swaths The gaps between GLORI-B swaths are filled by interpolating the merged GLORI-B and SeaBeam grid. Finally, the areas out of the survey region are filled with SeaMARC II and ETOPO 5 data. Becaus e the latter two data sets have much lower resolution than GLORI-B and SeaBeam 2000 data, we did not calibrate them to match the SeaBeam 2000 data. 42

PAGE 56

Figure 3.3 Flow chart of the bathymetry data processing Included in the dashed lines are SeaBeam 2000 data processing and GLORI-B data processing, respectively, which are detailed in Figures 3.4 and 5. The boxes indicate the bathymetry data at different stages of processing. The rounded corner boxes describe the actions applied to the data. The solid arrows point the flow directions of the data.

PAGE 57

Bathymetry Data Processing ,----------1 ----------------------, I I I I I I I I 1 I I I Proce ss and grid int o GMT GRD formal Merg e the tw o dat a sets in GMT GRD f ormal Grid i n to GMT GRD format Merge the navigation data, convert th e data into RA S formal and process in this form al lo r emove track noise into GRD format, m e rg e with th e best data, and proc ess t o fill data gaps and redu ce rand om nois e and tr ack noi se further I I I I (Thi s is the best data) data in GRD formal I L ______________________________ l This box indi c at es input dat a o r data at d i ffe rent s t age of proces si ng An arrow indi cates flow directi o n o f the data This box ind icates th e pr ocess applied t o th e input data sct(s) ETOP05 data Convert inlo GRD format and resampl e the data into fin e r grid 10 m a t ch the grid size of the merg ed SeaBeam and GLORI-8 data

PAGE 58

SEABEAM 2000 BATHYMETRY DATA PROCESSING SeaBeam 2000 data are stored in binary files (usually one file per day), each of which is composed of a number of records Each ping of the bathymetry data is stored in a data (or sensor) record, which consists of 120 beams across shiptrack Preceding each data record, there is a navigation record that provides information, such as heading and geographical location of the center beam. Fewer artifacts make processing the SeaBeam 2000 bathymetry data much easier than processing GLORI-B data (Bird et al., 1996; Caress and Chayes, 1993; Chavez, 1986; Cobra, 1990; Goff and Kleinrock, 1991; Hagen et al., 1996; Liu et al., 1994; Miller et al. 1991; Mitchell, 1991; Scheirer and Macdonald, 1993; Scheirer et al 1996) (Figure 3.4) The data are released to users after preprocessing. A sophisticated program sb2llz written by Scheirer et al. (Scheirer and Macdonald, 1993; Scheirer et al., 1996) provides a handy tool to process the data. With sb2llz, an ARC (Average Residual Cross-track) profile is calculated from the data. This profile is used to make correction for the cross track biases in the SeaBeam 2000 data. The geographical location of each data point are calculated and written into ASCII files as triplets (longitude, latitude, and depth). These triplets produced by sb2llz are subsequently gridded in a GRD file Correcting Cross-Track Bias There is a -30 m vertical bias across swaths in SeaBeam 2000 data. Data from the outer beams are apparently deeper than data collected by the center beams. This bias is stronger on the starboard side. If the data are not corrected, the bias will be significant 45

PAGE 59

SeaBeam Bathymetry Data Processing Calc ulat e an average c ross-track profile f o r each d a ta ftle Compute an average co rr ectio n profile fr o m the pr oftle Make the c r oss -track corrections and c onvert the data i n to U z format A S CIJ file o f triplets in liz f ormat (lo n girue. l atitude. and depth) Register the d ata i nt o GMT GRD format. An average d epth is used if m ore than one point appear.; within a gridding box. SeaBea m 2 000 d ata in GMT GRD f orma t a t a grid size of 0 .003 x 0.003 degree Figure 3.4 Flow chart of the SeaBeam 2000 bathymetry data proce s sing The same notations as in Figure 3.3 are u s ed for boxes and arrow s 46

PAGE 60

when looking at fine scale structures of the seafloor. The cause of this bias is unknown A simple empirical correction is devised based on an ARC profile A residual cross-track profile is calculated (using avgsbpj) by subtracting from each beam the average depth of all the beams of a ping Theoretically, averaging a large number of pings will be enough to produce a flat residual profile if there is no bias Figure 3 .6A is an ARC profile computed from all the SeaBeam 2000 data of Legs 6 and 7. The cross-track bias correction is made by subtracting the ARC profile from each ping for each corresponding beam. The cross-track bias appears relatively constant in both Legs 6 and 7 It is effective in reducing the systematic errors to an overall precision of less than 10 m. Although not exhaustively tested, correcting the SeaBeam depths in our study area by the average profile reduces the errors in bathymetry significantly. A few uncertainties in these corrections still remain (and should be investigated), such as, whether the cross-track errors are a function of average water depth, cross -track distance (not angle/beam-number), or cruise. Converting Data to GRD Format Longitude and latitude are computed for every depth point using sb2llz (Scheirer and Macdonald, 1993; Scheirer et al. 1996) The geographical locations are calculated using simple triangle geometry. At a given regional center location, the geographical offsets per meter in both longitude and latitude directions are computed and called unit offsets. There are two numbers associated with each point, a depth and a cross-track distance from the center beam (negative distance is for the points to the port side). These two numbers are used with the unit offsets to calculate the longitude and latitude offsets for each beam. The geographical location of each point is obtained by adding the longitude and 47

PAGE 61

latitude offsets to the center beam location. Therefore a triplet of longitude latitude and depth is calculated for each data point, which is written to an ASCll file. The data triplets are fmally gridded into a GRD file. The program xyzmgrd is used to register the triplets into a regular grid with a cell size of .003 by .003 degrees The cell size of the grid is limited by the resolution of the data. If a finer cell size is used during the gridding, there will be more data holes in the resultant grid. The triplets may not be evenly distributed in the grid. If this is the case, some cells will not have any depth data points while others may contain multiple data points For the former case, the value NaN is assigned to the cells indicating data holes in the grid. For the latter situation, xyzmgrd will calculate an average depth for each of the cells Up to this point, a SeaBeam grid is acquired, which contains only data collected using SeaBeam 2000 within the survey region. FILLING THE TOPOGRAPHY DATA OF THE ISLANDS INTO THE SEABEAM 2000 GRID The topographic data for the islands are acquired separately and patched onto the SeaBeam grid. There are two large data holes left in the bathymetry grid near Easter and Salas y Gomez islands. The topographic data of Easter Island are digitized from contour maps (Hagen et al., 1990). The topographic data of Salas y Gomez Island are digitized from a figure in a published paper (Fisher and Norris 1960). However, the mapped center of Salas y Gomez Island (26 27 8'S, 105 27.8'W) does not agree with our observation during Leg 6 of the Gloria Expedition (26"28.5'S, 105"21.7'W) The digitized data of the Island are shifted accordingly Gridded topography data for the two islands are pasted onto the SeaBeam grid using grdpatch This patched SeaBeam grid will be used to calibrate GLORI-B data and to replace the less accurate GLORI-B data in overlapped areas. 48

PAGE 62

GLORI-B BATHYMETRY DATA PROCESSING Although the GLORI-B system produces less accurate data than the SeaBeam 2000, it is a valuable system for reconnaissance in a large region due to the 24-km wide swaths of bathymetry it can produce. During Legs 5 and 6, the GLORI-B system was used to collect swath bathymetry for the first time. Many unique system artifacts make GLORI-B data processing more complex than any other part of the bathymetry data processing (Bird et al., 1996; Caress and Chayes, 1993; Chavez, 1986; Cobra, 1990; Goff and Kleinrock, 1991; Liu et al., 1994; Miller et al., 1991; Mitchell, 1991; Somerset al 1978; Somers and Hugget, 1993). There are four major system artifacts identified in the GLORI-B bathymetry data: cross-track bias, along-track bias, edge drop, and random noise. The cross-track bias in GLORI B bathymetry data is similar to that in the SeaBeam 2000 data, but has much greater amplitude (Figure 3.6B). There are significant differences between the two profiles from Leg 5 and 6, indicating that the bias is associated with water depth because the Leg 5 surveyed in the shallow area closer to the East Pacific Rise. Therefore, diff erent profiles are u sed to minimize the biases for the data collected from different regions. The along-track bias discussed earlier is one of the major artifacts presented in the GLORI-B bathymetry data (Fig ur e 3.2B). A lin e filter is developed to deal with the problem. During processing of the GLORI B data, it is applied twice and significantly reduces the biases First, the filter is applied to the data in RAS format to remove the biases along the axis of the swath. Second, the filter is applied to the data in GRD format to remove biases along the shiptracks, 070 in our survey. With information from nearby swaths, this second filtering is capable of removing or reducing the biases near the edges of the swaths. 49

PAGE 63

At shallow depths, the bathymetry data acquired with the GLORI-B system deflect downwards abnormally near the outer beams of the swaths. This is called the edge drop effect (see the area near the seamount in Figure 3.2B). This pr o blem is seen more often in the Leg 5 data because the seafloor of the surveyed area is much shallower The slope of the dropped edge faces away from the sonar system If the data were real, the slope surface should be in the shadow of the sonar beams In other words, the system could not possibly detect the direct reflections from that portion of seafloor. These signals are so noisy that we cannot determine if they are multiple reflections. These portions of unreliable data are cut using computer algorithms Otherwise they would affect the interpolation for the gaps between swaths and the following data proce s sing. The random noises become increasingly stronger from the center to the outer beams (Figure 3.2B). This may be because the outer beams travel longer in the water column. Compared with the SeaBeam 2000 data at the same range the GLORI-B data are still noisier. The noises are reduced with a median filter with a boxcar size of 5 by 5 cells Larger random noised, or speckles, are reduced with a clip technique suggested by Paul Johnson (personal communication, 1994) Procedure of GLORI-B Data Processing The artifacts in the GLORI-B bathymetry data make this portion of the data processing much more complex than any other. The data are processed in two different formats : RAS format (Figure 3.5A) and GRD format (Figure 3.5B) In the RAS format, the data are stored in a long array that maintains the lateral geometry f o r the data points in a swath In this format we focus on removal of biases associat e d with shiptrack direction and eliminating the dropped edges. In the GRD format, the information from nearby 50

PAGE 64

GLORI-B Bathymetry Data Processing Navigation data with smoothed beading data Use rasjiltu to smooth the data with a 20X I 0 mean filte r to obtain stab le surface trend A binary mask o r the swath with 0 in areas or slope >I :2 and I in remaining areas Along-tra c k bias S m oothed GLORI-B two smoo thed swaths Use rascut to remove some pings (200 pings ) of data from the beginning and the end of the GLORI-B swath which are the d u p licated data of the first or last few pings t o obtain reliabl e track information (suc h as b eading. track noise ) for subsequent processing Those pings were added to the swath during conversion of the data t o RAS format Use rasmllz t o calculate the geographical location f o r each depth point or the GLORI-B bathymetry data and write the data into an ASCil file or three columns (longirude. l atit ude and depth in each line) Use .ryvngrd to register the GLORI-B data int o grids in GMT GRD format. An average depth wiU be used i f there is more than one point at the same gridding box of 003X.003 degree (Conti nued on n ext pa ge) Further processing in GMT GRD format Figure 3.5A. Flow chart of the GLORI B bathymetry data processing in RAS format. 51

PAGE 65

Figure 3.5. (Continued ) A mosaic with all the bes t data sets in a GMT GRD fllc at a grid size of .003X.003 dcgrtt including SeaBeam swa lhs ccnt<:ml along th e shiputcks. GLORIB data alon g both sides of the SeaBeam swllh>. and int<:rpolated data filling the gaps between data swalhs Figure 3.5B. Flow chart of the GLORI-B bathymetry data processing in GRD format. 52

PAGE 66

swaths can be used to remove the track biases near the edges of the swaths and to reduce any noise remaining in the data further. The general flow of the GLORI-B bathymetry data processing is shown in Figures 3.5A and 5B. The data are converted to RAS format as the navigation information is merged into the data Random noises are reduced by using a 5 by 5 cell median filter. The dropped edges are cut based on topographic gradient in cross-track direction. An ARC profile is calculated and applied to remove cross-track biases. The line filter is used to reduce along-track biases Finally, the data are converted to GRD format (Figure 3.5A). In GRD format, the dropped edges are cut further using additional techniques (Figure 3 5B). The smooth and clip method is used to reduce the larger speckle noises. The SeaBeam 2000 data are used to calibrate the GLORI-B data before merging them. After most of the biases and noises are removed or reduced, the gaps between the swaths are interpolated. With additional information from nearby swaths, the line filter is used again to reduce the along track biases further. Eventually, we obtain a fine grid of bathymetry data covering the region in the large box of Figure 3.1. This procedure is detailed step by step below. Preparing Navigational Data GLORI B bathymetry data do not include the navigation information The shipboard navigational data are used to process GLORI-B data. The ship's navigational data are converted and merged into a file for processing GLORI-B image (Liu and Naar, 1996c) However, headings are still not included. The headings are computed from the navigation data using avgnav and stored in a file with other navigation information for merging with bathymetry data in the next step. Every two consecutive ship locations are 53

PAGE 67

used to compute a shiptrack orientation These orientations can be used as an approximation for the ship heading because most of the shiptracks are straight. There is a heading lag that the vehicle falls behind the ship heading because the GLORI-B vehicle is towed a few hundred meters behind the ship on a cable The vehicle headings should be smoother than the shiptrack orientation Smoothed track orientation (the average of several consecutive shiptrack orientations) would represent the real GLORI-B vehicle's heading more closely. In this example an average of 5 headings i s applied, which includes two headings before and two behind Converting the Data into RAS Format After preprocessing, the GLORI-B bathymetry data are stored in ASCIT files (usually 120 pings per file one hour's data) The header line of the data files provides information about date, time and the number of the first ping etc The date and time will be used to search in the navigation data file for the corresponding segment of navigation information Remaining lines in the GLORI-B data files are the triplets of bathymetry data points (cross track distance, ping number, and depth) The positive cross-track d i stance is to the starboard side The ping number usually starts from 1 and increases consecutively It is very difficult to process the bathymetry data in the raw data format To remove cross-track and along-track biases, the data need to be in a raster format that maintains the lateral geometry of the data points in a swath Therefore the data are converted into the RAS format. Each ping is registered into 601 bins (of 50-m-wide) and is stored in a row of the swath array (columns 0 through 600) The center bin is stored in column 300. The four additional columns 601-604 in the array contain the navigation data merged from the navigation file 54

PAGE 68

During registering the raw GLORI-B data into RAS files, the cross track distance of each data point determines the bin (or column) number in a row. Zero distance indicates the center beam and is registered in bin 300 The negativ e distance (port side) is registered to bins 0 through 299 and the positive bins 301 through 600 If more than one point belongs to one bin, an average depth is assigned to the bin In a RAS ftle a number of consecutive pings form a swath as a two-dimensional array. During the processing, we put about 10-days of data into a single RAS file This makes it easy to extract reliable average residual cross track proftles and also reduce the boundary effects during application of the line filter. The bathymetry data are converted to RAS format using glormras that also merges the navigation data. The GLORI-B vehicle is towed -300 m behind the ship. If the GLORI B vehicle is following the ship on a straight line and the ship travels at a constant speed, the space offset can be adjusted by a time lag. The speed of our cruise is about -13 kmlhr (or -7 kts) Therefore, a time lag of -90 seconds is applied to make up the distance offset. During merging the navigation data with the bathymetry data, the navigation records are shifted accordingly to match the corresponding pings Reducing Random Noises and Filling Small Holes Some high frequency noises are removed or reduced with a median filter using rasjilter. Compared with SeaBeam 2000 bathymetry data GLORI-B data have more high frequency noises (Figures 3.2A and 2B) A 5 by 5 boxcar median ftlter is used to reduce the noises. Some small data holes on the swaths are filled, too. Figure 3.7A shows the data without any corrections comparing with Figure 3 7B after the application of rasfilter. 55

PAGE 69

20 20 15 1 5 10 lO ---5 E '-" 5 .c .c a a Q) "0 0 o.S ca ca ::I ::I :'Q "' -5 Q) .... "0 c;; -5 e -10 -10 -15 -15 A: SeaBeam Average Residual Cross-track Profile -2 0 4 0 5 0 60 -60 -50 -40 -30 -20 -10 0 1 0 2 0 3 0 ----E '-" .c a Q) "0 ca ::I "0 c;; e 300 200 100 0 -100 port <--beam number acros s track --> stbd ;,,"/ / I / I I I / ,, ........ -------Le g 5 -Leg6 B: GLORI-B Average Residual Cross-track Profiles 400 300 200 0 -100 2<>?300 -250 -200 -150 -100 -50 0 50 100 150 200 250 30<5200 port<--bin numbe r a cross tra c k --> stbd Figure 3.6. Average residual cross-track profiles from SeaBeam 2000 bathymetry data (A) and from GLORI-B bathymetry data (B) 56

PAGE 70

Figure 3.7. Bathymetry maps in the test region showing the procedure of the data processing The same color scale for Figure 3.2 is used for all the bathymetry maps. The insets A through E illustrate the processing in RAS format; and the insets F through P demonstrate the processing in GRD format. A : The bathymetry map of the GLORI B data without any correction. B : The bathymetry data after filtering with a 5 by 5 cells median filter. C : The outward slope from the center track of the swaths. The dark areas are the slopes facing away from GLORI-B sonar system. D: The bathymetry data after removing cross-track biases and reducing along track biases in RAS format. E: The bathymetry data after dropped edges are cut. F : The bathymetry data after trimming G : The bathymetry data after shrinking H : The bathymetry data after clipping. 1 : The bathymetry trend calculated from SeaBeam 2000 data only. J : The difference trend between GLORI-B and S e aBeam 2000 bathymetry K: The GLORI B bathymetry data after adjusting using the difference trend. L: The bathymetry data after SeaBeam 2000 data is added and replaced the overlapping GLORI-B data M : The bathymetry data after the gaps are interpolated from the merged data N : The along track biases calculated from the interpolated data. 0: The bathymetry data after the along track biases are removed from data in Panel M using biases in Panel N. P: The final processed bathymetry data after SeaBeam 2000 data are restored.

PAGE 71

(Continued on next page) 58

PAGE 72

Figure 3 7. (Continued) 59

PAGE 73

Cutting the Dropped Edges Near shallow seamounts (<-2000 m) or shallow seafloor, the GLORI-B bathymetry data drop to unrealistic depths toward the edges of the swaths. The surface of the dropped edge faces away from the GLORI-B vehicle, actually in the shadow of the sonar beams (see the dark areas at the edges near seamounts in Figure 3.7C) It is better to cut off these portions of unreliable data because they will affect the nearby normal data in the following processing The gradient from the center beam in the cross-track direction is calculated with raslope. The dropped edges are then removed using rascull. This is done by cutting off the edge of a swath with a slope facing away from the center beam and a gradient greater than a given threshold During our data processing, this threshold is 1/2. Figure 3 .7E shows the swaths after cutting (compare with Figures 3.7 A and 7B) Removing Cross-Track Biases Like SeaBeam 2000 data GLORI-B data contain cross-track biases. These biases vary with depth. Figure 3 .6B shows the ARC profiles calculated from data collected during Legs 5 and 6. Instead of using a single profile for each leg, we calculate a profile from each RAS file and apply it to remove the biases from the data of the same file. The program rasprof is designed to calculate the profile and make the correction. 60

PAGE 74

Reducing Along-Track Biases The line filter is used to reduce the along-track biases. Instead of coding a single program, two multi-functional programs, ras.filter and rasmath, are combined to produce the line filter. The boxcar of the first median filter is 05 by 20 km (or 1 by 200 cells) with the long axis parallel to shiptrack The boxcar dimension of the second filter is 10 by 20 km (or 200 by 200 cells) with long axis parallel to the shiptrack too. Figure 3 7D shows the data after the filtering (compare with Figures 3.7A and 7B). Removing Duplicated Pings During converting the GLORI-B data to RAS format, about 10-days of raw data files are merged into a single RAS file. About 200 extra pings at the beginning and the end of each RAS file are duplicated from first and the last raw data files. The purpose is to reduce the boundary effects at the beginning and the end of swaths when using ras.filter in the processing Although these additional pings are identical data points to the first and last files, they put extra weight for these points during registering the data into GRD format. The program rascut is used to cut these pings Converting the Data from RAS to GRD Format The geographical coordinates (longitude and latitude) are computed for each data point in the RAS files using rasmllz. This is done in a similar way as was done for the SeaBeam 2000 data. Because the navigation information is in the same row as the 61

PAGE 75

bathymetry data, it is easier to code rasmllz than sb2llz. Here we calculate unit offsets for every ping instead of the entire region (in sb2llz). This will produce a more accurate location for each point However this will not make much diff ere nce if the region is very small, such as only a few degrees wide in latitude. The bathymetry triplets (longitude, latitude, and depth) calculated with rasmllz are registered into a GRD file (in the large box of Figure 3 .1) using xyzmgrd. The cell size of the grid is 003 by .003 degrees the same as in the SeaBeam grid Trimming the Data Most of the dropped edges have been cut in the RAS format. However, some erroneous data may still exist in the form of small isolated spots or peninsulas attached to data swaths. The program grdtrim is used to remove small data spots or narrow peninsulas with one of their dimen s ions l ess than a given number of cells, s uch as 5 cells in our data proce ssi ng. Some small patch es of normal data are removed in the process. The thre s hold of the trimming dimension should be chosen so that we gain mor e quality than we lose Large thresholds could cause more normal data to be removed while small ones may not be able to clea n large patches of erroneous data Figure 3.7F shows the data after applying grdtrim (compare with Figur e 3.7E). A few cells attached to the edges of all the swaths are cut further with the program grdshrink. This program removes a band of one cell width from the edges of the data swaths. It is applied twice to remove two cells (Figure 3.70). Finally, the remaining erroneous data are removed semi-automatically. An irregularly shaped patch can be represented by a number of overlapped boxes. Each box is defined by two comer coordinates (the lower left and upper right comers). The program 62

PAGE 76

grdlwling remove data points within boxes of a grid using their corner coordinates These corner pairs are digitized manually from a preliminary plot of the data Clipping Speckles The speckles are reduced further using a clipping technique. An averaged grid is calculated using a moving boxcar of 5 by 5 cells The difference between the data and the smoothed grid provides the magnitude of the noisy peaks The cells with a difference greater than 200 m are removed. This technique cuts the sharp speckles and funnels from the grid (Figure 3.7H). Most of the holes after the cut are near the edges of the swaths, indicating that the data were more noisy near the outer beams of the swath Calibrating GLORI-B Data Using the SeaBeam Grid The GLORI-B data are calibrated to match the SeaBeam 2000 data so that they can be merged smoothly on the borders Because the GLORI-B vehicle is towed about 50 meters below the sea surface, there should be an offset between GLORI-B and SeaBeam 2000 data During the data processing, we found that the offset is not a constant number Therefore, GLORI-B data cannot be well matched to the SeaBeam 2000 data by simply shifting up or down by a constant number for all the cells in the GLORI-B data grid. A smoothed general trend is created from the difference grid between the GLORI-B grid and the SeaBeam trend The trend is used to adjust the GLORI-B grid To calculate the difference trend a general surface trend is created from the SeaBeam grid. This is a smoothed and interpolated surface trend of the SeaBeam grid 63

PAGE 77

(Figure 3.71) A difference grid between GLORI-B grid and the SeaBeam trend is calculated by subtracting the two grids. The difference grid is smoothed using grdmean with a boxcar size of 50 by 50 cells to obtain the difference trend (Figure 3.7J). The GLORI-B grid is adjusted by subtracting the difference trend (Figure 3.7K). Merging GLORI-B and SeaBeam Grids The 10-km-wide SeaBeam 2000 swaths overlap -40% of the GLORI-B swaths along the center tracks These SeaBeam 2000 data replaced the GLORI-B data in the overlapping areas. The two data sets may not match perfectly on their borders e g., a small "cliff' may be observed on the borders A boxcar filter of 5 by 3 cells with its long axis inN-S direction, which is approximately perpendicular to the borders (070 ), is used to smooth the data. To retain the best data, the SeaBeam grid is again restored to the smoothed grid (Figure 3.7L). Interpolating the Merged Data to Fill the Data Gaps The data gaps between the swaths are filled by interpolating the merged grid using grdsurface (Figure 3.7M). The program grdsurface modified from a GMT program surface, put a surface on a grid using a GRD file instead of an ASCII file of triplets (longitude, latitude, and depth). The gaps are interpolated by solving (1 T) L(L(z)) + T L(z) = 0 64

PAGE 78

where Tis a tension factor between 0 and 1, and L indicates the Laplacian operator T = 0 gives the 'minimum curvature' solution and T = 1 gives a harmonic surface. A tension factor of 0 25 is chosen during our data processing The interpolated data outside of the densely surveyed area are not used. Those data are unreliable because there are no data points that control the interpolation. There are about 20-km-wide data gaps between the SeaBeam swaths to the northwest of the mosaicking area due to the absence ofGLORI-B swaths (Figure 3.1C). These gaps are left open because it may be unreliable to interpolate across such a large distance at so fme a grid size. Reducing the Along-Track Bias Further The line filter is applied again to further reduce the along-track biases in the gridded bathymetry data. It is possible to use the line filter on the gridded bathymetry data because most of the shiptracks (and thus the along track biases) are oriented in the same direction (070 ) It is necessary to use the line filter again because the additional information from nearby swaths will help to reduce the biases near the outer swaths. The line filter is applied to the data through a combination of three programs grdmedian, grdmean, and grdmath. The first filter is provided by grdmedian, in our example using a boxcar of 1.5 by 30 km (or 5 by 100 cells) with its long axis in 010 direction. And the second filter uses grdmean with a boxcar of 30 by 30 km The two filtered grids are subtracted to produce a grid of difference (Figure 3 7N). By subtracting the difference from the data, we acquire a new grid with fewer of the along-track biases (Figure 3.70). 65

PAGE 79

Restoring the SeaBeam 2000 Data The above processing affects the SeaBeam 2000 data, too. The SeaBeam 2000 data are restored to the new grid to retain the best data. A difference grid is obtained by subtracting the SeaBeam grid from the processed data. This difference grid is interpolated using the grdsuiface to obtain the difference surface. This surface is used to adjust the previously processed grid. This final touch will reduce the borders between the GLORI-B and SeaBeam 2000 data further, while restoring the SeaBeam 2000 data at the same time (Figure 3 7P) At this point, a grid of mosaicked bathymetry data is acquired in the EasterSalas y Gomez region, the west portion of the Easter Seamount Chain (Figure 3.1 C). The grid includes SeaBeam 2000 data swaths centered along the shiptracks, GLORI-B data along both sides of the SeaBeam 2000 data swaths, and interpolated data filling the gaps between swaths, covering the densely surveyed region completely. The data is stored in a GRD flle of the cell size 003 by .003 degrees. FILLING THE REST OF THE GRID There is no GLORI-B and SeaBeam 2000 data coverage beyond the densely surveyed region. These areas are filled with SeaMARC II (provided by R Hey) and ETOPO 5 data. Though their resolutions are much lower than the GLORI-B and SeaBeam 2000 data they still provide additional information about the seafloor beyond the coverage of the GLORI-B and SeaBeam 2000 data These data are filled in the areas out of the surveyed region without attempting to match their borders to the GLORI-B and SeaBeam data. Both of the SeaMARC II and ETOPO 5 grids are resampled to the same grid size 66

PAGE 80

( 003 by 003 degrees) The SeaMARC II data covers only the area to the west of 109w in the gridding box. However, it has higher resolution than ETOPO 5 data. Thus the final grid is composed of five data sets. In the surveyed region, there are SeaBeam 2000 data, GLORI-B data, and the interpolated data. Outside of the surveyed area are the SeaMARC IT data to the west of 109"W and the ETOPO 5 data to the east of 109w. The borders of the surveyed region are clearly identified (Figure 3.1C). ASSESSMENT OF PROCESSED BATHYMETRY A few single 3.5 kHz echo sounding profiles crossing this region are used to compare with the processed bathymetry data (Figure 3.8). As shown in Figure 3.8A the sounding systems provide apparently deeper bathymetry than the SeaBeam 2000 system (the GLORI-B data were calibrated to match the SeaBeam 2000 data in the data processing). Along these profiles, the average depth is 2805.06 m for the processed data and 3660.77 m for the sounding data It is normal to have some offset between different sonar systems We have seen offsets between SeaBeam 2000 and GLORI-B data, and between SeaBeam 2000 and SeaMARC II data. However, this offset (855.71 m) is significantly larger. In Figure 3.8B, two data sets are compared at exaggerated vertical scale after each is shifted to its mean level. The patterns of the profiles mimic each other very well. This observation confirms that the features observed from the processed data are reliable, and are not artifacts of the data collection systems or subsequent processing It appears that the processed data provide smoother topography than that from the sounding system The standard deviation is 679.15 m for the sounding data and 468.83 m for the processed data. The RMS is 285 75 m between the two data sets Although the median filter and the line 67

PAGE 81

Figure 3.8. Assessment of the processed bathymetry data. A : Comparison of the single sounding profiles with the processed bathymetry data along the profiles, showing an offset between the two data sets. B: Comparison of the deviation of the two data sets, showing that the processed data are smoother than the sounding data C : Comparison of the processed data with the sounding data after calibration to match the processed data.

