xml version 1.0 encoding UTF-8 standalone no
record xmlns http:www.loc.govMARC21slim xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.loc.govstandardsmarcxmlschemaMARC21slim.xsd
leader nam Ka
controlfield tag 001 001914286
007 cr mnu|||uuuuu
008 071022s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001749
Vanderbloemen, Lisa Anne.
Satellite analysis of temporal and spatial chlorophyll patterns on the West Florida shelf (1997-2003)
h [electronic resource] /
by Lisa Anne Vanderbloemen.
[Tampa, Fla] :
b University of South Florida,
ABSTRACT: The objective of this dissertation is to gain a better understanding of the environmental and climatic effects on the temporal and spatial variability of phytoplankton biomass along the West Florida Shelf. Chapter 1 examines temporal and spatial patterns in chlorophyll concentrations using satellite data collected between 1997 and 2003. Chlorophyll data derived from the SeaWiFS sensor are validated with in-situ data and analyzed. Wind, current, sea surface temperature, river, and rain data are used to better understand the factors responsible for the patterns observed in the satellite data. My question is whether the standard OC4 algorithm is adequate for studying short-term variability of chlorophyll concentrations along the WFS. I will examine temporal and spatial trends using the OC4 and compare them to the Carder semianalytical algorithm which uses remote sensing reflectances at 412nm, 443nm, 490nm,and 555nm to estimate chlorophyll concentrations separately from CDOM estimates. In Chapters 2 and 3 the potential problems due to CDOM and bottom reflectance are examined. In Chapter 2 I analyze the influence of riverine induced CDOM. Water leaving radiances are analyzed in an effort to discriminate true chlorophyll patterns from CDOM contaminated signals. Chapter 3 examines the impact of bottom reflectance on the satellite signal by using the percentage of remote sensing reflectance at a wavelength of 555 to differentiate between optically shallow waters and optically deep waters. Optically shallow waters are defined as those with the percentage of Rrs at 555 due to bottom reflectance greater than or equal to 25 percent, while optically deep waters have percent bottom reflectance less than or equal to 25 percent. These analyses will help assess the validity of the temporal and spatial patterns ofchlorophyll concentration observed with the SeaWiFS data described in Chapter 1.
Dissertation (Ph.D.)--University of South Florida, 2006.
Includes bibliographical references.
Text (Electronic dissertation) in PDF format.
System requirements: World Wide Web browser and PDF reader.
Mode of access: World Wide Web.
Title from PDF of title page.
Document formatted into pages; contains 112 pages.
Adviser: Frank Muller-Karger, Ph.D.
Color dissolved organic matter.
West Florida Shelf.
Remote sensing reflectance.
Chlorophyll specific absorption.
Cyclonic and anticyclonic eddies.
x Marine Science
t USF Electronic Theses and Dissertations.
Satellite Analysis of Temporal and Spatial Chloroph yll Patterns on the West Florida Shelf (1997-2003) by Lisa Anne Vanderbloemen A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy College of Marine Science University of South Florida Major Professor: Frank Muller-Karger, PhD Kendall L. Carder, PhD John Walsh, PhD Gabriel Vargo, PhD Mark Luther, PhD Date of Approval: October 13, 2006 Keywords: chlorophyll-a, color dissolved organic m atter, West Florida Shelf, phytoplankton absorption, remote sensing reflectanc e, bottom reflectance, chlorophyllspecific absorption, seawifs, cyclonic and anticycl onic eddies, pigment composition, red tide, el nio Copyright 2006, Lisa A. Vanderbloemen
ACKNOWLEDGEMENTS My success at the College of Marine Science (CMS) would not have been possible without the support of many, many people. First, I would like to thank my main advisor, Frank Muller-Karger, for his continuing su pport throughout my tenure at the CMS. In addition, the other members of my committe e provided a great deal of advice and support throughout the process. Dr. Jyotika Vi rmani and Jennifer Cannizzaro were also indispensable with their continuing advice and assistance, particularly during the completion of my dissertation. I also would like t o thank Dr. Chuanmin Hu, Dr. Ted VanVleet, Dr. Cindy Heil, Dr. Paul Carlson, Doug My re, all members of the IMARS lab, and the CMS staff and faculty for their assistance during my stay at the CMS. I could not have made this accomplishment without their continu ous help. Finally, I would like to thank my family. My mothe r was instrumental in providing me with the drive and determination to co mplete this Â“journeyÂ”. All of my family has been vital during this process, providin g never ending support and encouragement, and not allowing me to quit. My hus band Joe Vanderbloemen was indispensable. I definitely could not have succeed ed without his unending support, both through his technical expertise, and his love, enco uragement, and confidence in me over the years.
i TABLE OF CONTENTS LIST OF TABLES..................................... ................................................... .....................iv LIST OF FIGURES.................................... ................................................... ......................v ABSTRACT........................................... ................................................... .......................viii INTRODUCTION....................................... ................................................... .....................1 CHAPTER 1: SPATIAL AND TEMPORAL VARIABILITY IN CHLOROPHYLL CONCENTRATION ON THE WEST FLORIDA SHELF (1997-2003) USING SEAWIFS DATA....... ..................................2 1.1 Introduction................................. ................................................... ..................2 1.2 Methods...................................... ................................................... ...................8 1.2.1 Rainfall Data.............................. ................................................... ....8 1.2.2 Riverflow Data............................. ................................................... ..9 1.2.3 Wind Data.................................. ................................................... ....9 1.2.4 Current Data............................... ................................................... ..10 1.2.5 Temperature Profiles....................... ................................................10 1.2.6 CDOM Time Series........................... .............................................10 1.2.7 SeaWiFS Chlorophyll Concentrations......... ...................................11 1.2.8 Seasonal SeaWiFS OC-4 Chlorophyll Anomalies ..........................13 1.2.9 In-situ ECOHAB Chlorophyll Data........... ....................................13 1.2.10 Statistical Analyses....................... .................................................14 1.3 Results and Discussion....................... ................................................... ........15 1.3.1 Rainfall Data............................... ................................................... ..15 1.3.2 Riverflow Data.............................. ................................................... 16 1.3.3 NCEP Wind Data.............................. ...............................................17 1.3.4 Current Data................................ ................................................... ..18 1.3.5 Temperature Profiles........................ ................................................19 1.3.6 CDOM Time Series............................ .............................................20 1.3.7 Annual/Seasonal/Monthly SeaWiFS Chlorophyll Patterns........................................... ................................................... 22 1.3.8 Seasonal SeaWiFS Chlorophyll Anomalies...... ...............................25 1.3.9 OC-4 vs. in-situ EOCHAB chlorophyll-a data.. ..............................28 1.3.10 OC-4 algorithm vs. Carder semi-analytical algorithm.......................................... ..............................................31 1.3.11 Statistical Analyses....................... .................................................33
ii 1.3.12 Summary.................................... ................................................... .34 1.4 Conclusions.................................. ................................................... ...............35 CHAPTER 2: GELBSTOFF (CDOM) VARIABILITY ON THE WES T FLORIDA SHELF (1997-2003) AND IMPACTS ON BIO-OPTICA L ALGORITHMS......................................... ................................................... ...............42 2.1 Introduction................................. ................................................... ................42 2.2 Methods...................................... ................................................... .................44 2.2.1 SeaWiFS ag443 (CDOM) Climatologies.......... ...............................44 188.8.131.52 ECOHAB in-situ ag versus SeaWiFS ag_443............................................. ...................................44 184.108.40.206 Nutrients................................. ...........................................45 220.127.116.11 ECOHAB Regional ag_443.................... ..........................45 2.2.2 ag_443 Spatial Extension.................... .............................................47 2.2.3 ag_443 Seasonal Anomalies................... .........................................48 2.2.4 Riverine-Influence on CDOM concentrations... ..............................48 18.104.22.168 High and Low River Influence Regions...... .....................48 2.2.5 Statistical Analyses........................ ..................................................4 9 2.2.6 Seasonal / El Nio images................... ............................................50 2.2.7 Red Tides................................... ................................................... ...50 2.3 Results and Discussion....................... ................................................... ........51 2.3.1 SeaWiFS ag_443 (CDOM) Climatologies......... .............................51 22.214.171.124 ECOHAB in situ versus SeaWiFS ag_443............................................. ...................................51 126.96.36.199 Nutrients................................ ...........................................52 188.8.131.52 ECOHAB Regional ag_443................... ..........................55 184.108.40.206.1 High ag_443 Surface Area within ECOHAB...................................... .................59 2.3.2 ag_443 Seasonal Anomalies.................. .........................................60 2.3.3 Statistical Analyses....................... ..................................................6 4 220.127.116.11 ECOHAB Region............................ .................................64 18.104.22.168 Regions Exhibiting Influence of High vs. Low Riverflow................................. ...........................65 2.3.4 Seasonal / El Nio Images.................. ............................................67 2.3.5 Red Tides.................................. ................................................... ...68 2.4 Conclusions.................................. ................................................... ...............71 CHAPTER 3: BOTTOM REFLECTANCE CONCERNS ALONG THE WE ST FLORIDA SHELF Â– IDENTIFICATION OF TEMPORAL AND SPATIAL IMPACTS USING RRS RATIO ALGORITHM.......... ...........77 3.1 Introduction................................... ................................................... ................77 3.2 Methods........................................ ................................................... .................78 3.2.1 Remote Sensing Reflectances / Semi-Analytic Model...................78
iii 3.2.2 Algorithm Parameters....................... ..............................................79 3.2.3 Applying algorithms........................ ...............................................81 3.2.4 Satellite Algorithm Comparison: OC-4 versu s Blend...................82 3.2.5 In-situ versus satellite-derived chlorophyl l.....................................82 3.2.6 CURVE values............................... .................................................83 3.3 Results...................................... ................................................... ...................83 3.3.1 OC4 versus Blend-Derived Seasonal Chlorophyll concentrations......................... ...................................83 3.3.2 CURVE Values............................... ................................................84 3.3.3 In-situ versus satellite-derived chlorophyl l.....................................85 3.3.4 Imagery Analysis........................... .................................................86 3.3.5 Regional Analyses.......................... ................................................90 22.214.171.124 Suwannee River Region.................... ...............................90 126.96.36.199 Waccassassa River Region................. .............................93 3.4 Discussion................................... ................................................... ................94 REFERENCES......................................... ................................................... ....................102 ABOUT THE AUTHOR................................... ................................................... .End Page
iv LIST OF TABLES Table 2.1. Seasons as defined in study............. ................................................... ..............43 Table 2.2. El Nio and La Nia years between 1997-2 003..............................................45 Table 2.3. Elevated concentrations for NO2+NO3, PO4 and SiO4 along the WFL shelf. Elevated concentrations are defined as: [NO2+NO3]>0.2M, [PO4}>0.24M, [SiO4]>5.5M.. ..........................51 Table 2.4. Seasonal CDOM average extension areas within the ECOHAB region for El Nio and non-El Nio periods. .................................65 Table 3.1. Coefficients were determined empirically from cubic polynomial functions (Cannizzaro et al 2006)...... ...........................................81
v LIST OF FIGURES Figure 1.1. Rivers discussed in text: Upper (Ap alachicola, Suwannee), North (Waccasassa, Anclote), Central (Hillsbourgh, Alafia, Manatee), South (Myakka, Peace), Lower (Caloosahat chee).........................9 Figure 1.2. Inner (0-30m) and outer shelf (30-18 3m) regions defined along the West Florida Shelf and ECOHAB cruise sta tions........................13 Figure 1.3. Average monthly rainfall within three major regions along West Florida Shelf outlined in Figure 8........... .............................................16 Figure 1.4. Monthly riverflow for the river region s defined in text. ..............................17 Figure 1.5. Current profile for offshore station C 09AA (27 28.529' N, 83 26.894' W). Upwellling condit ion occurred from September-December 2001............. ......................................19 Figure 1.6. Temperature profiles for September 200 1.................................................. ....20 Figure 1.7a-b. (a) Monthly CDOM concentrations for 2 ECOHAB transects as determined from SeaWiFS ag_443 data using the Carder semianalytical algorithm......... ..........................................21 Figure 1.8. Seasonal CDOM determined from SeaWiFS a g_443 data using the Carder semi-analytical algorithm... .......................................23 Figure 1.9. SeaWiFS seasonal chlorophyll time serie s for the WFS Derived using OC-4................................ ................................................... ...23 Figure 1.10. 1997-2003 SeaWiFS chlorophyll positive anomaly images (mg m-3):.........27 Figure 1.11. 1997-2003 SeaWiFS chlorophyll negative anomaly images (mg m-3) ........28 Figure 1.12. Offshore and inshore chlorophyll value s for SeaWiFS and in-situ chlorophyll values within the ECOHAB region........ ...................................29
vi Figure 1.13. Monthly SeaWiFS and ECOHAB chlorophyll time series for the WFS........................................... ................................................... ..........30 Figure 1.14. K. brevis concentrations during October 1999 along the 10m isobath between Tampa Bay and Chalotte Harbor..... ..................................31 Figure 1.15. Comparison of monthly inshore chloroph yll concentrations for the ECOHAB region derived using the OC-4 and Carder sem i-analytic algorithms for the 1997-2003 period................ ............................................32 Figure 2.1. ECOHAB stations........................ ................................................... ..............47 Figure 2.2. Rivers emptying into coastal waters alo ng the WFS. .................................50 Figure 2.3. ECOHAB in-situ and SeaWiFS CDOM........ ...............................................52 Figure 2.4a-b. (a) 1997-2003 monthly time series o f SeaWiFS ag_443. Horizontal line represents ag_443 = 0.05 m-1 (b) T ime series of riverflow, and rainfall................. ................................................... .58 Figure 2.5. ag_extension area for ECOHAB region expressed in # pi xels and area.......................................... ................................................... ............61 Figure 2.6. ag_443 anomalies for WFL shelf, 1997-2003. Positive anomalies indicate areas of increased CDOM and negative anomalies indicate areas of reduced CDOM. .................................64 Figure 2.7. Linear regressions between satellite-de rived ag_443 and chlor for 2 river regions and between riverflow and ag_443.. .....................................67 Figure 2.8. Seasonal CDOM averages for El Nio peri ods (winter/ spring 1997-98 and winter/spring 2002-03) and non-El Nio periods (all other periods 1998-2003). .....................................68 Figure 2.9. Total riverflow for central region alon g the WFS versus red tide and El Nino events from November 1946December 2003..................................... ................................................... .....70 Figure 2.10. (a) Red Tide events along the WFS. Gr een dots represent all events between 1946-2002. Yellow dots represent events be tween 1997-2002 with counts 1000). (b) Red Tide events between 9/97-4/98 (El Nino)............................... ................................................... .....71 Figure 2.11. Total riverflow for central region alo ng the WFS versus CDOM, El Nino, and red tide events from September 1997December 2003.............................................. ................................................... .............72
vii Figure 3.1. Upper and lower threshold CURVE (y-ax is) values (dotted lines) were established based on the best-fit quadratic p olynomial function for optically deep data (%bt_555 < 25%)........... ..........................................82 Figure 3.2. Average seasonal chlorophyll concentrat ions for the WFS determined using the OC-4 algorithm and the Blend algorithm...................85 Figure 3.3. Differences between seasonal chlorophyl l for the WFS as determined by the OC-4 and the Blend algorithms (OC-4 minus Blend)................................ ................................................... ...86 Figure 3.4. ECOHAB in-situ chlorophyll values versu s OC-4 derived chlorophyll and blend-derived chlorophyll. .....................................87 Figure 3.5. Average seasonal chlorophyll estimates as determined by the OC-4 algorithm (a) and the blend algorithm (c). CURVE estimates are shown in (b) images........... .......................................91 Figure 3.6. Bathymetric image for the central WFS g enerated from SeaWiFS imagery using SeaDAS software............. .....................................92 Figure 3.7. Average seasonal chlorophyll estimates for the Suwannee River region (defined in text) as determined by the OC-4 and blend algorithms....... ......................................93 Figure 3.8. Differences between seasonal chlorophyl l for the Suwannee Region as determined by the OC-4 and the blend alg orithms (OC-4 minus Blend)................................ ................................................... ...94 Figure 3.9. Average seasonal chlorophyll estimates for the Waccassassa River region (defined in text) as determined by th e OC-4 and blend algorithms.................................. ................................................... ......95 Figure 3.10. Differences between seasonal chlorophy ll for the Waccassassa Region as determined by the OC-4 and the blend alg orithms......................96
viii SATELLITE ANALYSIS OF TEMPORAL AND SPATIAL CHLOROPH YLL PATTERNS ON THE WEST FLORIDA SHELF (1997-2003) Lisa A. Vanderbloemen ABSTRACT The objective of this dissertation is to gain a be tter understanding of the environmental and climatic effects on the temporal and spatial variability of phytoplankton biomass along the West Florida Shelf (WFS). Chapter 1 examines temporal and spatial patterns in chlorophyll concen trations using satellite data collected between 1997 and 2003. Chlorophyll data derived fro m the SeaWiFS sensor are validated with in-situ data and analyzed. Wind, current, sea surface temperature, river, and rain data are used to better understand the factors resp onsible for the patterns observed in the satellite data. My question is whether the standard OC-4 algorithm is adequate for studying short-term variability of chlorophyll concentration s along the WFS. I will examine temporal and spatial trends using the OC-4 and comp are them to the Carder semianalytical algorithm which uses remote sensing refl ectances at 412nm, 443nm, 490nm, and 555nm to estimate chlorophyll concentrations se parately from CDOM estimates (Carder et al 1999, Hu et al 2003).