PAGE 82

-111 .g -27 .... longitude -110 -109 -108 -107 I A: Compare with Other Bathymetric Data I [ 4000 depth (m) processed data ----------sounding data average offset: 855.71 (m) 2 .g -27 2 ]! ::-: ..... ..... 1 B: Deviation Comparison I ... . .. ..... . ......... 0 :. ... .. .:.. . ... ( 1 000 wiggle(m) processed data ----------. sounding data GLORI-B tra cks RMS: 285.75 (m) I C : after 1 . . . .... .... . longitude 69 ( IOQO.wiggle(m) processed d a ta ---------_-. so unding data GLORI-B tra cks RMS: 177.00 (m) . -106 .. ... ... -27 "0 2 C
PAGE 83

filter used in the data processing could produce a smoothing effect, they cannot account for such a large difference During the processing, the SeaBeam 2000 data were never smoothed Comparing the sections of the profiles near shiptracks, the SeaBeam 2000 data show the same deviation from the sounding profiles indicating that these differences are the results of the differences between the two types of sonar systems Assuming that the difference is due to the different systems or poor navigation, the sounding data are re-calibrated to match the SeaBeam 2000 data by multiplying the data with a factor of 0.76624 (=2805 06/3660.77). Now, they have the same average depth. The standard deviation for the sounding data is 520.39 m, which is much closer to that of the processed data (468 .83 m). And the new RMS is 177.00 m between them, comparing with the previous value (285.75 m) The calibrated sounding profiles agree with the processed data much better (Figure 3.8C) The sounding profiles are from older data and thus do not have GPS navigation The mismatch may be related to offset navigation. Although we cannot verify which data is more reli a ble without any additional information, the features detected from the two data sets are consistent. Therefore the processed data are reliable for feature identification in future data analysis and interpretation. SUMMARY We have successfully processed GLORIB and SeaBeam 2000 data comprehensively using the BP package to remove or minimize the syste m artifacts. The final result is a regular grid for the Easter-Salas y Gom ez region (104/llS"W, 25/29 S) with resolution of about 300 by 300 m The major features identified from the processed bathymetry data are reliabl e for data interpr etation. The data processing package BP is 70

PAGE 84

compatible with the IP (Image Processing) package (Liu and Naar, 1996c), WHIPS, GMT, and MB (MultiBeam) system (Caress and Chayes, 1993). The software BP package is available by anonymous ftp at sunset.marine.usf.edu (contact zjl@marine.usf.edu or naar@marine.usf.edu with questions). 71

PAGE 85

CHAPTER 4 FORMATION OF THE EASTER SEAMOUNT CHAIN AND IMPLICATIONS FOR DEEP EARTH STRUCTURE ABSTRACT The Easter Seamount Chain (ESC), located entirely on the Nazca plate, is a major feature in the southeast Pacific between the Easter microplate and the Nazca ridge We present major observations from the GLORI-B and SeaBeam 2000 bathymetry, side-scan sonar image and radiometric age data collected during Legs 5 6 and 7 of the 1993 Gloria Expedition. These data indicate a general age progression along the ESC which can be explained with a hot spot model. However the western portion of the ESC is composed of several volcanic ridges in a dextral en-echelon pattern, numerous small volcano e s, and young low-lying lava fields The overall trend of 085 combined with the trend of the Galapagos and Juan Fernandez islands give a new pole for Nazca-hotspot motion location at 85 9"N, 171.4 E. The individual ridges are oblique to the overall chain by 5 -15". The ridges trend sub-parallel to the relative plate motion Volcano ages increase to the east along the two major ridge Ea c h ridge is about 1-3 km high 200-500 km long and less than 80 km wide, and is spaced about 50 100 km from other ridges Larg e, robust volcanoes tend to be at the center or eastern portion of the ridges, while young lava flows are usually scattered at the western ends It is inferred from the observations that each of the ridges may tap from a long "footbal l -l i ke" e nriched mantle source at the base of the 72

PAGE 86

lithosphere, which may form from a mantle blob after being sheared by the relative motion between the lithosphere and the underlying mantle as the blob rises buoyantly from depth in the form of a hot bubble Geophysical and geochemical evidence indicates that the San Felix Island is not part of the ESC. INTRODUCTION Several hypotheses have been proposed to explain the formation and physical nature of the ESC (Easter Seamount Chain) {also referred as Salas y Gomez Ridge (Pilger and Handschumacher, 1981)), a major bathymetric feature in the southeast Pacific (Figure 4.1). None of these hypotheses predicts all of the geophysical observations. Legs 5, 6, and 7 of the Gloria Expedition aboard the RN Melville in early 1993 have been conducted along the ESC (Hey et al., 1995 ; Naar et al., 1993a; Naar et al., 1993b) (Figure 4.2). Two recently upgraded side-scan sonar systems, GLORI-B and SeaBeam 2000, were used to collect both swath bathymetry and side scan sonar image data. The processing of the data is discussed elsewhere (Liu and Naar, 1996c; Liu and Naar, 1996d; Liu et al 1994). We will summarize the previous work present the data, and interpret the observations in this paper. The data provide new constraints for the models of the origin of the ESC and improve the current understanding of the mantle processes responsible for the formation of this feature. The difficulty of fitting the new observations by the previous models requires the proposal of an alternative hypothesis, the Hot Bubble model (Liu et al., 1995) 73

PAGE 87

Figure 4 .1. Tectonic setting of the ESC (Easter Seamount Chain) and major features in the southeast Pacific. The image shows the bathymetry predicted from ETOP05 data (National Geophysical Data Center, 1988) and free-air gravity anomalies calculated from the Geosat altimetric data (Sandwell 1992 ; Sandwell and Smith 1992) which is the sum of long-wavelength regional depth and the passband prediction (Smith and Sandwell 1994). The long-wavelength component (>160 km) is filtered from ETOP0-5 data. The short-wavelength component of bathymetry is predicted from free-air gravity anomalies within the band of wavelength between 15 and 160 km. The bathymetric features of wavelength less than 160 km are flexurally supported by the plate with elastic thickness greater than 5 km Their vertical amplitudes are proportional to the downward continuation of free air gravity anomalies by a factor of the Bouguer constant 2npr, where p is the density of the seafloor relative to seawater and r is the Newtonian gravitational constant. We used a value of 15 m/mGal for the Bouguer constant during the bathymetric prediction The vectors show different models of absolute Nazca plate motion Solid arrows are calculated from the trends of the volcanic chains (see text) The dashed arrows were calculated from average Pacific hot spot pole of the last 10 m .y. (A12) (Minster and Jordan 1978; Naar and Hey 1989). The same rotation rate is used for calculating the v e ctors

PAGE 88

-..) VI Jso w 14o w ----13ow 12o w IIO w Ioow 9ow sow 1ow o 1os 2os 3os 4os

PAGE 89

Review of Previous Work It is clear from its appearance alon e, that the ESC is the result of long and complex interactions between the oceanic lithosphere and the underlying mantle However, the nature of these interactions is unknown Several hypotheses have been proposed to explain the origin of this feature, such as the "hot spot hypothesis" (Hagen et al. 1990 ; Handschumacher et al., 1981; Hey et al ., 1995 ; Morgan, 1971 ; Morgan, 1972 ; Okal and Cazenave, 1985; Pilger and Handschumacher 1981; Schilling et al., 1985a; Wilson, 1963a; Wilson, 1963b; Wilson, 1973) the "l e aky fracture zon e hypothesis (Clark and Dymond 1977; Epp, 1984), the "hot line" hypothesi s (Bonatti and Harrison, 1976 ; Bonatti et al., 1977), and the "diffuse extension" hypothe s is ( Sandwell et al. 1995; Winter e r and Sandwell, 1987) We will review these hypotheses under two categories, hot spot versus other models. Hot spot. A hot spot or m a ntl e plume, i s an upwelling thermal anomaly in the mantle (Morgan, 1971 ; Wilson, 1963b). Hot spots ar e proposed to b e r e latively stationary in the mantle As a plate moves over a hot spot, a chain of volcanism is formed on the plate surface. It has been suggested that a hot spot located under th e EPR ( East Pacific Ris e ) might be responsible for the formation of th e ESC ( Morgan, 1971; M o rgan, 1972; Wilson 1963a; Wilson, 1963b ; Wilson, 1973). Accordingly, the Crough-Ducie-Henderson Tuamotu Island Chain has been suggested to be the mirror imag e of the ESC extending to the west o f the EPR from the same hot spot Reconstructions of the Pacific and Nazca plates suggest that the Nazca and Tuam otu ridges originated from a melting anomaly und e r the Pacific-Farallon spreading center during the time interval between anomalies 19 and 11 (Handschumacher et al., 1981 ; Pilger 76

PAGE 90

and Handschumacher, 1981) However, the ESC formed entirely within Nazca plate from a hot spot located to the east of the EPR (Pilger and Handschumacher, 1981) while Crough-Ducie-Hendderson Island Chain resulted from another hot spot (Okal and Cazenave, 1985) Side-scan sonar image data and swath bathymetry data show there has been recent volcanism about 130 Ian to the west the Easter Island, referred as Ahu volcanic field (Hagen et al., 1990; Hey et al. 1995). This volcanic field is clearly off the ridge axis and might be the current location of the "Easter hot spot" (Haase and Devey, 1996 ; Hagen et al., 1990), or near Umu volcanic field less than 100 km south of the Ahu volcanic field (Stoffers et al., 1994) Geochemical and geophysical evid e nce supports that the East rift of the Easter microplate is under the influence of a hot spot located beneath either Salas y Gomez or Easter Island (Hey et al 1985; Naar and Hey 1986; Schilling et al., 1985a). Data on K20 and La/Sm variations along the East rift of the Easter microplat e culminate around 27 "S. Similar patterns are observed on the West rift but with small e r amplitudes indicating that this segment of the EPR is influ e nc e d by lateral flo w of plume-d e rive d material from east of the microplate (Schilling et al., 1985a). This hypothesi s is also supported by other geochemical studies (Fontignie and Schilling, 1991; Poreda et al ., 1993b). Studying the group velocity of Rayl e igh waves that propagated along either the Nazca ridge or the ESC suggests that the crustal thickness along the Nazca ridge is about 18 km, while it is normal (-6 km) along th e ESC (Woods and Okal, 1994). Flexural studies indicate that the elastic thickness of the lithosphere along the ESC is very thin ( -3 km) at the time when th e volcanoes were loaded (Liu and Naar l996a ; Liu et al. 1995 ; Woods et al., 1993), while the elastic thickness n e ar the S a n Felix Island (-13 km) is significantly greater (Liu and Naar, 1996a ; Liu et al. 1995). Different geochemical 77

PAGE 91

patterns have also been observed between the San Felix Island and the ESC (Gerlach et al., 1986). Other models. If the ESC is a hot spot trail, the age of volcanism should increase progressively to the east. Based on the rate of motion of the Nazca plate over hot spot reference frame (Naar and Hey, 1989a), Salas y Gomez should be about 4 m .y. older than Easter Island. The ages calculated from rock samples using K-Ar method show that the volcanism on Salas y Gomez is generally younger than that on Easter Island (Baker, 1966; Baker et al., 1974; Bonatti and Harrison, 1976; Bonatti et al., 1977 ; Clark and Dymond, 1977). To explain what was thought to be contemporaneous K-Ar ages along the chain, several alternative hypotheses have been advanced, assuming the K-Ar ages are correct A fracture zone (referred to as the Easter fracture zone) apparently offsets the Mendoza rise (Herron, 1972a ; Menard and Atwater, 1968), stretching from near 25s, 90"W to approximately 26s, 99w. The ESC lies north of and parallel to the fracture zone, extending from 9ow to 101w (Fisher and Norris, 1960). A leaky fracture zone might be responsible for the formation of the ESC (Clark and Dymond, 1977; Menard and Atwater, 1968). Instead of a single hot spot, a mantle hot line, corresponding to upwelling limbs of secondary mantle convective rolls, was proposed to be responsible for the origin of the ESC (Bonatti and Harrison, 1976 ; Bonatti et al., 1977). Geochemical data indicate that the sources of the volcanism along the ESC and the Tuamotu Chain are similar to that of the intraplate volcanism elsewhere (Bonatti and Harrison, 1976; Bonatti et al., 1977). San Felix, a volcanic island about 2800 km east of the Easter Island, has shown signs of volcanic activity after the Chilean earthquake of 1922 (Firth, 1943; Willis and Washinton, 1924). Based on theoretical consideration and laboratory experiments, Richter and Parsons (197 5) suggested that there could be secondary convection rolls in the upper mantle, 78

PAGE 92

approaching Rayleigh-Benard-type convection, which are parallel to the plate motion and reach depths of about 650 k:m (Richter and Parsons 1975). The thermal anomaly resulting from this type of convection would explain the apparent contemporaneous volcanism along the ESC which was based on K-Ar age data A diffuse extension model is proposed from the active off-axis volcan i sm on the south Pacific oriented roughly in the direction of absolute plate motion (Sandwell et al. 1995; Winterer and Sandwell, 1987). SeaBeam data along the Pukapuka ridges indicate the chain are composed of numerous ridge segments arranged in both end to-end and en echelon patterns En-echelon ridges are commonly sigmoidal, and most are in dextral at angles of 5 -20 to the overall trend of the Pukapuka Chain (Sand well et al 1995) The slab-pull from the trenches to the north may produce tensional stresses in this direction (Sandwell et al., 1995). A 10-20 % strain would be required for the formation of the extensional ridges. However the multibeam data show no evidence for widespread E W striking normal faults as predicted by a N-S extension. The flexure studies along the volcanic ridges (Goodwillie 1995) and previous studies indicate that the south Pacific region is characterized by lower than expected elastic thickness values Lithospheric stretching or small-scale convection model cannot satisfactorily explain all of the observations (Goodwillie 1995). Age measurements. The radiometric age of a rock sample may be different from the true initial age of a volcano due to sampling and contamination problems. Each seamount may be built by sporadic volcanic activities over several million years The rock samples are usually collected on the surface of the seamounts which represent the youngest volcanism. Therefore, it is likely that the calculated age is younger than the age predicted by the hot spot model. Contamination problems may also cause the age offsets, such as 79

PAGE 93

alteration of the rock due to exposure to seawater For example, the Sr contamination from seawater may cause the calculated ag e of a sample to be younger than the age of formation. More accurate 40Ar-39 Ar age data indicate that the volcanism becomes progressively older to the east along the ESC (Haase and Devey, 1996 ; O'Connor et al 1995). Samples most suitable for 40Ar3 9Ar dating were selected after careful thin section inspection. 40Ar39Ar and some earlier K-Ar ages decrease systematically between the Nazca ridge and the western end of the Salas y Gomez Island, supporting that at least this part of the ESC has formed by the Nazca plate migrating over a hot spot (Haase and Devey, 1996; O'Connor et al 1995) This age trend is also supported by additional age data (Stoffers and Hekinian, 1990) There is disagreement, however, regarding the location of the hot spot. Tectonic Setting The ESC, lying entirely on the Nazca plate, trends about 085 within a box with corners at 30 S/120W and 20S/80W (Figure 4.2). The chain runs about 3000 km from the shallowest part of the EPR, slightly west of the Easter Island, to its junction with the Nazca ridge, about 1000 km west of the Peru-Chile trench The Mendoza rise, a failed rift, can be identified north of the ESC near about 94 O W in both ETOPO 5 data, gravity anomalies calculated from satellite altimetry data (Sandwell 1992), and crossing during Legs 6 and 7 of the Gloria Expedition (Naar et al., 1993a ; Naar et al ., 1993b). To the west of the ESC is the Easter microplate The fastest seafloor spreading center in the world exists to the south of the Easter microplate (DeMets et al 1990 ; Handschumacher et al., 1981; Herron, 1972a ; Hey et al. 1995; Klaus et al., 1991; Naar and Hey, 1989a; Rea, 1981). Extending to the west of the EPR is the Crough-Ducie-Henderson and Tuamotu Island Chains. 80

PAGE 94

Figure 4 .2. A : Location map showing the shiptracks of Legs 5, 6, and 7 of the Gloria Expedition. The plate boundaries are from digitized global plate boundary. B: The predicted bathymetry along the ESC within the region indicated in the box in panel A. Shallow depth is in light gray. Short lines are the location of topographic profiles shown in Figure 4.6. The detail GLORI-B side-scan sonar image data and bathymetry data for areas A and B indicated by small boxes are shown in Figure 4.5. The large box indicates the Eater-Salas y Gomez region, where most of the data were collected during the Gloria Expedition. C : The shiptracks on the background of gray image of the predicted bathymetry for the area indicated by the large box in panel B.

PAGE 95

II.) -o 0 -10 g -20 ] -30 -40 -50 -130 20 II.) -o 3-25 ;:: -30 -25 -26 II.) -o .g -27 ] -28 -29 -115 A : Location Map Pacific Plate Nazca Plate Antarctic Plate -120 -110 -10 0 -90 -115 -110 -105 1 00 -9 5 -114 -113 -11 2 -Ill -110 -109 -108 l o ngitude 82 South Americ a 80 -70 -90 -85 -107 -106 -105 -104 10 0 -10 II.) -o -20 g ] -30 40 50 60 -20 II.) -o -25 g ] -30 -80 -25 -26 II.) -o -27 g ] -28 -29 -103

PAGE 96

The Easter and Salas y Gomez islands are in the western portion of the ESC. Many shallow seamounts were observed in the chain to the east along the transits of the Gloria Expedition (Naar et al., 1993a; Naar et al., 1993b). The western end of the ESC is near the Ahu volcanic field. Estimated from the side-scan intensity, the age of this volcanic field is about 0.4 Ma (Hagen et al., 1990), which is in close agreement with the age data from recent samples (Liu et al., 1996; Naar et al., 1993a; Stoffers and Hekinian, 1990). The volcanic field is on seafloor of 1.66 to 3.88 Ma (Hagen et al., 1990). The oldest volcanism, about 43 Ma, has been observed at the junction between the Nazca ridge and the ESC (Pilger and Handschumacher, 1981). Existing to the east of the ESC, the San Felix and San Ambrosio islands do not appear to be part of the ESC due to an observed volcanic gap and morphology difference (Naar et al., 1993a; Naar et al., 1993b). Flexural studies show that the e l astic thickness of the litho sp here near the islands is significant different from the ESC, suggesting they are different origins (Liu and Naar, 1996a; Liu et al., 1995) Recent volcanic activities occurred about 70 years ago at San Felix and San Ambrosio islands (Firth, 1943; Willis and Washinton, 1924). Between magnetic chrons 30 and 6c (67 and 24 Ma), the Pacific Farallon Spreading Center was active, trending NNW in a dextral offset (Searle et al., 1995). At 24 Mathe Farallon plate split into the Cocos and Nazca plates, which was followed by a major plate reorientation (Hey, 1977b; Hey e t al., 1977; Searle et al., 1995). The spreading direction rotated clockwise. During about 9-21 Ma, spreadi ng was active at the l ocatio n now known as the Mendoza rise (Herron, 1972a). Since about 9 Ma, the NNE-oriented EPR became active (Herron, 1972a) and persisted until initiation of the Easter microplate between 5 1 5.7 Ma (Bird and Naar 1994) that is in approximate agreement with other studies (Klaus et al., 1991; Naar and Hey, 1991; Okal and Cazenave, 1985; Rushy, 1992; Searle et al., 1993). 83

PAGE 97

DATA AND OBSERVATIONS ALONG THE ESC Legs 5, 6 and 7 of the Gloria Expedition focu sed on the western end of the ESC including Easter and Salas y Gomez islands (Figure 4.3) In this region, a grid of bathymetry and of side-scan sonar image data are compiled from the bathymetry swaths and side-scan sonar image data covering completely the surveyed region (Liu and Naar, 1996c ; Liu and Naar, 1996d ; Liu et al., 1994) Th e transits to the east of the Easter Salas y Gomez region along the chain from Legs 6 and 7 also collected swath bathymetry and side scan sonar image data, which provide important evidence and constraints for understanding the physical nature of the entire chain En-echelon Volcanic Ridges The ESC, inst e ad of being a single volcanic chain i s actually composed of a number of small individual ridges in an en-ech e lon pattern The predicted bathymetric features in the Easter-Salas y Gomez region (Figure 4.3A) were confirmed to be volcanic features (not fracture zones) by Gloria Expedition (Figure 4.3B). The high intens i ty values observed in the side-scan sonar data, presumably due to th e young lava flows and l o w sediment cover, exist to th e northwest and southw e st of Easter Island and northwest and northeast of Salas y Gomez Island (Figure 4.3C). However, these lava flows are broad horizontal constructs rather than vertical. Combination o f th e bathymetry and sid e-scan sonar image data (Figure 4.3D) shows both type s of volcanic f e atures, which lin e up as several small individual ridges. The ridge s ar e group e d in a d e xtral en echelon, with e ach ridge trending from 090" to 1 ()()", oblique to the overall trend of tl)e ESC 085". The former trends are close to the Nazca plat e motion with respect to the Pacific plate and the Easter 84

PAGE 98

Figure 4.3. Comparison of different data sets in the Easter-Salas y Gomez region indicated by the large box in Figure 4 2B. A : The predicted bathymetry. B : Observed bathymetry illuminated from southwest. The data are compiled into a grid of .003 by .003 degrees from SeaBeam 2000, GLORI-B, SeaMARC II, and ETOPO 5 data sets (Liu and Naar, 1996c; Liu et al., 1994). In the densely surveyed area, the data are composed of three sources: the SeaBeam 2000 swaths centered along the nadir and extending about 5 km to both sides, the GLORI-B data along both sides of SeaBeam swaths reaching out to about 24 km of total swath width, and the interpolated data filling the 6-km gaps between the swaths. Outside of the surveyed area, the SeaMARC II data is shown to the west of 109"W and ETOPO 5 data is shown to the east of 109"W. C: Observed side-scan sonar intensity. The dark areas near shallow seamounts and between shiptracks on shallow seafloor are artifacts of the GLORI-B sonar system. The data are compiled into a grid of 001 by .001 degrees from side scan sonar data collected with GLORI-B and SeaBeam systems (Liu and Naar, 1996b; Liu et al., 1994). D: A combination of bathymetry and side scan sonar image data showing the trends of the volcanic ridges The subsidence trend of the seafloor are removed from the bathymetry data based on the age grid calculated from the seafloor isochron. Then, the bathymetry data are scaled to the range from 1 to 255 to be compatible with the side-scan sonar image data. The bathymetry and image grids are merged by selecting the larger value (intensity or scaled bathymetry) for each cell.

PAGE 99

longitude -I 15 -114 -I 13 -112 -Ill -1 1 0 -109 108 -107 -1 06 -105 -1 04 1 03 -25 0 "0 -26 .l -27 E! 0 "0 -28 -29 -25 -26 -28 -29 26 .l-27 E! 0 "0 -28 -2 5 -26 .l -27 E! -28 -29 225

PAGE 100

microplate, while the latter trend indicates the Nazca plate motion with respect to the mantle plume that is responsible for the formation of the ESC The orientation of the ridges rotates counterclockwise from west to east. Each ridge is about 1-3 km high, 200-500 km long, and less than 80 km wide, and spaced about 50-100 km from other ridges Four major individual ridges can be identified in this region (Figure 4.4A) Two larger ones are clearly seen in the bathymetry data, containing Easter Island and Salas y Gomez Island which we refer to as Easter ridge and Salas ridge, respectively Two smaller ones exist, one running through Getu seamount between Easter ridge and Salas ridge (referred as Getu ridge), and the other running through Tupa volcanic field (Stoffers et al ., 1994) on the southern side of the SOEST FZ (Hey et al 1995) (referred as Tupa ridge) Detailed information for each ridge is listed in Table 4.1. The ridges trend subparallel in the NW -SE direction. Table 4.1. Information about individual ridges western end ridge lat. Ion. name ("S) CW) eastern end lat. Ion CS) ("W) dimension trend length width (azim.) (km) (km) total size volume sphere d (km3) (km) ---------------------------------------------------------------------------------------Tupa 27 6 110 8 28.0 109.0 100.5 180 50 3174 18.2 Easter 26.7 111.3 27 6 107 1 102.5 480 80 16420 31.5 Getu 26.7 108 5 27 1 106 0 98.5 270 60 4718 20. 8 Salas 26.4 108.8 26.5 103 8 91.0 540 50 16776 31.8 The Tupa ridge exist to the southwest of the Easter Island and the Ahu volcanic field (Figure 4 3) Volcanism increases to the west along the ridge The Easter ridge runs across the outer pseudofault of the East rift. Easter Island, the largest volcano in the ridge, formed near the pseudofault (Figures 4.3 and 4) The pseudofault may have acted as a magma conduit because the lithosphere has been weakened More low-lying volcanism exists along the western portion of the ridge than to the east of the pseudofault suggesting a difference in lithosphere structure between the two sides of the pseudofault. It appears to 87

PAGE 101

Figure 4.4. Interpretation maps. A: Major linear structures of seafloor interpreted primarily from bathymetry data. The combined of bathymetry and side-scan sonar image data from Figure 4.3D is shown in the background. B: The distribution of seamounts. The basal outlines of the seamounts were digitized (by Y. Rappaport) from bathymetry data (shown in Figure 4.3B) and are approximately represented by gray ellipses. Each profile in the shaded boxes illustrate the mean height of the seamounts across each of the boxes The total volume of the seamounts is calculated for each of the boxes (Table 1). C: Volcanism and lineations interpreted primarily from side-scan sonar image data. The rose diagram shows the orientation histogram of the spreading fabrics

PAGE 102


PAGE 103

be more difficult for the magma to penetrate the older, thicker lithosphere on the eastern side. The Getu ridge starts from the east of the Kahuna seamount (Figure 4.3B). However, it has a small trail of volcanoes and lava flows extending to its northwest (Figure 4.3C). Two deep troughs identified near the Kahuna seamount trending NNW may have also weakened the lithosphere forming a magma conduit of the very large Kahuna seamount and reducing magma supply northwest of the seamount (Figure 4 .3B). The Salas ridge appears more robust to the east of 106.5 W (Figure 4.3B) The western extension of the ridge consists of scattered lava flows (Figure 4.3C). The large volcanoes near 26.4"S, 106.5"W may also have resulted from magma flowing into the troughs In general, large and robust volcanoes tend to be in the central or eastern portions of a ridge, while the western end is comprised of scattered young lava flows The large area of lava flow to the northeast of Salas ridge near 25.5"S, 104"W appears to be the western end of another volcanic ridge (Figure 4.3C) In fact, several large volcanoes were observed along transits due east of this field On the other hand, the volcanism near 28.5"S, 111.2"W and 28.3 S, 110.7"W appears to be a young ridge in its early stage, and more volcanism may emerge and extend to the northwest from this area. This type of volcanic ridge probably forms the ESC to the east of the Easter-Salas y Gomez region, because similar patterns are identified in the Geosat altimetry data for that region (Figures 4 1 and 2C). These observations disagree with single classic hot spot and hot line models which predict a single chain of volcanism. En-echelon Ridges Elsewhere En-echelon ridges are not unique to the ESC. Similar patterns have been observed elsewhere in the altimetry data, e g ., the seamount chains extending to the west of the 90

PAGE 104

southern East Pacific Rise and the Easter microplate (Sandwell et al., 1995; Searle et al., 1995) and the Foundation Seamount Chain (Mammerickx, 1992) on the Pacific plate (Figure 4.1). Predicted ridge-like en-echelon patterns between Tahiti and the Easter microplate on Pacific plate were confirmed to volcanic ridges by the GLORI-B and SeaBeam data collected during the Charles Darwin and Jean Charcot cruises. There are six en-echelon ENE-trending ridges, between 121.5"W and 118 2 W (Searle et al., 1995). They are 60120 km long, 0.5-2.5 km high, and 10-25 km wide, and are clearly of volcanic construction, consisting of many coalesced cones of 1-3 km in diameter. Occasionally these ridges have more amorphous, knobby material on their flanks which display strong backscatter in side-scan sonar data. The tops of the individual cones do not produce such signals in the side-scan sonar data, suggesting that they have a moderate sediment cover and are relatively old. However, most of the ridges appear to have relatively young lava flows associated with them The ridges all have quite sharp, linear crests, as demonstrated by their acoustic shadows; these crests strike consistently 065"-076 The trend of the local spreading fabric is not consistent and often departs considerably from being orthogonal to the ridges (Searle et al., 1995). Similar volcanic patterns were observed along Pukapuka ridges. SeaBeam data were collected along a series of volcanic ridges that extend 2600 km northwest from the west flank of the EPR. The major structures in the chain are composed of numerous ridge segments arranged in both end-to-end and en echelon patterns (Sandwell et al. 1995). Individual ridge segments and sets of ridges are highly elongate in the approximate direction of present absolute plate motion at angles 5 -20" oblique to the overall Pukapuka trend. The largest gap in volcanism is about 130 km long with spreading fabric being generally buried. Ridges are generally lower (<1500 m) and highly elongate along the eastern end of the chain and higher ( < 4000 m) and more conical along the western end 91

PAGE 105

Individual ridges are typically about 35-50 km long and 15 km wide The ridges formed along a band 50to 70-km-wide in the trough of one of the more prominent gravity lineaments Radiometric age dating shows a general progressive increase in the age of volcanism from east to west along the Pukapuka chain Ages of these ridges range from 6 to 27.5 Ma, while the ages predicted with a single hot spot model are 6 9-33.8 Ma, and the seafloor ages vary 7.7-45.4 Ma (Sandwell et al. 1995) The Hawaiian-Emperor Chain is also composed of many linear volcanic ridges that lie along relatively short, curved trails which are sometimes of sigmoidal form (Jackson and Shaw 1975). These ridges lie at an angle to the overall orientation of all the volcanoes. The PrattWelker, Hawaiian, Tuamotu and Austral Chains consist of en echelon ridges with a dextral overlap, whereas the supposed older extensions of these chains that lie nearly north-south (the Emperor and Ellice-Gilbert-Marshall Chains) consist of sets of ridges with a sinistral overlap (Jackson and Shaw, 1975). Failed Propagators There are many deep troughs identified in th e bathymetry data (Figure 4.3). Some of them may be failed pro pagators and pseudofaults (Figure 4.4) The EPR ridge crest near 29 S, 112 W is the active northward propagator of an overlapping spreading center (Figure 4.3B) (Hey et al., 1995 ; Klaus et al., 1991; Naar and Hey 1991). A series of linear structures exist to the northeast of the propagator in a sinistral en-echelon pattern, which is interpreted as failed propagators (Figure 4.4A) (Hey et al ., 1995) It appears that the scale of the propagator inc reases from older to younger. Th e largest failed propagator in this group is near 28 5 S, 11 rw. Its southern portion is buried under young volcanoes, suggesting the rift cut deep into the lithosphere and became a volcanism site. Its 92

PAGE 106

northern portion is a deep rift, analogous to the Galapagos 95.5 W propagating rift tip (Hey et al., 1986), suggesting the level of faulting did not tap magma at depth. The large deep trough between Getu and Kahuna seamounts and the volcanic ridge extending to the south from the trough appears to be analogous to the failed propagator near 28.5S, 111.2 W (Figure 4.3B) Several shallower troughs fan to the south from the deep trough, suggesting it was a northward propagator. There is another large trough to the east of the Kahuna seamount, which may be the outer pseudofault associated with the failed propagator (Figures 4.3B and 4A). To the west of the Easter Island there are several pseudofaults formed by the East rift as it propagated to the north and south (Figure 4.4A) (Hey et al., 1995; Hey et al., 1985; Naar and Hey 1986; Naar and Hey, 1991; Rushy, 1992; Searle et al., 1989) To the northeast of the Easter Island, a small portion of a pseudofault is identified in the side scan data near 25 5S, 108W (Figures 4.3C and 4C). The spreading fabrics to the east of the pseudo fault are truncated, while those to the west are sub-parallel to the pseudofault or fanning to the south, suggesting the existence of a northward propagating rift prior to the northward propagator that formed the Easter microplate. There are several large troughs in the area between this pseudo fault and the outer pseudofault of the East rift, which may be associated with failed propagating rifts Fracture Zone A linear feature is clearly identified between the Easter and Tupa ridges on both bathymetry and side-scan sonar image maps (Figures 4.3B and 3C), which is interpreted as the SOEST FZ (SOEST fracture zone) (Hey et al., 1995) (Figures 4.4A and 4C). Starting from near 28 S, 107 W, the fracture zone can be identified in both side-scan sonar image 93

PAGE 107

and bathymetry data. It becomes more pronounced from 107 .5"W, extending to the WNW at a trend of 098". The western end of this fracture zone disappears under volcanism near 27 7"S, 110.6 "W. Starting a little south of the western end of the SOEST FZ, a smaller fracture zone extends further west, trending 107". This change of the fracture zone trend suggests a reorientation of the relative plate motion between the Na zca plate and the Easter microplate, which may be associated with a change (or initiation of) rotation of the Easter microplate. The western end of this fracture zone is near 27 5"S, 111.5 "W, near the junction of the two inner pseudofaults of a northward and southward propagator (Figures 4 3 and 4) (Hey et al., 1995; Naar and Hey, 1991). Clearly, the SOEST FZ extends to the east beyond the outer pseudofault of the Easter microplate, thus existing before the initiation of the Easter microplate (Bird and Naar 1994). The SOEST FZ (098 ) trends close to the Getu ridge (098 5"), but oblique to the two ridges surrounding it, the Easter ridge (102 5 ) and the Tupa ridge (100.5") (Figures 4.3 and 4A) (Table 4.1 ). The trend of the two latter ridges rotates clockwise and becomes close to the trend of the western extension of the SOEST FZ (107 ) The clearance between the SOEST FZ and the two nearby ridges suggests that they were of different origins. Based on the en-echelon and oblique ridge pattern, it is unlikely that there are fracture zones under the ridges; however we are unable to prove this with bathymetry and side-scan sonar image data only Seamounts In addition to the islands, there are numerous shallow seamounts in the ESC In the Easter-Salas y Gomez region, there are a total of 553 seamounts over 0 2 km high (above seafloor), including 383 seamounts with aspect ratios less than 2 (Figure 4.4B). Of 94

PAGE 108

the 383 seamounts, there are 45 seamounts of 1-2 km high and 15 seamounts of >2 km high including two islands Most of the large seamounts are distributed along the major volcanic ridges. The average seamount heights are calculated across the shaded boxes and shown as the profiles in Figure 4 4B. The peaks are in the center or to the east portion of the profiles. The total volume of seamounts on each ridge is calculated in the corresponding box and ranges from 4,000 to 16,000 km3 (Table 4.1) There are also many shallow seamounts to the east of the Easter-Salas y Gomez region which are apparent in the altimetry data and are confirmed along the transits by the bathymetry data from the Gloria Expedition. The number of seamounts increases to the east of 112"W which is about 100 km off the spreading center. The ESC style volcanism does not continue all the way to the EPR, indicating there is a volcanism gap between the ESC and the EPR. The shallow depths of these larger volcanoes suggest that they were tapping from magma sources of high hydraulic head and/or temperature The magma sources may be deeper than the volcanism along normal mid-ocean ridges. However, the diffuse extension, leaky fracture zone, and hot line models predict a magma source depth similar to that of mid ocean ridges, therefore, lack of "deep mantle" geochemical signature. Volcanism and Lineations The volcanic gap between the ESC and the EPR is also supported by the side-scan sonar image data. Areas where high intensity values were observed in the side-scan sonar image data, presumably areas of young volcanism, are distributed along the active spreading centers and the western ends of the volcanic ridges (Figures 4.3C and 4C). The former areas of young volcanism can be attributed to the activities of spreading center, while the latter areas are assumed to be related to the thermal anomalies that produced the 95

PAGE 109

ESC There is an area of normal seafloor between these two types of volcanism (as partially observed by Hagen et al., (1990)). The westernmost volcanism associated with the ESC the Ahu volcanic field, exists only east of 112W. Figures 4.5A and 5B show the volcanic fields surround Easter and Salas y Gomez islands, which are typical along the ESC and quite different from other volcanic flows, even the Hawaiian flexural flows (Moore et al., 1989) These lava fields are near large volcanoes that are usually formed along the individual volcanic ridges (Figures 4 5C and 5D) The volcanic fields are characterized by small circular shaped volcanoes (Figures 4.5E and 5F). The hot line model cannot explain the existence of the volcanic gap between the ESC and the EPR and the physical difference and gap from the ESC to the San Ambrosio and San Felix islands and eastward to the Chile trench. The abyssal hill fabric is clearly identifiable in the volcanic gap (Figures 4.3C and 4C) Along the ESC, the abyssal hills are buried under the lava flows. Due to what appears to be active episodes of rift propagation the orientation of the seafloor abyssal hill fabric may not well represent the original spreading pattern. As shown in the rose diagram (Figure 4.4), there are three major groups of fabric. They trend at about 001 005", and 007. Assuming the fabric perpendicular to spreading direction, these trends correspond to seafloor spreading directions of about 091, 095, and 097", respectively. The clockwise rotation of the spreading direction may result from the change of relative plate motion from Nazca/Pacific to Nazca/Easter or Nazca/Easter to a new Nazca/Easter as the Easter microplate rotates. A chang e of relative motion is also suggested by the change of the SOEST FZ trend from 098 to lOT near 27.TS, ll0.6W. 96

PAGE 110

Figure 4.5 Data near Easter Island (panels A, C, and E for area A indicated in Figure 4.2B) and Salas y Gomez Island (panels B, D, and F for area B). Panels A and B show the side-scan sonar image data Panels C and D show the bathymetry data Panels E and F show the interpretation of the data. Light gray represents young lava flows. Dark gray shows old lava flow and volcanoes. Circles represent the circular shaped young volcanoes Ellipses show the approximate basals of seamounts with aspect ratios less than 2 and heights greater than 200 m.

PAGE 111

Easter I s land Salas y Gomez Island A B c D E F 98

PAGE 112

Seamount Morphology Topography of the islands and seamounts in the chain suggests that the eastern ones have been exposed to longer periods of erosion than the western ones, indicating that, in general, the volcanism (associated with very large seamounts) becomes progressively older to the east along the ESC. Eight shallow seamounts or islands along the chain are selected to compare the changes in their morphology (Figure 4.6). Among them, the westernmost two run across the Easter and Salas y Gomez islands, respectively The indented or flattened slope of the profiles indicates some type of erosion and/or sedimentation after their formation. Salas y Gomez Island appears much smaller than Easter Island, even though they have a similar basal size The slope of the Easter Island is relatively constant below sea level, while the profile across Salas y Gomez Island is flattened between sea level and the 1000 m depth. Near shore of the Salas y Gomez Island, there are large recently formed underwater platforms near sea level (Fisher and Norris 1960) Five radial profiles show that there is a underwater platform that is terminated in a well-defined break west, south, and east southeast of the island. The uniformity in this shelf-break depth, 119-121 m, suggests that the platform was cut by waves East-southeast of the island, there is a steep break in slope at about 193 m. Above this break the bottom shallows gradually to the platform break at 119 121 m. This gently sloping surface may represent an older tilted shelf or the top of a flow originating east of the present island (Fisher and Norris, 1960). Although we do not have much bathymetry data control at th e depth between 200 and 1000 m, the major indentation on the slopes in the profiles indicates that some type erosion altered the original morphology of the island. Near surface wave erosion and/or landslide may cause the indentation of the slope which may be subsequently submerged due to the sea level rise and thermal and flexural subsidence of the basement. In contrast 99

PAGE 113

Figure 4.6 The topographic profiles of some islands and seamounts in the ESC at the locations indicated in Figure 4.2B. Profiles A and B are sampled from the processed bathymetry data along the base lines across Easter Island and Salas y Gomez Island, respectively (Liu et al., 1994) Profiles C through H are along transits of the Gloria Expedition.

PAGE 114

distance( degree) distance( degree) 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.0 0.1 0 2 0.3 0.4 0.5 0.6 0 A B 0 1 1 ------E E ...:..: ...:..: ......, ......, ..c ..c 0.-2 -2 0. C!) C!) -o -o -3 -3 -4 Easter Island Salas y Gomez Island 4 0 c D 0 ----1 -1 E E' ...:..: ...:..: ......, ......, ..c ..c 0.-2 -2 0. C!) C!) -o -o -3 -3 -4 -4 0 E F 0 ---1 1 ---E E ...:..: ...:..: ......, ......, ..c ..c 0.-2 -2 0. C!) C!) -o -o -3 -3 -4 -4 0 G H 0 ----1 1 E E' ...:..: ...:..: ......, ......, ..c ..c 0.-2 -2 0. C!) C!) -o -o -3 -3 -4 0.0 0.1 0.2 0 3 0 4 0.5 0 6 0.0 0.1 0.2 0.3 0.4 0 5 0 6 4 distance( degree) distance( degree) 101

PAGE 115

the Easter Island appears less eroded and, thus, presumed to be younger than Salas y Gomez Island. The side-scan sonar image of the seafloor surrounding Easter Island has higher intensity values in side-scan image data than that surrounding Salas y Gomez Island (Figures 4.5A and 5B), suggesting younger lava flows and/or less coverage of sediment surrounding Easter Island. It appears that the skirt of the Salas y Gomez Island is covered by thick sediments (suggested be 3 5 kHz data) which may be eroded from the island by mean of wave cut and landslide. The two profiles of the seamounts to the east of the Salas y Gomez Island have flattened tops at a depth of 200 m Because these shallow platforms are not circular and are considerably larger than normal craters they are unlikely to be filled craters. These flat tops may be submerged guyots or eroded surfaces This morphology indicates that these seamounts have been exposed for a longer time than the islands to their west. The profiles of the seamounts further to the east show either flattened tops or buried slope, also indicating older ages of volcanism. Progression of Side-scan Intensity To estimate the relative volcanic age (e g., Hagen et al. 1990), both average and modal intensities of the side-scan image along the ESC are compared (Figure 4.7). A series of locations along the chain (Figure 4.7 A) are selected where the backscatter intensity appears to be the local maximum in order to compare with young volcanism near the EPR. At each location, an average intensity and a modal intensity are calculated within a 0.2 by 0.2 degree box in order to obtain the reliable statistics. As shown in Figure 4 7B, the average intensity of the seafloor decreases from the EPR to the eastern flank with a jump to 102

PAGE 116

2o s 25 s 3o s 6500 -5800 -5100 -4400 -3700 -3000 -2300 -1600 -900 200 depth (m) 130 120 110 100 B : Average Intensity 0 90 ;;; c:: . c:: 70 60 50 40 130 1 2 0 110 100 0 ;;; 90 c:: . c:: 80 0 70 60 50 40 -80 Figure 4.7. Average (B) and modal (C) intensities calculated within each of the .2 by .2 degree boxes (A) from GLORI-B side-scan image data. 103

PAGE 117

high intensity at the western end of the ESC near the Ahu volcanic field as previously observed (Hagen et al. 1990). The intensity decreases along the ESC until to the area near the San Felix Island where the intensity jumps to a high value again The modal intensity displays a pattern similar to the average intensity (Figure 4 7C) The image acquired from a surface facing towards the sonar system will be biased towards higher intensity values The presence of old seafloor surrounding the volcanoes within the calculating box will reduce the average and modal intensities due to sediment cover absorbing the sonar signal. High intensities indicate strong backscatter from the seafloor presumably covered by young volcanic materials (based on their morphological shape, e.g. lava flow patterns, cones, etc. ) Therefore, the patterns of the intensity trend suggest that the volcanic age increases to the east along the ESC. If the San Felix Island was created by a s ingle hot spot along with the ESC, we would expect lower intensity near the Island. However, the side -scan intensity near the Island is relatively high with respect to th e volcanism near the eastern end of the ESC (Figure 4.7), suggesting volcanism is quite young and may be form e d from a diffe rent hot spot or offaxis process. Seamount Age Progression A major challenge facing the hot spot model is the apparent contemporaneous volcanism along the ESC based on the old K Ar age data (Bonatti e t al., 1977 ; Clark and Dymond, 1977). However, recent 40Ar_39Ar data indicate that th e volcano ages actually increase to the east along the ESC (R. Duncan, p e rsonal communication, 1996; Liu et al. 1996 ; O'Connor et al., 1995) New and old age data are compared in Figure 4 8 The new volcano ages are calculated from dredge samples collected during both Leg 7 of the 104

PAGE 118

Figure 4.8. Seamount ages along the ESC A : Dredge location. Stars from Leg 7 of the Gloria Expedition (R. Duncan, personal communication, 1995) circles from O'Connor (1995), Squares from Bonatti (1977), and triangles from Clark (1977). The background image is the predicted bathymetry (the same as in Figure 4.1) indicating the locations of the seamounts. B: Seamount ages versus longitude. The data are represented with the same symbol convention as in panel A. Uncertainti es for the volcano ages shown in th e plot represent standard deviation (e.g. one sigma) Some of the error bars lie within the dimensions of the symbols. No unc e rtainty rang e is provided from Bonatti (1977) data All his five samples from Easter Island shown atmospheric 4DAff36Af ratios; only upper limits are given for their age.

PAGE 119

II 15s 2o s 25 s 3o s 11 15s 2os 25s 3os -6500 -5800 -5100 -4400 -3700 -3000 -2300 -1600 -900 -200 500 depth (m) 35 35 B: Seamount Age 30 6 Clark 0 30 o Bonatti 0 Sonne80 25 25 Gloria07 N !11 Q) E ;:;E 0 ;:;E "--' C) "--' Q) >. Q) ... '-I.. 15 "' < "' c:: < < '-Ll Cl) Cl) (!) 10 0 10 0 IJ 5 5 !11 i ** 0 0 o 0 -115 -110 -105 -100 -95 -90 -85 8 0 longitud e (0 ) 106

PAGE 120

Gloria Expedition (Naar et al. 1993a ; Naar et al. 1993b; Pereda et al., 1993a) (calculated by R. Duncan pers onal communication 1995) and the Sonne 80 cruise (O'Connor et al ., 1995) Some of the ages from Bonatti (1977) are unreliable due to post alteration. Because the samples are usually taken on the surface of the seamount, their ages represent the last events of their volcanic activities and, therefore are younger than the age of the seamount emplacement. Except the age data from Bonatti (1977) the volcano ages increase generally to the east, supporting the hot spot model. The patterns near the western end of the ESC indicate a large overlap of the volcano ages as more data have been obtained. The seamount age from the sample near the San Felix Island is about 0.8 05 Ma (R. Duncan, personal communication, 1995) and is much younger than the seamount age near the eastern end of the ESC This data is in agreement with the in situ observation of the dredge samples, which had a young appearance (Naar et al. 1993a). Volcanic activities were also reported after the Chilean earthquake of 1992 (Firth, 1943) Figure 4 9 shows the volcano ages in the Easter-Salas y Gomez region The ages are calculated by R. Duncan (personal communication, 1995) from the dredge samples collected during Leg 7 of the Gloria Expedition (Naar et al., 1993a; Naar et al., 1993b ; Poreda et al., 1993a) The white line boxes in Figure 4.9A show the approximate range of the two major volcan i c ridges the Easter and Salas ridges The volcano ages calculated from the samples within the boxes fall into the two tippled regions respectively (Figure 4.9B) The two young ages calculated from the samples near the eastern end of the Salas ridge may associate with the young lava flow at the western end of another volcanic ridge. These data support the age progression along individual ridges indicated by the side-scan intensity. The noisier age progression (Figure 4 .8) may be explained by the lateral overlap of the volcanic ridges 107

PAGE 121

Figure 4 9. Age trends along the Easter and Salas ridges A: Leg 7 dredge locations of the Gloria Expedition on the combined bathymetry and image (Liu and Naar, 1996a). White line boxes indicate the two major volcanic en-echelon ridges B : Seamount ages calculated by R. Duncan (personal communication, 1996) The ages from the samples within the two boxes fall into the two stippled regions, respectively Uncertainties for the seamount ages shown in the plot represent standard deviation (e g one sigma). Some of the error bars lie within the dimensions of the symbols

PAGE 122

112 w 111w 110 w to9w 25 s A: Dredges on Volcanic Ridges 26s 21s 2s s 25 50 100 200 225 250 Image Intensity or Scaled Bathymetry B : Seamount Age Trends 3 3 'i it t ,..... I ,..... "' "' 62 Sal as Ridge 26 Easter Ridge bl) bl) "' "' 1ft f w 'i 1ft * 0 0 -112 -111 -110 -109 -108 -107 -106 -105 -104 longitude CO) 109

PAGE 123

HOT BUBBLE MODEL The hot spot model provides a general explanation for the observations along the ESC. However, it is difficult to apply this model to the en-echelon patterns observed from the side-scan image, bathymetry, and radiometric age data. An alternative hypothesis, the Hot Bubble model (Figure 4.10), has been proposed to explain the formation of the ESC (Liu et al. 1995) This modified hot spot model provides a better explanation for the origin of the chain and suggests a possible deep structure for unsteady mantle plumes. The term Hot Bubble is used to emphasize buoyant hot "blob" (Schilling and Noe-Nygaard, 1974), and the model is similar to a model proposed for Hawaii (lhinger 1995) Constraints for the Model The observations presented in this paper are major constraints for any model about the formation of the ESC. The lines of evidence are rephrased as follows: (1) The ESC is composed of a number of small individual ridges in a dextral en-echelon pattern. Large and robust volcanoes tend to occur at the central or eastern portion of the ridges, while younger, scattered low relief volcanism usually occurs near the western ends of the ridges (2) Many of the relic rifts have become sites of volcanism. Most of the rifts trend generally in an N-S direction, indicating EW extension. (3) The fracture zones identified near the western end of the ESC do not underly the nearby volcanic ridges and they have different trends. (4) There is a volcanic gap of about 100 km between the ESC and the EPR. (5) The elastic thickness of the lithosphere is about 3 km and consistent along the ESC, while this is -13 km near San Felix Island. There is also a volcanic gap between San Felix Island and the ESC and between the San Felix Island area to the Peru-Chile trench. (6) The 40Ar-110

PAGE 124

Figure 4.10. A model of the deep structure of Earth 's dynamic system. A mantle plume is coupled with and heated by an upwelling spurt of the convection cells in the outer core. The plume in the mantle advects in the form of a series of hot bubbles emanated from a diapir on the core/mantle boundary. The bubbles are sheared into long footballs" in the upper mantle by the relative motion between lithosphere and the upper mantle. These footballs are eventually captured and embedded in the base of the lithosphere. Each of the volcanic ridges is tapping from a magma source of the long footballs. The schematic velocity curve indicates that the major counterflow in the upper mantle takes place in the Low Velocity Zone (presumably hot and low viscosity). Therefore the maximum shear stress should occur between the Low Velocity Zone and the lithospher e. The lithosphere plays two roles in mantle convections, as the upper limb of the upper convection cell and as the carrying belt for transportation of the intraplate volcanic material s to the trenches for the deep mantle convection cell.

PAGE 125

Shear Stress Subd u ctio n Zone Hot Bubbl e M o d e l 112 Volcanic R idge S u bduc t ion Zone Velocity

PAGE 126

39Ar age data show that, in general, the volcanism becomes progressively older to the east along the ESC, especially along individual ridges (R Duncan, personal communication, 1995). The eastern seamounts have been exposed to longer periods of erosion than the western ones. The intensity values of the side-scan sonar data decrease to the east, while the alteration and sediment thickness increase (7) The following evidence suggests that the volcanism is d e rived from deep, enriched mantle sources Besides the islands there are numerous shallow seamounts over 3000 m height. Previous studies demonstrate that there are strong geochemical anomalies for the rock samples along the ESC and the nearby EPR suggesting a "plume signal nearby (Baker et al 1974 ; Bonatti et al. 1977 ; Poreda et al. 1993a ; Schilling et al 1985a). Descriptions of the Model The geochemical patterns of the Faeroes plateau suggest that the Faeroe Iceland plume was rising in the form of a series of mantle blobs rather than a continuous plume (Schilling and Noe-Nygaard, 1974). The s patial pattern of the rare-Earth elements through the Faeroes plateau basalt monitor the Faeroe-Iceland plume activities with time. An abrupt change from light Re-enriched to depleted patterns is observed near the boundary of the middle and upper series of the Faeroes pl a teau basalt (Sc hilling and Noe Nygaard, 1974) The discontinuity seems to reflect a change of volcanic regime from plume-derived to MORB-derived. The change also coincides with field evidence for beginning of subsidence of the plateau (Schilling and Noe-Nygaard, 1974). Magnetic reversal frequency correlates inversely with mantle plume activity for the past 150 m y (Larson and Olson 1991) Larson and Olson (1991) suggested th a t mantle plumes control magnetic reversal frequency by the following sequence of events. Mantle 113

PAGE 127

pulses rise from the bottom layer of the mantle, which increases core cooling by allowing heat to be conducted more rapidly acro s s the core/mantle boundary Outer core convection activity then increases to restore the abnormal heat loss causing a decrease in magnetic reversal frequency. When core convect i on activity increases above a critical level, a magnetic superchron results (Larson and Olson, 1991) It is inferred that, instead of a steady state plume, blobs of hot mantle may emanate from the mantle diap i rs that dome up on the core/mantle boundary and rise buoyantly up in the form of bubbles However, Larson and Olson (1991) refer to much larger volume blobs leading to super plumes and super swells. Studies of the Hawaiian-Emperor Chain indicate that the chain consists of en ech e lon sets of overlapped ridges (Jackson and Shaw, 1975). Thi s observation was interpreted as a result of the shearing of the relative motion between the lithosphere and the upper mantle (Ihinger 1995) Ihinger (1995) proposes that each of the ridges formed from a "plumelet" that emanate from ascending mantle diapirs that reach the shear zone in the upper mantle The plumelets are sheared just below the lithosphere resulted from the plate motion and its underlying upper mantle (Thinger, 1995; Whitehead, 1982; Whitehead and Luther 1975). His studies suggest that the upper mantle is flowing toward the EPR opposite and about 30 obliqu e to the motion of the overriding Pacifi c pla te. The hot bubble model is a combination of the above models. Th e model s uggests that hot lower mantle rises buoyantly in an unsteady plume in th e form of bu bb les (or blobs) Each bubble will be sheared into a long "football" at a shear zone. Ove rlap between two volcanic ridges can result from two consecutive bubbl e s if the shear i s produced by both plate motion and the counterflow in the upper mantle. Though the convection pattern in the mantle is not well unders too d the existence of lateral motion in the mantle is generally agreed upon ( Ih i nger 1 9 95 ; Morgan 1 9 71 ; Morgan, 1972; Wilson, 1963a; Wilson, 1965 ; Wilson. 1973 ) Because o f this l ateral 114

PAGE 128

motion, it is unlikely for an upwelling plume to rise through the mantle without being sheared. It is also commonly accepted that the lithosphere, moving from mid-ocean ridges toward subduction zones, acts as the upper limb of the cells. The relative motion between the lithosphere and the upper asthenosphere may result in a shear zone unless the upper mantle moves with the lithosphere Decrease of seismic velocity in the LVZ (Low Velocity Zone) may result from partial melting and/or low viscosity If the latter is true, the LVZ is possible a major zone of mantle counterflow toward the nearby spreading center because its low rheology can be easily taken to accommodate the shearing due to the relative motions. In this case, the shearing would be limited between the L VZ and the lithosphere. As a hot bubble rises in the mantle and crosses a shear zone, it will presumably be stretched into a long "football" (elongated ellipsoid) (Figure 4.10). The higher temperature of the bubble (presumably lower viscosity) makes it more vulnerable to the shearing than the surrounding mantle materials. The strong rheology of the lithosphere prevents the football from rising further Instead, it is captured and embedded in the base of the lithosphere, and eventually, becomes the magma source for volcanism The upper portion of the bubble is sheared away from the spreading center and results in robust volcanoes, while the lower portion of the bubbl e may be scattered at depth in a upper mantle counterflow toward the mid-ocean ridge. Near the spreading centers the counterflow in the mantle (the lower limbs of convection cells) are likely to be in the direct opposite direction of the lithosphere motion Therefore, the sheared footballs are likely perpendicular to spreading centers if they are formed near the mid-ocean ridges However, this is not always true in areas far away from the spreading centers. Presumably, the patterns of the counterflows are controlled by the depth of the counterflows and the global distribution of spreading centers and subduction zones. If hot bubbles rise in areas far away from the ridge axes, the footballs may be oblique to the direction of the plate motion 115

PAGE 129

Figure 4.11. Geometric models of hot bubble ridges on the seafloor. A : The mantle is stationary with respect to the plume. The shea..-ing is generated by the plate moving over the mantle The volcanic chain is consisted of end-to-end ridges in a line. B: The upper mantle counterflow is directly opposite the plate motion. Shearing results from the relative motion between the lithosphere and the underlying mantle The individual volcanic ridges generated by this process have larger aspect ratio compared with those in model A. C: The upper mantle counterflow is opposite but oblique to the plate motion as is the shearing direction. The individual ridges, therefore, are oblique to the overall trend of the volcanic chain. D: The upper mantle counterflow is directly opposite the plate motion and the plume migrates southwards; alternatively, the mantle and the lithosphere migrate northwards with respect to the plume Each of the individual ridges trends in the direction of the relative plate motion; however, the overall trend of the chain is oblique to the relative plate motion. E : Schematic diagram of major lineaments in the western ESC. The pattern of the ridges is similar to that produced by model D for the eastern ridges. The change in the trends of the ridges near the western end of the ESC is associated with the formation of the Easter microplate, which changed the local shearing direction.

PAGE 130

117

PAGE 131

Different patterns of volcanic ridges may result from various combinations of the relative motions between the lithosphere, the upper mantle, and the plume source (Figure 4.11). If shearing is in the same direction of the absolute plate motion a volcanic chain composed of end-to-end ridges will be formed in a line from a series of sheared bubbles (Figures 4.11A and llB). (1) If the shearing is produced by the plate moving relative to the stationary upper mantle, the aspect ratio of the individual ridge is small (Figure 4.11A). (2) If the mantle counterflow is close to the surface, for example in the L VZ, the shearing stress should focus between the lithosphere and the L VZ and, therefore, produce individual ridges with larger aspect ratios (Figure 4.11B). If the shearing direction is oblique to the absolute plate motion, en-echelon ridges will form (Figures 4.11C and llD). (3) If the counterflow in the mantle is oblique to the relative plate motion, the overall chain will trend perpendicular to the spreading center with individual ridges oblique to the chain (Figure 4.11C) (4) If lithosphere and mantle are sheared in the same direction as the relative plate motion and there is a relative southward migration of the plume source, the overall volcanic chain will be oblique to the relative plate motion while each individual ridge is parallel to the relative plate motion (Figure 4.11D). Applications of the Model The ridge pattern in the ESC (Figure 4.11E) appears most like the results shown in Figure 4 11D. This pattern suggests that there is a relative northward migration of the Nazca plate with respect to the plume This type of relative migration is also seen in the Foundation Seamount Chain and the Nazca ridge. All the chains are formed from plumes near or on the ridge axes The overall trends of the chains are oblique to the relative plate motion. 118

PAGE 132

Assuming plumes are fixed in the mantle, the northward migration of plates in the direction parallel to the ridge axes can be calculated from the spreading rate and the angles between the volcanic chains and the relative plate motion. The angles between the volcanic trends and the nearby fracture zones are about 8 for Foundation Seamount Chain and about 30 for Nazca ridge (Figure 4.1) Because there is no well mapped long fracture zone near the ESC, the relative plate motion for the Nazca plate (about 095) is estimated from the average of the present ridge direction (EPR about 010) and relic ridge direction (Mendoza rise about OO
PAGE 133

On the other hand, the plumes may not be fl.xed with respect to each other (Norton, 1995). The oblique trends of the volcanic chains may be a partial result of plume migration. Assuming that a rigid plate rotates about a rotation pole with respect to fixed plumes, the volcanic chains should form small circles. In other words, all the great circles perpendicular to the volcanic chains should intersect at or near rotation pole In addition to the ESC trend of 085, there are two other hot spot volcanic chains in the Nazca plate, the Galapagos ridge and Juan Fernandez Seamount Chain, trending 089 and 083, respectively (Figure 4.1 ). Based on these volcanic trends and the location of their western ends, an average rotation pole (85.9 N, 171.4 OE) is calculated for the Nazca plate with respect to the plumes using the Minster and Jordan (1978) algorithm. Using this pole, an absolute plate motion trend can be calculated with respect to a flxed point under the Nazca plate. Assuming that the plumes are fixed in the mantle, the three volcanic trends calculated from this rotation pole at the western ends of the volcanic chains are 085.5 085 9 and 085.2 for Easter, Juan Fernandez, and Galapagos, respectively (Figure 4 .1). Previous Euler poles only flt two of the volcanic trends (Gripp, 1994; Naar and Hey, 1989a). This new Euler pole matches all three of the volcanic trends (Figur e 4.1). The fonnation and motions of the Easter microplate may have changed the orientations of the younger ridges to the western end of the ESC Notice the following three observations (Figure 4.11E): (1) the trends of the westernmost four ridges in the ESC rotate clockwise from 091 at ridge 'd (Salas ridge) to 100 at ridge 'a'; (2) the trends of the two fracture zones also change in a clockwise direction from 098 (the SOEST FZ) to lOT; (3) although the SOEST FZ runs between the Easter and Tupa ridges, the fracture zone trends subparallel to Getu ridge, but the two surrounding ridges trend closer to western extension of the SOEST FZ (Figure 4.4A and Table 4.1). The relationship of the trends between the ridges and the fracture zones may be used to constrain the time frame of the ridge formation because both trends are associated with relative plate motion. This is 120

PAGE 134

possible in the area near the Easter microplate due to the change of the relative motion during the formation and evolution of the microplate which presumably effect the underlying mantle counter flow Assuming that the ridge trends represent shearing directions that are parallel to the relative plate motion, the change of the ridge trends describe the history of the relative plate motion changing from Nazca/Pacific to Nazca/Easter The Salas ridge and the ridges to the east trend in almost same direction, indicating a stable relative motion between the Nazca and Pacific plates during their formation From ridge c to the western ridges, the trend of the ridges suggest influence of the Easter microplate Because the SOEST FZ extends beyond the outer pseudofault, it existed before the initiation of the Easter microplate (5.15. 7 Ma) and lasted until about 2 Ma. The orientation of the bubble underneath ridge c appears to be partially controlled by the structure of the SOEST FZ because they have similar trends. The trends of the Tupa and Easter ridges are in the directions between the trends of the SOEST FZ and its west extension. Thus, we can determine a rough time limits for the shearing period of the bubbles that are responsible for the formation of the ridges The ridges are formed after the bubbles are sheared and therefore, should be younger than the shearing periods However, the volcanism could last for quite long period once it has begun The distance between Easter Island and Salas y Gomez Island is about 350 km. Using the average half spreading rates of the EPR north of the Easter microplate (72 km/m.y.) and south of the Easter microplate (74 5 km/m.y.) (Hey et al., 1995), the estimated age difference of the seafloor under the two islands is about 5 m y . There are two bubbles emerged at the distance of the two islands. Assuming there is no relative migration between the plume and the EPR, the time difference between the islands should equal to the time difference between the two bubbles which gives an estimate of emerging period of about 2 5 m.y. for a bubble, which is compatible with the rate estimated for 121

PAGE 135

Hawaiian Chain (3 m.y .) (Ihinger, 1995). This is within the time range determined by the above comparison between the fracture zones and the ridge trends The total volume of the individual ridges provides a minimum estimate for the bubble size of 20-30 km in diameter (Table 4 1 ). For Hawaiian Chain, the estimated size is about 110 km in diameter (Ihinger, 1995). The true size of the bubbles in both places could be much larger because not all of the mantle bubbles would reach the surface. Implications of the Model The Hot Bubble model is similar to an unsteady state mantle plume. Some implications for the deep Earth structure are discussed in Appendix 4.1. In addition to predictions from the hot spot model, there are several implications from the Hot Bubble model. (1) Recent volcanism is expected in the area extending to the northwest from the Tupa ridge. Based on the overall trend of the ESC, a new volcanic ridge may emerge to the southwest of the Tupa ridge (the proposed center location is indicated in Figure 4.11E), trending WNW from the failed rift near 28.5 S, 111 W. More volcanism is also expected to extend to the west along the Tupa ridge. (2) The age of a volcanic chain should decrease progressively towards the plwne (to the west for the ESC) in general. This prediction is generally supported by the recent radiometric age data along the ESC (R. Duncan personal communication, 1996; Liu et al ., 1996 ; O'Connor et al., 1995; Stoffers and Hekinian 1990) For each individual ridge, the age of volcanism should become systematically younger towards the plume. Thus, the oldest volcanism should locate near the eastern end of each individual ridge in the ESC, while the western end should be younger. Based on the radiometric ages dates referenced above, the oldest volcanism at the eastern end of the Tupa ridge is about 2 5 Ma. Using the 122

PAGE 136

above estimated bubble emerging period (2.5 m ylbubble), the oldest volcanism along the ridges should range from 0-2.5, 2.5-5 0, 5.0-7 5, and 7.5-10 Ma for Tupa, Easter, Getu, and Salas ridges, respectively At the middle of the Easter ridge the oldest volcanism for the Easter Island should be about 3.8 Ma. Similarly, the oldest volcanism for the Salas y Gomez Island should be about 8 8 Ma Note, in testing these predicted ages, one should keep in mind that the age data may appear younger due to young flows overlying older ones, and this method of age prediction is a gross approximation (3) The lateral overlap between two consecutive ridges suggests the existence of mantle counterflow that is partially responsible for the shearing of the hot bubbles. If the shearing is due to the relative motion of the plate over a relatively stationary upper mantle, there should be not overlap between two consecutive ridges unless the plume migrated in the same direction as the plate motion There is about 50 % of lateral overlap between major ridges in the Easter-Salas y Gomez region along the ESC. The overlaps is about 50% along the Ducie-Crough Seamount Chain (Figure 4.3 in Searle et al. (1995)) and >50% along Hawaiian Chain (Figures 4.3 4, and 5 in Jackson and Shaws paper (1975)). The overlap ranges from 0% to 50% along the Pukapuka Chain (Figures 4 2 and 3 in Sandwell et al. (1995)). We propose the amount of overlap is determined by three factors : the shear strains, rising rate of the hot bubbles, and the plate motion rat e with respect to the plume. In addition, overlying lithospheric structur e appears to play a rol e especially near the SOEST FZ. The large overlap between en-echelon ridges and the larg e aspect ratio ind i cate that the strong shearing between the litho s phere and the mantle takes place in a narrow zone, suggesting the shallow d e pth of the mantle counterflow If the mantle counterflow is at great depth, the shearing would be accommodated in a wid e zone, bubbles would not be significantly sheared (4) The existence of mantle counterflow at shallow depth may also be implied by the apparent rate of absolute plate motion even there is no overlap between volcanic ridges 123

PAGE 137

An apparent rate of absolute plate motion is calculated by dividing the length of an individual ridge by the age difference at the two ends of the ridge. If shearing is resulted from the relative motion of a plate over a relatively stationary mantle with respect to a plume, the apparent rate should be consistent with the true rate If the apparent rate is greater than the true rate, it is likely that the shearing is also resulted partially from a counterflow in the mantle because the apparent rate is actually a measurement of the relative motion of the plate motion with respect to the counterflow in the mantle, instead of the plume. CONCLUSIONS The general age progression along the ESC predicted by hot spot model is supported by the new radiometric age data, side-scan intensity, and morphology of the seamounts. Side-scan image and swath bathymetry data collected near the western end of the ESC indicate that the western portion of the chain is composed of a number of volcanic ridges in a dextral en-echelon pattern The trend of the individual ridges is close to the direction of relative plate motion, while the overall trend of the chain is about 10 oblique to the direction of relative plate motion indicating a relative migration between the Nazca plate and the Easter plume parallel to the spreading center. The trends of individual ridges closely follow the reorientations in the relative plate motion Volcano ages increase to the east along the Easter and Salas ridges. Large, robust volcanoes are formed in central or eastern portion of each ridge, while young lava flows are scattered near the western ends of each ridge. The failed propagators weaken the lithosphere and become sites of post volcanism. The orientation of the rifts and the seafloor spreading fabric suggest that lateral extension is oriented in an E-W direction (not in a N-S direction as suggested for the 124

PAGE 138

Pacific plate en-echelon ridges). Fracture zones imaged in this region are not responsible for the formation of the volcanic ridges, therefore the leaky fracture zone hypothesis is not valid here The western edge of the ESC is separated from the EPR by -100 km Several seamounts rise to a height of above 3000 m above seafloor. These observations cannot be explained by the hot line hypothesis. The Hot Bubble hypothesis provides an alternative explanation for the origin of the ESC. The large aspect ratio, large overlap between en echelon ridges, and the close relationships between the trends of the ridges and those of the fracture zones suggest that there is a strong shearing between the lithosphere and the underlying mantle due to their relative motion and that the mantle counterflow is probably at shallow depth. Each of the individual ridges may form from an elongated mantle bubble that is sheared as it rises across the mantle shearing zone. Geophysical and geochemical data show significant differences between the San Felix Island and the ESC, suggesting that they were formed from different sources 125

PAGE 139

CHAPTER 5 EFFECTIVE ELASTIC THICKNESS OF THE LITHOSPHERE ALONG THE EASTER SEAMOUNT CHAIN ABSTRACT Bathymetry and gravity data collected during Legs 5 6, and 7 of the 1993 Gloria Expedition and the recently released 2-minute altimetry-derived global gravity grid are used to determine the effective elastic thickness of the lithosphere along the Easter Seamount Chain (ESC). Forward, admittance, and coherence methods yield consistent results With the exception of the eastern and western ends of the ESC, the effective elastic thickness along the chain is approximately 3 km and fairly uniform. These results are consistent with seafloor ages and recent radiometric dating of seamounts. The elastic thickness southeast of the Nazca fracture zone is approximately 8-9 km, apparently due to the seafloor age discontinuity across the fracture zone. The elastic thickness near the San Felix Island is even greater (12-13 km) which is compatible with the radiometric ages suggesting that the relative age difference between the seafloor and the San Felix Island is greater than that for the ESC seamounts. At the western end of the ESC, approximately 100 km from the East Pacific Rise, the best-fitting elastic plate thickness from forward modeling is slightly higher than along the rest of the chain. However, when a large area is included, admittance modeling estimates the same elastic thickness as the rest of the ESC but with subsurface 126

PAGE 140

loading and non-isostatic compensation. This result may reflect a dynamic component in the compensation of recent loads on the young seafloor in this area. INTRODUCTION The Easter Seamount Chain (ESC) is a major bathymetric feature in the southeastern Pacific (Figure 5.1) The chain extends approximately 3000 km across the Nazca plate The Easter microplate and the fastest shallowest spreading segment of the East Pacific Rise are located 100 km west of the western end of the ESC (DeMets et al 1990; Hey et al 1995; Naar and Hey, 1989b) The eastern end of the ESC joins the southwestern end of the Nazca ridge on the southeastern side of the Nazca fracture zone near the San Felix Island (Liu et al., 1996 ; Pilger and Handschumacher, 1981). A simple hot spot model was initially proposed to explain the origin of this chain (Morgan, 1971; Morgan, 1972; Wilson, 1963a ; Wilson 1963b; Wilson, 1965; Wilson, 1973) Alternative hypotheses have been advanced over the last several decades to explain the apparent contemporaneous volcanism along the ESC (based on K-Ar age data). These hypotheses include a "hot line" model, in which the ESC formed along a thermal anomaly corresponding to an upwelling limb of a secondary convection roll (Bonatti and Harrison, 1976; Bonatti et al 1977) ; a "leaky fracture zone" model (Clark and Dymond 1977; Herron, 1972a; Herron, 1972b); and a "diffuse extension" model in which the ESC originated along tension cracks (Sandwell et al., 1995 ; Winterer and Sandwell, 1987) More recently a Hot Bubble model has been proposed to explain the en-echelon volcanic ridges (Figure 5.1C) near the western end of the ESC (Liu and Naar, 1996b ; Liu et al., 1995) The model suggests that an unsteady plume may ris e in the form of a series of mantle bubbles which are sheared, while rising, into football-shaped bodies The western 127

PAGE 141

Figure 5.1. Tectonic location and data. A: Tectonic location of the ESC. The ship tracks are from Legs 5, 6, and 7 of the Gloria Expedition in 1993. Seamounts are indicated approximately by light stippled areas although some volcanic flows extend to north or south of the stippled areas. The box shows the boundary of panel B. B: Bathymetry predicted (Liu et al 1996; Smith and Sandwell, 1994) from ETOP0-5 data (National Geophysical Data Center, 1988) and gravity anomalies calculated from newly declassified altimetric data collected with Geosat/ERM, Geosat/GM, ERS-1, and ERS-1/GM (Smith and Sandwell, 1995a; Smith and Sandwell, 1995b) The Te along the ESC is modeled using both forward and admittance methods for the nine areas indicated by white line boxes. Triangles are the locations of the islands along the ESC. The large black line box shows the area for panels C and D. C: Combination of side-scan image and bathym etry of the Easter-Salas y Gomez region showing two of the major volcanic ridges in light gray (black lines) referred to as Easter and Salas ridges (Liu and Naar, 1996a; Liu et al., 1995) Bathymetry data are scaled to the range of and merged with the side-scan image by selecting a larger value for each pixel from either the scaled bathymetry or side-scan image data collected (Hey et al., 1995; Liu and Naar, 1996a ; Naar et al., 1993). The five white line boxes show areas where Te is modeled along two major volcanic ridges Te is als o modeled along the five profiles indicated by white lines within the boxes. D : Free-air gravity anomaly in Easter-Salas y Gomez region calculated from the 2-minute global gravity grid (Smith and Sandwell 1995a ; Smith and Sandwell, 1995b). Te is modeled in areas A, B (black line boxes), and C (white line box). White diagonal line marks the locations of the profiles shown in Figure 5.3.