ix In Chapters 2 and 3 the potential problems due to C DOM and bottom reflectance are examined. In Chapter 2 I analyze the influence of riverine-induced CDOM. Waterleaving radiances (nLw) are analyzed in an effort t o discriminate true chlorophyll patterns from CDOM-contaminated signals. Chapter 3 examines the impact of bottom reflectance on the satellite signal by using the percentage of remote sensing reflectance at =555 (Rrs(555)) to differentiate between optically shallow w aters and optically deep waters. Optically shallow waters are defined as those with the percentage of Rrs(555) due to bottom reflectance (%bt_555 ) 25%, while optically deep waters have %bt_555 Â£ 25%. These analyses will help assess the validity of the temporal and spatial patterns of chlorophyll concentration observed with the SeaWiFS data described in Chapter 1.
1 INTRODUCTION The objective of this dissertation is to gain a be tter understanding of the environmental and climatic effects on the temporal and spatial variability of phytoplankton biomass along the West Florida Shelf (WFS). Chapter 1 examines temporal and spatial patterns in chlorophyll concen trations using satellite data collected between 1997 and 2003. Chlorophyll data derived fro m the SeaWiFS sensor are validated with in-situ data and analyzed. Wind, current, sea surface temperature profiles, river, and rain data are used to better understand the factors responsible for the patterns observed in the satellite data. NASA considers SeaWiFS ocean color data produced w ith the OC-4 algorithm are Â“climate-qualityÂ” for global analyses. However coastal areas such as the West Florida Shelf frequently fall under the broad categ ory of optical Case II waters (Morel and Prieur 1977) for which past and even present bi o-optical models are inadequate. In Case II waters, chlorophyll concentration estimated from satellite data are contaminated by the color of various non-chlorophyll constituent s such as Colored Dissolved Organic Matter (CDOM), suspended sediments, and frequently also by bottom reflectance. Within the Northeastern Gulf of Mexico (NEGOM), in regions of moderate to high chlorophyll concentration and at times river-i nfluenced, CDOM contribution to the absorption of light may be substantial (Del Castill o el al., 2000; Hu et al 2004). SeaWiFS-estimated chlorophyll concentrations using the Carder semianalytical algorithm
2 tend to be overestimated and CDOM absorption coeffi cients underestimated in this region (DÂ’Sa et al 2002; Carder et al 1999). Changes in the optical properties within th e WFS are frequently due to variations in the mixing of riverine and marine waters (Del Castillo et al ., 2000). Our question is whether the standard OC-4 algorithm is adequate for studying short-term variability of chlorophyll concentration s along the WFS. I will examine temporal and spatial trends using the OC-4 and comp are them to the Carder semianalytical algorithm which uses remote sensing refl ectances at 412nm, 443nm, 490nm, and 555nm to estimate chlorophyll concentrations se parately from CDOM estimates (Carder et al 1999, Hu et al 2003). In Chapters 2 and 3 the potential problems due to C DOM and bottom reflectance are examined. In Chapter 2 I analyze the influence of riverine-induced CDOM. Waterleaving radiances (nLw) are analyzed in an effort t o discriminate true chlorophyll patterns from CDOM-contaminated signals. Chapter 3 examines the impact of bottom reflectance on the satellite signal by using the percentage of remote sensing reflectance at =555 (Rrs(555)) to differentiate between optically shallow w aters and optically deep waters. Optically shallow waters are defined as those with the percentage of Rrs(555) due to bottom reflectance (%bt_555 ) 25%, while optically deep waters have %bt_555 Â£ 25%. These analyses will help assess the validity of the temporal and spatial patterns of chlorophyll concentration observed with the SeaWiFS data described in Chapter 1. In regions where bottom influence is large, OC-4 deriv ed chlorophyll concentrations will tend to be inaccurate. These are regions where the use of the Â“blendÂ” algorithm described
3 in Chapter 3 is recommended. In areas where CDOM c ontamination is a problem, use of the Carder semi-analytical algorithm described in C hapter 2 is recommended. The hypotheses posed for Chapter 1 are: (1.1) Elevated chlorophyll in areas along the WFS, adjacent to rivers, is due to riverine nutrient input. Rainfall and riverflow wil l impact the nutrient input, i.e. periods of elevated chlorophyll will be observ ed subsequent to periods of increased rainfall and riverflow. To test this h ypothesis rainfall and riverflow patterns, and their relationship to chlor ophyll patterns, will be examined. In addition, periods of elevated CDOM wil l serve as a freshwater indicator. (1.2) Elevated chlorophyll in areas along the WFS w here there is limited riverine influence is due to introduction of nutrients from upwelling. Appropriate wind and current flow resulting in upwelling will l ead to such periods of elevated chlorophyll concentrations. To test this hypothesis, temperature profiles, wind, and current patterns will be examin ed. Lower SSTs, in conjunction with upwelling-favorable wind and circu lation patterns, are indicators of upwelling. To achieve the objectives of identifying potential contamination due to CDOM and bottom reflectance, the subjects of chapters 2 and 3, the following hypotheses are identified: (2.1) In regions influenced by riverine input, sate llite chlorophyll concentration estimates will be erroneous due to influence of ele vated CDOM. Approach: CDOM concentrations can be estimated by satellite t hrough estimates of the
4 CDOM light-absorption coefficient. CDOM leads to hi gher absorption of blue light relative to what is effected if there we re chlorophyll alone. Seasonal maps of CDOM abundance will indicate perio ds when contamination is a problem in estimating chlorophyl l along the WFS. The results will help in the development of more effect ive environmental policies for this region. (3.1) In shallow waters of the WFS, enhanced satell ite chlorophyll concentrations estimates will be erroneous due to bottom reflectan ce. This may be particularly acute during winter months. Approach: Areas where nLw(670) approaches 0, and nLw pixel values at 510 and 555 a re high may be characterized as bottom reflectance. Identification of areas of bottom reflectance can then be used to determine color ano malies throughout the time series.
2 CHAPTER 1: SPATIAL AND TEMPORAL VARIABILITY IN CHLOROPHYLL CONCENTRATION ON THE WEST FLORIDA SHELF (1997-2003) USING SEAWIFS DATA 1.1 Introduction Several studies have demonstrated that the temporal and spatial patterns in chlorophyll concentrations in the deep Gulf of Mexi co (GOM) are dominated by a strong seasonal cycle and anomalies associated with El Ni o-Southern Oscillation events (Muller-Karger et al ., 1991; Gilbes, 1996; Melo Gonzales et al ., 2000. Very little, however, is still known about the short-term variab ility of chlorophyll on the West Florida Shelf (WFS). Winds, currents, tides, rainfa ll, nutrient inputs from rivers, as well as aeolian sources have all been identified as fact ors that may influence phytoplankton on this shelf (Lenes et al 2001), but their relative impact remains unclear. Phytoplankton growth is dependent on temperature (E ppley 1972), light, and nutrient availability. In sub-tropical waters such as the GOM where temperatures typically remain fairly high and light is available throughout the year, nutrient availability tends to be the limiting factor for primary product ivity (Parsons et al 1984). These warm waters are typically well mixed with limited thermo cline development. In the interior of the GOM, seasonality is observed in the primary productivity with the Â“spring bloomsÂ” that typically occur in la te fall Â– winter (Muller-Karger et al
2 1991). Destabilization of the water column at this time results in nutrient input toward the surface from the deeper depths and increased pr oduction. This destabilization may be due to wind mixing, convective overturn due to surf ace cooling, or in shallow waters (such as the WFS) it may be a factor of increased t urbulence due to bottom influence (Parsons et al 1984). As offshore water flows into shallow wate rs the bottom is Â“feltÂ” and there is an increase in water velocity and fric tion, and thus turbulence. The resulting increase in nutrients at the surface, combined with light, results in increased productivity. The source of the upwelled water, though, will dete rmine the impact on productivity. In shallow coastal waters such as th e WFS, upwelling of waters from midshelf does not increase available nutrients at the surface since bottom water concentrations are only slightly higher than surfac e waters (Ault 2006). Only shelf break upwelling of nutrient-enriched waters will have an impact on productivity along the WFS. The WFS is influenced by approximately 25 rivers th at empty into the coastal waters. These rivers introduce nutrients into the shallow coastal waters, which impacts coastal productivity. Therefore the riverflow patt erns may be an additional factor underlying the seasonality of the chlorophyll conce ntrations along the WFS which may or may not reflect the patterns previously observed fo r the entire GOM. Nababan et al (2005) examined the temporal and spatial variabil ity of bio-optical properties and the underlying mechanisms, including river discharge, in the Northeastern Gulf of Mexico. Joliff (2004) studied river discha rge and its impacts on the biology within the area of influence of the Suwannee River. The connectivity between the West Florida Shelf and the Florida Keys region was exami ned by Hu et al (2004). River
3 plumes and blooms that formed near the central port ion of the West Florida coast were transported southward due to winds and the ensuing water circulation. Fratantoni et al (1998) examined the sequence of events that leads t o the formation and propagation of the Tortugas Gyre through Florida Strait. While th ese studies have examined the role of rivers in the Gulf of Mexico and the WFS, none look ed at their influence on temporal and spatial chlorophyll patterns. Walsh et al (2003) examined the relationship between physical intrusions of slope water and the biological response by phytoplankton, specifically in the fall of 1998. Weisberg et al (2004) explained the increased coastal chlorophyll concentrations during the summer-fall 1998 period as the result of the tr ansport of deep-water nutrients towards the inner-shelf via currents within the bottom Ekma n (frictional) layer. Walsh et al (2001a, 2001b, 2002, 2005) also analyzed the possi ble underlying causes for Red Tides along the WFS. They hypothesi zed that red tide events are the result of a sequence of physical and ecological eve nts to include a phosphorus-rich nutrient supply, aeolian iron from the Saharan Dese rt, a subsequent Trichodesmium bloom with Dissolved Organic Nitrogen (DON) release build-up of shade-adapted Trichodesmium and toxic dinoflagellates at the bottom of the eup hotic zone, onshore currents to upwell these populations to CDOM-rich s urface waters in coastal waters, small initial red tide of Karenia brevis additional nutrient source from dead fish, and ultimately a large red tide (Walsh et al 2003). Other possible nutrient sources include riverine fluxes, benthic fluxes of remineralized nu trients, zooplankton excretion, and submarine groundwater discharge (Vargo et al submitted, Walsh et al in press, Hu et al 2006).
4 Numerous studies have examined upwelling events alo ng the WFS and their impact on the biology. Li et al (1999a, 1999b) and Weisberg et al (2001) studied the circulation patterns along the WFS that result from upwelling-favorable winds. Knowledge of these physical patterns is instrumenta l in explaining the biological patterns observed. On average, there appears to be a season al cycle of upwelling on the WFS Â– upwelling in the winter and downwelling in the summ er, with transitions in the fall and spring due to changes in the net surface heat flux (He et al 2003). Weisberg et al (2000) examined the factors underlying an upwelli ng event off Tampa Bay during May 1994. They showed that as a r esult of southward alongshore winds and a drop in coastal sea level, a pressure g radient develops which results in southward alongshore current flow. Offshore transp ort occurs within the surface Ekman layer and onshore transport within the bottom Ekman layer, resulting in upwelling. Unfortunately, often the underlying hydrography and /or wind events prevent a surface expression detectable via satellite imagery such as Sea Surface Temperature (SST) imagery (Weisberg et al 2000). While upwelling conditions may exist in th e inshore waters of the WFS, the nutrient concentration is no t increased due to the low nutrient concentrations within the upwelled waters (Ault 200 6). Some apparent increase in pigment concentration off shore may also be the result of the southward transport of the Mississippi River plume (Gilbes et al 2002; Gilbes et al 1996; Muller-Karger et al 1991), which occurred regularly during the July-Se ptember time frame between 1997-2004. Numerous additional studies have also examined the relationship between circulation, temperature patte rns, and ultimately the biology along the WFS (Weisberg et al 2004, He et al 2003).