PAGE 142

13ow 12ow Pacific Plate intensity 2.50 225 200 175 150 125 100 75 50 25 mGal 120 103 86 69 52 35 18 1 -16 -33 11 Ioo w Nazca Plate 2o s 3os 4o s 2os 29 s 112w 111 w 11 o w 109w 1 w wrw I06 w 1 os w 104 w 103 w 129

PAGE 143

ESC volcanic ridges may be tapping such elongated mantle sources at the base of the lithosphere (Liu and Naar, 1996b; Liu et al., 1995). In this paper we use new bathymetry and gravity data and both forward and spectral methods to determine the effective elastic thickness of the lithosphere CTe) supporting the ESC and the San Felix Island. The integrated lithospheric strength can be expressed in terms of a flexural rigidity, which can in tum be related to thickness of an elastic plate with flexural properties that approximate those of the lithosphere. Because this effective elastic plate thickness reflects the strength of the lithosphere at the time the seamounts form, it can be used as an indicator of thermal structure or age of the lithosphere at that time. Thus we can use estimates of Te to test the predictions of the various models for the formation of the ESC and San Felix Island. Both shipboard geophysical data (Hey et al., 1995; Naar et al., 1993a ; Naar et al 1993b) and the new altimetry-based 2-minute global gravity grid (Smith and Sandwell, 1995a; Smith and Sandwell, 1995b) are independently used to estimate Te. Shipboard data include swath bathymetry and gravity collected along the ESC during Legs 5, 6, and 7 of the 1993 Gloria Expedition (Hey et al. 1995; Naar et al., 1993b). In this paper gravity data collected along the shiptracks of the Gloria Expedition are referred to as shipboard gravity, while the new global gravity grid is referred to as altimetry gravity. We use a high resolution grid of bathymetry data (node spacing of-300m) which has been compiled along the ESC (Liu and Naar, 1996d; Liu et al., 1994). The data include the bathymetry swaths collected with both GLORI-B and SeaBeam 2000 systems with swath widths of -24 and -10 km, respectively. Outside the coverage of the bathymetry swaths the bathymetry grid is filled with the ETOP0-5 data set (National Geophysical Data Center, 1988) Near the Easter and Salas y Gomez i s lands (the large black line box in Figure 5.1B) bathymetry values along the -6-km gaps between the GLORI-B swaths are interpolated from both GLORI-B and SeaBeam swaths (Figure 5 2A). 130

PAGE 144

In this study we estimate Te along the ESC and near the San Felix Island, and in greater detail in the Easter-Salas y Gomez region Along the ESC and near the San Felix Island the best-fitting Te value is found for nine distinct areas (white line boxes in Figure 5.1B). In each area both forward and admittance methods are used with altimetry gravity and bathymetry grids. In the Easter Salas y Gomez region, we estimate Te in five areas (white line boxes in Figure 5.1C) along two of the major distinct volcanic ridges referred as Easter and Salas ridges (ridge axes are indicated by the black lines in Figure 5.1C) (Liu and Naar, 1996b; Liu et al., 1995). For each of the five areas both forward and admittance methods are used with the altimetry gravity and bathymetry grids and with shipboard data along track lines (white lines in Figure 5 1C). In this region the data resolution also permits comparison of admittance and coherence methods (white line box C in Figure 5 .10). Admittance is also calculated for areas A, B, and C (Figure 5.10) to compare and determine if there are subsurface loads and/or non-isostatic compensation of the young seamounts near the western end of the ESC. METHODS The methods used in deriving the best-fitting elastic plate thickness values are discussed briefly below These methods are described in greater detail in the Appendix. The Forward Method The forward model assumes that the topographic loads on th e seafl oo r are isostatically compensated at the Moho by flexure of a lithosphere with a uniform effective 131

PAGE 145

Figure 5.2 Easter-Salas y Gomez region A : Bathymetry data compiled from swath bathymetry data collected during the Gloria Expedition using both GLORIB and SeaBeam systems (Liu and Naar, 1996c) Outside of the survey region is filled with gridded data (to the west of 109"W from F. Martinez, personal communication, 1995) and the ETOP0-5 data (to the east of 109 W) B : Modeled free-air gravity anomaly from bathymetry shown in panel A using forward method with the default parameters (Table 1). C: Bouguer anomaly modeled from the bathymetry data using the Parker's model (1972), showing gravity lows beneath topographic loads

PAGE 146

Jo6 w 105 w 104 w 103 w 25 s 2 6 s 21 s 28 s 29 s -5000 4450 3900 3350 -2800 -2 250 1700 -1150 600 -50 25 s 26 s 2 1 s 28 s 29 s -50 -27 4 1 9 42 65 88 Ill 134 1 57 180 25 s 26 s 27'S 28 s 29 s 60 50 4 0 -30 -20 -10 0 1 0 2 0 30 40 133

PAGE 147

elastic thickness. Model free-air gravity anomalies are calculated from bathymetry data and the computed Moho deflections (Equation 8 in Appendix 5 1) For example, Figure 5.2B shows the model gravity calculated from bathymetry in Figure 5.2A. It i s clear from the Bouguer gravity anomalies (Figure 5.2C) that seamounts are supported by low density roots. The model gravity anomalies are compared to the data for a range of values of Te and crustal density. Other parameters (which have a much smaller effect on the model gravity) were held fixed at the values listed in Table 5.1. The best-fitting Te value for a given region is defmed as that for which the root mean square (RMS) difference between model gravity and data is a minimum, as in (Wolfe and McNutt, 1991) Table 5 1 Default parameters Quantity Symbol Value Unit Poisson's ratio v 0.25 Young's modulus E 7 0 X 1010 kgls 2 m Newton's gravitational constant G 6.67 X 1Q-ll m3/s2kg acceleration of gravity g 9.8 rn!s 2 crustal-water density contrast i\po 1600 kglm3 mantle-crustal density contrast i\pl 550 kglm3 density of the seafloor Pw 1030 kglm3 depth of the seafloor Zc 3000 m depth of the Moho Zm 10000 m effective elastic thickness of the lithosphere T e 3000 m weighted subsurface/surface load ratio f 0 compensation percentage of surface load c 100% compensation percentage of subsurface load c' 100 % ---------------------------------------------------------------------------------------------RMS differences between model gravity and data are computed both for shiptrack gravity profiles and for 2-D altimetry gr a vity grids When modeling shiptrack profiles, model gravity is computed using the 2-D bathymetry grid described above and interpolated to the shiptrack Figure 5.3A shows an example of observed and model gravity across the Easter and Salas ridges, along the shiptrack indicated by the white line in Figure 5.1D. We note that the model gravity becomes coherent with shipboard gravity at wavelengths greater 134

PAGE 148

Figur e 5.3 Comparison of obs e rved and modeled gravity along the profile i ndicated by th e white line in Figure 5 1D A : Gravity Shipboard gra vity is from the Gloria Expedition. Altimetry gravity is interpolated along the shiptrack Synthetic gravity is mod e l e d from 2-D bathymetry data assuming T e = 3 0 km. B : Bathymetry from SeaBeam 2000 system on the Gloria Expedition

PAGE 149

40 20 -20 40 shipboard gravity altimetry gravity model gravity (Te = 3 km ) ,-, ; f' 50 100 A: Free-air Anomaly !l B: Bathymetry 150 200 250 300 350 400 4 3 2 0 -I -2 -3 -350 -300 -250 -200 -150 -I 00 -50 0 50 100 150 200 250 300 350 400 di s tance km

PAGE 150

than about 10 km. The altimetry gravity is coherent with shipboard (and model) gravity only at wavelengths greater than approximately 50 km, as expected from the resolution of altimetry gravity (Smith and Sandwell, 1995a; Smith and Sandwell, 1995b). The Admittance Method We compute admittance (Equation 42 in Appendix 5 1) as the ratio of the Fourier transform of the free-air gravity anomaly to the Fourier transform of bathymetry. Free-air admittance is convenient to compute because the uncertainties in crustal densities associated with the Bouguer correction can be avoided. In this study we explicitly include the ratio of subsurface to surface loads (f) as a variable in the admittance model. We also consider the possibility that bathymetric and subsurface loads may not be fully isostatically compensated by including in the model parameters ( c and c') that specify the percentage of compensation of each type of load. Figure 5.4 shows predicted admittance curves as a function of the model parameters. The location and amplitude of the central peak in the admittance curves is particularly sensitive to both T e and the ratio of subsurface to surface loads (f) (Figures 5.4A and 4B). On a stiffer plate (higher Te), the wavelength of Moho deflections compensating bathymetric loads increases, producing a longer wavelength gravity signal (Figure 5.4A). If the subsurface/surface load ratio f is high, most of the loads are found near the depth at which they are compensated, and the positive gravity signal of the loads and the negative signal of their compensation have similar spectra and tend to cancel (Figure 5.4B ) The long wavelength portion of the admittance curve is quite sensitive to the percentage of compensation of the surface load (c) if there are no subsurface loads. This sensitivity decreases as the subsurface/surface load ratio f increases. If significant 137

PAGE 151

Figure 5.4. Admittance curves as a function of various parameters. The default values for parameters not listed with each graph are shown in Table 1. In panel D, f = 1 is used because c' is associated with subsurface loads.

PAGE 152

A r. = 8 4 2. o 1cm E 6p0 = 1.8, 1.7, 1.6 1.5 g/cm3 f=O T=3km f = 0 E o .040 E 0 (.) c ..... <;::::: E "'0 0 .00 B f=O, 1,2,4 F 6p1 = 0 4 0.5. 0 .6 0 7 g/cm3 T = 31cm T = 31cm f= 0 E ...._ "@ o .040 E 0 (.) c ..... ..... a "'0 0.00 c c = 100,75, 50 25 % zc = 2, 3 4 5 lcm T = 31cm T =3km f = 0 f = 0 E ...._ "@ o.040 E 0 (.) c ..... ..... a "'0 0 .00 D c' = 1 00 75. 50,25% H Zm= 1 2, I I 10,9km T=3km T =3km f = I f = 0 E ;::::; o.040 E 0 (.) c ..... ..... a "'0 0 00 1 03 1 02 1 01 102 1 01 wavelengt h km w a ve len g th km 139

PAGE 153

subsurface loads are present, long wavelength portion of the admittance signal is also sensitive to the percentage of the compensation of subsurface loads (c') (Figure 5.40). Variations in other model parameters have smaller effects on the admittance (Figures 5.4E-4H). Theoretical admittance patterns are compared with observed admittance to determine the best estimate forTe. We compute admittance for 2-D grids using altimetry gravity and the bathymetry grids described above, and also directly along shiptrack profiles using shipboard gravity and bathymetry Grid size and profile lengths are set through a compromise of (a) sampling long enough wavelengths to capture the peak in the admittance curve and (b) restricting the study area to a limited tectonic province. In a large region admittance may place primary emphasis on the areas with the greatest topographic relief (Forsyth, 1985). The Coherence Method The advantage of the coherence method is that it is less sensitive to subsurface loads (Forsyth, 1985) and thus provides a reasonable estimate of elastic thickness without high certainty of the amount of subsurface loads. The coherence model assumes there are subsurface loads and that surface and subsurface loads are statistically independent, however, in many cases, they are likely to be tectonically related processed and, therefore, spatially correlated (Forsyth, 1985) Our coherence mod e ls follow the m e thod of Forsyth (1985), with the inclusion of possible non-isostatic compensation (Equation 49 in Appendix 5.1) Coherence modeling in this study is limited to regions near the Easter and Salas ridges where high resolution bathymetry is available over a region large enough to capture wavelengths of hundreds of kilometers. 140

PAGE 154

RESULTS Along the ESC and near San Felix Island Te is estimated for nine distinct areas (Figure 5.1B) that span the ESC using altimetry gravity and bathymetry grids. The western eight (areas 1-8) lie along the ESC, while the easternmost (area 9) is near San Felix Island. Area 7 is near or on the Nazca fracture, while area 8 is across the fracture zone on the same side with San Felix Island (Figures 5.1A and 1B). The results of forward modeling are shown in Figure 5 5. Panels 1 through 9 correspond to the nine areas with 1 representing the westernmost and 9 the easternmost region. The figure shows contours of the RMS difference between the model and altimetry gravity grid in each area, for a range of values of crustal density and elastic plate thickness. Lighter gray indicates smal ler RMS values. The patterns of RMS contours show that best fitting Te values are generally fairly well con s trained despite uncertainties in crustal densities. Assuming a crustal density of 2650 kg/m3, the estimated values forTe are 2.6, 2.4, 2.9, 3.4, 2.8, 4.0, 2.5, 8.1, and 12.6 km, for the nine areas, respectively. Visual inspection of forward modeling profiles suggests that the uncertainty in the estimation of Te is about km. Models with T e values that differ from the be s t-fitting value by more than 1 km produce noticeably inferior fits. Admittance modeling of the same regions gives consistent results, shown in Figure 5.6. The difference between the best-fitting T e values derived from the forward and admittance methods are less than 1.5 km, with the exception of region 3. The origin of th e region 3 discrepancy is unclear. Figure 5.6 also indicates that models with no subsurface loads (f = 0) yield better fits to the data than models with significant subsurface loads (f >= 1). (The low admittance values observed at short wavelengths reflect the limitation of 141

PAGE 155

Figure 5.5. Forward modeling of Te along the ESC. Panels 1 through 9 correspond to the nine areas, from west (1) to east (9) in Figure 5 .1B. The value of each pixel in the figure is a RMS difference between altimetry and model gravity grids of the corresponding area. Model gravity is computed for a range of values of Te and crustal density. RMS differences range from 5 to 24 mGal The contour interval is 0 2 mGal Calculations were done varying T e and crustal density by intervals of 0.2 km and 10 kgtm3, respectively Red color indicates lower RMS values, e g better fit Te and density values.

PAGE 156

00 r-\0 C'i C'i C'i 00 C'i r--; \0 N C'i 00 C'i 00 C"l !SU;}p r-\0 C'i C'i r--; \0 N C'i !SU;}p 143 00 r-\0 C'i C'i C'i 00 C'i r--; \0 N C'i

PAGE 157

Figure 5.6. Admittance modeling of Te along the ESC, for the same nine areas as in Figure 5 .5. Circles are the admittance data calculated from the gridded bathymetry and altimetry gravity. Curves are the model admittance for the T e indicated in each panel. Solid lines are calculated with f = 0; dashed line with f = 1. Other parameters are given in Table 1. Uncertainties for some of the admittance data lie within the dimensions of the symbols on the plots.

PAGE 158

0 08 Te =4.0 km Te=3. 5 km 0 07 2 Te= 7.5 km 3 0 08 0 07 0 06 0 06 0 0 05 0 05 E -----0 04 / 0 04 II) f=O ,_--I u / I c:: 0.03 / f= I I 0.03 j I I I 0.02 I I 0 02 / I / I "' 0 .01 -----/ 0 .01 ---------+ + + 0 00 0 00 0 08 5 6 0 08 Te = 3.5 km 4 Te= 2.5 km Te =4. 5 km 0 07 0 07 E 0 06 0 06 -... "@ 0 05 0 05 0 E II) 0 04 ,_--0 04 u ,_-/ c:: 0 03 I I 0.03 I I I / I 0 02 / I / I 0.02 / I / ... "0 -/ "' t,.. __ ..... ... 0 .01 --'-"'------0 .01 + 0 00 0 .00 0 08 7 8 9 0 08 Te= 1 5 km Tc=9.0km e=l2km 0.07 0.07 0 06 0 06 0 0.05 ----0 05 E / II) 0.04 I 0 04 I u I c:: "' 0 03 I 0 03 I 0 .02 I 0.02 I "0 I "' 0 .01 I 0.01 / ---0 00 0 00 J02 JOI J02 JQI J02 JOI wavel e n gth km wave l e n gt h km wave l engt h km

PAGE 159

resolution of the altimetry gravity at these wavelengths .) Similar results have been reached using both forward and admittance methods with shipboard gravity (Liu et al., 1995). In summary, the Te estimates along the ESC to the west of the Nazca fracture zone from all the methods provide a fairly consistent value of about 3 km It appears that the strength of the lithosphere at the time the ESC volcanic loads formed was roughly comparable along most of the chain. This result is compatible with radiometric age of seamounts and magnetic isochron age of seafloor and is discussed further below. The relatively low Te value estimated in area 7 (-2 km) may be due the weakness of the lithosphere near or on the Nazca fracture zone. The higher T e values for the easternmost two areas (8 and 9) suggest loads formed on stronger, presumably older seafloor in this region. The -9-km T e estimate in area 8 supports thi s conclusion, as the Nazca fracture zone separates this older seafloor from the younger seafloor to the west along the ESC. Area 9 lies on the older side of the Nazca fracture zone, but the presence of very recent volcanism on and near San Felix Island suggests the age disparity b et ween seafloor and volcanic loads may be much greater here than on the ESC. The high Te values we find in area 9 around San Felix Island support this conclusion and corroborate that San Felix Island is not part of the ESC. In the Easter-Salas y Gomez Region Near the western end of the ESC the bathymetric signal i s dominated by two major volcanic ridges, referred to as the Easter and Salas ridges (black lines in Figure 5.1C) With nearly complete coverage of swath bathymetry data in this region ( Figure 5.2A ) we can estimate both a regional average and local variations in the effective elastic plate thickness, and more accurately assess admittance and coherence at long wavelengths 146

PAGE 160

Figure 5.7. Comparison of admittance and coherence models Admittance (circles in two upper panels) and coherence (circles in lower two panels) are calculated for area C (while line box in Figure 5 1D) using altimetry gravity and bathymetry grids Uncertainties for the data lie within the dimensions of the symbols on the plots Upper left: admittance as a function of Te with fixed f (0.2); Upper right: admittance as a function off with fixed Te (3 km); Lower left: coherence as a function of Te with fixed f (0.2); and Lower right : coherence as a function off with fixed Te (3 km). Forsyth's (1985) coherence model (the coherence between bathymetry and Bouguer anomaly) is used (Appendix Equation 53).

PAGE 161

0.06 f= .2 admittance for area C T e = 3 km admittance for area C 0 0 .0 5 0 0 .04 0 0 .03 0 0 .02 0 0 .01 0. 0 .00 0. 1.0 coh ere n ce f o r area C coh e renc e for area C I. 0 .8 0. 0.6 0.1 2 0.4 0. 0 .2 o.: 0 0 f = 2 T e = 3 km 0 ( J0 2 JOI J0 2 JOI wavelength km wavelength km

PAGE 162

Figure 5.7 shows the results of admittance and coherence computed in a large region spanning the westernmost ESC (white line box C in Figure 5 1D) Coherence is calculated using Bouguer anomalies with an assumed crustal density of 2650 kglm3 and following the method of Forsyth (1985) (Equation 53 in Appendix 5 .1). Both admittance and coherence methods give a best-fitting Te value of about 3 km, similar to that found along most of the ESC. The coherence plots show the transition from high to low coherence takes place over a broader range of wavelengths than does the model for any specific elastic plate thickness (Figure 5.7 lower left panel). The observed pattern probably arises from averaging over several tectonic provinces with different flexural rigidities. As mentioned earlier coherence method is much less sensitive than admittance to th e effects of subsurface loading (Figure 5.7 right side), and indicates that, even with uncertainties in f, the estimate of an elastic plate thickness of 3 km is accurate to within about one kilometer. The admittance results (upper plots), in turn, suggest that subsurface loads are considerably less than surface loads (f < -0.5). The consistent results from the coherence and admittance methods imply the no subsurface load (f = 0) assumption used elsewhere in this study is reasonable The origin of the Easter and Salas ridges has been proposed to be a result of rising hot mantle blobs or bubbles undergoing shear while rising through mantle and providing an elongated football shaped source (Liu and Naar, 1996b; Liu et al., 1995) This Hot Bubble model predicts that the seamounts in the western portion of each of the ridge are younger than those in the eastern portion, which is confirmed by recent 40 Ar-3 9 Ar age data (Liu et al. 1996). By examining smaller regions covering parts of these ridges, we can test whether local variations in Te reflect the proposed age progression. We estimate Te for five areas, shown as boxes a through e in Figure 5.1 C. Figure 5.8 shows results from forward modeling. Figures 5.8A through 8E show the RMS differences between the altimetry and model gravity for the areas a through e in Figure 5.1C, respectively. Figures 5 8A' through 8E' show the RMS differences between 149

PAGE 163

Figure 5.8. Forward modeling of Te along the two volcanic ridges (Figure 5.1 C) in Easter-Salas y Gomez region A through E and A' through E' correspond to areas a through e in Figure 5 1C As in the Figure 5.5, panels A through E show the RMS differences between altimetry gravity and model gravity grids for the five areas Panels A' through E' show RMS differences between model and shipboard gravity profiles within the corresponding boxes Other parameters as in Figure 5.5

PAGE 164

N !'-> N !'-> i..n "' 00 density(g/cm 3 ) N i.e N !V N !'-> N i..n "' 00 i.e !'-> !'-> !'-> N VI 0\ -...J Oo density(g/ c m3) den s ity(g/cm 3 ) !'-> !'-> !'-> N VI 0\ -...J Oo density(glcm 3 ) !'-> N !'-> !'-> VI 0, -...J 00 density(g/cm 3 ) N i.e

PAGE 165

shipboard and model gravity along shiptracks across each of the five areas (diagonal lines in Figure 5.1C). Assuming that the crustal density is 2650 kg/m3, the best estimations of Te from the altimetry gravity fit are 5.6, 2.8, 2 .6 1.6, and 2.2 km, and from shipboard gravity fit are >10, 3 6, 1.8, 1.4, and 2.2 km The average of the best fitting Te values is about -3 k.m, with consistently higher T e values found in the western regions a and b. Very similar patterns of Te values result from admittance modeling of the same data (Figure 5.9). (The low Te values found in this region are also consistent with previous modeling of satellite altimetry (Calmant, 1987).) If the Easter and Salas ridge s increase in age from west to east as does the seafloor, we would expect the age of the seafloor at the time of loading (and hence the elastic thickness) to vary relatively little along the ridges This is observed in areas c, d, and e The higher Te estimates in areas a and bon the western part of Easter ridge could reflect a greater seafloor age at time of loading Howeve r the age disparity between the -0 Ma Ahu volcanic field near the western end of the Easter ridge and the -2 Ma underlying seafloor is small (Liu et al., 1996 ; O'Connor et al., 1995). It does not seem probable that the seafloor age at time of loading on the western end of the chain is significantly greater than that of the eastern portion of the ridges An alternative explanation for the higher apparent elastic thickness is that the young western end of the Easter ridge is in part dynamically supported, and thus not completely isostatically compensated. With dynamic support, the local low-density root of the ridge would be smaller. A greater elastic plate thickness supporting the ridge would similarly produce a lower amplitude local root. Thus incomplete compensation may appear as an overestimate of plate thickness in gravity modeling. To test this hypothesis we calculate admittance, assuming both complete and incomplete compensation, over two regions of intermediate size (boxes A Bin Figure 5 .1D). Area A covers the western end of the Easter ridge; area B the eastern ends of the ridges These larger grids are needed compute 152

PAGE 166

Figure 5.9 Admittance modeling of T e for the same five areas as in Figure 5 8 Following the convention in Figure 5 6, circles are the admittance data while curves are the admittance models. Uncertainties for some of the admittance data lie within the dimensions of the symbols on the plots. Admittance data for panel A through E are calculated from altimetry gravity and bathymetry grids for areas a through e in Figure 5 1C The admittance data for panels A' through E' are calculated from shipboard gravity and bathymetry profiles along the five white lines within the areas a through e in Figure 5.1C.

PAGE 167

0.05 __.... / f""" E f=O !-/-----I I + I I I I E If= 1 I + 0.03 'rr I I I I u I c I 0.02 / I _.., --- I / -------+ 0.01 "0 "' 0.00 Te = 3.5 (km) B 1 Te = 3.5 (km) B'l 0.06 0.05 __.... E -. --/' ... r I I E I I 0.03 'rr I I I I u / c / I 0.02 I -/ ----0 .01 "0 "' 0 00 Te = 2.0 (km) C 1 T e = 2.0 (km) C'f 0.06 0.05 __.... E -- y 0 ; /--/-..... E I 0.03 'rr / / I ..... -I u .... / c ................ / 0 .0 2 --- -.... -+ 0.01 "0 "' 0 00 Te = 2.5 (km) D 1 Te = 2.5 (km) D f 0 06 0 .05 __.... /-E I + / 0 03 'rr I / I u .... + c .... --I 0.02 ........ y .,_ - 0.01 "0 "' .........__ I I 0.00 Te = 2.5 (km) E f j Te = 2.5 (km) E 0 06 + 0.05 __.... r 1 E -t /'!" /,. ......... I E I I 0.03 'rr I I / I u ...... ..... .... I c _../ 0 02 !S y ..... ....-...... .,_ :;..-'---" 0.01 "0 "' 0.00 1 (12 1 (II 102 101

PAGE 168

admittance with sufficient accuracy at long wavelengths to distinguish compensation mechanisms The resulting admittance patterns are compared with models in Figure 5.10. For clarity in distinguishing models both Bouguer anomaly (Equation 43 in Appendix 5.1) and free-air anomaly admittance are shown. Panels A and B correspond to areas A and Bin Figure 5.1D The uncertainties in the admittance data lie within the dimensions of the dots on the plots. We find that both areas A and Bare best fit by models that assume Te = 3 km, that some subsurface loads (f=0.4 ), and that a small component of bathymetric load that is not compensated (15% in area A and 10% in area B) The similarity between areas A and B suggests that incomplete isostatic compensation alone is not a full explanation of the slightly higher elastic thickness observed in area A. Revisiting the larger area C which encompasses most of A and B we find no evidence for incomplete isostatic compensation Area C includes Easter Island, the largest edifice in the ESC. In this region a large range of seamount ages near the western end of the ESC (Liu et al. 1996) indicate the co-existence of young and old seamounts, which can be explained by lateral over-lapping of the enechelon volcanic ridges in the Hot Bubble model (Liu and Naar, 1996b; Liu et al 1995) It appears that some young seamounts and flows may not be fully compensated in this region (area C); however in the admittance modeling the signal from the larger, mature, well compensated features is more heavily weighted. Dynamic support of young volcanism near the western end of the ESC may come from distributed mantle upwelling associated with the nearby Easter hot spot (Haase and Devey, 1996; Haase et al., 1996; Hagen et al., 1990; Hey et al., 1995; Liu and Naar, 1996b; Liu et al., 1995; Liu et al., 1996; Lonsdale, 1989; O'Connor et al., 1995; Okal and Cazenave, 1985; Pilger and Handschumacher, 1981; Schilling et al 1985a; Schilling et al., 1985b; Searle et al., 1995; Stoffers et al., 1994). 155

PAGE 169

Figure 5.10 Estimating subsurface load and compensation. Circles are admittance calculated using gridded bathymetry and altimetry gravity data in the three large areas (A B, and C in Figure 5.1D) Uncertainties for the data lie within the dimensions of the symbols on the plots Curves are the best-fitting admittance models. Panels A through C and A' through C' correspond areas A through C in Figure 5 1D. The admittance show in panels A, B, and Cis calculated as a ratio of the Fourier transform of Bouguer anomaly to the Fourier transform of bathymetry The admittance show in panels A', B', and C' is calculated as a ratio of the Fourier transform of free-air anomaly to the Fourier transform of bathymetry The best fitting values for f and c indicated in each of the panels are used for the models. Other parameters are given in Table 1.