5 Numerical modeling studies explain the development of upwelling conditions by suggesting that both offshore winds to the southwes t, and alongshore winds to the southeast, result in a reduction in sea level along the West Florida Shelf and an increase in sea level along the Panhandle coast. The result ing generation of an anticyclonic ocean gyre off the south-central West Florida coast leads to upwelling-favorable wind conditions, which lead to increased nutrients and i ncreased biological activity (Li and Weisberg 1999a & b; He and Weisberg 2002). Weisberg and He (2003) concluded that both local f orcing due to tides, wind and buoyancy (heating and fresh water input), and deepocean forcing due to the Loop Current, play a role in defining the circulation of shelf waters. The Loop Current can force material isopleths up and onto the shelf, and local forcing can move them shoreward beyond the shelf break. May 1998, for ex ample, was a clear example of the local forcing necessary for upwelling conditions al ong the central west-florida coast under southwestward winds. Muller-Karger (2000) studied the northeastern Gulf of Mexico (NEGOM) during the summer of 1998 using multiple satellite data se ts to show that a combination of winds and mesoscale eddies resulted in upwelling along th e northeastern coast of the Gulf of Mexico. During this period, approximately 32% of t he time winds were eastward, while during other years eastward winds occurred only abo ut 20-26% of the time. These strong eastward winds, combined with the high river discha rge due to elevated rainfall during the 1997-1998 El Nio, and the northward movement o f an anticyclonic eddy originating from the Loop Current resulted in regional upwellin g and cooling events in the coastal waters along the Florida Panhandle. The baroclinic structure due to the eddy contributed
6 to the development of the upwelling, which was enha nced by the strong eastward winds (Jochens, et al 2002). The cooler, nutrient-rich waters resulted in increased phytoplankton growth, and minor hypoxia along the s hores of the Florida panhandle. Similar events may occur along the WFS. Clearly, combining satellite and in situ data may h elp us better understand the dynamics of phytoplankton blooms off central Florid a. In the present study I employ monthly and seasonal climatologies of SeaWiFS data to examine the effects of multiple factors on the temporal and spatial variability of chlorophyll concentration along the WFS between 1997-2003: specifically rainfall, river flow, and winds. Annual variability between non-El Nio and El Nio years is analyzed a s well due to the weather changes that occur during El Nio events which affect both rainfall and riverflow, ultimately affecting the CDOM and chlorophyll concentrations. Rivers have a strong seasonal impact on the eastern GOM (Gilbes et al 1996; Gilbes et al 2002; Del Castillo et al 2001; Morey, et al 2003; Weisberg et al 2003; Weisberg et al 2006). During the spring, approximately between F ebruary and March, a tongue of low-salinity waters originating in the no rtheastern Gulf of Mexico extends to the southeast along the WFS. This is commonly refe rred to as the Â“green riverÂ” and is thought to be the result of the local forcings, of increased riverine nutrients, and upwelling (Gilbes et al 1996; He et al 1999). During summers, there appear to be regular episodes of east/southeastward transport of Mississ ippi plume water farther offshore as a result of entrainment along the eastern edge of the Loop Current. Del Castillo et al (2001) analyzed in-situ and satellite chlorophyll and colored dissolved organic matter (CDOM) data on the West Fl orida Shelf. Their results revealed
7 regional variability in both high CDOM and chloroph yll, most likely associated with variability in the summer Missisissippi River plume extension. Hu et al (2004) also revealed relationships among rainfall, nutrients, and river runoff on bloom formations along the southwestern c oast of Florida. During October 2003, periods of increased rainfall produced elevat ed riverflow and nutrient input, resulting in elevated chlorophyll and color dissolv ed organic matter (CDOM) concentrations. During El Nio years, the chlorophyll concentration s in the GOM may be different from non-El Nio years (Melo Gonzalez et al 2000). During the 1982-1983 winter El Nio, stronger winds associated with an i ncrease in the number of atmospheric fronts led to intense mixing of the upper water col umn over deep waters in the central GOM. Elevated chlorophyll concentrations were obser ved using CZCS images (Melo Gonzalez et al 2000). In recent years, several field studies have collec ted in-situ hydrographic and chlorophyll data on the WFS over limited spatial an d temporal domains. Satellite data complements these programs with frequent, repeated, and synoptic coverage. Specifically, NOAAÂ’s AVHRR (Advanced Very High Reso lution Radiometer) and OrbimageÂ’s SeaWiFS (Sea-viewing Wide-Field-of-view Sensor) offer sea surface temperature and ocean color data, respectively, for the entire WFS region, on a near-daily basis. These datasets contribute to a more comprehe nsive understanding of the temporal and spatial variability of physical and biological patterns within the region. NASA considers ocean color data produced from SeaWi FS using the OC-4 algorithm is Â“climate-qualityÂ”. We will examine whe ther this algorithm is adequate for
8 identifying and analyzing short-term climate effect s such as El Nio on chlorophyll concentrations along the WFS. Contamination probl ems due to CDOM and bottom reflectance in the shallow waters of the WFS reduce the effectiveness of this algorithm. The Carder semi-analytical algorithm is designed to correct for CDOM contamination, particularly in waters where chlorophyll concentrat ion is Â£ 1.5 to 2.0 mg m-3. Therefore, comparisons between the Carder semi-analytical algo rithm and the OC-4 will indicate when or where the OC-4 is inadequate in this enviro nment. 1.2 Methods 1.2.1 Rainfall Data Both annual and monthly rainfall data for the peri od 1997-2003 were obtained from the Southwest Florida Water Management Distric t (SFWMD) for the Northern, Central, and Southern regions as shown in Figure 1. 1. Periods of increased rainfall were identified as those being above the 75th percentile for the entire dataset (16 cm Â– central region; 17 cm Â– northern region; 18 cm Â– southern r egion). South Central Upper Lower North
9 Figure 1.1. Rivers discussed in text: Upper (Apal achicola, Suwannee), North (Waccasassa, Anclote), Central (Hillsbourgh, Alafia Manatee), South (Myakka, Peace), Lower (Caloosahatchee). Rainfall regions ( North,Central, South) as defined by SFWMD are present. Upper and Lower regi ons were defined for this study as described in text. 1.2.2 Riverflow Data Riverflow data were obtained from SFWMD (USGS stat ions) for ten rivers along the WFS. Rivers were separated into the 3 rainfall regions defined by SFWMD (Northern, Central, and Southern) along the coast, in addition to an Upper region (Apalachicola and Suwannee Rivers) and a Lower regi on (Caloosahatchee River) (Figure 1.1). Monthly averages for 1997-2003 were analyzed for spatial and temporal (monthly, seasonal) variability along the WFS to determine if a relationship exists between riverflow and chlorophyll patterns. Periods of inc reased riverflow were identified as those periods when flow was above the 75th percenti le (903 m3/s Â– upper region; 6 m3/s Â– northern region; 24 m3/s Â– central region; 20 m3/s Â– southern region; 90 m3/s Â– lower region). 1.2.3 Wind Data Monthly mean wind fields from National Centers for Environmental Prediction (NCEP) 2.5 gridded data for the area 20-31 N, 80-100 W were used to examine monthly and interannual variability in winds along the WFS (data courtesy of Dr. J. Virmani, USF Ocean Circulation Group). Wind data f rom NOAA and Southeast US Atlantic Coastal Ocean Observing System (SEACOOS) b uoys positioned along the WFS were also used to identify upwelling-conducive cond itions. Winds in the southward or southwestward direction where considered upwellingfavorable.
10 1.2.4 Current Data Current data profiles from the USF Ocean Circulatio n Group (courtesy of Dr. Weisberg, http://ocgweb.marine.usf.edu/ ) were analyzed for offshore buoys in which data was collected during the 1997-2003 study period to identify upwelling conditions. Periods of upwelling were identified when surface c urrent flow was in the westward direction and current flow at depth was in the east ward direction. There was also typically a southward component to the currents dur ing these times. Upwelling conditions at offshore buoys were analyzed in con 1.2.5 Temperature Profiles Temperature profiles from the Ecology and Oceanogr aphy of Harmful Algal Blooms (ECOHAB) period (June 1998-September 2001) were analyzed for indications of upwelling. Periods when temperatures were coole r at the surface than at depth, and an upward sloping of the isotherms was present were id entified as upwelling periods. 1.2.6 CDOM Time Series ag_443 (CDOM) data were derived from SeaWiFS data usi ng the Carder semianalytical algorithm (Carder et al 1999). These data were derived for 2 transects w ithin the imagery covering the ECOHAB region (stations 110 and 40-70; figure 1.2b) during the 1997-2003 period to examine temporal variabilit y of CDOM along the coast. Elevated CDOM (>0.05 m -1) was used as another indicator of potential riveri ne influence. Images of the ag_443 data were generated in the Se aWiFS Data Analysis System (SeaDAS) for the entire WFS. The coastal area wher e ag_443 0.5 m-1 was traced in the
11 images to identify riverine-influence. These areas are regions of possible CDOM contamination in the satellite-derived chlorophyll concentrations. 1.2.7 SeaWiFS Chlorophyll Concentrations Chlorophyll-a estimates were derived from the SeaW iFS sensor, launched in 1997 by NASA and Orbimage aboard the Orbview-II spacecra ft. Daily SeaWiFS data were collected with a SeaSpace High Resolution Picture T ransmission (HRPT) antenna installed at the University of South Florida in St. Petersburg, Florida. Level 2 files (unmapped 1 km resolution calibrated and atmospherically-corrected radiance fields) were processed to Level 3 (mapped chlorophyll and water-leaving radiance fields) using SeaDAS v4.0 developed by sci entists within the SeaWiFS Project at NASA. The algorithms described in Gordon and Wan g (1994), Ding and Gordon (1995), and OÂ’Reilly et al (2000) were used. Chlorophyll concentrations wer e derived using the OC4 version 4 bio-optical algorithm devel oped by OÂ’Reilly et al (2000). During the Level 2 processing, pixels contaminated with clouds, land, stray light, sun glint, and excessively high total radiances due to uncertainties in the bilinear gains and thus calibration were masked and not processed. The daily chlorophyll and spectral water-leaving ra diance (Lw) centered at 412, 443, 490, 510, 555, and 670 nm Level 3 images were mapped to cover the Gulf of Mexico, and averaged to derive monthly, seasonal, a nd annual composites covering September 1997 through October 2003. IDL (Interact ive Data Language) programs developed within the Institute of Marine Remote Sen sing (IMaRS) were utilized to develop the climatologies for the above parameters. Time-series of chlorophyll concentrations were developed for the entire West F lorida Shelf (WFS), specifically the
12 region from 24N-31N, 80W-90W between the coast and the edge of the continental shelf (approximately the 180 m isobath). The shelf region was divided into inner shelf (0-30m) and outer shelf (30-183m) (Figure1.2a) to b etter understand the spatial variability along the WFS and the underlying causes Inshore was defined as the 30m isobath since estuarine flux extent occurs out to a pproximately that depth (Vargo et al submitted). The stations of the ECOHAB program are shown below (Figure 1.2b). (a) (b) Figures 1.2a-b. Inner (0-30m) and outer shelf (30183m) regions defined along the West Florida Shelf and ECOHAB cruise stations. Seasonal composites were developed according to ast ronomical seasons. Spring is defined as March 21-June 21, Summer as June 21-S eptember 22, Fall as September 22December 22, and Winter is defined as December 22-M arch 21. Monthly and seasonal composites were also generated for the WFS using the Carder semi-analytical algorithm. This algorithm u ses both a semi-analytical and an empirical algorithm to estimate chlorophyll based o n remote sensing reflectances at 412nm, 443nm, 490nm, and 55nm (Carder et al 1999). In waters with chlorophyll Inshore Offshore
13 concentrations Â£ 1.5-2.0 mg m -3 the semi-analytical algorithm was used and when concentrations exceeded that, the empirical form (a default to NASAÂ’s OC4v4) was used. Since this algorithm is designed to help eliminate CDOM contamination in the chlorophyll estimates, these composites were used t o evaluate the effectiveness of the OC-4. When chlorophyll estimates derived using the OC-4 algorithm were higher than chlorophyll estimates using the Carder algorithm th is would indicate CDOM contamination in the OC-4 estimates. 1.2.8 Seasonal SeaWiFS OC-4 Chlorophyll Anomalies Chlorophyll concentration anomalies were developed for each season and year (i.e. Spring 1998, Fall 1999, Winter 2000, etc.). A nomalies were also generated for each season, based on climatological seasonal averages c onstructed over the 1997-2003 study period (i.e. Spring 97-03, Summer 97-03, Fall 97-03 Winter 97-03). Positive anomalies were indications of elevated chlorophyll for a spec ific season relative to the norm for that season over the entire study period. Negative anom alies were indications of reduced chlorophyll for a specific season relative to the n orm. Images of the seasonal anomalies were developed for analysis of the spatial variabil ity along the WFS. 1.2.9 In-situ ECOHAB Chlorophyll Data In-situ chlorophyll-a and HPLC (High Performance Li quid Chromatography) pigment concentration are available from the ECOHAB (Ecology and Oceanography of Harmful Algal Blooms) program, which focused on wat ers between Tampa Bay and Charlotte Harbor on a monthly basis between 1997-20 02 (Figure 1.2). The ECOHAB data were collected and processed by staff, faculty and graduate students within the
14 University of South Florida (USF) College of Marine Science. Such efforts were led by Dr. Cynthia Heil (Florida Marine Research Institute ) and Dr. Gabriel Vargo (USF) (Ault 2006). There are only few concurrent SeaWiFS and in situ o bservations at the ECOHAB station locations. I collected all satellite-derive d chlorophyll-a observations for the ECOHAB stations for the 1997-2000 period and derive d monthly image means on a per pixel (1 x 1 km2) basis for the ECOHAB region. To examine chloroph yll variability between inshore and offshore I partitioned the pixe ls into shelf regions deeper than and shallower than 30 m, and derived monthly means for these two subregions within the ECOHAB region. ECOHAB in-situ chlorophyll-a data co llected during each monthly cruise (3-5 days duration) were stratified into ins hore versus offshore based on the same depth criteria, and averaged (Figure 1.2b). These averages were then compared to the satellite-derived means. 1.2.10 Statistical Analyses Correlations and linear regressions were performed on the SeaWiFS 1997-2003 monthly chlorophyll data and multiple independent v ariables using Statistix v.8. These independent variables included in situ rainfall and riverflow data. Monthly rainfall data for the three regions defined by the SFWMD (norther n, central, and southern) and riverflow data for the regions shown in Figure 8 we re compared against the chlorophyll data using linear regression analyses. Cross-corre lation analyses were performed using Statistix v.8 to determine whether there were lags between rainfall and chlorophyll, and between riverflow and chlorophyll.
15 Root mean square (RMS) estimates have been used to describe the performance of ocean color algorithms in numerous studies (Zhan g et al 2006, in press; OÂ’Reilly et al 2000). These error estimates (as percentages) are calculated as follows: RMS = ( (x)2 )/ (n-1) x 100 where x= (S-I)/I (S=satellite data; I=in-situ data ) RMS estimates were derived between satellite and in -situ data, and between the OC-4 and Carder satellite algorithms. 1.3 Results and Discussion 1.3.1 Rainfall Data Annual rainfall records from the Southwest Florida Water Management District (SWFMD) for 1997-2003 illustrate elevated rainfall during 1997 and 2002 (Figure 1.3). The drought of 2000 is apparent with extremely low numbers throughout the year, with the exception of the summer months. Monthly records during 1997-2003 indicate peak rainfall typically occurred during the summer (June -September), with the peaks for the entire period occurring during 2003 (Figure 1.3). During the entire 1997-2003 period, peak rainfall (top 25th percentile) occurred in all three regions either 2 -3 months prior to, or during, all peak chlorophyll periods (Figure 1.3 ).
16 Monthly Rainfall Southwest Region0 5 10 15 20 25 30 35 40 45Sep-97Dec-97 Mar-98 Jun-98 Sep-98Dec-98 Mar-99 Jun-99 Sep-99Dec-99 Mar-00 Jun-00 Sep-00Dec-00 Mar-01 Jun-01 Sep-01Dec-01 Mar-02 Jun-02 Sep-02Dec-02 Mar-03 Jun-03 Sep-03DateRainfall (centimeters)0 1 2Chlorophyll (mg/m3) Northern Region Central Region Southern Region Chlorophyll Fall Fall Fall Fall Fall Fall Fall Figure 1.3. Average monthly rainfall within three major regions along West Florida Shelf outlined in Figure 8. Seasons are delineated by horizontal gridlines. SOURCE: Southwest Florida Water Management District (SFWMD) 1.3.2 Riverflow Data To understand the overall impact of riverflow on ch lorophyll concentrations along the WFS, the ten rivers studied were aggregated int o the SFWMD regions defined for rainfall. An Upper region and a Lower region were a lso included, to gain a better understanding of riverflow impacts along the entire WFS (Figure 1.1). Average riverflow was calculated for the study peri od, and the level of flow determined for each region. During the 1997-2003 p eriod, the upper region had the highest level of flow (avg flow=810 m3/s) and thus the greatest potential for impact on chlorophyll concentrations during periods of elevat ed flow. The northern region had the lowest flow (avg flow=5 m3/s) and would have the least impact on chlorophyll. The central, southern, and lower regions have medium le vel flow (avg flow=19 m3/s, 16 m3/s,
17 60 m3/s respectively) and thus elevated riverflow in the se regions could have an impact on the chlorophyll concentrations. All five river regions had similar patterns over t he 1997-2003 period as seen in Figure 1.4. The highest overall flow was in the Up per Region (Figure 1.4; notice different scale on secondary x-axis). During the e ntire 1997-2003 study period, peak riverflow (top 25th percentile) occurred in 3-4 of the regions during 11 of the 24 peak (top 25th percentile) monthly chlorophyll periods (46%) (Fig ure 1.7). These periods were November 1997-March 1998, October 1998, December 20 02-January 2003, and JulyOctober 2003. Regional Riverflow0 20 40 60 80 100Sep-97Dec-97 Mar-98 Jun-98 Sep-98Dec-98 Mar-99 Jun-99 Sep-99Dec-99 Mar-00 Jun-00 Sep-00Dec-00 Mar-01 Jun-01 Sep-01Dec-01 Mar-02 Jun-02 Sep-02Dec-02 Mar-03 Jun-03 Sep-03 m3/s0 500 1000 1500 2000 2500 3000 3500 4000Upper Region (m3/s) NORTHERN Region CENTRAL Region SOUTHERN Region UPPER Region LOWER Region Figure 1.4. Monthly riverflow for the river region s defined in text. 1.3.3 NCEP Wind Data
18 Analysis of the NCEP monthly mean wind data for the WFS between 1997-2003 (Virmani 2005) reveals that during periods of incre ased chlorophyll concentrations the winds were predominantly from the west or northwest which typically are not upwellingfavorable unless they have evolved during a front. While during fall/winter 1997 winds were predominantly from the west, this was an El Ni o period and rainfall and riverflow were elevated. These results suggest that the latt er factors caused the elevated chlorophyll concentrations observed at that time. 1.3.4 Current Data Current profile data from buoys positioned along t he WFS out to the 200m isobath do reveal upwelling conditions at multiple locations during 11 of the 24 chlorophyll peak periods (46%). During October 199 8 upwelling-favorable currents (westward at surface, eastward at depth) and southw estward winds were present at multiple buoys within the 30m isobath, between Tamp a Bay and Charlotte Harbor. Similar current patterns were observed in August 19 99, with winds towards the southsoutheast. From September-December 2001 (Figure 1. 5), and during January 2003 upwelling conditions were present at multiple buoys with winds predominantly to the southwest. Upwelling conditions were also present from August-October 2003, with an extended area of upwelling occurring in October 200 3. During October 2003 upwelling conditions extended beyond the 30m isobath out to t he 100m isobath and further north to Waccasassa Bay (USF OCG, http://ocgweb.marine.usf.edu/ ).
19 Figure 1.5. Current profile for offshore station C 09AA (27 28.529' N, 83 26.894' W). Upwellling condition occurred from Sep tember-December 2001. Data courtesy of Dr. Wiesberg, USF OCG. Upwelling conditions occurred at individual buoys l ocated beyond the 30m isobath and between the Waccasassa and Apalachicola Bays during October 1997, January 1998, March-April 1998, and May 1998. Winds were strong (15-30 m/s) and towards the south-southwest during these times. 1.3.5 Temperature Profiles During the ECOHAB period, June 1998-September 2001 temperature profiles revealed upwelling conditions during only one of th e six chlorophyll peaks during this time period. In September 2001, the profiles for t he Tampa Bay, Sarasota, and Ft. Myers transects all revealed upwelling between 40-125 km offshore (Figure 1.6). These results
20 corroborate the current and wind data for that peri od which indicate upwelling was prominent at these times of increased chlorophyll c oncentrations (Figure 1.9). Figure 1.6. Temperature profiles for September 200 1. Data courtesy of Dr. Wiesberg, USF OCG. During winters of 1999 and 2000 temperature profile s reveal well-mixed conditions and the absence of a thermocline. These were period of reduced chlorophyll concentrations (Figure 1.7) (USF OCG, http://ocgweb.marine.usf.edu/shelf/Ecohab/hydro/ECO -hydro.shtml ). 1.3.6 CDOM Time Series Monthly CDOM concentrations determined from SeaWIF S ag_443 concentrations were elevated ( 0.05 m-1) during the fall/winter months for 1997-2003, particularly October-December (Figure 1.7a). Thes e results are coincident with elevated riverflow during these periods as well (Figure 1.7b ), potentially leading to increased nutrients and ultimately increased chlorophyll conc entrations (Figure 1.9). Peak riverine nutrient loadings into Tampa Bay and Charlotte Harb or occur during late summer into the fall (Vargo et al submitted). Tampa Transect Fort Myers Transect Sarasota Transect
21 Monthly ag_443(Stations 1-10, 40-70)0 0.02 0.04 0.06 0.08 0.1 0.12 0.14Sep-97Dec-97 Mar-98 Jun-98 Sep-98Dec-98 Mar-99 Jun-99 Sep-99Dec-99 Mar-00 Jun-00 Sep-00 Nov-00 Feb-01 May-01 Aug-01 Nov-01 Feb-02 May-02 Aug-02 Nov-02 Feb-03 May-03 Aug-03 Nov-03ag_443 (m-1) Sta 1-10 Sta 40-70 0 5 10 15 20 25 30 35 40 45Sep-97Dec-97 Mar-98 Jun-98 Sep-98Dec-98 Mar-99 Jun-99 Sep-99Dec-99 Mar-00 Jun-00 Sep-00Dec-00 Mar-01 Jun-01 Sep-01Dec-01 Mar-02 Jun-02 Sep-02Dec-02 Mar-03 Jun-03 Sep-03Rainfall (cm)0 1 10 100 1000Riverflow (m3/s) N. Rainfall C. Rainfall S. Rainfall Central Rivers El Nino El Nino Figure 1.7a-b. (a) Monthly CDOM concentrations for 2 ECOHAB transects as determined from SeaWiFS ag_443 data using the Carde r semianalytical algorithm. Horizontal line represents where ag_443 =0.05 m-1. (b) Time series of riverflow, and rainfall. Rainfall regions are defi ned by Southwest Florida Water Management District. Central rivers include Hillsb orough, Alafia, and Manatee Rivers. During these periods of elevated riverflow and CDO M, potential contamination by CDOM in the satellite-derived chlorophyll estima tes is increased. The coastal areas where CDOM concentrations are 0.05 m -1 (indicator of riverine influence) are shown in Figure 1.8. These areas fall within the 30m iso bath during all seasons, indicating potential CDOM contamination in satellite-derived c hlorophyll concentrations occurs
22 only in inshore waters. The area of potential cont amination is greatest during the fall and winter months when the areal extent of waters with CDOM 0.05 m -1 is largest (Figure 1.8). On average, during these seasons, the CDOM 0.05 m -1 threshold is located close to the 30 m isobath. 1997-2003 Spring ag_443 1997-2003 Summer ag_443 1997-2003 Fall ag_443 1997-2003 Winter ag_443 Figure 1.8. Seasonal CDOM determined from SeaWiFS ag_443 data using the Carder semi-analytical algorithm. White line delin eates where ag_443=0.05 m-1. Contour lines represent 30 and 100m depths. 1.3.7 Annual/Seasonal/Monthly SeaWiFS Chlorophyll P atterns Analysis of the chlorophyll concentration time ser ies within the WFS (24-31 N, 0-183 m) (Figure 1.2a) reveals that between 1997 an d 2003, maxima (0.97-1.44 mg m-3)
23 occurred in the fall, typically in October of each year (0-183m region, Figure 1.9). In the 0-30 m region the largest average concentrations we re seen in fall and winter 1997. This region showed the highest apparent chlorophyll conc entrations throughout the entire time series. There appears to be a similar seasonal var iation in the inshore and offshore averages, with inshore values a factor of 2 higher than offshore values (30-183m; Figure 1.9). SeaWiFS Seasonal Chlor Averages (1997-2003) WFL Shelf 0.5 1 1.5 2 2.5 3Fall 97 Winter 97Spring 98 Summer 98 Fall 98 Winter 98 Spring 99 Summer 99 Fall 99 Winter 99Spring 00 Summer 00 Fall 00 Winter 00Spring 01 Summer 01 Fall 01 Winter 01Spring 02 Summer 02 Fall 02 Winter 02Spring 03 Summer 03 Fall 03 Chlor (mg/m3) 0-183m 0-30m 30-183m AVERAGES Region mg/m3 0-30m = 1.43 0.3730-183m = 0.74 0.190-183m = 1.06 0.23 Figure 1.9. SeaWiFS seasonal chlorophyll time seri es for the WFS derived using OC-4. Standard deviations are represented by error bars. Horizontal line represents 75th percentile for the 0-183m region over the time seri es. Points above the line are considered peak periods. Throughout the period 1997-2003 the lowest mean co ncentrations for the entire shelf (0-183m) typically occurred during the spring ranging between 0.67-0.93 mg m-3. A minimum was observed in spring 2000. During spri ng, the difference between inshore and offshore average concentrations is smallest (ra nge: 0.24-0.61 mg m-3). In fall/winter, the difference is largest (range: 0.70-1.11 mg m-3). This large difference is likely in part due to an increase in riverine influence inshore du ring the fall (Figure 1.4).
24 The chlorophyll pattern from late 1997 Â– early 1998 (December 22-March 21) was different than for other years. Concentrations were higher throughout this period with peak chlorophyll concentrations across the ent ire shelf (0-183m) in winter 1997 (December 1997) and high concentrations during spri ng (late March-June 1998) as well, which we did not see in other springs. Late 1997-early 1998 was a strong El Nio season. Li and Weisberg (1999a, 1999b) have shown that during periods of upwellingfavorable winds (alongshore or offshore), the response is a combination of both lo cal and larger-scale processes. While the wind initiates the response, the pressure gradi ent, Coriolis force, and vertical friction are all factors along the inner shelf. The geometr y of the coastline and isobath structure along the WFS impact the responses observed along t he middle and outer shelves. Weisberg et al (2001) examined the conditions during April 1998 a nd showed that the overall circulation along the WFS at this time was a combination of the local wind forcing and the density structure resulting from of fshore and local buoyancy forcing. During El Nio fall, winter, and spring seasons, ra infall is elevated in Florida (Hanson and Maul 1991, Sittel 1994a, Livezey et al 1997, Schmidt 2001). Riverflow is affected by rainfall due to the impact on runoff an d groundwater inputs (Schmidt et al 2001). During El Nio years, increased rainfall le ads to increased riverflow with a possible 1-2 month lag in riverflow (Schmidt et al 2001). During spring 1998 when chlorophyll concentrations were high, riverflow for the Upper, Central, South, and Lower regions was elevated (Figure 1.4). In December 199 7, approximately three months earlier, rainfall was elevated (Figure 1.3). This suggests a relationship among rainfall,
25 riverflow, and chlorophyll during this period. Chl orophyll concentrations either coincide with, or lag 2-3 months behind, rainfall and riverf low throughout the time series. Analysis of the variability between inshore and off shore chlorophyll concentrations along the WFS reveals the largest di sparity was not during the winter months (December 1997-March 1998) when overall conc entrations were greatest, but during fall 1997 (Figure 1.9). The difference betw een inshore and offshore during fall 1997 was 1.76 mg m-3 with extremely high inshore concentrations (2.35 m g m-3) relative to offshore (0.55 mg m-3). The smallest difference between inshore and off shore, 0.24 mg m-3, was during Spring 1999 when inshore concentratio ns were 0.84 mg m-3 compared to 0.60 mg m-3 offshore (Figure 1.9). Throughout the 1997-2003 study period there were 24 peak monthly chlorophyll events. These were defined as the top 25th percentile ( 1.18 mg/m3). They occurred during fall/winter 1997, spring/summer/fall 1998, f all 1999, fall 2001, fall/winter 2002, and summer/fall 2003 (Figure 1.9). 1.3.8 Seasonal SeaWiFS Chlorophyll Anomalies The seasonal SeaWiFS chlorophyll anomaly images, de rived relative to seasonal climatological means (1997-2003) provide additional insight into the temporal and spatial variability within the WFS (Figure 1.10). During w inter 1997-98 and spring 1998, an El Nio period with increased rainfall and riverflow, there was a positive chlorophyll anomaly along the entire WFS, while during summer a nd fall 1998 the largest anomaly was within approximately 28.5-29.5 N, 82.5-84 W (1998 red box, Figure 1.10). Particularly pronounced was a nearshore bloom to th e north of Tampa Bay. This area is impacted by the Suwanee and Waccasassa rivers (Figu re 1.1). During winter 1998-99
26 there was again a positive chlorophyll anomaly alon g the entire WFS, with the largest anomalies occurring in the Big Bend region. 1998 Spring 1998 Fall 1998 Winter 1999 Spring 1999 Summer 2001 Summer 2001 Fall 2002 Spring 2002 Summer 2002 Fall 2002 Winter 2003 Spring 2003 Summer 2003 Fall 1997 Winter 1998 Summer
27 Figure 1.10. 1997-2003 SeaWiFS chlorophyll positiv e anomaly images (mg m-3): Regions outlined in red and yellow represent anomal ies discussed in text. In 1999 the positive anomalies were much smaller in magnitude and spatial extent. Small patches of high concentrations during spring and summer 1999 relative to 1998-2003 spring and summer were observed along the southernmost tip of the WFS. In 2001 the greatest anomalies occurred during the sum mer and fall along most of the coast, while fewer anomalies occurred during spring. In s pring and winter 2002 there were large anomalies along the southernmost tip of the W FS, while during summer an anomaly occurred nearshore along the central and southern W FS (Figure 1.10). During fall 2002 positive anomalies occurred along the northern WFS. In 2003, large anomalies were present along the northern portion of the coast, wi th highest levels during the spring and summer. During the summers of 1998, 1999, 2001, and 2003 ch lorophyll anomalies appear inshore and offshore (yellow regions, Figure 1.10). The location of these offshore anomalies relative to the anomalies present in the northeastern GOM during these times would suggest they are an extension of the chloroph yll-rich waters from the Mississippi plume that have been observed to flow towards the s outh-southeast during the summer months (Gilbes et al 2002; Gilbes et al 1996; Muller-Karger et al 1991). Summer 1998 was also a period of upwelling favorable winds alon g the WFS and strong thermal stratification which resulted in southward offshore current flow and may explain the chlorophyll anomalies observed inshore (Weisberg an d He 2003; Joliff et al 2003).