PAGE 170

Bouguer Anomaly/Bathymetry Free-air Anomaly/Bath ymetry -0.08 f= .4 A f = .4 A' 0.06 -0.07 E -0.06 0.05 E ;::::, -"' -.;:; 0 0.05 E g -0.04 u c=85% 0.03 -0.03 '0 0.02-g "' -0.02 -0.0 1 O.Ql 0.00 0.00 -0 08 f= 3 B f = .3 B 0.06 -0.07 E -0.06 0.05 E ;::::, -"' 0-0.05 E -.;:; g -0.04 u 0.03 0.03 '0 0 02 -g "' -0.02 -0.01 O.Ql 0.00 0 .00 -0.08 f= 1 c f= .1 C' 0.06 0.07 E 0 06 0.05 E ;::::, ;::::, "' "' 0.040 0 0.05 E E u g -0.04 c = 100% 0.03 -0.03 '0 0.02 -g "' 0 02 0.01 0.01 0.00 0 00 102 101 102 1 01 wavelength ian wavelength km 15 7

PAGE 171

DISCUSSION The consistently low values of Te (<5 km) found along the ESC are compatible with values obtained in previous studies (Calmant 1987; Liu et al., 1995; Woods et al., 1993) in this region. These elastic thicknesses are slightly smaller than values of 4-8 km found on young seafloor (Blackman and Forsyth, 1991). However, lithospheric reheating at the time of seamount or ridge emplacement may reduce the flexural rigidity of the lithosphere. Figure 5.11 shows the relationship between elastic plate thickness isotherms, and seafloor age at time of loading for the ESC and for other seamounts formed on young lithosphere (Table 5 .2). Most elastic plate thickness estimates for seafloor loaded by seamounts lie in the depth range of the 200 to 500 C isotherms of the cooling lithosphere (Wessel, 1992). Thus, the ESC lithospheric strength falls within the broad range of values observed near other seamounts Table 5.2. Elastic thickness estimated along the ESC Age (Ma) Te (km) 4 4 4 5 1 4 1 8 30 2.6.0 2.4.0 2 .9.0 3 4.0 2.8.0 4.0.0 2.5.0 8.1.0 12.6.0 Area Tectonic Site 1 2 3 4 5 6 7 8 9 Easter I. on pseudofault Salas y Gomez I. on normal seafloor seamount on normal seafloor seamount on normal seafloor seamount near Mendoza rift seamount on normal seafloor seamount on Nazca FZ seamount on older seafloor south of Nazca FZ San Felix I. on older seafloor south of Nazca FZ ESC seafloor ages at time of loading used in Figure 5 .11 are computed by subtracting the age of seamounts from seafloor age. Seafloor ages are taken from (Liu et al., 1996). Ages for bathymetric loads are derived by interpolating between radiometric ages (Liu et al 1996; O'Connor et al., 1995) calculated from the dredge samples collected 158

PAGE 172

Figure 5.11. Te vs age of seafloor at time of loading. Circles are data from this study. Crosses are data compiled by Wessel (1992) The horizontal error bar indicates uncertainty in seafloor age at the time of seamount formation Uncertainty of estimated e lastic thi c kn ess is about km based on visual inspection of forward modeling profiles. Curves are isotherms for the cooling plate model (Parson and Sclater, 1977)

PAGE 173

Easter Seamount Chain --+ + + 5 + :: 10 + 2 ., ., ) 2 Eastern End of the ESC ) ; 20 2ooc 0 5 I 0 15 20 25 30 35 40 sea tl oo r ag e at lime of l oa ding m y

PAGE 174

along the ESC (Haase and Devey, 1996; Naar et al., 1993b; Poreda et al., 1993a; Stoffers et al., 1994). These data show a systematic increase in seamount age towards the east along the chain (Liu et al., 1996; O'Connor et al., 1995). Combined with reconstructions of the tectonic history of southeastern Pacific near the ESC, we find most seamounts along the ESC formed on seafloor of less than -7 Ma (Liu et al., 1996). These results are compatible with the fairly uniform plate thicknesses found in this study. Together, these observations favor the model in which the ESC is produced by a single hot spot. The complex history of seafloor under the ESC may explain the observation that Te values for the region lie near the low end of the expected range (young age points on Figure 5.11 ). This area includes a pseudofault associated with the East rift near area 1 (Figure 5.1B) (Hey et al., 1985; Naar and Hey, 1991; Rushy, 1992; Searle et al 1995), the failed Mendoza rift near area 5, and the Nazca fracture near area 7 (Figures 5.1B, 5, and 6) (Herron, 1972a; Liu et al., 1996; Lonsdale, 1989; Okal and Cazenave, 1985). The uniform Te values estimated along the ESC cannot be explained by the "hot line", "leaky fracture zone", and "diffuse extension" models. These models predict simultaneous volcanism along the ESC on the seafloor of different ages, and therefore predict larger T e values on the eastern portion of the chain. The leaky fracture zone and diffuse extension models both predict lithospheric weakening near the ESC but cannot explain the uniform flexural strength we fmd along the ESC. In our modeling we assume a crustal thickness of 6 km, compatible with the results of a study using the group velocity of fundamental mode Rayleigh waves (Woods and Okal, 1994). The high Te value found near San Felix Island (area 9, Figure 5 1B) indicates the San Felix Island is very young, and hence is not part of the ESC. This result is consistent with other geophysical and geochemical observations (Liu et al., 1996) These include a very young radiometric age ( -0.8 Ma) (Liu et al., 1996) and volcanic activities were 161

PAGE 175

reported in 1922 (Firth, 1943). There is also a gap of volcanoes between the ESC and the San Felix Island (Liu et al., 1996; Naar et al., 1993a; Naar et al 1993b). The volcanism style (Liu et al., 1996) and geochemical patterns (Gerlach et al., 1986) are different between the ESC and the San Felix Island. Finally, the intensity of side-scan images of young volcanism decreases along the ESC to the east, but jumps to a high value near the San Felix Island (Liu et al., 1996). CONCLUSIONS The effective elastic plate thickness along the ESC is found to be close to 3 km along most of the chain, -9 km on older seafloor east of the Nazca fracture zone, and -13 km near the San Felix Island. Forward, admittance, and coherence methods with both shipboard and altimetry gravity data yield consistent results. Admittance and coherence methods together suggest that the ratio of subsurface loads to surface loads is less than 0 5 Thus the assumption of no subsurface loads in this region used in the admittance calculations is reasonable. T e values along the chain lie between the 200 and 500C isotherms for lithosphere at the time of loading These results lie within the range observed in other studi e s of lithosphere loaded by seamounts The elastic thickness values are compatible with a model assuming a single hot spot origin for the ESC They are also consistent with other geophysical and geochemical evidence pointing to an independent source and mode of fo nnation for the San Felix Island than that of the ESC A slightly higher Te value (5 km) at the far western end of the ESC may reflect in part a small component of dynamic support for bathymetric loads. Admittance modeling in the Easter Salas y Gomez region suggests smaller features may be partially (10-15%) non-isostatically compensated Dynamic uplift could be a product of mantle upwelling associated with the Easter hot spot. 162

PAGE 176

ABSTRACT CHAPTER 6 EVOLUTION OF THE SOUTHEAST PACIFIC AND THE EASTER SEAMOUNT CHAIN A schematic tectonic history o f the south e rn part of the Nazca plate and its corresponding part of the Pacific plate is reconstructed for the time period since magnetic anomaly 10 (28.5 Ma) This part of the southeastern Pacific basin includes the Easter Seamount Chain (ESC) extending eas twards from the Easter microplate, the Nazca ridge bending northeastwards from the east e rn end of the ESC, the San Felix islands to the east of the ESC, the Crough volcanic chain extending to the west of the Easter microplate, the Tuamotu ridge, and the Juan Fernand e z volcanic chain Three active and some inactive microplates are within this region The M e ndoza rise, a failed propagator locates to the east of the East Pacific Rise (EPR), intersecting the ESC near 25 S 94 W All the available magnetic profiles within this region from National Geophysical Data Center are used for the reconstruction. The recently declassified Geosat altimetry data and the swath bathymetry and side-scan image data from Legs 5, 6, and 7 of the Gloria Expedition along the ESC are used to constrain the tectonic reconstruction. The southern Pacific-Nazca spreading center was developed from a southward propagator after the Farallon plate broke into the Nazca and Cocos plates at about chron 6c (23 7 Ma) To the east of the EPR, the Mendoza paleo-microplate was formed as the Mendoza rift propagated northwards between 163

PAGE 177

chrons 5c (16.4 Ma) and 5a (12.3 Ma), probably in response to a change in plate motion and/or the Easter hot spot interaction A fracture zone near the ESC may have existed since chron 6 (19.6 Ma) based on the offset of the identified isochrons. As part of the Easter hot spot chain, the Nazca ridge bent to the ESC after the Nazca plate motion changed due to the break-up of the Farallon plate The ESC and the Crough volcanic chain are forming from two separate hot spots (under the Nazca and Pacific plates, respectively), which were -700 km apart before chron 7 (25.0 Ma) and offset by -200 km in latitudinal direction before chron 5 (10.6 Ma). The apparent shortening of distance between the Easter and Crough hot spots suggests they are migrating towards each other. Geophysical and geochemical evidence indicates that the San Felix Island is not part of the ESC and was not formed by a Easter hot line INTRODUCTION The formation and evolution of the Nazca plate (Figure 6 .1) since chron 7 play a major role in the tectonic history of the southeastern Pacific basin, which includes plate motion changes, plate boundary reorganizations, microplate formations, and hot spot interactions. Comprehensive tectonic history of the southeastern Pacific was reconstructed for the time periods between chrons 30 and 7 and after chron 5 (Handschumacher, 1976; Herron, 1972a; Herron, 1972b; Lonsdale, 1989; Mayes et al., 1990; Okal and Cazenave, 1985; Searle et al 1995). Between magnetic chrons 30 and 7 (67 and 25 Ma), the Pacific and Farallon plates were diverging along a NNW spreading center offset dextrally by the Galapagos-Gri jalva, MarquesasMen dana, AustralNazca, and Agassiz-Challenger fracture zones (Figure 6 2) After chron 7, probably at chron 6c ( -24 Ma) (Searle et al., 1995), the Farallon plate broke into the Nazca and Cocos plates along the Galapagos-Grijalva fracture 164

PAGE 178

Figure 6.1. Predicted seafloor age and bathymetry of the southeastern Pacific basin. The bathymetry are predicted from ETOP0-5 data (National Geophysical Data Center, 1988) and free-air gravity anomalies calculated from the Geosat altimetry data (Smith and Sandwell, 1995a; Smith and Sandwell, 1995b) using the method described by Smith and Sandwell (1994) (see text for detail) The seafloor age is calculated from the predicted bathymetry based on Heestand and Crough's (1981) seafloor subsidence model (see text for detail) In this method islands and seamounts are shifted to younger age because they are shallower than normal seafloor. The swath bathymetry and side-scan image data collected during the Gloria Expedition within the labeled boxes will be presented in separate figures.

PAGE 179

soo w 145o w 140 o w 135o w 'l 10 s 15

PAGE 180

zone (Hey, 1977b; Hey et al., 1977), creating a major plate boundary reorganization. The spreading direction for the Nazca plate rotated clockwise from a ENE to a ESE direction. Since chron 5 the new and relatively stable spreading system was established again along an ENE direction between the Pacific and Nazca plates. The so-called Easter fracture zone (Herron, 1972a) of the EPR has been located and mapped south of the Easter Island and is called the SOEST fracture zone (Hey et al., 1995) The SOEST counterpart on the Pacific plate, FZ 2 (Okal and Cazenave, 1985), was also imaged (Searle et al., 1995) The spreading in this configuration persisted until just before chron 3 (Klaus et al., 1991 ; Naar and Hey, 1991 ; Okal and Cazenave, 1985 ; Rushy, 1992; Searle et al., 1993). At -5.6 Ma both Easter and Juan Fernandez microplates were initiated by rift propagating from within transform fault zones (Bird and Naar, 1994; Naar and Hey, 1991) The tectonic history of the southeastern Pacific between chrons 7 and 5 has been reconstructed for the Galapagos (Nazca-Cocos) spreading center (to the north of the study region) (Hey, 1977b; Hey et al., 1977; Wilson and Hey, 1995) and the Nazca-Antarctic spreading center (to the south of the study region) (Tebbens and Cande, 1996; Tebbens et al. 1995). However, th e tectonic history of the Pacific-Nazca spreading center during this period is still under debate. Herron (1972) suggests that the Pacific and Nazca plates were spreading along the Mendoza rise between 9 and 21 Ma and the spreading about the present EPR was initiated only during the last 10 m .y. Okal and Cazenave (1985) believe that a microplate was formed as a 500-km westward ridge jump from the Mendoza rise at -18 Ma. According to Lonsdale (1989), after the break-up of the Farallon plate, spreading between Pacific and Nazca plates was reoriented to an ENE direction and persisted until the formation of the Mendoza microplate around 15 M a by a northward propagating of the East Pacific Rise (EPR). Hot spot interaction s which not only buried the evidence of the seafloor spreading but might also initiate the rift propagating and microplate formations, appears to have further complicated the tectonic history of this area. 167

PAGE 181

One of the major volcanic features in this region is the Easter Seamount Chain (ESC), which has attracted scientific interest for more than 30 years (Bonatti et al., 1977; Clark and Dymond, 1977; Haase and Devey, 1996 ; Hagen et al., 1990 ; Herron, 1972a; Herron 1972b; Hey et al 1995 ; Liu and Naar, 1996b; Morgan, 1972; Naar et al 1993b ; O'Connor et al., 1995; Pilger and Handschumacher, 1981; Poreda et al 1993a; Poreda et al., 1993b ; Schilling et al., 1985a; Schilling et al., 1985b; Stoffers et al., 1994; Wilson, 1963a; Wilson, 1973). The chain, lying entirely within the Nazca plate (29 S 112 W and 23 s. 82" W) trends -085" from -100 km east of the Easter microplate (Liu and Naar, 1996b) to its junction with the Nazca ridge (Figure 6 .1). Studying the group velocity of Rayleigh waves that propagated along either the Nazca ridge or the ESC suggests that the crustal thickness along the Nazca ridge is -18 km, whereas it is -6 km along the ESC (Woods and Okal, 1994) Extending to the w est of the Easter microplate on the Pacific plate are the Crough-Ducie Henderson island chain and Tuamotu ridge The San Felix Island existing to the east of the ESC does not appear to be part of the ESC because of a volcanic gap and morphology difference between them (Naar et al., 1993a; Naar et al., 1993b) Flexural studies show that the effectiv e elastic thickness of the lithosphere near the Island (-13 km) is significantly greater than that along the ESC (-3 km) (Liu and Naar, 1996a ; Liu et al. 1995) The latest volcanic activities at the San Felix Island occurred about 70 years ago (Firth, 1943 ; Willis and Wa s hinton, 1924). Geochemical evidence (Gerlach et al ., 1986) also shows different NdSr and NdSr-Pb patterns between the ESC and the San Felix Island It has been sugge s ted that a single hot spot or mantle plume, on the EPR may be responsible for the formation of the volcanic chains extending to both sides from the Easter microplate (Morgan, 1971; Wilson 1963b ) Reconstruction of the spreading history between the Pacific and Nazca plates suggests that the Nazca and Tuamotu ridges originated from a melting anomaly under the Pacific-Farallon ridge during the time interval 168

PAGE 182

between anomalies 19 and 11 (Handschumacher et al ., 1981; Pilger and Handschumacher, 1981). However, the ESC fonned entirely within the Nazca plate from a hot spot located to the east of the EPR (Pilger and Handschumacher, 1981), while the Crough-Ducie Henderson Island chain resulted from another hot spot to the west of the EPR (Okal and Cazenave, 1985). Side-scan sonar image data and swath bathymetry data from the Gloria Expedition show that there are large area of recent volcanism existing about 100 km to the east of the EPR (Hey et al., 1995 ; Liu and Naar, 1996b) named Abu volcanic field (Hagen et al., 1990), Umu and Tupa volcanic fields (Haase and Devey, 1996; Haase et al., 1996; Stoffers et al 1994). These volcanic fields are clearly off the ridge axis (Liu and Naar, 1996b), suggesting the c urrent location of the Easter hot spot (Hagen et al ., 1990; Liu and Naar, 1996b). Analyzing the morphology of the seamounts and the volcanic distribution in the Easter-Salas y Gomez region (at the western end of the ESC) from the swath bathymetry and side -scan image data indicates that the chain is composed of several volcanic ridges in a dextral en -ec helon pattern with an angle of about 5 -15. oblique to the overall trend of th e ESC (Liu and Naar 1996b ; Liu et al 1995) It is inferred that each of the ridges may tap from an elongated mantle source at the base of the lithosph e re which may fonn from a mantle blob after being sheared by the relative motion between the lithosphere and the underlying mantle as the blob rises buoyantly from depth (referred to as the Hot Bubble model) (Liu and Naar, 1996b ; Liu et al., 1995 ) The model suggests that an unsteady mantle plume may be unsteady and rise in a fonn of a series of bubble s which are sheared, while rising, into football shaped sour ce s fonning the en-echelon volcanic ridges Several other hypotheses were proposed to explain the apparent contemporaneous volcanism along the ESC (based on old K Ar age data). (1) Instead of a single hot spot, a mantle "hot line", corresponding to upwelling limbs of secondary mantle convective rolls (Richter and Par s ons, 1975), may be respons i ble for the simultaneous volcanism at several 169

PAGE 183

places along the ESC (Bonatti and Harrison, 1976; Bonatti et al., 1977) (2) The volcanism may form along a "leaky fracture zone" (Clark and Dymond, 1977; Menard and Atwater, 1968). (3) A "diffuse extension" model suggests that the slab-pull from the trenches may produce tensional stresses and en-echelon ridges may form along the weakened lithosphere (Sandwell et al., 1995; Winterer and Sandwell, 1987). More data were acquired from recent surveys, especially Legs 5 (Hey et al ., 1995), 6, and 7 (Naar et al 1993a; Naar et al. 1993b) of the Gloria Expedition along the ESC from the EPR to the San Felix Island. With recently declassified Geosat altimetry data (Smith and Sandwell, 1995a; Smith and Sandwell, 1995b) it is now feasible to reconstruct the tectonic history for this region. Incorporating all the recent data available, in addition to old magnetic data from the National Geophysical Data Center in this region, we identify new tectonic features and new magnetic isochrons The tectonic history of the southern Nazca plate and its corresponding part of the Pacific between Marquesas-Mendana and Agassiz-Challenger fracture zones since chron 10, particularly b e tw ee n chrons 7 and 5, is reconstructed. Based on thi s reconstruction and new 4 0 Ar-39 Ar age data derived from dredge samples collected in recent cruises, w e are abl e to discuss the evolution of the ESC and the hot spot activities and its interactions with the oceanic lith o sphere. DATA INTERPRETATION The bathymetry was used as general constraints of th e tectonic reconstruction (Figure 6 .1). The bathymetry are predicted based on ETOP0-5 data (National Geophysical Data Center, 1988) and gravity anomalies calculated from the recently declassified Geosat altimetry data (Smith and Sand well, 1995a; Smith and Sand well, 1995b) The depth of the seafloor is the summation of two components. The long-wavelength component (>160 170

PAGE 184

km) is filtered from ETOP0-5 data, while the short-wavelength component (15-160 km) is predicted from gravity anomaly. It is assumed that the bathymetric features of wavelength less than 160 km are flexurally supported by the elastic plate with effective thickness greater than 5 km. The vertical amplitudes of the bathymetric features are proportional to the downward continuation of free-air gravity anomalies by a factor of the Bouguer constant 2npr, where p is the density of the oceanic crust relative to seawater and r is the Newtonian gravitational constant (also known as G). We used a value of 15 m/mGal for the Bouguer constant during the bathymetric prediction The method of the bathymetry prediction has been discussed in more detail elsewhere (Smith and Sandwell, 1994). Figure 6.1 can also be viewed as approximate age using the age scale because of the simple relationship between the depth and age of the seafloor This age prediction is intended for normal seafloor. The seafloor age near seamounts or the reheated oceanic lithosphere near mantle plumes will be under estimated due to the shallow depths The relationship between the d e pth and age of the seafloor is d escri bed by D = r + s..fi, where D is the depth of the seafloor, Tis the age of the seafloor r is the depth of mid ocean ridge, and s is the seafloor subsidence factor. Param e ters for r and s from several subsidence models (Heestand and Crough, 1981; Levitt and Sandwell 1995; Richardson et al., 1995; Stein and Stein, 1992 ; Turcotte and Schubert, 1982) have been tried; and we found that Heestand and Crough's (1981) param eters (r = 2700 m and s = 295 m/my0 5 ) produces the least residual between the predicted seafloor age and magn etic isochron age for the Nazca plate. Figure 6.1 presents the age of the seafloor predicted using these parameters. The resolution of the predicted age and depth of th e seafloor is the same as Sandwell and Smith's (1995) gravity data (-2 minute) The predicted gravity, bathymetry, and age are used as a guideline for the reconstruction of the tectonic history of the southeastern Pacific Many pseudofaults and fracture zones can be identified from the diff ere nc e of the bathymetry and seafloor age on 171

PAGE 185

both sides of the features For instance, the Mendoza rift (near 94 OW in Figure 6 1) is bounded by two pseudofault-like features on both sides that merge at th e northern tip with the Mendoza rift, indicating northward pr opaga tion and then abandonment Comparing the seafloor age near the Mendoza rift and to the eastern frank of the EPR (where reliable isochrons have been identified) suggests that M e ndoza rift failed around 12 Ma. It is inferred from these data that the northern portion Pacific-Nazca spreading center developed from a northward propagator (Goff and Cochran 1996), which initiated the Bauer paleo microplate and turned off the Galap agos rise at about 5 Ma. A pseudofault like feature between 15 S and 20 S is clearly seen to the west of the EPR, suggesting that the southern portion Pacific Nazca spreading center may have developed from a southward propagator. The location and age of the surrounding seafloor suggest that this propagator must have initiated prior to the northward propagator and the Mendoza rift A major fracture zone can be identified all the way across the southeastern Pacific basin from the east the Challenger fracture zone, the Chile fracture zone and the Agassiz fracture zone (Figure 6.1). Ther e are several fracture zone-like features on both sides of the EPR near 15 S Figure 6.2 shows shiptracks of Legs 5, 6, and 7 of the Gloria Expedition (dotted lines) plus the available transits from the National Geophy sical Data Center, along which the magnetic anomaly data are used for the isochron identifications. The id e ntification s are constrained by the two-dim e nsional magnetic mod els (Wilson and Hey 1995 ) based on the revised calibration of the geomagnetic polarity tim escale (Cande and Kent, 1995). Previously identified isochrons befor e chron 7 (Cande et al., 1989) and after chron 3 (Bird and Naar, 1994; Hey et al., 1995 ; Naar and Hey, 1986 ; Naar and Hey, 1991) are incorporated. We also incorporat ed some of the isochrons (chrons 3a through 12) identified by Searle et al. (1995) along the Crough and Tuamotu chains based on the data from two separate cruises. The isochron s and seafloor lineations along the ESC are also constrained by the swath bathymetry and side-scan image data of the Gloria Expedition for 172

PAGE 186

Figure 6.2. Available magnetic data and identified isochrons. Thin solid wiggles are the magnetic proflles available from National Geophysical Data Center and from the Gloria Expedition with its shiptracks in dotted lines. Labeled dash lines are the identified isochrons including the previously identified isochrons (Cande et al 1989; Searle et al., 1995). Previously identified fracture zones, failed rifts, pseudofaults, and ridge axes are in thick lines labeled with their name if any Synthetic magnetic anomaly is generated for the seafloor of 4-km-depth at 27s using the Magbath programs provided by D. Wilson (1995). Different synthetic profiles are created using different spreading rates at various latitudes to model the observed magnetic anomalies in different regions.

PAGE 187

1 3 5 W 130 W 125 W 120 W ll5" W llO" W o -IO" S IS S 20 S 40 S

PAGE 188

the areas within the boxes A through F in Figure 6.1 that are presented in Figures 6.3 through 8, respectively. Figure 6 9 shows the tectonic features and isochrons interpreted from the data. Legs 5 and 6 of the Gloria Expedition focused on the western part of the ESC, including the Easter and Salas y Gomez islands (box A in Figure 6.1 ) In this region, grids of image (Figure 6.3A) and bathymetry (Figure 6 3B) in the survey area are compiled from the side-scan sonar image data and bathymetry swaths, respectively (Liu and Naar, 1996c; Liu and Naar, 1996d; Liu et al. 1994). The fracture zones, pseudofaults and ridge axes (Figure 6.3C) are constrained primarily by the side-scan image and bathymetry data The SOEST fracture zone (Hey et al., 1995) is well-defined within the coverage of both side scan image and bathymetry data. It is suggested from the offset of the magnetic isochrons that the fracture zon e may extend further to the east. There is a deep trough near 26 7 s, 106.sw with steep wall bounding to the eastern side (Figure 6.3B). There are several pseudofault-like features fanning southwards from the southern part of the trough It is proposed that the trough may represent a failed northward propagating rift (Liu and Naar, 1996b). A grid of magnetic anomaly (Figure 6 3D) is also compiled for this region from the available magnetic profiles (Figure 6.3C) and hand-contours (constrained by swath bathymetry and GLORI-B side-scan imagery). With the gridded bathymetry and magnetic field, an attempt was made to compute the magnetization of the seafloor based on the three dimensional magnetic inversion method (Carbotte and Macdonald, 1992; Cormier and Macdonald, 1994; Macdonald et al., 1980; Miller and Hey, 1986; Parker, 1972 ; Parker and Huestis, 1974; Perram et al 1993; Sempere et al., 1989). However, the inversion result was unsatisfactory due to large track spacing and NNE and NNW trending artifacts of unknown origin (M. -H Cormier personal communication, 1996). 175

PAGE 189

Figure 6 3. GLORI-B data and interpretation for the Easter-Salas y Gomez region (box A in Figure 6.1). A: GLORI-B side-scan image data SeaBeam 2000 image data are incorporated covering -10-krn along the ship tracks of Leg 6 (to the east of 1 09W) (Liu and Naar, 1996b). B: GLORI-B and SeaBeam 2000 swath bathymetry data. Out of the survey areas are filled with the predicted bathymetry from ETOP0-5 and Geosat data. Within the survey region, the data are composed of three parts, the SeaBeam 2000 swaths covering -10-krn along the shiptracks, the GLORI-B swaths extending to -12-krn on both sides of the SeaBeam swaths, and the interpolated bathymetry based on the SeaBeam and GLORI-B data filling the gaps between the swaths (Liu and Naar, 1996c). C: Magnetic data and isochrons. Thin wiggles show the available magnetic proflles Thin labeled lines are the identified isochrons. Ridge axes, pseudofaults, and fracture zones are in thick lines D: Gridded magnetic anomaly field based on the existing magnetic profiles (Figure 6.3C) and educated hand contours. E: Identified isochrons, tectonic structures, and seafloor lineaments. The isochrons are in labeled lines. Thick solid lines represent tectonic structures, mid-ocean ridges, pseudofaults, and fracture zones. Seafloor fabrics are in dotted lines. F: Seafloor age interpolated from the identified isochrons.

PAGE 190

26s 26 s 27"S 2s s 29 s 29s -5000 -4450 -3900 -3350 -2800 -2250 -1700 -1150 -600 -50 26s 21s 2s s (Continued on next page) 177

PAGE 191

Figure 6.3. (Continued) p4 w 113 w 112 w 111w 25 s 26 s 26s 2TS 21s 2s s 2ss 29 s 29s 26s 2TS \ 4a \ : f f 2ss 2& isochron .......... .......................... lineation 26s 21s 2ss 29s 0 2 3 4 5 6 7 8 9 10 11 178

PAGE 192

Figure 6.3E shows seafloor lineations and major tectonic features presented in the side-scan image and bathymetry data (Figures 6 .3A and 3B) and the isochrons identified based on the magnetic data from the Gloria Expedition in addition to the data from the National Geophysical Data Center (Figure 6 3C). The seafloor ages (Figure 6.3F) are interpolated from the identified isochrons (Figure 6 3E) The isochrons and seafloor lineations to the east of the outer pseudofault of the East rift suggest a more complex seafloor spreading history than previously thought (Bird and Naar, 1994; Hey et al., 1985; Naar and Hey 1986 ; Naar and Hey, 1991 ; Rushy 1992; Searle et al 1989 ; Zukin and Francheteau, 1990). Several groups of seafloor fabrics exist at different locations with various trends. A group of seafloor fabrics near 25.5 S, 108 W truncate another group of fabrics to their east (Figure 6.4). It appears to be an eastern pseudofault of a failed propagator (Liu and Naar, 1996b). Unfortunately, the magnetic isochrons near the wide spread lava flows are not well identified The isochrons on both sides of the volcanism of the ESC appear to be offset and in different trends. These observations suggest the existence of an OSC which may be responsible for the formation of the complex patterns of the seafloor lineations to the east of the outer pseudofault and the north of the SO EST fracture zone. The eastern boundaries of the Mendoza paleo-microplate (which is referred to as Mendoza rise (Herron, 1972a) and will also be referred to as eastern rift, or eastern Mendoza rift, or Mendoza propagator with respect to the western rift, the western boundary of the microplate) can be well identified in the alt.imetric gravity and predicted bathymetry age data (Figure 6.1). However, most of its southwestern boundaries may be buried under the volcanism of the ESC. To the south of the ESC at the location near 27.5"S 97 5 "W, a group of seafloor lineations merge to the southeast, which is interpreted as the southern tip of the failed western rift of the Mendoza pal eomicroplate To the north of the ESC within box C in Figure 6.1, some corresponding seafloor lineations were imaged by the GLORI-B 179

PAGE 193

Figure 6.4. GLORI-B data showing a pseudofault near 18DW between two groups of seafloor lineations (box B in Figure 6.1). A: GLORI-B image data. B: GLORI-B and SeaBeam 2000 bathymetry data. Out of the survey area is filled with predicted bathymetry from ETOP0-5 and Geosat altimetry data.

PAGE 194

1 25 26 s 26 s A: Side-scan Image 0 25 50 75 100 125 150 175 200 225 -4500 -4250 -4000 -3750 -3500 -3250 -3000 -2750 -2500 -2250 181

PAGE 195

Figure 6.5. GLORI-B data showing the seafloor lineations near the failed western propagator of the Mendoza paleo-microplate (box C in Figure 6.1). A: GLORI-B image data The inset shows magnetic wiggles in solid lines, isochrons in dash lines and shiptracks in dotted lines. The imaged area is indicated by the stippled region. B: SeaBeam 2000 bathymetry data Out of the survey area is filled with predicted bathymetry from ETOP0-5 and Geosat altimetry data.

PAGE 196

2s s 2s s A: Side-scan Image 0 25 50 75 100 125 150 175 200 225 24s 2s s -6207 -6200 -5620 -5040 -4460 -3880 -3300 -2720 -2140 -1560 -980 183

PAGE 197

System (Figure 6.5) Notice the orientation change of the seafloor fabric along the transit from a NNW to ENE d ire ction between lOl"W and lOO" W The change of the isochron trends is also observed in this area along several closely spaced EW -trending magnetic profiles (inset of Figure 6.5A). The patterns of the seafloor lineations observed from the altimetric gravity data suggest that the inner pseudofault of the western rift of the microplate is nearby, probably cross the transit at 100.2"W As shown in Figure 6 .1, the southern end of the eastern rift curves westwards to the interpreted southern tip of the western rift This trend of the seafloor lineations is also supported by the GLORI-B side-scan image and SeaBeam 2000 bathymetry data (Figure 6.6). In the center of Figure 6.6B, a NNE-trending bathymetry high is centered between two troughs. A distinct inward facing scarp is shown in the SeaBeam 2000 data on the eastern wall of the eastern trough (Figure 6.6B). The magnetic and bathymetric profiles show symmetric patterns about the bathymetry high (inset of Figure 6.6A) The SeaBeam System recorded a shallow d e pth of 916 m when crossing the summit of the bathymetric ridge. These observations suggest the feature is volcanic origin probably due to the interaction of the hot spot with the Mendoza rift, and may be an analogy of th e volcanic ridge along the failed rift near 28.5 "S, 111.2 W (Hey et al., 1995; Liu and Naar, 1996b). Our preferred location of the failed eastern rift axis is just to the east of th e shallow ridge near 24.5"S, 94.4"W although the morphology and magn etic pattern are not typical of other failed rifts and thus it may actually und erlie the shallow volcanic ridge Figure 6.7 shows the transits of Legs 6 and 7 of the Gloria Expedition n ear the intersection of the ESC and the Nazca ridge. This is also the area where Nazca ridg e cross the Nazca fracture zone. However, no obvious fracture zones were imaged possibly due to thick sediments or widespread volcanic flows on the seafloor. Comparing with Figure 6 7, the style of the volcanism near the San Felix I s land appears quite different (Fig ur e 6.8). A seamount gap of about 200 km exists betw een the San Feli x Island and the ESC. Th ere is 184

PAGE 198

Figure 6.6 GLORI-B data showing the southern portion of the failed eastern Mendoza rift (box D in Figure 6 .1). The ENE trend bathymetry high in the center of this area is interpreted as the ridge location because the magnetic profiles are symmetric about the feature A : GLORI-B image data. The inset shows magnetic wiggles in solid lines, bathymetry profiles in dash lines, and shiptracks in dotted lines The imaged area is indicated by the stippled region B: SeaBeam 2000 bathymetry data. Out of the survey area is ft.lled with predicted bathymetry from ETOP0-5 and Geo s at altimetry data

PAGE 199

2s s 9s w 0 25 50 75 100 125 150 94 w 93 w A: Side-scan Image 175 200 225 24s 2s s inten s ity 250 24 s 2ss -5295 -5000 -4500 -4000 -3500 -3000 -2500 -2000 -1500 -1000 186

PAGE 200

187

PAGE 201

Figure 6.7. G L ORI-B data near the intersection that the Nazca ridge and the Nazca fracture zone and the bend fro m t he Nazca ridge to the ESC (box E in Figure 6.1) A: GLORI B image data. B: SeaBea m 2000 bathymetry data. Out of the survey area is fllled with predicted bathymetry from ETOP0-5 and Geosat altimetry data.

PAGE 202

Figure 6 8 GLORI-B data near the San Felix Island showing the volcanic gap between the ESC and the Island (box F in Figure 6 1 ) A small portion of the volcanism related to the ESC is shown at the northwestern comer. The volcanism extending to the west of the San Felix Island shows a different style from that along the ESC A : GLORI-B image data B : SeaBeam 2000 bathymetry data. Out of the survey area is filled with predicted bathymetry from ETOP0-5 and Geosat altimetry data.