28 During winters of 1999, 2000, and 2001 negative ano malies were prevalent along the WFS (Figure 1.11). The area north of Tampa Bay (circled in red) had large negative anomalies averaging -2 mg m-3 thus indicating low chlorophyll concentrations dur ing these times relative to the overall seasonal averag e for 1997-2003. Examination of temperature profiles during these times reveals wel l-mixed conditions, suggesting increased convective overturn and increased nutrien t input from deeper waters. Midshelf nutrient concentrations are only slightly hig her than surface concentrations along the WFS, though, so upwelling would not result in i ncreased nutrients (Ault 2006). Riverflow in this area was also lower during these times (Figure 1.4) which would suggest reduced riverine nutrients. These conditio ns help explain the negative anomalies observed. Figure 1.11. 1997-2003 SeaWiFS chlorophyll negativ e anomaly images (mg m-3) 1.3.9 OC-4 vs. in-situ EOCHAB chlorophyll-a data In general, SeaWiFS OC-4 chlorophyll estimates wer e in agreement with the insitu ECOHAB chlorophyll concentration estimates thr oughout the year although the OC4 overestimated the chlorophyll in both inshore and offshore waters (Figure 1.12). For the June 1998-December 2000 period, the average RMS error for inshore waters was 199 9 Winter 2000 Winter 2001 Winter
29 22% with the smallest error occurring in June and S eptember 1999 and the largest occurring during January-March 2000. The average R MS for offshore waters was 11% with the smallest error occurring in November 1998 and the largest occurring during November 2000. Some of the error may be due to CDO M contamination or due to the difficulty in obtaining completely concurrent SeaWi FS and in-situ values. While the insitu data were collected at multiple times during t he daily cruises, the satellite data were collected during one daily pass over the region. T he in-situ data were also collected at point locations (stations), while the satellite dat a were collected on a pixel basis (1km x 1km). These factors also would contribute to the lower R2 values observed (Figure 1.12). Inshore Chlorophyll0.1 1 10 0.1110 ECOHAB In-Situ ChlorSeaWiFS (OC-4) R2 = 0.38, n=26 Offshore Chlorophyll0.01 0.1 1 0.010.11 ECOHAB In-Situ Chlor SeaWiFS (OC-4) R2 = 0.56, n=25 Figure 1.12 (a-b). Offshore and inshore chlorophyl l values for SeaWiFS and in-situ chlorophyll values within the ECOHAB region. Pleas e note the scale differences in axes. The greatest agreement was observed during spring a nd early summer, while the worst agreement occurred during the fall and winter (Figure 1.13). Riverflow was high during the fall months which would cause increased CDOM and increase the potential for CDOM contamination in these shallow waters. This m ay be a factor underlying the low correlation, in addition to the patchiness of field samples during red tides (see below). The large disparity during October 1999 (Figure 1.1 3) occurred when a major red tide
30 bloom was observed (Lenes et al 2001). Due to the method used to average the sat ellite data over the ECOHAB region, the effect of the red tide bloom was muted in the satellite time series. SeaWiFS versus ECOHAB Chlorophyll (1998-2000)0 1 2 3 4 5 6 7Jun-98 Sep-98Dec-98 Mar-99 Jun-99 Sep-99Dec-99 Mar-00 Jun-00 Sep-00Dec-00 Chlorophyll (mg/m3) SeaWiFSInshore SeaWiFSOffshore ECOHABInshore ECOHABOffshore Figure 1.13. Monthly SeaWiFS and ECOHAB chlorophyl l time series for the WFS. (Spring=Mar to Jun, Summer=Jun to Sep, Fall=Sep to Dec, Winter=Dec to Feb) Examination of the daily ECOHAB data for 2-12 Octob er 1999 revealed maximum chlorophyll concentrations at inshore stati ons 76 (26.94 N, 82.47 W) and 80 (27.24 N, 82.63 W), with elevated chlorophyll at station 30 as well (27.22 N, 82.82 W) (Figure 1.14). The SeaWiFS (OC-4) daily data for 2 -12 October 1999 reveal elevated chlorophyll at the same locations. These stations were sites of elevated Karenia brevis concentrations during this period. Normal levels o f K. brevis are Â£ 1,000 cells/l while concentrations >1,000,000 cells/l are considered hi gh, leading to risk of respiratory irritation and probable fish kills. Concentrations of 5,270,000 cells/l were found at station 80 on 10/5/99.
31 -83.50-83.00-82.50-82.00-81.50 26.00 26.50 27.00 27.50 28.00 28.50 0 500000 1000000 1500000 2000000 2500000 3000000 3500000 4000000 4500000 5000000ECOHAB: FLORIDAOctober 5-7, 1999G. breve ConcentrationG. breve (c/l)Longitude (W)Latitude (N) Figure 1.14. Karenia brevis concentrations during October 1999 along the 10m i sobath between Tampa Bay and Chalotte Harbor. Figure cour tesy of D. Ault (2006). 1.3.10 OC-4 algorithm vs. Carder semi-analytical al gorithm Chlorophyll concentrations derived from the OC-4 al gorithm were compared to those derived using the Carder semi-analytical algo rithm. During the study period the OC-4 algorithm consistently overestimated chlorophy ll concentrations relative to the Carder semi-analytical algorithm, particularly duri ng late summer into fall (Figure 1.15a). The worst agreement occurred during October 2001 (F igure 1.15a). This was the period of a moderate-high red tide bloom off St. Petersbur g and Charlotte Harbor (avg conc = 5,637,636 cells/l). In red tide blooms where the Karenia brevis concentrations are >104 cells/l, there is a factor of 3-4 decrease in Rrs() due to the lower particulate backscattering coefficients (Cannizzaro 2004). Sin ce both algorithms are based on Rrs ratios, the decrease in Rrs during a red tide may account for the increased er ror between the two algorithms during October 2001 (Figure 1.15 a). The best agreement was during
32 March of each year when CDOM concentrations were lo w (Figure 1.7), with the exception of March 1998 which was the end of the 19 97-1998 El Nio. Analysis of the performance of these two algorithms specifically within the ECOHAB inshore region (0-30m) of the WFS produced a n R2=0.84 (Figure 1.10b) and an average RMS error of 27%. The error between the two algorithms was slightly higher for the entire WFS with an average RMS error of 31% and an R2=0.20. ECOHAB Region Inshore Chlor (0-30m)0 1 2 3 4 5Sep-97Dec-97 Mar-98 Jun-98 Sep-98Dec-98 Mar-99 Jun-99 Sep-99Dec-99 Mar-00 Jun-00 Sep-00 Nov-00 Feb-01 May-01 Aug-01 Nov-01 Feb-02 May-02 Aug-02 Nov-02 Feb-03 May-03 Aug-03 Nov-03 Chlor (mg m-3) OC-4 Carder_chlor (a) ECOHAB Region Inshore Chlor (0-30m) 0.1 1 10 0.1110 Carder Chlor OC-4 Chlor R2 = 0.8357, n=30 (b) Figure 1.15 (a-b). Comparison of monthly inshore c hlorophyll concentrations for the ECOHAB region derived using the OC-4 and Carder sem i-analytic algorithms for the 1997-2003 period. Arrow in (a) delineates October 2001, discussed in text.
33 1.3.11 Statistical Analyses In addition to calculation of RMS errors discussed previously, linear regressions were performed on monthly SeaWiFS OC-4 chlorophyll data for the entire WFS and monthly mean rainfall data for the 3 SWFMD regions (northern, central, southern). Correlations were low between rainfall and chloroph yll with R2 Â£ 0.055 for all three rainfall regions. Such low correlations are most li kely due to the need for rainfall seepage to occur before the water reaches the rivers. Correlations between riverflow and chlorophyll vari ed, with lower correlations in the upper and northern regions (R2 Â£ 0.051). The correlations between riverflow and chlorophyll in the central and southern regions wer e higher with R2 0.18. but still indicating no correlation. These correlations were still fairly low, so cross correlations were performed between chlorophyll and rainfall, and chlorophyll a nd riverflow to determine if there is a lag time for an observable effect. These correlati ons reveal a lag time of 2-3 months for rainfall, suggesting that there has to be a factor accommodating the need for seepage, flow of the water to the coast and dispersal along the coastal zone. Cross-correlations for riverflow reveal the strongest correlations (R=0.58 to 0.63) occur between chlorophyll and the rivers in the central and southern regions, both considered medium-flow river regions, with a 0-1 month lag time (Feinstein 1956, Dixon 2003, Dixon et al 2004).
34 1.3.12 Summary During the 1997-2003 study period I identified 24 periods of increased chlorophyll ( 1.18 mg m-3) along the WFS using the OC-4 algorithm. Along the WFS the major underlying factor for increased chlorophyll i s the availability of nutrients which originate from either riverine input, atmospheric i nput, upwelling, or convective overturn. These factors can be inferred according to the foll owing conditions. Riverine: High rainfall, high riverflow, high CDOM Upwelling: Current profiles, southward/southeastwar d winds, temperature profiles Convective Overturn/Vertical Mixing: Winds, temper ature profiles (lack of thermocline) During the 24 periods of increased chlorophyll, riv erflow has a major impact on chlorophyll during November 1997-December 1997, Feb ruary 1998, December 2002, and July 2003. Current and temperature profiles in dicate upwelling is the major factor from April-May 1998, July 1998, August 1999, Septem ber-December 2001, and October 2003. Indicators are that both riverine influence and upwelling may be factors in the elevated chlorophyll concentrations during October 1997, January 1998, March 1998, October 1998, January 2003, and August-September 20 03. Fall/winter 1997-1998 was an El Nio period. In addition to increased rainfall a nd riverflow during El Nios, the increase in atmospheric fronts results in intense m ixing of the upper water column within the GOM interior and offshore upwelling. This woul d explain the role of both riverflow and offshore upwelling in the elevated chlorophyll concentrations during the October 1997-March 1998 period. During October 1997, January 1998, and March 1998 u pwelling was indicated at one offshore station, while riverflow was well abov e the 75th percentile in the Upper,
35 North, and Central regions during these periods. I n addition, the inshore chorophyll concentrations accounted for up to 85% of the total chlorophyll concentration (0-183m) (Figure 1.9) which would indicate that while upwell ing was a factor, riverflow was a slightly stronger factor. During January, August, and September 2003, upwelling was observed at at least two stations (Table 1.1), and riverflow was either right at the 75th percentile or slightly above it. The inshore chlor ophyll concentrations were only 60% of the total chlorophyll concentration for the WFS whi ch would suggest that upwelling was more of a factor than riverflow during these times. Upwelling was observed along the edge of the inshore region (30m), though, which wou ld suggest limited increases in nutrient concentrations from upwelling (Ault 2006). Date Stations Locations Isobath (m) Jan-03 C10AH C11AL 27 09.84'N, 82 55.53' W; 2712.673' N, 8249.238' W 30 25 Aug-03 Sept-03 C10AJ C15AM 27 09.84' N. 82 55.53'W; 2717.949' N, 8238.360' W 30 20 Table 1.1. ECOHAB stations where upwelling was obs erved. Comparisons between the OC-4 and Carder semi-analyt ical algorithms reveal similar chlorophyll patterns along the WFS between 1997-2003. Therefore, I conclude that the OC-4 is adequate for revealing large-scale temporal (monthly/seasonal) and spatial patterns of chlorophyll along the WFS as we re produced in this study. 1.4 Conclusions SeaWiFS imagery over the past 5 years reveals a te mporal and spatial pattern in chlorophyll concentration along the West Florida Sh elf. While peaks in primary productivity within the GOM have been shown to occu r typically in the winter months (Muller-Karger et al 1991, El Sayed 1972), our results show that along the WFS peak
36 chlorophyll concentrations tend to be more variable with peaks occurring in the spring, summer, and fall. Examination of periods of elevated chlorophyll duri ng 1997-2003 reveal that during fall, riverflow is the predominant factor af fecting concentrations, while during spring/summer months, the importance of upwelling i n controlling concentrations may be higher. During extreme climate periods such as El Nio, both may play a role in the elevated chlorophyll concentrations observed as dis cussed previously. Late 1997early 1998 was a very strong El Nio wit h increased rainfall and riverflow, which could explain the elevated chlorop hyll values during winter 1997-98 (1.44 mg m3) when the inshore values accounted for 80% of the overall concentration. The excessive chlorophyll concentrations during win ter 1997 (12/97) and spring 1998 (3/98), relative to the 1997-2003 seasonal averages are apparent in the large chlorophyll anomalies along the entire WFS. Riverflow for this period (12/97, 3/98) was particu larly high for the Upper and Lower regions which include the Apalachicola, Suwan ee, and Caloosahatchee rivers, 3 major rivers providing large volumes of freshwater along the WFS. CDOM levels were high during this period, and the winds were from th e west (Virmani 2005) and not upwelling-favorable, providing additional evidence that riverine influence was the predominant factor at this time. Riverflow patterns for all five regions were simila r during 1997-2003 with peaks occurring during the fall and winter months of 1997 and 2002, and the summer months during the other years. Examination of the individ ual rivers within the river regions reveals a spatial variability in riverine impact ba sed on the individual rivers present and
37 their average flow-rate. The influence of the Suwa nee River within the Upper Region, a high flow rate region, can be seen during summer 19 98 and fall 2003 when the greatest anomaly occurs where it discharges onto the shelf, introducing increased nutrients and consequently biomass (chlorophyll) (Figure 1.10, re d boxes). During the summer of both years rainfall was increased, and riverflow was hig h through May 1998 and throughout 2003, with concomitant positive chlorophyll anomali es. Our results indicate that during the winters of 199 9, 2000, and 2001 the waters off the WFS were well mixed with low chlorophyll concen trations. Peak chlorophyll concentrations were observed in the fall of 1999 an d 2001 when convective overturn is common (Weisberg et al 1999a, 1999b), which would suggest that convective overturn and vertical mixing may have occurred at these time s. By the winter months the nutrients would have been depleted and primary prod uctivity reduced, resulting in the lower chlorophyll concentrations. Offshore upwelling is another factor that may resul t in elevated chlorophyll concentrations. Out to approximately the 30m isoba th on the WFS bottom water nutrient concentrations are just slightly higher than surfac e values and therefore the impact of inshore upwelling on chlorophyll concentrations is limited. Wind, current, and SST patterns may be used as indicators of possible upwe lling. During April-May 1998, July 1998, August 1999, September December 2001, and O ctober 2003, the currents were upwelling-favorable and the winds were predominantl y towards the south-southwest. Temperature profiles also indicated upwelling condi tions during September 2001. Riverflow was not elevated during all of these peri ods, therefore these other observations support the conclusion that upwelling led to the in creased chlorophyll concentrations.
38 There is evidence of both riverine and upwelling in fluence on chlorophyll during October 1997, January, March and October 1998, Janu ary and August-September 2003. Riverflow was elevated during these periods, in add ition to the occurrence of upwelling conditions at multiple buoys as indicated by curren t and temperature profiles. Our results prove our initial hypothesis that eleva ted chlorophyll in areas along the WFS that are adjacent to rivers is due to river ine nutrient input. Periods of elevated chlorophyll were observed subsequent to, or during, periods of increased rainfall and riverflow. Elevated chlorophyll concentrations du ring periods of offshore upwelling proves our secondary hypothesis in areas along th e WFS where there is limited riverine influence, increased chlorophyll is due to increase d nutrients from offshore upwelling. The apparently higher chlorophyll concentrations in inshore waters relative to offshore waters may be due to increased CDOM concen trations in inshore waters, and possible contamination in the OC-4 satellite-derive d chlorophyll concentration. Both the inshore and offshore concentrations have similar pa tterns during the study period, though, so analysis of their overall patterns can provide u seful insights into the true chlorophyll patterns for the entire WFS (Figure 1.9). In addition, comparisons of chlorophyll concentrati ons calculated with the OC-4 and with the Carder semi-analytical algorithm, whic h corrects for CDOM in low chlorophyll waters, revealed similar overall patter ns with an RMS<25%. The largest disparities between the two occur in the late summe r to fall months when riverflow is high and when CDOM is increased, for which the OC-4 algorithm does not correct. Future studies will look at the temporal and spatia l patterns of CDOM contamination
39 along the WFS to identify when and where the contam ination is greatest and to allow improved interpretations of chlorophyll patterns de rived from the OC-4 algorithm. To help assess the accuracy of the SeaWiFS OC-4 dat a, SeaWiFS observations within the ECOHAB region were compared to ECOHAB in -situ data. There was better concurrence during spring and summer than fall and winter within this region. There may be several reasons for these results. In the E COHAB region, elevated rainfall typically occurs during the summer months. Maximum riverflow for many of the rivers in this region occurs during the late summer to fal l. The overall effect of freshwater is therefore not as great during the spring and summer resulting in reduced contamination due to CDOM. The performance by the SeaWiFS algori thm is improved, producing better agreement with the insitu data during these times. Considering the limitations of the OC4-derived chlo rophyll data within coastal regions the present study does reveal a definite te mporal and spatial pattern of chlorophyll concentration along the West Florida Sh elf. During the periods of elevated rainfall and riverflow such as El Nio periods (win ter 1997-spring 1998) areas adjacent to major rivers (Hillsborough and Suwannee Rivers) experience increased chlorophyll concentrations. Intense mixing in the upper water column during El Nio periods may also produce upwelling as was suggested in January 1998 and 2003. Therefore during such times increased chlorophyll concentrations may be the result of both rivers and upwelling. During periods of normal riverflow, or in areas not impacted as greatly by riverflow, upwelling conditions appear to be the ma jor underlying factor for increased chlorophyll concentrations (August 1999; SeptemberDecember 2001; October 2003).
40 Knowledge of these underlying mechanisms enables a much better understanding of the variability in chlorophyll concentrations al ong the West Florida Shelf, thus facilitating the development of improved management practices in these waters. In those high or medium flow regions where riverine influenc e is the greatest (Upper or Central regions; Figure 3) periods of increased rainfall, s uch as during El Nio periods, will have much more of an impact on chlorophyll concentration s and ultimately the fisheries and shellfish industries. In future studies I will ide ntify problems in the satellite data due to both CDOM and bottom reflectance in an effort to im prove the understanding of the true chlorophyll patterns and assist managers to a great er degree.
41 CHAPTER 2: GELBSTOFF (CDOM) VARIABILITY ON THE WEST FLORIDA SH ELF (19972003) AND IMPACTS ON BIO-OPTICAL ALGORITHMS 2.1 Introduction In coastal areas, chlorophyll concentrations deriv ed from satellite data may be contaminated by numerous factors. Contributors inc lude Colored Dissolved Organic Matter (CDOM) or gelbstoff, suspended sediments, bo ttom reflectance, seagrasses, benthic micgroalgae and macroalgae. These can chan ge the perceived reflectance of the water and interfere with the atmospheric correction of the satellite data and with biooptical inversion algorithms (Hu et al 2000). CDOM is derived from the breakdown of organic mater ial in terrestrial and aquatic ecosystems (Kirk, 1994). CDOM orginates fr om either land drainage or as a byproduct of algal cell degradation during primary pr oductivity, cellular DOC excretion, and zooplankton grazing (Fogg et al 1958; Yentsch et al 1962; Carder et al 1989, 1991, 1993; Momzikoff et al 1994; Siegel et al 1996). There is evidence that in upwelling regions the upwelled water may contain elevated CDO M concentrations thus impacting the color and satellite signal (Carder et al ., 1991; Coble et al ., 1998; Siegel et al ., 2002). Therefore, rivers, upwelling and other processes su ch as local remineralization of organic matter in the water column or underlying sediments may influence the CDOM concentrations in particular coastal waters.
42 Along the West Florida Shelf (WFS), approximately 2 5 rivers release CDOM into shallow coastal waters. Studies within the northea stern Gulf of Mexico (GOM) show that in both moderate to high chlorophyll and river-infl uenced waters, CDOM influences may be substantial (Hu et al 2004). Fluorescence and absorption spectroscopy s tudies have characterized the important influence of riverine i nputs on the optical properties along the WFS (Del Castillo et al 2000). The seasonal variability in river discharg e and the variability in biological activity resulting from v ariable riverine nutrients both impact the CDOM concentrations (Del Castillo et al 2001, Del Castillo et al 2000, Coble 1996). In the shallow, coastal waters of the WFS, satellit es such as SeaWiFS provide good temporal and spatial resolution in observation s of parameters such as chlorophyll and CDOM, but the actual estimated values can have problems because of the factors mentioned above. Studies within Florida Bay reveal that SeaWiFS-estimated chlorophyll concentrations tend to be overestimated with the Se aWiFS OC4 algorithm (DÂ’Sa et al 2002). It is possible to flag and often mask contaminated pixels as images are processed from calibrated radiances to bio-optical products. Hu et al (2000) also derived a method for atmospherically-correcting images over turbid c oastal waters such as near the coast of the WFS. The method is time-consuming and computer -intensive because it searches for valid atmospheric correction parameters in nearby o pen ocean regions and then applies them near the coast, but future studies need to inc orporate new approaches for improved analysis of the variability in chlorophyll concentr ation inherent to coastal areas. There is a need for improvements in the standard SeaWiFS alg orithms to enable application in case-2 environments.
43 This study examines the temporal and spatial variab ility in CDOM along the WFS and identifies impacts on satellite derivation of c hlorophyll concentrations. Identification of areas of elevated CDOM will allow better interpr etation and analysis of remotelysensed chlorophyll estimates. The core hypothesis is that regions along the shelf most impacted by riverflow will experience the greatest influence by CDOM. Periods of elevated rainfall and riverflow would lead to eleva ted CDOM and greater contamination of coastal chlorophyll signals, as suggested in the analysis by Jolliff et al (2003). To test the hypothesis, I based the analysis of sa tellite data on the assumption that CDOM may be identified as areas where nLw(443) valu es are depressed relative to those expected based on the absorption effects of chlorop hyll alone. Chlorophyll concentrations are derived from SeaWiFS data using the Carder semi-analytical algorithm (Carder et al 1999). While there are potential contaminants wit hin these estimates due to other factors such as bottom refle ctance or suspended sediments, the assumption is that the chlorophyll concentrations a re reasonably accurate. The Carder CDOM semi-analytical algorithm also will be used to assess areas of CDOM influence. Seasonal maps will be created to indicate temporal periods when CDOM contamination is a problem in the chlorophyll signal along the WF S. A secondary hypothesis involves a possible relatio nship between riverflow, CDOM, and Red Tides. I hypothesize that during per iods of increased rainfall, riverflow, and CDOM there is a higher incidence of Red Tides. Therefore, we should observe more Red Tides during El Nio periods. To test this hyp othesis I will examine the historical records of Red Tides during the study period (19462003), the riverflow/rainfall data, and CDOM data to determine if a correlation exists
44 2.2 Methods 2.2.1 SeaWiFS ag443 (CDOM) Climatologies Monthly SeaWiFS Level 3 processed files (mapped and binned) for the period 1997-2003 were derived by the Institute of Marine R emote Sensing (IMARS) at the USF College of Marine Science, based on SeaWiFS data pr ocessed using CarderÂ’s semianalytic MODIS algorithm (Carder et al ., 1999). This semi-analytic algorithm is based on a bio-optical model of remote sensing reflectanc e, Rrs( l ). Rrs( l ) values are inserted into the model, the model is inverted, and the vari ous absorption coefficients are computed, including phytoplankton absorption coeffi cients, af( l ), and gelbstoff, ag( l ). Chlorophyll concentrations are also calculated as d escribed by Carder et al (1999). 188.8.131.52 ECOHAB in-situ ag versus SeaWiFS ag_443 In-situ ag data were collected from ECOHAB cruises during Mar ch (1-4), July (58), November (6-8) 1999, and January (11-14), March (1-4), and October (4-6) 2000. The data were collected aboard the R/V Suncoaster and R/V Bellows (Florida Institute of Oceanography) using a Perkin-Elmer Lambda 18 spectr ophotometer according to methods described by Mueller and Austin (1995). Da ta were collected and processed by Jennifer Patch-Cannizaro, Jim Ivey, and Dan Otis, m embers of Dr. Kendall CarderÂ’s lab, and by Robin Conmy and Carlos Del Castillo, members of Dr. Paula CobleÂ’s lab. Monthly in-situ ag_442 and ag_444 data for all ECOHAB stations were averaged for each cruise to simulate ag_443. The satellite ag_443 data were derived from SeaWiFS data using CarderÂ’s MODIS algorithm. Individual sta tion data then were compared to satellite-derived ag_443 at each station. Pearson correlation coeffici ents were calculated using Statistix v.8 to determine the level of agree ment between the satellite estimates and
45 the in-situ estimates. The higher the correlation, the more a ccurate the satellite estimates of ag_443. 184.108.40.206 Nutrients Nutrient samples were collected during all ECOHAB c ruises by members of Drs. Gabriel Vargo and Kent Fanning teams at the Univers ity of South Florida. These samples were collected at the surface and at multiple depth s along the cruise transects, and analyzed for concentrations of phosphate (PO4), silicate (SiO4), nitrite (NO2), and nitrate (NO3), using the WOCE protocol (Gordon et al 1992). Daily surface concentrations were averaged for eac h station on a monthly basis to examine nutrient temporal variability as it relates to the temporal variability in CDOM. Periods when multiple nutrients are elevated above their normal concentrations would suggest either a riverine or upwelling influx. Spa tial patterns observed in satellite imagery can help determine the probable source. Cl ose proximity to the coast and rivers, along with a low salinity signal, would suggest riv erine. High concentrations seen offshore in association with a front, or nearshore under upwelling wind conditions would suggest upwelling. Nutrient concentrations were pl otted for each transect within a cruise and the results then analyzed for overall spatial p atterns and variability. 220.127.116.11 ECOHAB Regional ag_443 Data files of the ECOHAB region (26N, 85W to 29N, 8 2W; Figure 1.1) were created from the SeaWiFS data for the WFS. Monthly ag_443 values were extracted for each pixel along the ECOHAB transects for stations 1-10 (Tampa Bay transect) and 4070 (Fort Myers transect). Seasonal climatological images of CarderÂ’s MODIS
46 chlorophyll and ag_443 were derived and values extracted along the tr ansects (seasons defined in Table 1.1). Figure 2.1. ECOHAB stations. Table 2.1. Seasons as defined in study. An ag_443 value of 0.05m-1 was used as the cutoff point for freshwater riveri ne influence and variations in the distance of this po int from the coast were observed on a seasonal and annual basis. Nababan (2005) reported ag_443 values for several rivers and regions within the Gulf of Mexico ranged between 0. 010.21 m-1 from July 1998-July Season Months Spring Mar 21-Jun 20 Summer Jun 21-Sep 22) Fall Sep 23-Dec 21 Winter Dec 22-Mar 20
47 2000. All river regions studied, including the Mis sissippi, Apalachicola, Suwanee, Tampa Bay, Alabama, Escambia, and Choctawhatchee re gions, had average ag_443 concentrations 0.05 m-1, while within the Central Gulf region average conc entrations were 0.04 m-1. Therefore 0.05 m-1 was established as the cutoff point for river influ ence. The minimum average regional concentration in the a bove study, 0.04 m-1, was found in the Central Region of the northern Gulf aw ay from the coast, while the maximum average concentration, 0.14 m-1 (3-fold increase), was found in the Mississippi River region. The Mississippi River region is defi ned as a high flow rate and large drainage basin region (Stovall-Leonard 2004) which would suggest the 3-fold increase in ag_443 in that region was the result of riverine la nd drainage rather than the result of phytoplankton decomposition or zooplankton grazing (Carder et al 1989, 1991; Vargo et al submitted). 2.2.2 ag_443 Spatial Extension IDL routines were developed within SeaDAS to identi fy the surface area of high ag values along the entire West Florida coast, using t he ag_443 = 0.05m-1 threshold. Monthly and seasonal images were derived to examine temporal changes along the WFS shelf. Images were created for all seasons in the 1 997-2003 period. This period covered multiple climate events of interest (2 El Nios, 1 draught) which may have influenced the bio-optical parameters estimated (Table 2.2).