PAGE 203

A: S i d e scan Image 0 25 50 75 100 125 150 175 200 225 26 S -5000 -4450 -3900 -3350 -2800 -2250 -1700 -115 0 6 0 0 -50 190

PAGE 204

Figure 6 9. Interpreted tectonic features and isochrons Dotted lines indicate the isochrons. Mid-ocean ridges, failed propagators, fracture zones, pseudofaultli, and other linear features are in solid lines I s land and seamount chains are represented by light stippled areas

PAGE 205

Location o ss tos 15"S 20" S 25"S 30 S 35" S o a ctive h o t spo t 40"S x failin g hot spo t C Crough h o t spot E Easter hotspot Juan Fernandez S San Felix h otspot 135 w tosw toow ............ --._ ........ .. "" / .. /' ./ ... ?"'{ ){ Cllain s ...... . ... .. . \ i QI; I2 \

PAGE 206

scattered volcanism existing to the west of the San Felix Island. This volcanism appears young (based on intensity), however the lineations of the seafloor are still visible. Assuming the young volcanism is associated with the San Felix Island, the volcanic gap between the ESC and San Felix volcanism is about 30 km No evidence indicates that the volcanism near the San Felix Island is associated with the ESC VOLCANIC AGE PROGRESSION ALONG THE ESC A major challenge facing the hot spot model is the apparent contemporaneous volcanism along the ESC based on the old K-Ar age data. However, recent 40Af_39Af data indicate that the volcanism actually ages to the east along the ESC (R. Duncan, personal communication, 1996) The variation and the pattern of the seamount and island morphology along the ESC also supports the general age progression along the ESC (Fisher and Norris, 1960; Liu and Naar, 1996b). Dredge samples were collected during both Leg 7 of the Gloria Expedition (Table 6.1) (Naar et al., 1993a; Naar et al., 1993b ; Poreda et al. 1993a) and the Sonne 80 cruise (O'Connor et al., 1995) (Figure 6.10A). Unlike the old K-Ar age data (Baker et al., 1974; Bonatti et al., 1977; Clark and Dymond 1977), the new age data calculated using 40Af_39Af method indicate a systematic increase of the seamount age to the east along the ESC (Figure 6.10B). The seamount age from the sample near the San Felix Island (about 0.8.05 Ma) is much younger than the seamount age near the eastern end of the ESC (Figure 6 10 and Table 6.1). This data is in agreement with the in situ observation of the dredge samples, which had a young appearance (Naar et al., 1993a). Volcanic activities were also reported after the Chilean earthquake of 1992 (Firth, 1943). 193

PAGE 207

Figure 6.10. Seamount and seafloor ages along the ESC A: Dredge location Squares from Leg 7 of the Gloria Expedition and triangles from Leg 80 of the Sonne cruise. The background image is the predicted bathymetry (the same as in Figure 6.1) indicating the locations of the seamounts The seafloor age is interpolated from the identified magnetic isochrons along the line, which is near the center of the stippl e pattern on Figure 6.9. B: Seamount ages of the rock samples are represented by stars (for the Gloria Expedition) and circles (for Sonne cruise). Uncertainties for the seamount ages shown in the plot represent standard deviation (e.g. one sigma). Some of the error bars lie within the dimensions of the symbols. The solid line show the seafloor age from isochrons.

PAGE 208

-6500 -5800 -5100 -4400 -3700 -3000 -2300 -1600 -900 -200 500 depth (m) 35 35 B: Age of Seafloor and Seamount 30 0 sample from Gloria07 30 A sample from Sonne80 c 25 "' :: :: 25 E c c .., "' "' 0 i "' "0 0 .., c "0 "' "0 "' c u '220 c ::E "' :;! 20 '2 ::E 6 E E z 6 "' E :; .c "' Q) .f! :; Q) 15 0 0 15 "0 "' "' ... "' 0. 10 10 ... "' 0 CD 5 5 8 88 8 8 88 8 0 0 -115 -110 -105 -100 -95 -90 -85 -80 longitude e) 195

PAGE 209

Table 6 .1. Volcanic age of dredge samples from Leg 7 of the Gloria Expedition measured by R. Duncan (personal communication, 1996) longitude CW) latitude CS) ageone-sigma (Ma) depth (m) sample ------------------------------------------------------------------------------------80.599 26.211 0.800.050 3571 D-23 93.387 25.426 12.6 1 3481 D -25 101.109 25.047 6 8 2 2931 D-28 102.234 25 893 2 .9. 05 2354 D-29 103.690 26.067 0.51 0.0 10 2808 D 30 104.353 26.158 0 080.015 2308 D-31 104.304 26.353 1.05 05 2656 D-32 105.827 26.377 2.6.1 2133 D-36 106.255 26.547 2.50.04 2184 D-37 106.149 26.762 0.90.20 2045 D-38 105.957 26 704 1.30.03 2312 D 39 106.613 27.185 1.10 .04 2260 D-42 107.490 26.392 1.12.02 3350 D-45 107.710 26.206 0.260.040 3367 D-46 108 287 26 .628 0.270.030 2692 D-47 108.559 26.425 0.650.030 3271 D-48 108 748 26.275 0.240.025 2340 D-49 109 453 26.752 0.120.020 3273 D-50 110 118 26.738 0 .27 5 050 3145 D-51 110.840 26 702 0.020.030 2757 D-52 110.317 27.039 0.270.040 2655 D-55 109.845 27.155 1.14.04 3174 D -56 108.694 27.315 1.45.04 2463 D-57 108.030 27.621 2.85 05 2891 D-59 ----------------------------------------------------------------------------------------For comparison, the ages of the seafloor (Figure 6.10B) along the center of the ESC (line in Figure 6.10A) are interpolated from the identified magnetic isochrons (Figure 6 9) Except the west most sample (the last triangle to the west in Figure 6.10), the seamount are generally younger than seafloor age by -7 m.y. or less, suggesting the Easter hot spot was off the ridge axis at distance -800 km or less during this period if the hot spot model is correct (Figure 6.10B) Because the dredge samples were collected on young volcanism from the top layer of the seamounts or lava flows, their ages are likely younger than the initial age of the volcanism. Thus, the hot spot should be closer to the spreading center than predicted from these age trends. If the Easter hot spot is near the Ahu volcanic field (Hagen et al 1990), the current distance between the hot spot and the spreading 196

PAGE 210

center is -200 km. From the GLORI-B side-scan image, the young lava flow associated with the ESC begins -100 km to the east of the East rift (Hey et al., 1995; Liu and Naar, 1996b). The variability of volcanic ages can be explained by the proposed Hot Bubble model (Liu and Naar, 1996b ; Liu et al., 1995). The age calculated from the west most sample is identical to the seafloor age at the corresponding latitude, suggesting that the sample was probably dredged from normal seafloor The age of the seafloor is not continuous near the failed rifts and fracture zones (Figure 6 10B). The ages of the seafloor and seamount are very close near the eastern Mendoza rift, suggesting that the Easter hot spot is close to the rift. The eastern Mendoza rift may form and propagate in response to a change in plate motion at or near the hot spot where the lithosphere is weakened. Based only on the seafloor age profile (Figure 6.10B) the average full spreading rate is -180 km/m. y. The full spreading of the western Mendoza rift (90 krn/m.y.) is less than the average, whereas that of the eastern Mendoza rift (210 km/m y.) is greater than the average, indicating that the eastern rift was dominant spreading center during the time it was active. However, both of the rates are over estimated because the sample line is not parallel to spr e ading due to the microplate rotation. The apparent full spreading rate to the west of the outer east pseudofault is -108 km/m y., which is in agreement with the estimation made by Naar and H e y (1986) (assuming they used the old magnetic timescale). RECONSTRUCTION OF THE TECTONIC HISTORY The tectonic reconstructions are bas e d on the identification of the magnetic isochrons and the tectonic features of the seafloor (Figure 6 9). Each of the reconstruction time frames is created by cutting and removing magnetic anomalies of a certain age and 197

PAGE 211

moving the two plates together to match the remaining isochrons and tectonic lineations. The interior of the Mendoza paleo-microplate was reconstructed in a similar manner to the method of Naar and Hey (1991). The Challenger, Chile, and Agassiz fracture zones provide a major constraint in latitudinal direction during the reconstructions Linear features, such as the Bauer fracture zones and the Garrett fracture zone are also present in the northern part of this region near 15"S. However, they were not used as constraints for the reconstruction because they are either short lived features or displaced by crustal compression or extension during the formation of the Mendoza and Bauer paleo microplates (Goff and Cochran, 1996) The reconstructions of the hot spot locations are constrained by the seamount ages (Figure 6.10B), bendings of the volcanic chains, and their present relative location Assuming that the Nazca ri!Jge and the ESC are formed from a single hot spot, the bending from the Nazca ridge to the ESC is probably due to the Nazca plate reorientation after the break-up of the Farallon plate. Thus, the bending was formed at chron 6c (24 Ma) which is in agreement with the seamount ages (Figure 6.10B). The current location of the Easter hot spot is still in question. For the recon st ructions we choose the center of the large volcanic field (including the Ahu, Umu, and Tupa volcanic fields) to the southwest of the Easter Island as the current location of the Easter hot spot. In each step of the reconstructions, the location of the Easter hot spot with re spec t to the ridge axis is interpolated according to the age trends of the seafloor versus seamounts discussed earlier (Figure 6.10). However, due to the wide range of the sea mount ages along the chain, the interpolation is not well -co nstrained Figure 6.11 shows the schematic reconstructions of the tectonic history between chron 10 and present. The time frames are selected to focus on major tectonic events. 198

PAGE 212

Figure 6 11. Schematic reconstructions of the southern portion of the Nazca plate and its corresponding part on the Pacifi c plate based on the identified tectonic features and isochrons (Figur e 6 9) (see text for more detail of the reconstruction method) Spreading centers, fracture zones and pseudofaults ar e r e presented by solid lines Isochrons are in labeled dotted lines Volcanic chains are indi c ated by stippled zones Small numbers indicate the anomaly names of the isochrons Not well constrained area or feature is mark e d with "?". The time for each reconstruction is indicated by the chron name and the age at the top right comer of each frame.

PAGE 213

E Chron 5ab (13.4Ma) F Chro n 5b (15 0Ma) G Chron 6 (19. 6Ma) H Chron 6b (22.8Ma) I Chron 7 (25.0Ma) I Chron 10 (28 .5Ma) 200

PAGE 214

Chron 10 The Pacific and Farallon plates were spreading along a NNW mid-ocean ridge offset dextrally by the Austral-Nazca fracture zone (Figure 6.111). The Nazca and Tuamotu ridges were forming from a single hot spot, the Easter hot spot, on the spreading center. The two ridges are symmetric about the spreading center but oblique to the nearby fracture zones, indicating that there is a relatively southward hot spot migration. The Crough hot spot exists to the west of the Pacific Farallon spreading center, but its location is not well constrained. The San Felix chain may have already formed to the east of the ESC. Again, its previous location is not well constrained because its trace has been subducted Likewise, we cannot determine the extent of the volcanic trace formed by the Juan Fernandez hot spot Chron 7 The Pacific-Farallon plate boundary is the same as before (Figure 6.111) The Easter hot spot migrated across the eastern Austral Nazca fracture zone and, therefore, was underlying only the Farallon plate. The Nazca ridge extends to the interior of the Nazca plate whereas the formation of the Tuamotu ridge ceased apparently because the hot spot moved far enough away from the spreading center it was feeding Chron 6b After the break-up of the Farallon plate at about chron 6c (Hey 1977b; Hey et al. 1977; Searle et al., 1995), there was a major reorientation of the spreading between Pacific 201

PAGE 215

and Nazca plates (Figure 6.11H). The spreading direction rotated clockwise. The Pacific Nazca spreading center broke into many small segments offset by OSCs or small offset transform faults A few failed propagators were left to the east of the spreading center, suggesting that western rifts dominated The length of the spreading segments increased as some long-lived propagators replaced shorter segments. Volcanic chain orientations changed in response to the plate motion change. Sharp bends formed along both the Easter-Nazca chain and the San Felix chain. On the Pacific plate, the Crough chain curves smoothly. The timing of this bend is not well constrained. Chron 6 More failed propagators were left to east of the spreading center as the western rift propagated southwards further (Figure 6.11 G). The configuration of the spreading center in the southern part of this region appears quite simple but is not well constrained from the data. It is inferred from the isochron offset that the Easter fracture zone may have become more organized The other two fracture zones to the north are not well constrained because the offsets of the isochrons are relatively small. The hot spot chains extended further without major change in their configuration. Chron Sb The eastern (Mendoza) rift began to propagate northwards at about chron 5c (Figure 6.11F) The nearby Easter hot spot may have been responsible for the initiation of the propagator A hot spot interaction has also been proposed for the Initiation of the East rift 202

PAGE 216

of the Easter microplate (Hagen et al. 1990 ; Naar and Hey, 1991 ; Searle et al., 1989) The western (Mendoza) rift continued to propagate southwards into old lithosphere. The old portion of the Juan Fernandez chain appears on the Nazca plate in our reconstru c tions The other hot spot chains were extending along their previous trends Chron Sab The eastern rift continued to propagate northwards (Figure 6 11E). The western rift propagated s outheast to the nearby fracture zone and slowly curved to the east. The curving of the southern portion of the western rift may associa te with the rotation of the Mendoza microplate (Schouten et al., 1993) The southern tip of the eastern rift propagated towards the southwest, much like the present Easter microplate area (Bird and Naar 1994; Hey et al., 1995; Naar and Hey, 1991) The Mendoza mi c roplate was m ost likely under compression, particularly in the northern part by analogy to the Easter microplate (Naar and Hey 1991). There is no major change in the configuration of the hot spot locations Chron Sa The eastern rift ceased activity by this time while the western rift propagated southeastwards further probably along a preexisting fracture zone, and met with the southern tip of the eastern rift (Figure 6 11D). Then, similar to the West rift of the Easter microplate (Liu et al., 1991 ; Searle et al. 1989 ) the western rift propagated across its own outer pseudofault and extended southwards decapitating the southeastward propagating rift 203

PAGE 217

tip This propagation might have crossed another fracture zone and formed an OSC with the previous spreading center to the south of the fracture zone Chron 5 An OSC appears to have existed north of the Easter fracture zone, which would help explain the complex patterns of the seafloor fabrics observed east of the outer pseudofault of the East rift which are oriented oblique to the spreading direction (Figure 6.11C). Two magnetic anomalies 5 within 100 km have been identified along a single magnetic profile to the east of the EPR (Figure 6 9), which also support the existence of an OSC at this time. It is inferred from the GLORI-B data (Figure 6 3) and the curvature of the isochrons that this OSC migrated southwards (Pollard et al., 1982), suggesting the western rift was still dominant. Chron 3 It has been proposed that an intratransform spreading center began to propagate northwards and initiated the Easter microplate at about chron 3 (Figure 6 11B) (Bird and Naar, 1994; Naar and Hey, 1991; Rushy, 1992; Searle et al., 1989) Since then, the Easter microplate increased its dimension as lithosphere was transferred from the Nazca plate to the interior of the microplate and due to seafloor spreading along the microplate boundaries. The West rift of the Easter microplate, the southern portion of the previous western rift, was segmented by small offset transform faults and rotated counterclockwise. 204

PAGE 218

The Juan Fernandez microplate formed at about the same time Since chron 5, the ESC has been bending southwards, suggesting a relative southward migration of the Easter hot spot. Present Both the Easter and the Juan Fernandez microplates are well pronounced (Figure 6.11A). The comprehensive reconstructions of the microplates have been presented elsewhere (Bird and Naar, 1994; Hey et al., 1995; Hey et al. 1985; Larson et al., 1992; Naar and Hey, 1991; Rusby, 1992; Searle et al., 1989; Zukin and Francheteau, 1990) The West rift extended southeastwards along the FZ 2, approaching the southern end of the East rift of the Easter microplate. A new propagator near southern tip of the East rift is propagating to the south and the north of the Easter fracture zone. The previous spreading segment to the south of the Easter fracture zone is being replaced during this southward propagation event (Hey et al., 1995) DISCUSSION Both inner and outer pseudofaults of the eastern Mendoza propagator are clearly visible in the bathymetry-age map (Figure 6 .1). The eastern rift initiated at about chron 5c (16.4 Ma) and ceased between chrons 5a and 5aa (12.3-13.1 Ma). The propagating rate is about 165 km/m.y., which is faster than the Galapagos propagator (70 km/m.y.) (Wilson and Hey, 1995) and the East rift propagator (150 km/m y.) (Naar and Hey 1986 ; Naar and Hey, 1989b): but slower than rates documented elsewhere (Caress et al., 1988; Cormier and Macdonald, 1994; Hey et al., 1980; Hey et al., 1988) The ratio of propagation rate to 205

PAGE 219

full spreading rate (-0.83) is comparable to these documented areas. However, the spreading rate is very fast, especially between chrons Sac and Sad The full rate is about 200 km/m.y based on the revised geomagnetic timescale (Cande and Kent, 199S). High full spreading rates (180-210 km/m y.) during this period are also reported on central part of the Cocos plate and the corresponding part of the Pacific plate (Wilson, 1996) This coincidence suggests a change in plate motion among the Pacific, Nazca, and Cocos plates at this time Near the intersection between the eastern Mendoza rift and the ESC, the seafloor age is very close to the seamount ages (Figure 6 10B), suggesting that the Easter hot spot was nearby or under the rift and may have initiated the rift in response to the change of the plate motion. An OSC will migrate if one of the rifts is dominant and propagating (Macdonald et al., 1992; Shemenda and Grokholsky, 1991) and two pseudofault-like traces may be left on either side of the OSC. To distinguish from a normal pseudofault (Hey, 1977a), this type pseudofault is referred to as an OSC pseudofault. Like a regular pseudofault, an OSC pseudofault is created by a propagating rift and there is an age offset between two sides of the fault although the offset is r e latively small However, the pseudofault on the dominant propagator side appears more pronounced than the other side The inner OSC pseudofault is usually not well preserved, probably, due to frequent decapitati o ns o f the propagator tips that destroy the linear structures of the inner pseudofault (Macdonald and Fox, 1988 ; Macdonald et al., 1988a; Macdonald et al. 1987; Shemenda and Grokholsky, 1991) An OSC pseudofault tends to be more complicated due to propagation decapitation overlap basins, etc. (Macdonald and Fox, 1983; Macdonald et al. 1992; Macdonald et al. 1988a ; Macdonald et al 1988b; Macdonald et al ., 1991 ; Macdonald et al. 1984 ; Macdonald et al., 1987; Shemenda and Grokholsky, 1991). We interpret the feature between 1S"S and 22"S to the west of the EPR near 12S" W as an OSC outer pseudofault. However, the corresponding inner pseudofault is not as 206

PAGE 220

obvious. We propose the OSC was initiated due to the change of the plate motion after the break up of the Farallon plate The OSC migrated southwards because the western rift was dominant and propagating southwards, leaving at least two failed northward propagating OSC rift segments near 20" S, 90"W in Figure 6 9 and just to the west of anomaly 7 in Figure 6.13G The inner pseudofault of the western rift of the Bauer paleo-microplate, or referred to as the Bauer scarp (Goff and Cochran, 1996), is less clear than its outer pseudofault. Yet, it can be identified from an average bathymetry offset (Figure 6.1). The southward propagating rift near 29 S, 113"W is a similar active example (Hey et al., 1995). The (new) East rift of the Easter microplate is expected to replace the West rift and become the new Pacific-Nazca s pr eading center (Handschumacher et al. 1981 ; Hey et al., 1985; Naar and Hey, 1991). In the Mendoza case, the new propagator, the eastern Mendoza rift, looked very similar to the present Easter microplate pattern. However western propagator curved southeastwards and met the southern tip of the eastern Mendoza rift, and apparently terminated the spreading of the eastern rift. Then, the western rift propagated southwards across its own outer pseudofault. A similar history may occur at the Easter microplate in the future, by this analogy, because the Easter West rift is presently propagating across its own pseudofault (Liu et al. 1991; Searle et al., 1989) Mter studying the tectonic evolution of the Juan Fernandez microplate, it has been suggested microplate tectonic system could repeat if the spreading centers obtained the same configuration again (Anderson-Fontana et al., 1986). In addition to the Mendoza and Bauer microplates, two other paleo-microplates are identified southeast of the Juan Fernandez microplate near 35"S 123" W and 38" S, 100 W (Figure 6.1) (Tebbens and Cande, 1996; Tebbens et al., 1995). Though the causes of microplate formation and abandonment are not well understood, the frequent appearance of microplates in this region suggests that microplate may form repeatedly in fast spreading environments where 207

PAGE 221

transform faults appear less stable (Fox and Gallo 1984; Naar and Hey, 1989b) and may play an important part in triple junction migration. After the Mendoza microplate, the Pacific-Nazca spreading boundary became a simpler two plate system and then back to a three plate system when the Easter microplate formed. The change of the plate motion and/or hot spot interaction may cause the repeated initiation and extinction of the microplates A peculiar feature near 27 S, 87"W (Figure 6.1) appears to be the southwestern extension of the Nazca ridge (Woods and Okal, 1994). Based on our reconstruction (Figure 6 11H), the Nazca ridge should bend to the ESC at the location within box F because the change of the plate motion at about chron 6c prevents the projected southwestwards extension. The trends of the seafloor lineaments in this region change from a WNW direction in the eastern area to a ENE direction in the west This feature may have formed from being near a "leaky" transform fault (Menard and Atwater, 1968) The area is near the transform fault of the Austral-Nazca fracture zones. The transform fault opened and became a spreading center as the plate motion changed after chron 6c This leaky transform fault may have tapped the nearby Easter hot spot in manner described by Epp (1984) It was suggested the ESC and the Crough chain ar e formed from a single hot spot under the EPR (Morgan 1972; Wilson, 1963a) However, subsequent reconstructions (Okal and Cazenave, 1985; Pilger and Handschumacher, 1981) had difficulty with just one hot spot. Our reconstruction also indicates that the ESC was offset to the north of the Crough chain by >200 km before chron 5. After chron 5 the ESC bent southwards They approached the same latitude after chron 3 or later. The two hot spots were at least 700 km apart at chron 7 (Figure 6 .111). Assuming the Crough hot spot is near the Easter microplate, the distance between the two hot spots appears much smaller now unless channeling from the Crough hot spot to the Easter microplate is occurring, contrary to 208

PAGE 222

previous models (Haase and Devey, 1996; Haase et al., 1996 ; O'Connor et al., 1995; Poreda et al. 1993b; Schilling et al., 1985b) The volcanism along the FZ 2 may also tap the Crough hot spot (Okal and Cazenave 1985). Or the Easter hot spot is near the Easter Island or Salas y Gomez Island and the volcanism to the west of the islands is due to westward channeling (Haase and Devey, 1996; Haase et al., 1996; Hagen et al., 1990 ; O'Connor et al., 1995; Poreda et al., 1993b; Schilling et al. 1985a; Schilling et al., 1985b) Further analyses of geochemical data should help resolve these two possibilities If there is no channeling, then there is relative migration between the Easter and Crough hot spots. The channeling hypothesis has difficulties to explain two observations ( 1) The volcanism of the western end of the ESC is -100 km to the east of the EPR, instead of intersecting with the EPR as expected by the channeling model and observed at the Galapagos islands (Morgan, 1978) where the islands trac e right back to the ridge axis (2) The channeling hypothesis does not explain the apparent north-south decrease in distance between the two hot spots which is in a direction perpendicular to the spreading direction. Therefore, relative migration between the two hot spots appears to have occurred to some extent. A nonstationary hot sp o t has been proposed to explain the Hawaiian-Emperor bend (Norton, 1995). The 60 change from Emperor to Hawaiian chain at 43 m.y. ago, referred to as "43 Ma bend", implies some major tectonic changes (Duncan and Clague, 1985 ; Engebretson et al., 1984) However, no major changes in plate motion have been observed within and along the edge of the P a cific plat e at 43 Ma (Norton, 1995) nor has there been strong geodynamic evidence for such a change (Richards and Lithgow Bertelloni, 1996). S e veral lines of evidenc e suggest that the San Felix Island o riginated from a source different from the ESC. (1) The radiometric age of a volcanic rock sample near the San Felix Island is about 0.8 Ma, which is much young e r than predicted if part of the ESC The predicted seamount age near the Island would be about 30 Ma based on the age trend 209

PAGE 223

along the ESC assuming one hot spot (Figure 6.10B) (2) The effective elastic thickness of the lithosphere at the time of seamount loading near the San Felix Island is -13 km while it is about 3 km along the ESC (Liu and Naar, 1996a; Liu et al 1995) (3) Geochemical evidence (Gerlach et al., 1986) indicates there are different Nd-Sr and Nd-Sr-Pb arrays between the ESC and the San Felix islands (4) And there is a volcanic gap between the ESC and the San Felix Island (Figure 6.8). Where is the current location of the San Felix hot spot? If the bends of the NazcaESC and the San Felix chain were formed at the same time the distance between the two hot spots is about 800 km. Therefore, the current location of the San Felix hot spot should be near lOOOW if there is no significant relative migration between these two hot spots However, no sign of another hot spot chain near 100 W is observed along the predicted trend of the San Felix chain. We consider four possible explanations (1) The San Felix Island represents the end of a weak failing hot spot The recent volcanic activities are result of residual heat from the hot spot (2) The San Felix Island may be the current location of the San Felix hot spot, but the bend of the chain may be not correctly identified because its weak trace is not well defined in the altimetry data. (3) The San Felix hot spot is not stationary with respect to the Easter hot spot. (4) Or, San Felix Island is not formed by a hot spot, i.e it is just an off-axis volcano. Without further age control we cannot speculate which explanation is viable CONCLUSIONS Based on the tectonic reconstruction, the southern Pacific-Nazca spreading center, or EPR, was a southward propagator referred to as the western rift, which became dominant since a major plate reorientation at chron 6c. The rift has existed through a history of interactions with OSCs and the Mendoza microplate Currently, the western rift 210

PAGE 224

interacts with the East rift of the Easter microplate and is predicted to overtake the East rift by analogy to the Mendoza paleo-microplate The Mendoza paleo-microplate was formed as the eastern Mendoza rift propagated northwards between chrons 5c and 5a The eastern rift was diverging at a very high full rate -200 km/m.y during chrons 5ac and 5ad and propagating at -165 krnlm.y The propagator may have initiated in response to a plate motion change and/or hot spot interaction The activities of an OSC may explain the complex seafloor structures to the east of the Easter microplate, which are oblique to spreading direction. An apparent microplate cycle on the Pacific-Nazca spreading boundary suggest the repetition of tectonic conditions, such as change of plate motion and/or hot spot interaction, which cause the initiation of the microplates in this area. The SOEST fracture zone and the FZ 2 are part of the Easter fracture zone which may have existed since chron 6 based on the isochron offsets although a non-transform offset may have existed. A leaky transform fault may explain the origin of what appears to be a southwestward extension of the Nazca ridge. Both Nazca and Tuamotu ridges were formed from the Pacific-Farallon spreading center before chron 7 (25 Ma) with probable channeling of plume material from the Easter hot spot. Th e Nazca ridge bends to the west to form the ESC as the Nazca plate motion changed after chron 6c in response to the break up of the Farallon plate. Our data support that the ESC and the Crough chain originated from (at least) two separate hot spots off the EPR, instead of a single hot spot on the EPR. The Easter and Crough hot spots appear to have been separated -700 km at chron 7 in EW direction and offset by -200 km in latitudinal direction before chron 5 The channeling can explain the apparent shortening of distanc e in the E-W direction but not in the N-S direction The reconstructions suggest that the hot spots are moving toward each other unless there is an error in our prop osed tectonic recon structio n that could bring these two hot spot closer Several lines of evidence suggest that the San Felix I s land formed from a source different from the Easter hot spot. 211

PAGE 225

REFERENCES Anderson-Fontana, S., J. F. Engeln, P Lundgren, R. L. Larson, and S. Stein, 1986, Tectonics and Evolution of the Juan Fernandez Microplate at the Pacific-Nazca Antarctic Triple Junction, Journal of Geophysical Research 91, 2005-2018. Baker, P. E., 1966, Preliminary Account of Recent Geological Investigation on Easter Island, Geol Mag. 104, 116-122 Baker, P. E., F. Buckley, and J. G. Holland, 1974, Petrology and Geochemistry of Easter Island, Contrib Mineral. Petrol. 44, 85-100. Banks, R J R. L. Parker, and S. P. Huestis, 1977, Isostatic Compensation on a Continental Scale: Local Versus Regional Mechanisms, Geophys. J R. Astron. Soc. 51, 431-452. Bercovici, D., G. Schubert, and G. A. Glatzmaier, 1989, Three-dimensional Spherical Model of Convection in the Earth's Mantle Science 244, 950-955 Bird, R. T and D. F. Naar, 1994, lntratransform Origins of Mid-ocean Ridge Microplates Geology 22, 987-990 Bird, R. T., R. C. Searle, V Paskevich and D. C. Twichell, 1996, Merged GLORIA Sidescan ana Hydrosweep Pseudo-sidescan : Processing and Creation of Digital Mosaics, Marine Geophysical Researches in press Blackman, D. K and D. W. Forsyth, 1991, Isostatic Compensation of Tectonic Features of the Mid-Atlantic Ridge : 25-2T30'S, Journal of Geophysical Research 96 11741-11758. Bonatti, E and C G. A. Harrison, 1976, Hot Lines in the Earth's Mantle, Nature 263, 402-404 Bonatti, E., C. G. A. Harrison, D E. Fisher, J. Honnorez, J. G Schilling, J. J. Stipp, and M. Zentilli, 1977, Easter Volcanic Chain (Southeast Pacific): A Mantle Hot Line, Journal of Geophysical Research 82, 2457-2478. Calmant, S., 1987, The Elastic Thickness of the Lithosphere in the Pacific Ocean, Earth Planet Science Letters 85, 277-288 Cande, S C and D. V. Kent, 1995, Revised Calibration of the Geomagnetic Polarity Timescale for the Late Cretaceous and Cenozoic, Journal of Geophysical Research 100, 6093-6095. 212

PAGE 226

Cande, S C., J. L. Labrecque, R L. Larson, W. C Pitman ID, X Golovchenko, and W. F. Haxhy, 1989, Magnetic Lineations of the World s Ocean Basins. Carbotte, S. and K. Macdonald, 1992 East Pacific Rise 8-10'N: Evolution of Ridge Segments and Discontinuities From SeaMARC II and Three-Dimensional Magnetic Studies, Journal of Geophysical Resear c h 97, 6959-6982 Caress, D. W and D N. Chayes, 1993, Status of Hydrosweep Data Processing and Display on the RN Maurice Ewing. Caress, D W., H. W. Menard, and R N Hey, 1988, Eocene Reorganization of the Pacific-Farallon Spreading Center North of the Mendocino Fracture Zone Journal of Geophysical Research 93, 2813 2838. Chavez, P S., Jr 1986, Processing Techniques for Digital Sonar Images from GLORIA, Photogrammetric Engineering and Remote Sensing 52 1133-1145. Clark, J. G. and J. Dymond, 1977, Geochronology and Petrochemistry of Easter and Sala y Gomez Islands : Implications for the Origin of the Sala y Gomez Ridge, Journal of volcanology and Geothermal Research 2, 29-48 Cobra, D. T 1990, Estimation and Correction of Geometric Distortions in Side-scan Sonar Images, 556 Cormier, M -H. and K. C M a cdonald, 1994, East Pacific Rise 18. -19.S: Asymmetric Spreading and Ridge Reorientation by Ultrafast Migration of Axial Discontinuities, Journal of Geophysical Research 99,543-564 Davies, G F 1995, Penetration of Plates and Plumes through the Mantle Transition Zone, Earth and Planetary Science Letters 133, 507-516 DeMets, C., R G. Gordon D F Argus, and S Stein, 1990 Current Plate Motions, Geophys Journal InternationallOl, 425-478. Dorman L. M. and B T. R. Lewis, 1970 Experimental Isostasy, 1, Theory of the Determination of the Earth's Isostatic Response to a Concentrated Load, Journal of Geophysi cal Research 75 3357 3365 Duncan, R. A and D A. Clague 1985 Pacific Plate Motion Recorded by Linear Volcanic Chains, The Pacific O c ean. New York, Plenum. Engebretson, D. C., A. Cox, and R G. Gordon, 1984, Relative Plate Motions b e tween Oceanic and Continental Plates in the Pacific Basin Spec Pap. Geol Soc. Am. 206, 59. Epp D., 1984, Possible Perturbation s to Hotspot Traces and Implications for the Origin and the Structures of the Line Islands, Journal of G e oph y sical Research 89 1127311286. Firth, R., 1943, Pacifi c Islands, Oxford University Press, New Y v. 2 p 739 213

PAGE 227

Fisher, R. L and R. M Norris, 1960 Bathymetry and Geology of Sala y Gomez, Southeast Pacific, Bulletin of the Geological Society of America 71, 497-502. D and J -G. Schilling 1991, Sr (87/86) and REE Variations along the Easter Mtcroplate Boundaries (South Pacific) : Application of Multivariate Statistical Analyses to Ridge Segmentation, Chemical Geology 89 209-241. Forsyth, D. W., 1985 Subsurface Loading and Estimates of the Flexural Rigidity of Continental Lithosphere, Journal of Geoph y sical Research 90, 12623 12632 Fowler, C. M. R 1990, The Solid Earth : an Introduction to Global Geophysics, Cambridge University Press, New York, 472. Fox, P J. and D G Gallo, 1984, A Tectonic Model for Ridge-Transform-Ridge Plate Boundaries: Implications for the Structure of Oceanic Lithosphere, Tectonophysics 104, 205-242 Gerlach, D. C., S. R Hart, V W J Morales, and C. Palacios 1986, Mantle Heterogeneity beneath the Nazca Plate : San Felix and Juan Fernandez islands, Nature 322, 165-169 Goff, J. A. and J R. Cochran, 1996, The Bauer Scarp Ridge Jump: A Complex Tectonic Sequence Revealed in Sat e llite Altimetry, Earth and Planetary Science Letters 141, 21-33. Goff, J. A and M. C. Kleinrock, 1991, Quantitative Comparison of Bathymetric Survey Systems, Geophysical Research Letters 18, 1253-1256. Goodwillie, A. M., 1995, Short-wavelength gravity lineations and unusual flexure results at the Puka Puka volcanic ridge system Earth and Planetary Science Letters 136 297-314. Gripp, A. E. 1994, Current Plate Motions: Reference Frames and Uncertainties, Ph.D. Dissertation, Northwestern University. Haase, K. M. and C. W Devey, 1996, Geochemistry of Lavas from th e Ahu and Tupa Volcanic Fields, Easter Hotspot, Southeast Pacific : Implications for Intraplate Magma Genesis near a Spreading Axis, Earth and Planetary Science Letters 137 129-143 Haase, K. M., C. W Devey, and S L. Goldstein, 1996, Two-Way Exchange between the Easter Mantle Plume and the Easter Microplate Spreading Axis Nature 382, 344-346. Hagen, R. A ., N. A. Baker, D. F. Naar, and R N Hey, A TI Survey of Recent Submarine Volcamsm near Easter Island Manne Geophyszcal Researches 12, 297-315. Hagen R. A., H Vergara, and D. F. Naar, 1996 Morphology of San Antonio Submarine Canyon on the Central Chile Forearc Marine Geology 129 197-205. 214