48 Table 2.2. El Nio and La Nia years between 19972003. Winter is defined as Dec 22-Mar 21. Year Event 1997-1998 Strong El Nio 1998-1999 La Nia 1999-2000 none 2000-2001 La Nia 2001-2002 none 2002-2003 El Nio The area of high ag_443 influence within the ECOHAB region (defined as 26N, 85W to 29N, 82W; Figure 2.1) was estimated by count ing the number of pixels within the area where ag_443 0.05 m-1. Each pixel was 1.1 km x 1.1 km. 2.2.3 ag_443 Seasonal Anomalies. In addition to the seasonal ag_443 composites created for each year (1997-2003), seasonal ag_443 anomalies were computed by comparing each year 's seasonal data to overall (1997-2003) seasonal means (i.e. spring 199 7-2003, summer 1997-2003, etc. Positive anomalies were indications of elevated ag_443 for a specific season. The images were used to identify areas and times where CDOM is a major contaminant in the SeaWiFS signal and thus the chlorophyll concentrati ons. 2.2.4 Riverine-Influence on CDOM concentrations The Apalachicola and Suwanee Rivers have an import ant influence on coastal waters along the northern section of the WFS shelf. Average peak streamflows ranges from 428 m3/sec (Suwanee River) to 1298 m3/sec (Apalachicola River). The Wacassassa and Anclote Rivers have much smaller average peak f lows of 9 to 11 m3/sec. 18.104.22.168 High and Low River Influence Regions
49 To determine the degree of riverine-influence on C DOM concentrations, two areas were chosen. They represent a region with Â“Hi gh-River InfluenceÂ” and a region with Â“Low-River InfluenceÂ”. Based on analysis of s treamflow data from the U.S. Geological Survey (USGS), the Apalachicola/ Suwanee River region (28.5-31 N, 8286 W; Figure 2.2) will be considered one of Â“High-Rive r InfluenceÂ” compared to the Waccasassa/Anclote River region (28-28.5 N, 82-85 W; Figure 2.2), which will be referred to as a Â“Low-River InfluenceÂ” region. Whi le these two regions are in close proximity they were selected due to data availabili ty. SeaWiFS WFS seasonal products were examined to study variability in these regions Figure 2.2. Rivers emptying into coastal waters al ong the WFS. Circled rivers are those discussed in text (Apalachicola/Suwanee River regio n and Waccasassa/Anclote River region). These regions represent High and Low Rive r Influence regions as discussed in text. Rainfall regions (North, Central, South) are also shown (source: Southwest Florida Water Management District (SFWMD)). 2.2.5 Statistical Analyses Central North South
50 Statistical analyses were performed on the dependen t variable, ag_443, and the independent variables, rainfall and riverflow. Cor relations, cross correlations, and multiple regressions were performed using Statistix v.8. 2.2.6 Seasonal / El Nio images To analyze variability between El Nio and non-El N io periods, SeaWiFS data for winters (Dec 22-Mar 20) of 1997-98 and 2002-03, representing two El Nio-So uthern Oscillation periods, were averaged to produce an av erage El Nio winter image. Data for springs of 1998 and 2003 were averaged to produce a post-El Nio spring image (Mar 21-Jun 20). Data for winters of 1998-99, 1999-00, 2000-01, 2001-02, and 2003-04 were also averaged, to produce a non-El Nio winter imag e, and similarly, data for springs of 1999, 2000, 2001, and 2002 were averaged to produce a post-non-El Nio spring image. All summers (Jun 21-Sep 22) during the 1997-2003 period were averaged to produc e a typical summer image, and all fall periods (Sep 23Dec 21) were averaged to produce a typical fall image. 2.2.7 Red Tides Records of Red Tide events occurring within the EC OHAB region of the WFS between November 1946December 2003 were obtained from the Fish and Wildlife Research Institute and Mote Marine Laboratories. R iverflow data were obtained from the USGS and examined for this period to determine if t here was a relationship between riverflow and Red Tide events (counts106 cell/l; Singh 2005). The rivers within the central region, including the Anclote, Alafia, Calo osahatchee, Hillsborough, Manatee, Myakka, and Peace rivers, were analyzed. Seasonal patterns for all seven rivers were similar, so total riverflow was used in the analysi s. In addition, SeaWiFS-derived CDOM
51 concentrations from September 1997December 2003 w ere analyzed relative to the riverflow and Red Tide events during that period. 2.3 Results and Discussion 2.3.1 SeaWiFS ag_443 (CDOM) Climatologies 22.214.171.124 ECOHAB in situ versus SeaWiFS ag_443 Results from comparisons between in-situ and SeaWiFS CDOM concentrations show good agreement for all periods except July 199 9 and January 2000 (Figure 2.3). The correlation coefficient (R) for all data was 0. 63 (N=270). Considering July 1999 and January 2000 as outliers and excluding them from th e analysis increases R to 0.75 (N=193). In both cases P<0.01, indicating these co rrelations were significant. ECOHAB In situ and SeaWIFS ag_4430.02 0.03 0.04 0.05 0.06 0.07 0.08Jan-99Apr-99 Jul-99 Oct-99Jan-00Apr-00 Jul-00 Oct-00 Dec-00 ag (m-1) Insitu ag_443 SeaWiFS_ag_443 Figure 2.3. ECOHAB in-situ and SeaWiFS CDOM. Phytoplankton species exhibit variability in absorp tion and scattering properties thus impacting Rrs calculations. July 1999 was a period of elevated Trichodesmium sp. concentrations at ECOHAB stations 16-22 (Figure 2.1 ) averaging 17.11 colonies/l. Average Trichodesmium concentrations along the WFS during the period 6/9 9-10/00
52 ranged between <1 to 5 colonies/l (Dr. Cindy Heil, personal communication, January 2006). The typical chlorophyll-specific absorption spectru m for Trichodesmium is different from the typical phytoplankton chlorophyl l-specific absorption spectrum. a*(490) and a*(550) for Trichodesmium are approximately 2x those for other phytoplankton (Westberry et al 2005). In addition, the chlorophyll-specific backscattering coefficient for Trichodesmium is approximately 1.4x that for the Â“other phytoplanktonÂ” from 400-600 nm (Westberry et al 2005). In eutrophic waters, CarderÂ’s semi-analytical algor ithm for calculating chlorophyll and CDOM is used in which Rrs(490) and Rrs(555) are employed. Rrs is based on the absorption (ap) and backscattering (bb) coefficients Â– Rrs () = bb / at() + bb() (Del Castillo 2005). Differences in absorption and backscattering coeffi cients as observed for Trichodesmium would increase the Rrs calculations. The Carder a lgorithm employs spectral Rrs ratios to determine ag(400) values, which in turn are used to derive ag(443) values (Carder et al 1999). Therefore the satellite-derived ag(443) values, and chlorophyll concentrations, would be increased in t he presence of Trichodesmium This would account for the disparity between the in-situ and SeaWiFS ag (443) observed in July 1999. 126.96.36.199 Nutrients The annual background surface inorganic nitrogen c oncentrations along the WFS average between 0.02 and 0.2 m M and typical phosphate (PO4) concentrations in the region range from 0.025 to 0.24 m M (Vargo et al submitted). Riverine SiO4
53 concentrations are approximately 10 m M, while oceanic SiO4 concentrations are much lower (Vargo et al submitted). Typical SiO4 concentrations within the WFL region range between 0.4 and 5.5 m M. Analysis of the nutrient concentrations reveal cert ain periods when multiple nutrients (NO2+NO3, PO4 and SiO4) are elevated above their normal concentrations (Table 2.3), suggesting either a riverine or upwell ing influx. Spatial patterns observed in satellite imagery can help determine the probable s ource. Close proximity to the coast and rivers, along with a low salinity signal, would suggest riverine. High concentrations seen offshore in association with a front, or insho re under upwelling wind conditions, would suggest upwelling. Inshore bottom nutrient c oncentration along the WFS, though, are only slightly higher than surface concentration s (Ault 2006) so increased nutrients inshore would likely be riverine. During March 1999, PO4 and SiO4 surface concentrations (0.26 and 5.78 m M respectively) were elevated above the norm at the i nshore station 70 (Table 2.3). PO4 and SiO4 surface concentrations were also elevated in July 1 999 at multiple coastal ECOHAB stations. During November 1999, concentrations of all nutrients examined were elevated at multiple coastal stations (Table 2.3) as well as CDOM concentrations (Figure 2.4a), suggesting a riverine source.
54 Table 2.3. Elevated concentrations for NO2+NO3, PO4, and SiO4 along the WFL shelf. Elevated concentrations are defined as: [NO2+NO3]>0.2 m M, [PO4}>0.24 m M, [SiO4]>5.5 m M. Month/Year Nutrient Elev Conc Station Dist from coast (km) Isobath (m) March 1999 PO4 0.26 70 13 10m 0.28 82 6 10m SiO4 5.78 70 13 10m July 1999 PO4 0.36 72 29 10m 0.26 74 4 10m SiO4 8.63 32 6 10m 5.53 82 6 10m Nov 1999 NO2+NO3 0.24 74 4 10m 0.29 76 8 10m 0.36 78 5 10m PO4 0.26 76 8 10m 0.42 78 5 10m 0.29 80 3 10m SiO4 9.87 51 18 10m 9.00 70 13 10m 6.30 74 4 10m 5.97 76 8 10m Jan 2000 SiO4 7.07 50 26 10m SiO4 13.75 51 18 10m SiO4 11.48 70 13 10m SiO4 8.32 72 29 10m SiO4 8.05 78 5 10m SiO4 7.65 79 4 10m SiO4 5.53 80 3 10m March 2000 NO2+NO3 0.36 11 228 200 PO4 0.28 11 228 200 Oct 2000 NO2+NO3 0.23 9 86 50 0.39 88 117 50 0.41 94 96 50 SiO4 9.89 94 96 50 During October 2000 elevated NO3 + NO2 and SiO4 concentrations were found at station 94 (0.41 and 9.89 m M respectively) located approximately 96 km from th e coast (Table 2.3). SeaWiFS imagery for this period reve als chlorophyll concentrations were also elevated (0.7-7 mg/m3) from the coast out to this station. CDOM concent rations also
55 were elevated (ag_443>0.05 m-1) in this region as observed in the Fall 2000 anoma ly image (Figure 2.6). The nutrients, chlorophyll, an d CDOM results suggest that October 2000 was a period when rivers were the source for a ll. Rainfall and riverflow data for the central region (Figure 2.4b) show that both were el evated 1-2 months prior to October 2000 which would represent the lag time required fo r the rain and riverflow to drain onto the shelf. 188.8.131.52 ECOHAB Regional ag_443 Analysis of satellite-derived CDOM concentrations over time reveals persistent high values along the transect defined by ECOHAB st ations 1-10, in water depths ranging from 183m to 0-30 m inshore. The distance from the coast where ag values 0.05 were found (Figures 2.5, 2.6) varied between E l Nio and non-El Nio years (Table 2.2). During El Nio year, values of 0.05 m-1 were found between 46 km to 95 km from the coast, with the greatest distance during winter 1997. During non-El Nio years the distance of the 0.05 m-1 threshold from the coast was much shorter, averagi ng between 29 and 46km. El Nio years are periods of elevated rainfall and riverflow during the winter and spring months (Hanson and Maul 1991, Sittel 1994a, Livezey et al 1997, Schmidt et al 2001). During El Nio years, increased rainfall le ads to increased riverflow with a possible 1-2 month lag in riverflow (Schmidt et al 2001). The winter of 1997-1998 was a strong El Nio period with increased rainfall and riverflow (Schmidt et al 2001). These events would produce elevated runoff and land drainage thereby introducing increased CDOM into th e coastal waters. The increased riverflow during the periods, as well as wind and c urrent activity, would also cause
56 greater dispersion of the CDOM to greater distances resulting in increased CDOM at multiple locations, not just coastal stations. Thi s increased CDOM may also contribute to contamination of satellite-derived chlorophyll. We seek to test the hypothesis that periods of incr eased riverflow and CDOM result in Red Tide bloom activity (106 cell/l; Singh 2005). Between October 1997March 1998 there were moderate Red Tide blooms from Charlotte Harbor to Venice (Mote 2006). This was during the strong 1997-1998 El Nio when riverflow was high (Schmidt et al 2001). Examination of the monthly ag_443 concentrations fo r the ECOHAB stations 110 and 40-70 reveals elevated CDOM concentrations ( ag_443>0.05 m-1) occurred from October 1997-February 1999, October 1999-January 20 00, August Â–December 2000, August 2001-February 2002, July 2002-April 2003, an d August Â–December 2003 (Figure 2.4a). Peaks were seen during December 1997; March September, and December 1998; November 1999; October and December 2000; October 2 001; December 2002; and December 2003. The longest period of elevated ag_4 43 concentration during the 19972003 period occurred between October 1997 and Febru ary 1999, during which period the 1997-98 El Nio occurred (Figure 2.4b). Comparisons of chlorophyll estimates derived from t he standard NASA OC-4 algorithm (OÂ’Reilly et al 2000) and the Carder semi-analytical algorithm whic h corrects for CDOM in low to moderate chlorophyll concentrati ons ( 1.5-2.0 mg m-3), reveals that the OC-4 estimates average 1.8 times the Carder est imates. During the periods of elevated CDOM listed above, the OC-4 chlorophyll es timates average 2.0 times the
57 Carder estimates. Estimates of chlorophyll using t he standard NASA algorithm will therefore be increased by at least a factor of two during periods of increased CDOM. Monthly ag_443(Stations 1-10, 40-70)0 0.02 0.04 0.06 0.08 0.1 0.12 0.14Sep-97Dec-97 Mar-98 Jun-98 Sep-98Dec-98 Mar-99 Jun-99 Sep-99Dec-99 Mar-00 Jun-00 Sep-00 Nov-00 Feb-01 May-01 Aug-01 Nov-01 Feb-02 May-02 Aug-02 Nov-02 Feb-03 May-03 Aug-03 Nov-03ag_443 (m-1) Sta 1-10 Sta 40-70 El Nino El Nino (a) 0 5 10 15 20 25 30 35 40 45Sep-97Dec-97 Mar-98 Jun-98 Sep-98Dec-98 Mar-99 Jun-99 Sep-99Dec-99 Mar-00 Jun-00 Sep-00Dec-00 Mar-01 Jun-01 Sep-01Dec-01 Mar-02 Jun-02 Sep-02Dec-02 Mar-03 Jun-03 Sep-03Rainfall (cm)0 1 10 100 1000Riverflow (m3/s) N. Rainfall C. Rainfall S. Rainfall Central Rivers El Nino El Nino (b) Figure 2.4a-b. (a) 1997-2003 monthly time series o f SeaWiFS ag_443. Horizontal line represents ag_443 = 0.05 m-1 (b) Time series of riverflow, and rainfall. Rainf all regions are defined in Figure 2.2. Central rivers include Hillsborough, Alafia, and Manatee Rivers. Temporal analysis of CDOM along ECOHAB stations 110 and 40-70 revealed similar patterns for both transects (Figure 2.4a). Analysis of CDOM along ECOHAB
58 stations 1-10, rainfall for the Northern, Central, and Southern regions (Figure 2.4b), and riverflow for the Central Region rivers (Hillsborou gh, Alafia, and Manatee Rivers) shows agreement between periods of elevated CDOM and elev ated rainfall and riverflow, particularly during El Nio periods (Figure 2.4a-b) This is apparent for both the 19971998 and the 2002-2003 El Nio events. While the C DOM remained high during January 2003, the rainfall and riverflow levels dro pped. During non-El Nio years there appears to be a lag before there is an observable effect on the CDOM concentrations in the coastal wa ters (Figure 2.4a) during these periods. From JuneSeptember 1999 peaks in rainfa ll and riverflow (Figure 2.4b) occurred during which times the CDOM concentrations were low (avg=0.033 m-1). CDOM did not peak until October 1999 (0.067 m-1). In September 2000 there was a peak in rainfall and riverflow while CDOM did not peak u ntil October 2000-January 2001 (avg=0.059 m-1). In 2001 the peak CDOM concentrations (0.081 m -1) occurred from October-November while peak rainfall and riverflow occurred earlier, July-September 2001 (Figures 2.4a-b). Finally, in 2003 peaks in r ainfall and riverflow occurred between June and August while CDOM concentrations peaked in December (0.099 m-1). During El Nio periods rainfall and riverflow are i ncreased (Figure 2.4b) and remain high for 5-6 months, while during non-El Ni o years peaks are only 1-3 months (Figure 2.4b). The increased rainfall and riverflo w over an extended period of time during El Nios would allow faster seepage and peak CDOM signals to occur quicker and for extended periods and help explain the lack of a lag period at these times. During the 1997-2003 El Nio, CDOM concentrations remained elevated for 5 months (November-March).
59 The largest ag_443 peak for stations 1-10 occurred in September 1 998 (0.12 m-1). This was a period of increased riverflow in the cen tral region (75.5 m3/s) and rainfall in all three regions (North Â– 28.83cm, Central Â– 30.68 cm, South Â– 26.77 cm) (Figure 2.4b). A Red Tide was also recorded two months later, Nove mber-December 1998 (69,1681,104,373 counts) (FWRI et al 2006). 184.108.40.206.1 High ag_443 Surface Area within ECOHAB To quantitatively analyze the spatial and temporal variability of the ag_443 extension (coastal regions with ag_443 0.05 m-1) in the WFS, we focused on the ECOHAB region (26N, 85W to 29N, 82W). The largest high_CDOM area occurred during the fall and winter seasons for most years ( Figure 2.5). Over the 6 year period the largest areas occurred during winter 1997 (5.6 x 104km2), spring 1998 (4.9x104km2), winter 2002 (4.9x104km2), and winter 2003 (5.2x104km2) (Figure 2.5 ). All periods except winter 2003 were El Nio years when rainfall and riverflow were elevated (Figure 2.4b). Comparisons of CDOM areas reveal that sprin g 1998 (El Nio) had the largest difference over non-El Nio springs (3.4x), while t he spring 2003 El Nio had the smallest difference (1.7x). The winters of 1997 a nd 2002 averaged 2.8x non-El Nio winters. During the weaker 2002-2003 El Nio the size of the area where ag concentrations were high dropped much quicker than seen during the 1997-1998 El Nio. The spring 2003 area of high CDOM along the coast (1.8x105 km2) was smaller than spring 1998 (2.5x105 km2) (Figure 2.5). The largest areal extent did occur during the El Nio periods, winters of 1997 and 2002, as well as the winter of 2003.