PAGE 228

Handschumacher, D. W ., 1976, Post-Eocene Plate Tectonics of the Eastern Pacific, GeophysicalMonograph 19, 177. Handschumacher, D. W R. H. Pilger Jr, J. A. Foreman, and J. F. Campbell, 1981, Structure and Evolution of the Easter Plate, Geological Society of America Memoir 154, 63-76. Heestand, R. L. and S. T. Crough, 1981, The Effect of Hot Spots on the Oceanic Age depth Relation, Journal of Geophysical Research 86, 6107-6114. Herron, E. M., 1972a, Sea-floor Spreading and the Cenozoic History of the East-Central Pacific, Geological Society of America Bulletin 83, 1672-1692 Herron, E. M., 1972b, Two Small Crustal Plates in the South Pacific near Easter Island, Nature 240, 35-37. Hey, R., 1977a, A New Class of "Pseudofaults" and Their Bearing on Plate Tectonics: A Propagating Rift Model, Earth and Planetary Science Letters 37 321-325 Hey, R., F K. Duennerbier, and W. J. Morgan, 1980, Propagating Rifts on Midocean Ridges, Journal of Geophysical Research 85, 3647-3658 Hey, R.N., 1977b Tectonic Evolution of the Cocos-Nazca Spreading Center, Geological Society of America Bulletin 88, 1404-1420. Hey, R.N., G. L. Johnson, and A. Lowrie, 1977 Recent Plate Motions in the Galapagos Area, Geological Society of America Bulletin 88, 1385-1403. Hey, R N., P D. Johnson, F. Martinez, J. Korenaga, M L. Somers, Q. J Huggett T. P. Lebas, R. I. Rushy, and D F Naar, 1995, Plate Boundary Reorganization along the Fastest Seafloor Spreading Center, Nature 378, 167-170 Hey, R.N., M. C. Kleinrock, S. P Miller, T. M Atwater, and R. C. Searle 1986, Sea Beam/DeepTow Investigation of an Active Oceanic Propagating Rift System Galapagos 95.5 W, Journal of Geophysical Resear c h 91, 3369-3393. Hey, R. N., H. W. Menard, T. M. Atwater, and D. W. Caress 1988 Changes in Direction of Seafloor Spreading Revisited Journal of Geophysical Research 93, 2803-2811. Hey, R N., D. F. Naar, M. C. Kleinrock, W. J P. Morgan, E. Morales, and J.:G Schilling, 1985, Microplate Tectonics Along a Superfast Seafloor Spreadmg System near Easter Island, Nature 317, 320-325. Ihinger P. D 1995, Mantle Flow beneath the P.acific Evidence from _Seamount Segments in the Hawaiian-Emperor Cham, Amencan Journal of Sczence 295, 1035-1057. Jackson, E D. and H R Shaw, 1975 Stress Field in Central Portions if the Pacific Plate : Delineated in Time by Linear Volcanic Chains, Journal of Geophysi ca l Research 80, 1861-1875. 215

PAGE 229

Klaus, A., W Icay, D. F. Naar, and R. N. Hey, 1991, SeaMARC II Survey of a Propaga?ng I:imb of a Large Non-transform Offset along the Fastest Spreading East Pacific Rise Segment, Journal of Geophysical Research 96, 9985-9998 Kleinrock, M. C., 1992, Capabilities of Some Systems Used to Survey the Deep-sea Aoor, CRC Handbook of Geophysical Exploration at Sea. Boca Raton, CRC Press. Kleinrock, M. C., R. N. Hey, and J. Theberge, A. E., 1992, Practical Geological Comparison of Some Seafloor Survey Instruments, Geophysical Research Letters 19, 1407-1410. Larson, R. L. and p. Olson, 1991, Mantle Plumes Control Magnetic Reversal Frequency, Earth and Planetary Letters 107,437-447. Larson, R L., R. C. Searle, M. C. Kleinrock, H. Schouten, R. T. Bird, D F Naar, R.I. Rushy, E. E. Hooft, and H. Lasthiotakis, 1992, Roller-Bearing Tectonic Evolution of the Juan Fernandez Microplate, Nature 356, 571-576. Levitt, D A. and D. T. Sandwell, 1995, Modal Depth Anomalies from Multibeam Bathymetry: Is There a South Pacific Superswell?, Earth and Planetary Science Letters 139, 1-16. Lewis, B. T. R. and L. M. Dorman, 1970, Experimental Isostasy, 2, An Isostatic Model for the U.S.A. Derived from Gravity and Topographic Data, Journal of Geophysical Research 75, 3367-3386. Liu, Z. J. and D. F. Naar, 1996a, Effective Elastic Thickness of the Lithosphere along the Easter Seamount Chain, Journal of Geophysical Research, under internal review. Liu, Z. J and D. F. Naar, 1996b, Formation of the Easter Seamount Chain and Implications for Deep Earth Structure, Journal of Geophysical Research, submitted. Liu, Z J. and D. F. Naar, 1996c, Side-scan Processing of GLORI-B and SeaBeam 2000, Marine Geophysical Researches under revision. Liu, Z. J. and D. F. Naar, 1996d, Swath Bathymetry Processing of GLORI-B and SeaBeam 2000, Marine Geophysical Researches, under revision. Liu, Z J., D. F. Naar, R. Beale, M. Somers, and C Demoustier, 1994, GLORI-B Data Processing, EOS Trans. AGU 75, 341. Liu, Z. J., D. F. Naar, F. Martinez, and R N. Hey, 1991, SeaMARC II Data Show Easter's SW Rift Propagated Across Its Own Pseudofault, EOS Trans AGU 72, 456. Liu, z J., D F. Naar, Y. Rappaport, S E. Kruse, and R. N. Hey, 1995, Hot Bubbles: Formation of the Easter Seamount Chain, EOS, Trans., AGU 76, 586 Liu, z. J., D. F. Naar, S F. Tebbens, R. N. Hey, and R. A. 1996, of the Southeast Pacific and the Easter Seamount Cham, Journal of Geophyszcal Research, under internal review. 216

PAGE 230

Lonsdale, P 1989, Segmentation of the Pacific-Nazca Spreading Center, 1 N-20 S Journal of Geophysical Research 94, 12197-12225. Louden, K. E., 1981, A Comparison of the Isostatic Response of Bathymetric Features in the North Pacific Ocean and Philippine Sea Geophys. J. R. astr. Soc. 64, 393-424. Louden, K. E. and D. W Forsyth, 1982, Crustal Structure and Isostatic Compensation near the Kane Fracture Zone from Topography and Gravity Measurements -I. Spectral Analysis Approach, Geophys J R Astr. Soc 68, 725-750 Macdonald, K. C and P. J. Fox, 1983, Overlapping Spreading Centers : New Accretion Geometry on the East Pacific Rise, Nature 301, 55 58. Macdonald, K. C. and P. J Fox, 1988 The Axial Summit Graben and Cross-Sectional Shape of the East Pacific Rise and Indicators of Axial Magma Chambers and Recent Volcanic Eruptions, Earth and Planetary S c ien c e Letters 88, 119 131. Macdonald, K. C., P J Fox, S. Miller, S. Carbotte, M. H. Edwards, M. Eisen, D J Fornari, L. Perram R. Po c kalny D Scheirer S. Tighe, C Weiland, and D Wilson, 1992 The East Pacific Rise and Its Flanks 8-18 N : History of Segmentation Propagating Direction Based on SeaMARC II and Sea Beam Studies, Marine G e oph y sical Resear c h e s 14 299-344 Macdonald K C., P J. Fox L. J Perram, M F. Eisen, R. M Haymon, S. P. Miller S M Carbotte, M .H. Cormier, and A. N. Shor, 1988a, A New View of Mid-Ocean Ridge from the Behaviour of Ridge Axis Discontinuities, Nature 335, 217-225 Macdonald, K. C. R. M Haymon, S. P Miller, J C Sempere, and P J Fox, 1988b, Deep-tow and S e a Beam Studies of Dueling Propagating Ridges on the East Pacific Rise near 20 40'S Journal of Geophysi c al Resear c h 93, 2875-2898 Macdonald, K. C S P Miller, S. P Huestis, and F. N. Spiess, 1980, Three Dimensional Modeling of a Magnetic Reversal Boundary from Inversion of Deep Tow Measurements, Journal of Geophysical Research 85, 3670-3680 Macdonald, K. C., D S. Scheirer, and S M. Carbotte, 1991 Mid-Ocean Ridge: Discontinuities, Segments and Giant Cracks, Science 253, 986-994. Macdonald, K C. J. C. Semp e re, and P. J. Fox, 1984 East Pacific Rise from Siqueiros to Orozco Fracture Zon e s : Along-Strike Continuity of Axial Neovolcanic Zone and Structure and Evolution of Overlapping Spreading Centers Journal of Geophysical Research 89 6049-6069. Macdonald, K C. J. C Sempere, P J. Fox and R C. 1987, Tect onic of Ridge-Axis Discontinuities by the Meeting, Leakmg, or Self-Dec apitatiOn of Neighboring Ridge Segments, Geology 15, 993-997 Mammerickx, J ., 1992 The Foundation Seamounts : Tectonic Setting of a Discovered Seamount Chain in the South Pacific, Earth and Planetary S c 1ence Letters 113, 293-306. 217

PAGE 231

Mayes, C. L., L. A Lawver, and D. T. Sandwell, 1990, Tectonic History and New Isochron Chart of the South Pacific, Journal of Geophysical Research 95, 85438567 McKenzie, D. and C. Bowin, 1976, The Relationship between Bathymetry and Gravity in the Atlantic Ocean, Journal of Geophysical Research 81, 1903-1915. McNutt, M 1980, Implications of Regional Gravity for State of Stress in the Earth's Crust and Upper Mantle, Journal of Geophysical Research 85,6377-6396. McNutt, M. K., 1983, Influence of Plate Subduction on Isostatic Compensation in Northern California Tectonics 2, 399-415 Menard, H. W and T. Atwater, 1968, Changes in Direction of Sea-floor Spreading, Nature 219, 463-467. Miller, R. L., F S. Dwan, and C.-F. Cheng 1991, Digital Preprocessing Techniques for GLORIA IT Sonar Images, Geo-Marine Letters 11,23-31. Miller, S. P. and R. N. Hey, 1986, Three-Dimensional Magnetic Modeling of a Propagating Rift, Galapagos 9Y30'W, Journal of Geophysical Research 91, 33953406. Mitchell, N C 1991, Improving GLORIA Images Using Sea Beam Data, Journal of Geophysical Research 96, 337-351 Mitchell, N.C. and M L. Somers, 1989, Quantitative Backscatter Measurements with a Long Range Side-scan Sonar, IEEE J. Oceanic Eng. 14, 368 374. Moore, J. G., D. A. Clague R. T. Holcomb, P. W. Lipman W. R. Normark, and M E. Torresan, 1989, Prodigious Submarine Landslides on the Hawaiian Ridge, Journal of Geophysical Research 94, 17 465-17 484. Morgan, W. J., 1971, Convection Plumes in the Lower Mantle, Nature 230, 197-198. Morgan, W J., 1972, Plate Motions and Deep Mantle Convection, The Geological Society of America 132, 7-22. Morgan, W. J 1978 Rodriguez Darwin, Amsterdam A Second Type of Hotspot Island, Journal of Geophysi c al Researches 83, 5355 5360. Naar, D. F., R. Batiza, R. Poreda, and J.-G. Schilling 1993a, Final Cruise Report for the RJV Melville Gloria Expedition Legs 6 and 7. Naar, D. F. and R.N. Hey, 1986, Fast Rift Propagation along the East Pacific Rise near Easter Island, Journal of Geophysical Researches 91, 3425-3438 Naar, D. F. and R.N. Hey, 1989a, Recent Pacific-Easter-Nazca Plate Motions, Evolution of Mid Ocean Ridges, IUGG Symposium 8. Washington, AGU Geophysical Monograph. 218

PAGE 232

Naar, D F. and R.N. Hey, 1989b, Speed Limit for Oceanic Transform Faults, Geology 17' 420-422. Naar, D. F. and R.N. Hey, 1991, Tectonic Evolution of the Easter Microplate, Journal of Geophysical Researches 96, 7961-7993. Naar, D. F., Z. J. Liu, Y. Rappaport, R. Batiza, R. Hagen, R. Hey, R. Nelson, T. Plake, R. Stefani, J.-G. Schilling, C Kincaid, G. Xu, R. Poreda, L. Joseph, C. Jacobs, R. Beale, D. Bishop, A. Harris, R. Rushy, D. Fontignie, A. Woods, S. Kruse, J Korenaga, N. Seama, H. Vergara, and R Guarda, 1993b, GLORI-B and Geochemical Investigation of the Easter Seamount Chain: EPR to San Ambrosio Island, EOS, Transactions, AGU 74, 672 National Geophysical Data Center, 1988, ETOP0-5 bathymetry/topography data, Data Announc. 88-MGG-02, Natl Oceanic and Atmos. Admin., U .S. Dep Commer., Boulder, Colo Norton, I. 0 1995, Plate Motions in the North Pacific: The 43 Ma Nonevent, Tectonics 14, 1080-1094. O'Connor, J. M., P. Stoffers, and M. 0. Mcwilliams, 1995, Time-space mapping of Easter Chain volcanism, Earth and Planetary Science Letters 136, 197-212. Okal, E. A. and A. Cazenave, 1985, A Model for the Plate Tectonic Evolution of the East central Pacific Based on SEASAT Investigations, EPSL 72, 99-116. Parker, R. L., 1972, The Rapid Calculation of Potential Anomalies, Geophys. J. R. astr. soc. 31, 447-455. Parker, R. L. and S. P. Huestis, 1974, The Inversion of Magnetic Anomalies in the Presence of Topography, Journal of Geophysical Research 79, 1587-1593. Paskevich, V., 1992, Digital Mapping of Side-scan Sonar Data with the Woods Hole Image Processing System Software, United States Department of the Interior Geological Survey Open-File Report 92-536. Perram, L. J., M.-H. Cormier, and K. C. Macdonald, 1993, Magnetic and Tectonic Studies of the Dueling Propagating Spreading Centers at 20" 40'S on the East Pacific Rise: Evidence for Crustal Rotations, Journal of Geophysical Research 98, 13835-13850. Pilger, R. H., Jr. and D. W. Handschumacher, 1981, The Fixed h?tspot a_nd the Origin of the Easter-Sala y GomezNazca Trace, Geologzcal Soczety of Amerzca Bulletin 92, 437-446. Pollard, D. D., P Segall, and P. T. Delaney, 1982, Formation and Interpretation of Dilatant Echelon Cracks, Geological Society of America Bulletin 93, 1291-1303. Poreda, R. J., J.-G. Schilling, R. Batiza, and D. Naar, 1993a, Geochemistry of Volcanism along the Easter Seamount Chain, EOS Trans. AGU 74,672. 219

PAGE 233

Poreda, 1 J G. Schilling, and H Craig, 1993b Helium Isotope Ratios in Easter Microplate Basalts Earth and Planetary Scien c e Letters 119 319-329 Rea, D. K., 1981, Tectonics of the Nazca-Pacific D i vergent Plate Boundary, Mem. Geol. Soc Am. 154, 27-62. Ribe, N M and A. B. Watts, 1982, The Distribution of Intraplate Volcanism in the Pacific Ocean Basin: A Spectral Approach, Geophys. J R astr Soc 71 333-362. Richards, M. A. and C. Lithgow-Bertelloni, 1996, Plate Motion Changes, the HawaiianEmperor Bend, and the Apparent Success and Failure of Geodynamic Models, Earth and Planetary Science Letters 137, 19-27 Richardson, W. P., S. Stein C A. Stein, and M. T Zuber, 1995 Geoid Data and Thermal Structure of the Oceanic Lithosphere, Geophysical Research Letters 22, 1913 1916. Richter, F. M and B. Parsons 1975, The Interaction of Two Scales of Convection in the Mantle, Journal of Geoph y sical Re se arch 80 Rushy, R. 1., 1992, Tectonic Pattern and Evolution of the Easter Microplate, Based on GLORIA and Other Geophysical Data, Ph D Dissertation, University of Durham. Rushy, S., 1970, A Long Range Sidescan Sonar for Use in the Deep-sea, Int. Hydrogr. Rev. 47, 25 39. Sandwell, D. T., 1992, Antarctic Marine Gravity Field from High Density Satellite Altimetry, Geophys J Int 109, 437-448 Sandwell, D. T E. L. Winterer, J. Mammerickx, R. A. Duncan, M A. Lynch, D. A Vevitt, and C L Johnson 1995, Evidence for Diffuse Extension of the Pacific Plate from Pukapuka Ridges and Cross-grain Gravity Lineations Journal of Geophysical Research 100, 15087-15099. Scheirer, D S and K. C. Macdonald 1993, Variation in Cross-sectional Area of the Axial Ridge along the East Pacific Rise : Evidence for the Magmatic Budget of a Fast Spreading Center, Journal of Geophysical Research 98,7871-7885 Scheirer, D. S., K C. Macdonald, D. W Forsyth, and Y Shen, 1996 Abundant Seamounts of the Rano Rahi Seamount Field near the Southern East Pacific Rise, 15 S to 19"S, Marine Geophysical Researches 18, 13-52 Schilling, J.-G. and A. Noe-Nygaard, 1974, Faeroe-Iceland Plume : Rare-Earth Evidence, Earth and Planetary Sci e nce Letters 24, 1-14 Schilling, J .-G. H. Sigurdsson A. N Davis and R. N. Hey, 1985a Easter Microplate Evolution, Nature 317,325-331. Schilling, J G., G. Thompson, R Kingsley, S Humphris, 1985b Hotspotmigrating Ridge Interaction m the South Atlanuc, Nature 313, 187-191 220

PAGE 234

H., _K. D. Klitgord, and D. G Gallo, 1993, Edge-Driven Microplate Kinemat.tcs, Journal of Geophysical Research 98, 6689-6701. Searle, R. C., 1992, Near Real-time Merging of GLORIA and Hydrosweep Pseudo Sidescan Image, Geophys. Res. Lett. 19, 1137-1140. Searle, R C., R. T Bird, R I. Rushy, and D. F Naar, 1993, The Development of Two Oceanic Microplates : Easter and Juan Fernandez Microplates East Pacific Rise, Geological Society of London Joumal150, 965-976 Searle, R. C., J. Francheteau, and B. Comaglia, 1995, New Observations on Mid-plate Volcanism and the Tectonic History of the Pacific Plate, Tahiti to Easter Microplate, Eanh and Planetary Science Letters 131, 395-421. Searle, R. C., T. P. Lebas, N.C. Mitchell, M L. Somers, L. M Parson, and P. Patriat 1990, GLORIA Image Processing: the State of the Art, Marine Geophysical Research 12, 21-39. Searle, R. C., R. I. Rushy, J. Engeln, R. N. Hey, J. Zukin, P. M Hunter, T P LeBas H.-J Hoffman, and R Livermore, 1989, Comprehensive Sonar Imaging of the Easter Microplate, Nature 341, 701-705 Sempere, J .C., J Gee, D. F. Naar, and R N. Hey, 1989, Three-Dimensional Inversion of the Magnetic Field Over the Easter-Nazca Propagating Rift Near 25 S, 112 25'W, Journal of Geophysical Research 94, 17409-17420 Shemenda, A. I and A. L. Grokholsky, 1991, A Formation and Evolution of Overlapping Spreading Centers (Constrained on the Basis of Physical Modelling), Tectonophysics 199, 389-404. Smith, W. H. F. and D T. Sandwell, 1994, Bathymetric Prediction from Dense Satellite Altimetry and Sparse Shipboard Bathymetry, Journal of Geophysical Research 99, 21803-21824 Smith, W. H. F. and D. T Sandwell, 1995a, Marine gravity field from declassified Geosat and ERS-1 altimetry, EOS, Trans AGU 76, 156. Smith, W H. F and D. T. Sandwell, 1995b, Oceanographic "pseudogravity" in marine gravity fields derived from declassified Geosat and ERS-1 altimetry, EOS, Trans ., AGU 76, 151. Smith, W. H F. and P. Wessel, 1990, Gridding with Continuous Curvature Splines in Tension, Geophysics 55, 293-305. Somers M L., R. M Carson, J. A. Revie R. H Edge B. _J. Barrow, A G. Andrews 1978 Gloria II --An Improved Long Range S1descan Sonar., m Proc. IEEE/IERE s'ub-conference on Offshore Instrumentation, Oceanology International'78, Technical Session J. London, BPS Publications, Ltd., London, V. 16-24. 221

PAGE 235

Somers, M. L. and D. Q. J. Hugget, 1993, From GLORIA to GLORI-B: New Upgrades Add Swath Bathymetry to 20-year-old GLORIA Reconnaissance Sonar Images, Sea Technology June, 64-68. Stein C. A and S. Stein, 1992, A Model for the Global Variation in Oceanic Depth and Heat Flow with Lithospheric Age, Nature 359, 123-129. Stoffers, P. and R. Hekinian, 1990, Hot Spot Volcanism in the Central Southpacific, Geomet Cruise Report 40 Stoffers, P., R. Hekinian, K. M. Haase, and Scientific Party, 1994, Geology of Young Submarine Volcanoes West of Easter Island, Southeast Pacific, Marine Geology 118, 177-185. Tackley, P. 'J., D. J. Stevenson, G. A Glatzmaier, and G. Schubert, 1993, Effects of an Endothermic Phase Transition at 670 km Depth in a Spherical Model of Convection in the Earth's Mantle, Nature 361,699-704. Tebbens, S. F. and S.C. Cande, 1996, Southeast Pacific Tectonic Evolution from Early Oligocene to Present, Journal of Geophysical Research, in press. Tebbens, S. F S.C. Cande L. Kovacs, J. C. Parra, J L. LaBrecque, and H. Vergara, 1996, The Chile Ridge: A Tectonic Framework, Journal of Geophysical Research, in press. Turcotte, D. L and G. Schubert, 1982, Geodynamics, John Wiley & Sons, New York, p. 449. Tyee, R. C., 1987, Deep Seafloor Mapping System--A Review, MTS Journal20, 4-16 Unidata Program Center, 1991, NetCDF User's Guide : An Interface for Data Access, University Corporation for Atmospheric Research. Vogt, P.R. and B. E. Tucholke, 1986, Imaging the Ocean Floor: History and State of the Art, The Western North Atlantic Region. Boulder, Geological Society of America, Inc. Walcott, R.I. 1970a, Flexure of the Lithosphere at Hawaii, Tectonophysics 9, 435-446 Walcott, R. I., 1970b, Flexure Rigidity, Thickness and Viscosity of the Lithosphere, Journal of Geophysical Research 75, 3941-3954. Wessel, P., 1992, Thermal Stresses and the Bimodal Distribution of Elastic Thickness Estimates of the Oceanic Lithosphere, Journal of Geophysical Research 97, 1417714193. Wessel, P. and W. H. Smith, 1991, Free Software Helps Map and Display Data, EOS Trans AGU 72, 5-446. Whitehead, J. A., 1982, Instabilities of Fluid Conduits in a Flowing Earth-Are the Lubricated by the Asthenosphere?, Geophysical Journal of the Royal Astronomzcal Society 70, 415-433. 222

PAGE 236

Whitehead, J. A. and D. S Luther, 1975, Dynamics of Laboratory Diapir and Plume Models Journal of Geophysi ca l Resear c h 80, 705-717 Willis, B. and H. S. Washinton 1924 San Felix and San Ambrosio: Their Geology and Petrology, Bullotin of the Geological Society of America 35, 365-384 Wilson, D. S., 1996, Fastest known spreading on the Miocene Cocos-Pacific plate boundary, Geophysical Research Letters, in press Wilson, D. S and R. N Hey, 1995, History of Rift Propagation and Magnetization Intensity for the Cocos-Nazca Spreading Center, Journal of Geophysical Research 100, 10041-10056. Wilson, J. T., 1963a, Evidence from Island on the Spreading of Ocean Floors, Nature 197' 536-538. Wilson, J. T., 1963b, A Possible Origin of the Hawaiian Islands, Canadian Journal of Physics 41, 863-870. Wilson, J. T., 1965, Convection Currents and Continental Drift, Philos op hical Transaction of the Royal Societs of London A. 258, 145-167 Wilson, J. T. 1973, Mantle Plumes and Plate Motions, Tectonoph ys i cs 19, 149-164 Winterer, E L and D. T. Sandwell, 1987, Evidence from En-echelon Cross-grain Ridge for Tensional Cracks in the Pacific Plate, Nature 329, 534-537. Wolfe, C. J. and M. K. McNutt, 1991, Compensation of Cretaceous Seam o unts of the Darwin Rise, Northwest Pacific Ocean, Journal of G eophysical Resear c h 96, 2363-2373. Woods, A., E. Plank, S. Kruse, D. F. Naar, and Z J. Liu, 1993, Compensation of Seamounts along the Easter Seamount Chain, EOS Trans. AGU 74,687. Woods, M. T. and E A. Okal, 1994, The Structure of the Nazca Ridge and Salay Gomez Seamount Chain from the Dis persion of Rayleigh Waves Geophysical J o urnal Int e rnational 117, 205-222. Zukin, J. and J. Francheteau 1990, A Tectonic Test of Instantan eous Kinematics of the Easter Microplate, O ce anoligi c a Acta spec 10, 183-198. 223

PAGE 237

APPENDICES 224

PAGE 238

APPENDIX 2.1. IMAGE PROCESSING PACKAGE Below is a list of the programs in IP (Image Processing) package with short description of the functions for each of the programs. An on-line help is provided to help users for the usage when the program is entered without any parameter or with only parameter -H. The author of each program in the list is indicated by the small character to the upper right of each program name. @ 6 avg_heading" dxj2whips" flip" Z. Liu. P. Wessel and W. H. F. Smith. V Paskevich. R. Hagen et al. smooth the headings by applying a running average to the heading values contained in the header of a CDF ftle display a selected region of a grid in an IMG file on an X-window. A color palette can be specified to the program to display a pseudo color image instead of gray scale image convert navigation data from DXF format to WHIPS format for merging into GLORI-B image data apply a low-pass, high-pass, zero replacement or divide filter to an image flip swaths in a CDF file. convert a file of GRD format to or from IMG format If required by users, it can rescale the values in the GRD file to the range 0-255. 225

PAGE 239

APPENDIX 2.1. (Continued) grdinfoc gsse gss_ve[e gssve histequ[e img2pse imgfiltere imghistgram* imgmath* imgmeane imgmodee imgmos* imgpastee imgscale* list header information of an IMG file compute geographic coordinates for the cells in a CDF file compute geographic coordinates for the cells in a CDF ft.le and apply the necessary velocity correction to regulate the aspect ratio. compute geographic coordinates for the cells from a velocity corrected CDF file apply a histogram equalization to an IMG file. produce a postscript image map from an IMG file extract a sub-region from an IMG file apply a boxcar, cosine arch, gaussian, or median fllter to an IMG file. calculate a histogram from an IMG me. apply one of the 17 operations on two IMG files at each cell to produce a new grid of IMG file. smooth an image using an averaging boxcar. The boxcar may orient in any direction specified by users. apply a median fllter to an IMG ft.le. apply one of the 4 operations to mosaic two IMG files at each cell to create a new IMG ftle paste two IMG flies with a common border to produce a large grid. rescale DNs according to two pairs of numbers. Each pair divides the 0-255 range into three sectio ns Each section is stretched or compre sse d to match the corresponding section from the other pair set cells of a certain value in an IMG me to another value. 226

PAGE 240

APPENDIX 2.1. (Continued) listhdr lowpass2b2'* median3'* minmaxdn'* mode3'* mode5'* mrgnav'* quickview'* rawps2whips'* restorehdr'* sbnav2cJx.r sbss2cd.f' shade'* shade3'* shade3gen'* slant'* sshead'* sumpror set the cells on boundaries of data patches to 0 A boundary cell is defmed as the a cell with at least one neighbor cell of value 0 list the contents of a header in a CDF file. apply a 2-by-2 low pass filter to a CDF file apply a 3-by 3 median filter to a CDF file compute the min and max pixel value values contained in a CDF file. apply a 3-by-3 mode filter to a CDF file. apply a 5 by-5 mode filter to a CDF file merge navigation data into a CDF file display an 8-bit image in a CDF file on an X-window convert a file of raw GLORI-B image data format to CDF format restore header to a CDF file from another image file convert navigation data from SeaBeam format to DXF format convert a file of SeaBeam sonar image data format to a number of CDF files Each of the files has a constant resolution. apply a simple shading correction to an 8-bit CDF image file apply a shading correction to an 8-bit CDF image using a shading profile shade3.dat in the current directory. generate a shading profile from a CDF file. correct CDF image for slant-range geometry distortions compute a simple heading for a CDF file using navigation information. accumulate the shading profiles (shade3 dat) to shades.dat for calculating an average shading profile. 227

PAGE 241

APPENDIX 2.1. (Continued) wtcombo8 xyz2img* swnmarize the infonnation contained in the header of a CDF file. correct CDF image for changes in ship's velocity and any aspect ratio distortions that exist combine 2-4 WlllPS images with weight. convert image triplets to an IMG file. If there is more than one data points in a cell, the average value is used for the cell. 228

PAGE 242

APPENDIX 2.2. IMAGE PROCESSING SCRIPT #!lbinlcsh -f #Processing SeaBeam Image Data set grid set range set datdir I s -1 awk set files set log set outdir mkdir echo "" = 0.001 = -110/ -103/-2 9/-24.5 = /disk3/SSS/GLOR06MV/SSDAT NSSMED $datdir/SSmed.93* I \ '{print$8}' > tmp_list.asc = 'cat tmp_listasc = resolution.asc = cdf $outdir > $log #Converting image data to CDF files @ i = 1 while ($i <= $#flles) 229

PAGE 243

APPENDIX 2.2. (Continued) sbss2cdf @ i ++ end -i $files[$i] -o $outdir/$files[$i] >> $log #Creating shading proflle Is -1 awk set files @ i = 1 $outdir/* I \ '{print $8}' > tmp_list.asc = 'cat tmp_listasc while ($i <= $#files) shade3gen sum prof @ i ++ end awk awk set files awk set resolu -i $files[$i] -t 1 ,255 '{print $1/$#files}' shades dat > shade3 dat '{print $1 } > tmp_list.asc = 'cat tmp_listasc' '{print $3 }' > tmp_list.asc = 'cat tmp_listasc 230

PAGE 244

APPENDIX 2.2. {Continued) set avgn set discr set width set sbimg echo '"' xyz2img @ i = 1 = 5 = degreeldegree/0-255/1/0/GLORIA06/SB_sidescan = 15000 = sbss6.img I \ -G$sbimg -I$ grid -R$range -D$discr V while ($i <= $#flies) #Shading the image shade3 -i $files[$i] -o tmp_ shd.cdf #Removing striping noises filter filter wtcombo restorehdr -i tmp_shd cdf o tmp_hpf.cdf -b 1,71 -h -v 1 ,2 53 -i tmp_shd .cdf -o tmp_lpf.cdf b 9 ,71 -1 -v 1,253 -i tmp_hpf cdf,tmp_lpf.cdf -o tmp_wco.cdf -c 1,1 -a -128 -i tmp_ wco.cdf h tmp shd .cdf 231

PAGE 245

APPENDIX 2.2. (Continued) #Smoothing heading avg_heading -i tmp_wco .cdf -o tmp_avg.cdf -1 $avgn #Flipping th e swaths flip restorehdr -i tmp_avg .cdf -o tmp_flp.cdf -i tmp_flp.cdf -h tmp_avg .cdf #Converting from CDF to IMG format gss_vel xyz2img -i tmp_flp .cdf -r $resolu[$i] -d $width I \ -Gtmp_dat.img -I$grid -R$range -D$discr -: -V #Mosaicking the images into a single grid imgmos @ i ++ end -A$sbimg -Btmp_datimg -G$sbimg -M2 232

PAGE 246

APPENDIX 2.2. (Continued) #Processing GLORI-8 Image Data set datdir 1s -1 awk set files echo '"' = /data5/g1or6data/raw_nav6 $datdir/GLOR06merge 93* I \ '{print $8}' > tmp_li s t.asc = cat tmp_1i s t.asc > tmp_n a vigation.dxf #Preparing navigation data @ i = 1 while ($i <= $#flie s) sbnav2dxf cat @ i++ end dxf2whips set datdir ls 1 -i $fll e s[$i] -o tmp_day dxf tmp_d a y dxf >> tmp navigati o n.dxf -i tmp_n a vi g ati o n dxf -o navigation whips = /data6/mlvl0693/img/rawps dat $datdir/rawps* I \ 233

PAGE 247

APPENDIX 2.2. (Continued) awk set files set avgn set discr set width set resolu set glimg echo 1111 xyz2img @ i = 1 '{print $8}' > tmp_list.asc = 'cat tmp_list.asc' = 41 = degree/degree/0-255/1/0/GLORIA06/IMAGE = 16000 = 50 = glor6.img I \ -G$glimg -I$grid -R$range D$discr -V while ($i <= $#files) cp $file[$i] tmp_dat.rawps.Z uncompress tmp_dat.rawps Z #Converting image data to CDF format rawps2whips -i tmp_dat.rawps -o tmp_whips .cdf #Incorporating the navigation into the image data files mrgnav -i tmp_whips.cdf -o tmp_mrg .cdf -n navigation whips -f 234

PAGE 248

APPENDIX 2.2. (Continued) sshead avg_heading -i trnp_mrg.cdf -o trnp_ssh.cdf -i tmp_2b2.cdf -o tmp_avg.cdf -1 $avgn #Computing the true ground ranges slant -i trnp_ssh.cdf -o trnp_slt.cdf #Shading the image shade3 -i trnp_slt.cdf -o trnp_shd.cdf #Removing striping noises filter filter wtcombo restorehdr -i trnp_shd.cdf -o trnp_hpf .cdf -b 1,71 -h -v 1,253 -i trnp_shd .cdf -o trnp_lpf.cdf -b 9,71 -1 -v 1,253 -i trnp_hpf.cdf,tmp_lpf.cdf -o tmp_ wco.cdf -c 1,1 -a -128 -i trnp_ wco.cdf -h tmp_shd.cdf #Correcting the aspect-ratio velocity -i tmp_wco.cdf -o trnp_vel.cdf 235

PAGE 249

APPENDIX 2.2. (Continued) lowpass2b2 restorehdr -i trnp_ vel.cdf -o tmp_2b2.cdf -i tmp_2b2.cdf -h trnp_ vel.cdf #Registering the image into grids gssv xyz2img -i tmp_avg.cdf -r $resolu -d $width I \ -Gtmp_dat.img -I $grid -R$range D$discr : -V #Mosaicking the IMG flies into a single grid imgmos @ i++ end -A$glimg -Btmp_dat.img -G$glimg M2 #Merging The Image Data #Calculating histograms imgshrink imgshrink imgshrink imgshrink $sbimg -Gtmp_sb.img tmp_sb.img -Gtmp_sb.img tmp_sb.img -Gtrnp_mask.img tmp_mask.img -Gtrnp_mask.img 236

PAGE 250

APPENDIX 2.2. (Continued) imgshrink imgshrink imgshrink imgmath imgmath imgmath imghistgram imgmath imghistgram tmp_mask.img -Gtmp_mask.img tmp_mask.img -Gtmp_mask.img tmp_mask.img -Gtmp_mask img tmp_sb.img X tmp_mask.img = tmp_mask .img tmp_mask.img 0 $glimg = tmp_mask.img $sbimg 0 tmp_mask.img = tmp_ s bmk.img tmp_sbmk .img > sb_histgram.asc $glimg 0 tmp_mask.img = tmp_glmk.img tmp_glmk.img > gl_hi s tgram asc #Merging the ima ges imgscale imgscale imgmos rm #end of the script tmp_sb.img Gtmp_sb.img -S12/20/161/220 $glimg -Gtmp_gl.img -S41.5/20/155/220 -Atmp_gl.img -Btmp_sb.img -Gglor+sb.img -Ml tmp_* 237

PAGE 251

APPENDIX 3.1. BATHYMETRY PROCESSING PACKAGE The programs of the BP (Bathymetry Processing) package are listed with a short description of the functions for each of the programs An on-line usage is provided when the program is entered without any parameter. The authors are indicated by the small character to the upper right comer of each program name. @ 0 avgpror dxj2whips,. Z. Liu P. Wessel and W H. F. Smith. V. Paskevich D Scheirer et al calculate headings from the navigation data and smooth them if required It also converts navigation data file from WHIPS format to a format ready to merge with GLORI-B bathymetry data smooth a profile using a mean or median filter. The input and output profiles are in ASCII format of two columns (x y) pairs calculate an average proflle f r om a number of SeaBeam or GLORI-B cross-track residual profiles convert navigation data from DXF format to WHIPS format for merging with GLORI-B image data compute longitude and latitude for each data point in raw GLORI-B bathymetry data fll.es using navigation data in the format provided by avgnav and write the bathymetry data points as 3-column triplets into ASCII files 238

PAGE 252

APPENDIX 3.1. (Continued) glorpror grd2xyzc grdclipc grdcutc grdfilterC grdinfoc merge navigation data (calculated from avgnav ) into raw GLORI-B bathymetry data while converting to RAS format. Each ping is rectified to 601 bins at a bin size of 50 m If there is more than one point within a bin, the average depth is used for the bin calculate an average residual cross-track profile or a set profiles for different depths from a number of raw GLORI-B bathymetry data files. convert data from GRD format to 3-column triplets (longitude, latitude, and depth). set the cell value above a given value to another value and/or set the cell value below a given value to another value. extract a sub-grid from a GRD file. filter a GRD file using a boxcar, cosine arch, gaussian, or median filter, and compute distance by Cartesian or Spherical geometry. set cells within given boxes (defined by lower left and upper right comers) to NaN. The comer pairs i s provided in an ASCII file. list the header information of a GRD file. interpolate a grid along Xand Y -directions for each row and column, and write th e Xand Y-component t o two GRD files, respectively create a grid and set cells inside outside, or on the polygons (given in an ASCII file) to different values 239

PAGE 253

APPENDIX 3.1. (Continued) grdmath0 grdpaste0 grdsample0 perform a number of mathematical operations to each pair of the cells between two GRD files or a GRD file and a number. This is a modified version from the same GMT program by adding four other functions. calculate an average value for each cell centered within a given boxcar. The axes of the box can be aligned to any direction If the axes are parallel to grid, a faster algorithm is used similar to grdmean but searches for a median value for each cell centered within a given boxcar merge two GRD files into a new grid Each cell in the new grid is calculated from the corresponding pair of the two input grids. The choices of operations are averaging, using the second values if it is not a NaN, using the larger values, or using the smaller values. paste two GRD files with a common border to produce a large GRD file paste a small grid onto a larger grid. This program is useful to merge a small portion to a large GRD file resample a GRD file to produce a new grid of different grid size (usually finer grid size). set the cells of a certain value to another value. These values are specified by users. set the boundary cells of data patches to NaN. A boundary cell is defmed as a cell with at least a neighbor of value NaN. 240

PAGE 254

APPENDIX 3.1. (Continued) rasjilter raspro,? sb2llz!' put a surface on a given grid by interpolating values for the cells of value NaN Modified from a GMT program surface to read a GRD file direct! y. set cells of small data patches and stripes to NaN. The size of patches or stripes is limited in any direction by a threshold of cell number specified by users. set cells with an outward slope greater than a given threshold to NaN and remaining cells to 1. This produces a mask for clipping dropped edges remove a given number of pings from the beginning and the end of a RAS file similar to grdfilter but take RAS flies calculate the outward slope of the data swath in a RAS file. That is the slope facing away from the center track of each ping. calculate geographical coordinates for each point of bathymetry data in RAS format and write out the bathymetry triplets to ASCII flles similar to grcinulth, but operate on RAS files compute an average residual cross-track profile from a RAS file and subtract the data swath using the profile to reduce the cross-track biases in GLORI-B bathymetry data calculate longitude and latitude for each bathymetry data point in files of SeaBeam format and write the triplets of the bathymetry points to ASCII or BINARY files. 241

PAGE 255

APPENDIX 3.1. (Continued) sbnav2dx.r sbpro.P convert navigation data from SeaBeam format to DXF format. The ship navigation data is stored in the SeaBeam format calculate an average cross-track residual profile from a SeaBeam bathymetry data flle. register bathymetry triplets into a grid in GRD format. This program is similar to the GMT program xyz2grd. The difference is that if there is more than one point within a cell, the average value is used for the cell instead the last value as used in xyz2grd 242

PAGE 256

APPENDIX 3.2. BATHYMETRY PROCESSING SCRIPT #!/bin/csh -f #Processing SeaBeam 2000 Bathymetry data set datdir Is -1 awk set files set tmpdir set range set orlat = /data5/glor6data/seabeam6/SBDATNsbf $datdir/* I \ '{print $8}' > tmp_list.asc = 'cat tmp_listasc' = tmp_prof = -115/ 103/-29/-24.5 = 27 #Correcting cross-track bias mkdir $tmpdir @ i = 1 while ($i <= $#flies) sbprof $flles[$i] V mv avg.proflle $tmpdir/prof_$i @ i ++ end 243

PAGE 257

APPENDIX 3.2. (Continued) Is -1 awk avgsbpf set grid set disc set sbgrd $tmpdir/* I \ I {print $8} I > tmp_list.asc $tmp_list.asc > sb_prof asc = 0.003 = "degree/degree/meter/1/0/SB _DATA/ESC_ W est_Mosiac" = SeaBeam.grd #Converting data to GRD format echo "" > SeaBeam.llz @ i = 1 while ($i <= $#files) sb2llz blockmedian @ i ++ end xyzmgrd grdmath $files[$i] -R$range -Csb_prof.asc -O$orlat -A V I \ -R$range -I$ grid V >> SeaBeam. liz SeaBeam llz R$range -I$ grid G$sbgrd -D$disc V $sbgrd N = $sbgrd 244

PAGE 258

APPENDIX 3.2. (Continued) #Filling The Topography Data Of The Islands Into The SeaBeam 2000 Grid set easter set salasr set eastdata set saladata set disc xyzmgrd grdpatch xyzmgrd grdpatch = -110/-109/-27.5/-26.5 = -106/-105/-27/-26 = /data5/pubdat/other/eastxyz = /data5/pubdat/other/sala xyz = II degree/ degree/meter/ 1/0/ISLAND/ESC_ West_Mosiac II $eastdata -R$easter -I$grid -Gtmp_island grd -D$disc V $sbgrd -Atmp_island.grd -G$sbgrd -M5 -V tmp.xyz $saladata -R$salasr -I$ grid -Gtmp_island grd -D$disc V $sbgrd -Atmp_island grd -G$sbgrd -M5 -V #Processing GLORI-B Bathymetry Data set datdir ls -1 awk set files = /data5/glor6data/raw_nav6 $datdir/GLOR06merge.93* I \ I {print $8} I > tmp_list.asc = 'cat tmp_listasc' 245

PAGE 259

APPENDIX 3.2. (Continued) #Preparing navigational data echo "" > tmp_navigation .dxf @ i = 1 while ($i <= $#files) sbnav2dxf cat @ i++ end dxf2whips avgnav -i $flles[$i] -o tmp_day dxf tmp_day dxf >> tmp_navigation .dxf -i tmp navigation .dxf o navi g ati o n whips navigation.whip s -A5 V > navigation asc #Converting into RAS format set datdir Is 1 awk set files set disc set gbgrd = /data6/mlvl0693/bath/xyz $datdir/* I \ '{print$8}' > > tmp_list.asc = cat tmp_list.asc = "d e gree/d e gree/meter/ 110/GWRIA/bathym e try" = GLORIA.grd 246

PAGE 260

APPENDIX 3.2. (Continued) glormras $files Ny/navigation.asc -T90 -V > GLORIA.ras #Reducing random noises and filling small holes rasfllter rasfllter GLORIA.ras -Fm5/5 V > tmp_5X5.ras tmp_5X5.ras -Fb20/l0 V > tmp_fil.ras #Cutting dropped edges rascull rasmath tmp_fil.ras -S25 V > tmp_mask ras tmp_5X5.ras x tmp_mask.ras = tmp_cull ras #Removing cross-track biases rasprof tmp_cull.ras V > tmp_prof.ras #Reducing along track biases rasfllter rasfllter tmp_prof.ras -Fb200/200 V > tmp_avg.ras tmp_prof ras -Fml/200 -V > tmp_trk.ras 247

PAGE 261

APPENDIX 3.2. (Continued) rasmath rasmath tmp_trk.ras tmp_avg ras = tmp_bias.ras tmp_prof.ras tmp_bias.ras = tmp_filtered.ras #Removing duplicated pings rascut tmp_filtered.ras -C200 V > tmp_cut.ras #Converting data from RAS format to GRD format rasmllz xyzmgrd grdmath tmp_cutras V I \ -G$gbgrd -I$ grid -R$range D$disc V $gbgrd N = $gbgrd #Trimming the data grdtrim grdshrink grdshrink grdholing $gbgrd -Gtmp_trim.grd -N5 V tmp_trim.grd -Gtmp_shk grd V tmp_shk.grd -Gtmp_shk.grd V tmp_shk grd -Gtmp_hole.grd -Oholes.xy V 248

PAGE 262

APPENDIX 3.2. {Continued) #Clipping speckles grdmean grdmath grdclip grdmath tmp_hole grd -Gtmp_mean.grd V tmp_hole.grd tmp_mean.grd = tmp_diff grd tmp_diff.grd -A200/NaN -B-200/NaN -Gtmp_mask.grd -V $gbgrd 0 tmp_mask.grd = clip.grd #Calibrating GLORI-B data using the SeaBeam grid grdfllter grdmath grdsurface grdmath grdmean grdmath $sbgrd -DO -FgO.l -Gtmp_trend grd -V tmp_trend.grd 0 $sbgrd = tmp trend grd tmp_trend.grd -Gtmp_trend.grd T0.8 V clip.grd tmp_trend.grd = tmp_diff.grd tmp_diff.grd -A0/50/50 -Gtmp_dftd.grd V clip.grd tmp_dftd.grd = tmp_adjt.grd #Merging GLORI-B and SeaBeam grids grdmath grdmean grdmath grdmath $sbgrd A tmp_adjtgrd = tmp_mrg.grd tmp_mrg.grd -A0/5/3 -Gunp_fitgrd V unp_fit.grd 0 clip .g rd = tmp_masked.grd $sbgrd A tmp_masked grd = mrg grd 249

PAGE 263

APPENDIX 3.2. (Continued) #Interpolating the merged data to fill the data gaps grdsurface mrg grd -Gsurf.grd T0.25 V #Reducing the along-track bias further grdmedian grdmean grdmath grdmath surf.grd -A 701100/5 -Gtmp_trk grd V surf grd -A0/1001100 -Gtmp_avg.grd V tmp_trk.grd tmp_avg.grd = tmp_bias.grd surf.grd tmp_bias.grd = filtrk grd #Recovering the SeaBeam 2000 data grdmath grdsurface grdmath grdmath filtrk.grd $sbgrd = tmp_dif.grd tmp_dif.grd -Gtmp_surf.grd T0.5 V filtrk.grd tmp_surf.grd = tmp_match grd $sbgrd A tmp_match grd = glor+sb.grd #Filling Remaining Area of the Grid set etopo = /data5/pubdat/etopo5/etopo5 grd 250

PAGE 264

APPENDIX 3.2. (Continued) set seamarc set west grdcut grdsamp1e grdcut grdsamp1e grdpatch grdmath #end of the script = /data5/pubdat/other/seamarc.grd = 115/-109/-29/-25 $etopo -Gtmp_etopo grd -R$range -V tmp_etop.grd -Gtmp_etop grd I$ grid -V $seamarc -Gtmp_sea grd -R$we s t -V tmp_sea grd -Gtmp_sea grd -I$grid -V tmp etop.grd -Atmp_sea grd -Gtmp_res.grd -V glor+sb.grd A tmp_res grd = e i gs_final.grd 251

PAGE 265

APPENDIX 4.1. IMPLICATIONS FOR DEEP EARTH STRUCTURE Activities inside of the Earth such as convections and advections, are controlled by unevenly distributed heat, or th e earth's thermal budget. Heat arrives at the earth's surface from its interior and from the sun Virtually all the heat coming from the sun is eventually radiated back into space The mean rate of heat flux from the Earth to space is about 4 x 1013 W (or 8 x 1Q-2 W/m2) (Fowler, 1990) A more conservative estimate of the figure is 3.55 x 1013 W (7 x 1Q-2 W/m2) (Turcotte and Schub e rt, 1982) The radioactive dacay of abundant isotopic elements results in a total radioactive heat production for the crust and mantle of 1.4 x 10132 7 x 1013 W, with a best guess value of 2.1 x 1013 W (Fowler 1990). There is a net loss of the Earth's internal heat of 1.45 x 1013 -1.9 x 1013 W Therefore, the Earth has been experiencing a long period of cooling since its formation The dominant cooling system controls that the Earth's heat is transferred from the interior out. Convection, advection conduction, and radiation are the four mechanisms of heat transportation from within the Earth to space Convection and advection are efficient means of heat transportation but are confined within the major physical boundaries, the surface and the core/mantle boundary. Heat can only be conducted or radiated across these boundaries. The heat flux out of these boundaries causes passive convections below the boundaries, such as convections in the core and mantle, and active advection above the boundaries, such as the mantle plumes. These heat transportation mechanisms form three types of convection or advection systems in the Earth (Figure 4.1 0). Near the surface of the Earth, the convection is carried out by plate motion and the counterflow in the mantle. The lithosphere moving from spreading centers toward subduction zones becomes the upper limb of a convection cell. 252

PAGE 266

APPENDIX 4.1. (Continued) The depth extension of the lower limb is not well known. The en-echelon pattern of the Hawaiian Emperor Chain leads lhinger to suggest that the counterflow in the upper mantle is immediately beneath the lithosphere (lhinger, 1995). The large aspect ratios of the volcanic ridges in the ESC and the close relationships between the trends of the en-echelon ridges and those of the fracture zones also suggest that the shearing that elongates the hot bubbles is likely to be caused by a combination of the relative plate motion and the counterflow in the upper mantle. The major shearing most likely occurs between the Low Velocity Zone and the lithosphere. However, the convection cells may be not closed circles. The subducted material may not always return to the same spreading center that created the slab. Part of the material may flow to other spreading centers. The patterns of the return flow are controlled by the depth of the limbs and the distribution of spreading centers and subduction zones Therefore, the mantle counterflow may not be directly opposite to the plate motion, which would result in shearing oblique to the direction of the plate motion which would produce en-echelon ridges (Figure 4 .11 C). The function of these convections is to bring heat from the mantle to the surface of the Earth Convection also takes place in the outer core where shear waves cannot penetrate, indicating that the material is in liquid phase As heat is conducted across the core/mantle boundary, the gradient of temperature increases downwards. The hotter and lighter material in the base of the outer core becomes unstable and buoyant. The cooler and heavier material sinks down and refills the space. This convection redistributes the heat and maintains a balance in the outer core The faster the heat is conducted across the boundary, the higher the speed of convection. 253

PAGE 267

APPENDIX 4.1. (Continued) This type of convection is called passive convection because it is not driven by heat increase from below but by the heat flux from above The function of these convections is to transport heat between the inner core and the core/mantle boundary. The passive convections in the outer core may be closely related to plumes in the lower mantle. As the lower layer of mantle material heated from below, it expands and become light The buoyant material rises in the form of diapirs. Some hot bubbles emanate from the diapirs and float upwards The colder lower mantle material flows in horizontal direction to fill the resulting space. This increases the temperature grad ient between the outer core and the lower mantle and therefore, accelerates heat conduction across the core/mantle boundary and increases the convection within the outer core In other words, each of the major plumes in the mantle may be coupled by an upwelling limb of a convection cell in the outer core We call this the Coupled Plume Model. The largest but week convection in the Earth may extend from the lower mantle to the surface of the Earth. The upwelling plumes in the form of hot bubbles may be considered active limbs of convection cells The hot bubbles rise all the way up through mantle to the lithosphere and become magma sources of intraplate volcanism (Davies, 1995). Carried by lithosphere the volcanism material will eventually return to mantle at subduction zones and be remelted Most of the material may flow to the mid-ocean ridges as counterflows to accomplish the upper mantle convections Part of the material melted from the intraplate volcanism possibly penetrate the transition zone (Bercovici et al., 1989; Tackley et al 1993), and sink to the lower mantle The remelted, colder material may flow laterally to nearby mantle plumes to refill the space left by the rising bubbles. Thus, lithosphere acts as a commuting belt in carrying the volcanism material to the subduction zones and therefore become partially the upper limbs of these large convection cells 254

PAGE 268

APPENDIX 4.1. (Continued) These convection cells may not be closed circles, either, because the downwelling material from the subduction zones does not necessarily return to the same plumes The intraplate volcanism in continental lithosphere may never return to the lower mantle. The function of these convections is transportation of heat from core to the surface of the Earth. Compared with upper mantle convections and outer core convections, these convections are probably weaker and, therefore, its upwelling is in the form of advection Both of the former convections are passively driven by cooling from above. However, lower mantle convection is driven by heat transferred into the mantle from the core below and by cooling of the lithosphere on the surface (which is part of upper mantle convection cells). If mantle plumes are driven mostly by the heat in the core, their distribution will be determined by the heat distribution in the core If the heat is evenly distributed in the core, the plumes should also be evenly distributed on the surface of the Earth. Thus, the larger plumes may be spaced farther away from each other, while smaller plumes may be closer together. The plumes should be distributed evenly beneath oceanic and continental crust. Some small hot bubbles may never penetrate thick crust, and either remain at the base of the lithosphere or result in intrusive volcanism. Therefore, more intraplate volcanism is seen on the relatively thin oceanic lithosphere, especially near mid-ocean ridges, while more intrusive volcanism is observed in continental crust. If continental lithosphere is stationary relative to the lower mantle, hot material from plumes may accumulate and cause doming and rifting. The break-up of Pangaea may have resulted from this mechanism The continent of South Africa may be currently underlain by accumulated hot material The coupled plumes do not have to be stationary. Migration of either component of a couple may lead the other component to follow it. 255

PAGE 269

APPENDIX 4.1. (Continued) Uneven distribution of heat within the core may result in the migration of the convection cells in the outer core and therefore, their corresponding plumes in the mantle On the other hand the migration of plumes in the mantle may also lead to changes in the convective structure within the outer core 256

PAGE 270

APPENDIX 5.1. SPECTRAL MODELS Spectral methods are useful tools in flexural modeling using bathymetry (or topography) and gravity data Gravity anomaly can be modeled from bathymetry data using forward model, which in turn can be used to estimate the effective elastic thickness of the lithosphere (Te) Both admittance and coherence methods are used to model the Te with gravity and bathymetry directly. The spectral models we present below use free-air gravity anomaly and bathymetry data with consideration of non-isostatic compensation by including factors of compensation percentage of volcanic loads. The Forward Model Free air gravity anomalies are assumed to result from seafloor topography and from deflection of the Moho in response to topographic loads. The gravity signal from each interface (seafloor and Moho) are computed following Parker (1972) who show that the Fourier transform of the gravity field produced at an observation plane at elevation z = Zo due to the layer of material at z = 0 is given by e-' 9\[.1g(r)] = 2nGpe-k7.o I-9\[h"(r)]. n=l n! (1) where R represents the Fourie r transform, .1g is the free-air gravity an o maly, r is the position, G is Newton's gravitational constant, p is the crustal density, k is the wave number and h(r) is the vertical deflection of the interface Using capital letters to denote the Fourier transforms the above equation can be simplified as (2) 257

PAGE 271

APPENDIX 5.1. (Continued) It is easy to generalize equation (2) to include both the seafloor and the Moho: AG(K") = H(K") + (3) where Ap1 = Pm -Pc and Ap0 = Pc (or Pc -Pw if under water), Pc, Pw and Pm are the densities of the crust, water, and mantle, respectively, Zc and Zm are the regional depths of the seafloor and Moho, respectively, H(K") is the Fourier transform of the bathymetry, and W(K") is the Fourier transform of the deflection of the Moho. Equation (3) thus gives the free-air gravity anomaly while the W term alone would represent the Bouguer anomaly Assuming the lithosphere behaves like a thin elastic plate, the deflection of the plate in response to the topographic loads is described by (furcotte and Schubert 1982; Walcott 1970a; Walcott, 1970b) D"'rw + Ap,gw = -Ap0gh (4) where hand ware topography and deflection after the loads are isostatically compensated, g is acceleration of gravity, and Dis the flexural rigidity of the plate, which is defmed as D = ETe3 (5) 12(1-v2 ) where E is Young's modulus, Te is the thickness of the elastic plate, and v is Poisson's ratio. Applying a Fourier transform to equation (4) yields a simple relation between deflection (W) and topography (H) in frequency domain W(K") =!J.po H(K") '!J.p, with '= 1 + vej !J.p,g. (6) (7) Therefore, equation (3) can be simplified for calculating synthetic free-air gravity anomaly from bathymetry data !J.G/K") = 2n!J.p0G( .. /' + )H(K"). (8) 258

PAGE 272

APPENDIX 5.1. (Continued) The Admittance Model The admittance or response function is defmed as a ratio of the Fourier transfonn of the Bouguer anomaly (L\Gb) to the Fourier transform of topography (H) (Banks et al., 1977; Donnan and Lewis 1970; Forsyth, 1985; Lewis and Dorman, 1970; Louden, 1981 ; McKenzie and Bowin, 1976; McNutt, 1980; Ribe and Watts, 1982) (9) It can be also defmed as a ratio of the Fourier transform of the free-air anomaly (L\Gc) to the Fourier transform of topography (H) F(k) = L\G1(K:)jH(K:). (10) Free-air gravity anomalies are usually collected during oceanic surveys. Therefore, the latter definition is more convenient in marine geophysical studies because the Bouguer correction can be avoided. The free-air anomaly is a summation of the Bouguer anomaly and the gravity field due to topography (L\Gt) L\G/K:) = L\Gb(K:) + L\G,(K:), (11) where L\G,(K:) = 2nll.p0Ge-kt. H(K:) according to equation (2). Therefore, the relationship between the two admittance definitions is (12) To avoid bias by noise, in practice, admittance F(k) is estimated as (ll.G1 H) F(k) = ( ) HH (13) where angle brackets indicate an average over discrete wave number bands using map data or over a number of individual profiles, and asterisks indicate the complex conjugate. 259

PAGE 273

APPENDIX 5.1. (Continued) Assuming that the topographic loads are isostatically compensated, the admittance for a thin elastic plate (Louden and Forsyth 1982) given by equation (8) is F(k) = 2nfip0G( -e-kt. .. I'+ e-kt., ). (14) However, the loads may be not isostatically compensated. If the actual deflection of the plate due to the loads is W the percentage of compensation is defmed as c = W/W. (15) A non-isostatic admittance can be derived from equations (3), (6), and (15) as F(k) = 2nfip0G( -c e -kt. .. I'+ e -kt., ), (16) or F(k) = 2nGa (17) (18) Similarly, we can derive admittance for subsurface variations in density which are, in effect, subsurface loads. Consider a load with density contrast Apb applied at depth zt, to a plate with flexural rigidity D (McNutt, 1983). If the only other density contrasts occur on the seafloor (Zc) and the Moho (Zm), the isostatic deflection of the plate is given by the following equation for a thin plate : DV4h' + (pm-Pw)gh' = -fipbgw' (19) where h' and w' are the deflection on the seafloor and the Moho, respectively, after the load is isostatically compensated. In the expression the seafloor topography is assumed to develop in response to the subsurface load. The Fourier transform of this equation is (20) Similarly, if the actual topography produced by the deflection in response to the subsurface load is H', the percentage of the compensation is d e fmed as c' = H'IH' (21) 260

PAGE 274

APPENDIX 5.1. (Continued) The gravity anomaly resulting from subsurfa c e loads is = -t:z.. W' + -k:z..,JP + -t:z., H') (22) After substituting equation (21) for the W term, the above equation becomes 2trG[-(k4Dig+ PmP w)e-k:z. H' + H' + F]. (23) For convenience, we represent the load at depth zb as relief on a density interface, but it could equally well be created by lateral variations i n density. For a simple example, considering the load on the Zm), the gravity anomaly becomes 2na[ -cev1 g + PmPw)e-kt., H' + .. H + H'] (24) or The non-isostatic admittance is F'=2na[ .. c' or F = 2na[-f!.p0e -t:z.., I c' + (1-1/ c')D.p,e -t:z.., + flp0e -t:z., ] with = 1 +vel flpog or F' = 2na{3 with f3 = -k:z., flp0e -t:z. .. I c' + (1-1/ c )flp1e -t:z. ... (25) (26) (27) (28) (29) (30) If we label the real and imaginary parts of the Fourier transform of the topography with subscripts r and i, respective ly, e g H = H, + iHi and H' = H?. + i"H:, the topography is T(K:) = H + H' = H, + H; + i(Hi + Ht) (31) and the combined free-air anomaly is = FH + F'H' = F(H, +iH)+ P(H; +iHt) (32) 261

PAGE 275

APPENDIX 5.1. (Continued) and the power for the bathymetry is E0 = (T 1) = (H? + 2H,H; + "If:2 + H ;2 + 2H}lJ + Assuming random phase, after eliminating cross terms, C and Eo are reduced to C= FH 2 + F'H' 2 and The admittance will be C FH2 + F'H'2 F =-= Eo H 2 +H'2 (33) (34) (35) (36) (37) The weighted ratio of the load on the subsurface to the load on the surface is defined as (Forsyth, 1985) != Ap1H'' Ap0H (38) Then the actual topography ( H') is related to the induced topography ( H') by the expression H' = c'H' = c'fllPo H = A.H (39) 'Ap, with A = c'fllP o ( 40) 'Ap, Combining equations (17), (29), and (39), the admittance is reduced to a+A2{3 F(k) = 2nG ).,2 (41) 1+ With equations (18), (30), and (40) the function can be rewritten as F(k) = 2nllp o G( c'f2 Apo ( l/>APo + Ap, -c' Ap,) + e -tz .. + e -tz ) ( 42) 262

PAGE 276

APPENDIX 5.1. (Continued) Notice the last term in the equation corresponds to the gravity field due to topography. According to equation (12), after removing the last term from the above equation, the non isostatic response function Q(k) is Q(k) 2 .A G c'f2 /).Po ( l/J/).Po +!::..pi -c' !::..pi) + c -kl =1Co.Po e .. c'2 12 t:..p; + '2 t:..p: (43) Assuming the load are isostatically compensated, e g c = 1 and c' = 1, the equation is reduced to A>if2!::.. 2 + 2 Q(k) = -2n!::..p G 'I'. Po '=' PI e-ll .. 0 !2 + c (44) which is compatible with Forsyth's admittance model (Forsyth, 1985). If there is no subsurface load, e.g f = 0, the above equation is further simplified to (45) which is Banks' admittance model (Banks et al., 1977). The Coherence Model The elastic thickness of the lithosphere can also be estimated using coherence between the Fourier transforms of gravity and bathymetry. Coherence is defmed as (46) where Eo, E1 and C are the power of bathymetry and gravity and the cross power between them, respectively. Assuming the surface and subsurface loads are statistically random, the expected powe r of the free-air gravity anomaly is 263

PAGE 277

APPENDIX 5.1. (Continued) (47) The power for the other terms can be simplified similarly. Therefore, the coherence can be calculated by (FH2 + F'H'2 ) 2 T -( F2 H2 + F'2 H'2 )( H 2 + H'2 ) (48) With equations (17), (29), and (39), it is reduced to (49) If the loads are isostatically compensated e .g. c = c' = 100%, A., a, and f3 are reduced to (50) (51) (52) Notice that the first term ( e -kzc ) in a and f3 is due to the gravity field produce by topographic loads 264

PAGE 278

APPENDIX 5.1. (Continued) If we ignore these teillls, the coherence is further reduced to + f+ +iP2 f 2D.p;) (53) which can be derived from the Forsyth's (1985) coherence definition between Bouguer gravity anomaly and topography. 265

PAGE 279

VITA Liu, Zhengrong EDUCATION MS in Environmental Science, July, 1991, Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, Chengdu, Sichuan, PRC Thesis: Landscape Ecology Classification and Geographical Information System. BS in Geography, July, 1983, Department of Geography, Nanjing University, Nanjing, Jiangsu, PRC. Thesis : Sedimentary Process in Northern Jiangsu Coast. PROFESSIONAL EXPERIENCE Researchffeaching Assistant, 1991-Present, Department of Marine Science, University of South Florida, St. Petersburg, Florida, USA. Research Assistant, 1983-1988, Institute of Mountain Hazards and Environment, Chinese Academy of Sciences, Chengdu, Sichuan, PRC. RECENT PUBLICATIONS Liu, Z. J., D. F. Naar, S. F. Tebbens, R N. Hey and R. A. Duncan, Evolution of the southeast Pacific and the Easter Seamount Chain, under internal review (to be submitted to the Journal of Geophysical Research), 1996 Liu, Z. J., S. E. Kruse, and D. F. Naar, Effective Elastic Thickness of the Lithosphere along the Eas ter Seamount Chain, under internal review (to be submitted to the Journal of Geophysical Research), 1996. Liu, Z. J. and D. F. Naar, Formation of the Easter Seamount Chain and Implications for Deep Earth Structure, submitted to Journal of Geophysical Research, 1996. Liu, Z. J and D. F Naar, Swath Bathymetry Processing of GLORI-B and SeaBeam 2000, under revision for Marine Geophysical Researches, 1996 Liu, z. J. and D. F. Naar, Side-scan Processing of GLORI B and SeaBeam 2000, under revision for Marin e Geophysical Researches, 1996 Liu z J and D F. Naar, Seasonal Variation of Drainage Networks Extracted from Digital Elevation Models, submitted to Geology 1996