60 Ag Extension Area within ECOHAB Region0 5000 10000 15000 20000 25000 30000 35000 40000 45000 50000Fall 97 Winter 97 Spring 98 Summer Fall 98 Winter 98 Spring 99 Summer Fall 99 Winter 99 Spring 00 Summer Fall 00 Winter 00 Spring 01 Summer Fall 01 Winter 01 Spring 02 Summer Fall 02 Winter 02 Spring 03 Summer Fall 03 Winter 03# Pixels0 10000 20000 30000 40000 50000 60000Area (km2) Figure 2.5. ag_extension area (ag 0.05m-1) for ECOHAB region expressed in # pixels and area (km2) 2.3.2 ag_443 Seasonal Anomalies Positive anomalies of the ag_443 seasonal mean along the WFS shelf, i.e. those areas where the ag_443 concentrations were elevated for one season wi thin a single year relative to the seasonal average for the entire stu dy period (1997-2003), were present along the entire WFS shelf from winter 1997 to spri ng 1998, an El Nio period. During these anomalies, concentrations ranged from 0.08-0. 16 m-1 above normal. During the winter 1997 the highest concentrations were present in the Suwannee River and Charlotte Harbor regions (0.16 m-1). By Spring 1998 the highest concentrations had s hifted south from the Suwannee area to the ECOHAB region. High concentrations had also shifted further south to the Florida Bay area (0.13 m-1). A Red Tide was observed in these areas of high CDOM from Charlotte Harbor to Venice betwee n October 1997 and March 1998 (FWRI 2006). El Nio El Nio
61 Positive anomalies also were present in winter 200 2 (El Nio) (0.08-0.12 m-1) and in all seasons during 2003 (Figure 2.6). During th ese periods, though, anomalies were not as widespread and were localized in the Suwannee River region. The results provide additional evidence of the relationship between inc reased riverflow and rainfall, such as during El Nios, and increased ag_443 concentration s. While the 2002-2003 El Nio was ending in spring 2003, the rainfall concentrati ons remained fairly high throughout 2003 which could explain the positive anomalies obs erved throughout 2003. During the summers of 1998 and 1999, positive anom alies (0.05 m-1) occurred offshore approximately between the 30m and 100m iso baths, extending south of the Mississippi River and paralleling the WFS (Figure 2 .6). This is indicative of the Mississippi plumes typically present during the sum mer months (del Castillo et al 2000). The positive anomalies previously present along the coast were replaced by negative anomalies (-0.02 m-1) indicating a reduced influence by WFL rivers at t hese times. By the falls of 1998 and 1999, positive anomalies (0.0 8-0.12 m-1) had developed to the south within Florida Bay. By the winter of 1998 a thin band of positive anoma lies (0.03 m-1) remained along the 30m isobath, with stronger anomalies present wi thin the Suwannee River region (0.06-0.12 m-1). During December 1998 a Red Tide was observed of f the central WFS coastal region. During fall 1999 positive anomalies (0.12 m-1) were present within Florida Bay, extending from Charlotte Harbor to Key West. Red T ide was observed in this region as well from December 1999-May 2000. Red Tide was als o observed from August 2001September 2002 between St. Petersburg and Charlotte Harbor. CDOM anomalies during
62 this time ranged between 0.04-0.12 m-1 (Figure 2.6). From February-October 2003 Red Tide was observed between Tarpon Springs and Marco Island. During all periods of positive CDOM anomalies the areas with the largest anomalies are those adjacent to river regions (Figu re 2.6). These include the Suwannee River region and along the ECOHAB region (central W FS) where the Hillsborough and Manatee Rivers are present in combination with seve ral other rivers (Figure 2.2). The Florida Bay region is another area where large posi tive anomalies occur. Florida Bay is the largest estuary in south Florida and its waters hed extends as far north as Lake Okeechobee. Increased rainfall and riverflow would result in runoff into Florida Bay and produce the positive CDOM anomalies observed. Alon g the WFS in areas where riverine and freshwater influence is high, the potential for elevated CDOM concentrations and thus contamination in the satellite signals is incr eased. 1997 Fall 1997 Winter 1998 Spring 1998 Summer 1998 Fall 1998 Winter
63 1999 Spring 1999 Summer 1999 Fall 1999 Winter 2000 Spring 2000 Summer 2000 Fall 2000 Winter 2001 Spring 2001 Summer 2001 Fall 2001 Winter 2002 Spring 2002 Summer 2002 Fall 2002 Winter 2003 Spring 2003 Summer 2003 Fall 2003 Winter Figure 2.6. ag_443 anomalies for WFL shelf, 1997-2003. Positive anomalies indicate areas of increased CDOM and negative anomalies indi cate areas of reduced CDOM. Negative anomalies, periods of reduced CDOM, were present during the falls of 1997, 1998, 1999, winter of 1999, and all of 2000 ( Figure 2.6). During these times rainfall and riverflow were declining (Figure 2.4b) Overall, the year 2000 was classified as a drought, explaining the low CDOM observed thro ughout the year. While the summer of 2000 experienced a short peak in rainfall and riverflow, as is typical for the summer months, the magnitude and duration were lowe r relative to other summers (Figure 2.4b).
64 2.3.3 Statistical Analyses 220.127.116.11 ECOHAB Region To determine the relationships among riverflow, rai nfall, and CDOM within the overall ECOHAB region, Pearson Correlation analyses were performed between river discharge rates within the central region, rainfall for all 3 regions, and CDOM for the ECOHAB region of the WFS. All 3 rainfall regions w ere included due to the potential impact of the seepage of adjacent rainfall on CDOM concentrations. While the strongest correlation was between the river discharge and CDO M (R=0.37; N=72, P=0.0012), the correlation was still low possibly due to a lag eff ect such as observed in earlier studies between rainfall and chlorophyll concentrations. La gged cross correlations performed between the three rainfall districts and CDOM showe d an increase in correlation with lag times of 4 months, i.e. central rainfall and CDOM (R=0.52, N=72, P<0.0001), northern rainfall and CDOM (R=0.50, N=72, P<0.0001), and sou thern rainfall and CDOM (R=0.52, N=72, P<0.0001). Least squares linear regressions were also performe d on untransformed data using multiple independent variables. These variables in cluded the seven central region rivers and the North, Central, and Southern rainfall distr icts. Results from regressions run with these independent variables and CDOM as the depende nt variable produced an R2=0.47, indicating that less than 50% of the variability in CDOM was explained. Cross correlations were run between the central rivers an d CDOM with the highest correlation occurring when a 3 month lag period was introduced. The CDOM data were then lagged by 3 months and regressions were run again. The R2 was increased to 0.87 indicating that
65 87% of the variability in CDOM can be explained by rainfall and rivers when the lag effect is considered. 18.104.22.168 Regions Exhibiting Influence of High vs. Low Riverf low To determine the degree of riverine influence on C DOM concentrations, I examined a region of high river influence (Apalachi cola/Suwanee) and of low river influence (Waccassassa/Anclote). Correlations betw een ag_443 and chlorophyll derived from SeaWiFS data for both of these regions were ex amined. Both CDOM and chlorophyll contribute to light absorption at 443 n m, and clearly CDOM also leads to higher apparent satellite chlorophyll retrievals (D el Castillo et al 2001). Studying this relationship is of use to help understand how stron gly biased the chlorophyll estimates are. In those areas where there is high river infl uence there would be larger bias in the chlorophyll estimates due to increased CDOM concent rations in those areas relative to low river influence areas. The correlation between satellite-derived chlorophy ll and ag_443 concentrations was slightly higher for the Waccassassa/Anclote reg ion (R=0.54) relative to the Apalachicola/Suwanee region (R=0.29; Figure 2.7). While the Waccassassa/Anclote correlation was higher, results indicate little cor relation for either region. Riverflow for individual rivers was was compared with ag_443 for these regions (Figure 2.7). Riverflow for the Apalachicola and Suwanee rivers h ad greater correlation with ag_443 (R=0.81) than riverflow for the Waccassassa/Anclote region (R=0.43) (Figure 2.7). These results suggest that the higher the riverflow the larger the ag_443 concentration in the adjacent coastal waters.
66 Waccassassa/Anclote River Regiony = 10.36x + 0.1352, R2 = 0.29 R=0.540.00 0.50 1.00 1.50 2.00 0.030.050.070.09 ag_443Chlor Apalachicola/Suwanee River Regiony = 7.3971x + 0.6373, R2 = 0.085 R=0.290.00 0.50 1.00 1.50 2.00 2.50 0.060.070.080.090.10 ag_443Chlor Waccassassa/Anclote Rivers Region y = -0.0005x + 0.0559, R2 =0.021 (Wacc) y = 0.0019x + 0.0555, R2 = 0.19 (Ancl) 0 0.02 0.04 0.06 0.08 0.1 02468101214 Riverflow (m3/s)ag_443 Wacc R Ancl R Apalachicola/Suwanee Rivers Regiony = 2E-05x + 0.0721, R2 = 0.68 (Apal) y = 6E-05x + 0.0743, R2 = 0.66 (Suw) 0.04 0.06 0.08 0.1 0.12 0.14 05001000150020002500 Riverflow (m3/s)ag_443 Suw R Apal R Figure 2.7. Linear regressions between satellite-d erived ag_443 and chlor for 2 river regions and between riverflow and ag_443. Note the differences in scales between graphs.
67 2.3.4 Seasonal / El Nio Images Images produced for the El Nio seasons and non-El Nio seasons reveal increased riverine CDOM areas (ag_443 0.05m-1) during El Nio seasons (Figure 2.8). During winter El Nios within the ECOHAB region, CD OM areas were 51% larger than during non-El Nio winters, while during spring El Nios CDOM areas were 61% larger than during non-El Nio springs. Lowest areas occu rred during the summer for the entire 1997-2003 period, while the average fall concentrat ions (1997-2003) were intermediate between the El Nio years and the summer averages ( Table 2.4). Figure 2.8. Seasonal CDOM averages for El Nio per iods (winter/spring 1997-98 and winter/spring 2002-03) and non-El Nio periods (all other periods 1998-2003). Pixels where ag_443 0.05m-1 are shown in white. Winter non El Nio Winter El Nio Spring El Nio Spring non-El Nio Nno 1997_2003 Fall 199 7_2003 Summer
68 Table 2.4. Seasonal CDOM average extension areas w ithin the ECOHAB region for El Nio and non-El Nio periods. Period CDOM Extension Area (km2) Winter El Nio (1997, 2002) 52,360 Spring El Nio (1998, 2003) 36,347 Winter non-El Nio (1998-2001, 2003) 25,878 Spring non-El Nio (1997, 1999-2002) 14,265 1997-2003 Summer (all) 17,435 1997-2003 Fall (all) 28,689 Once again this would suggest the importance of El Nio periods in terms of CDOM concentrations and the potential impact on the accuracy of chlorophyll concentrations determined from SeaWiFS data. Durin g these periods of increased rainfall and riverflow the propensity for CDOM cont amination in the signal is high and must be considered when evaluating chlorophyll conc entrations. 2.3.5 Red Tides Historical Red Tide events for the ECOHAB region of the WFS between November 1946 and the end of this study period, Dec ember 2003, were compared with riverflow data for the seven rivers within the cent ral region of Florida (as defined earlier). From November 1946-December 2003, 21 Red Tide event s occurred ranging in length from one month or less to eight months (FWRI 2006). The average WFS central region riverflow for the entire period was 58.5 m3/s. Periods when riverflow > 60 m3/s were considered periods of elevated flow. Of the 21 Red Tide events, 13 occurred during or im mediately after peak riverflow periods (62%) (Figure 2.9). In addition, of the 16 El Nios during this time,
69 Red Tide events were reported after the onset of on ly five of them (31%) (arrows, Figure 2.9), suggesting no cause-effect relationship. TOTAL RIVERS vs. RED TIDE vs. El Nio ECOHAB REG ION (1946-1976)0 50 100 150 200 250 300 350 400 450 Nov-46Nov-48Nov-50Nov-52Nov-54Nov-56Nov-58Nov-60Nov -62Nov-64Nov-66Nov-68Nov-70Nov-72Nov-74Nov-76Monthm3/s Red Tide El Nino TOTAL RIVERS vs. RED TIDE vs. El Nio ECOHAB REGI ON (1977-2003)0 50 100 150 200 250 300 350 400 450 500 Jan-77Jan-79Jan-81Jan-83Jan-85Jan-87Jan-89Jan-91Jan -93Jan-95Jan-97Jan-99Jan-01Jan-03Monthm3/s Red Tide El Nino Figure 2.9. Total riverflow for central region alo ng the WFS versus red tide and El Nino events from November 1946December 2003. Rivers i ncluded are described in text. Arrows show where El Nios precede Red Tides. Red tides during 1997-2003 were concentrated betwee n 26-28 N and extending out from the coast to the 100 m isobath (Figure 2.1 0a). From October 1997-March 1998 moderate-high blooms (peak = 7,539,464 counts) were recorded along the WFS from Charlotte Harbor to Venice (Figure 2.10b). This wa s during the strong 1997-98 El Nio when riverflow was elevated (Figure 2.4b). From October 1999-May 2000, moderate blooms (peak = 534,381 counts) occurred from Charlotte Harbor to Key West. In 200 1, a large Red Tide began in August 2001 and continued through September 2002, extendin g from St. Petersburg to Charlotte
70 Harbor producing massive fish kills (Mote 2006). I n August 2001, K. brevis cell counts were as high as 43,430,000 in Charlotte County and by September 2001, counts reached 105,000,000 in Sarasota County (FWRI 2006). This w as not an El Nio period, but rainfall and riverflow were elevated from during th at period from August 2001 Â– October 2001. From November December 2002 there were moderate t o high blooms recorded on the east coast, from Vero Beach to Cape Canavera l. Numerous studies have shown that east coast Red Tide blooms are initiated on th e west coast and then transported to the west coast via the Loop Current or spinoff eddies ( Murphy et al 1975, Roberts 1979, Tester et al 1991, Tester et al 1997). November-December 2002 was another El Nio period when rainfall and riverflow were increased. At the end of this El Nio period in February 2003, medium to high blooms remained betwe en Tarpon Springs and Marco Island and until October 2003. (a) (b) Figure 2.10a-b. (a) Red Tide events along the WFS. Green dots represent all events between 1946-2002. Yellow dots represent events be tween 1997-2002 with K. brevis counts 1000). (b) Red Tide events between 9/97-4/98 (El Nino). Data source: FWRI.
71 During the 1997-2003 period, riverflow patterns cor responded with CDOM patterns. CDOM peaks occurred during, or slightly lagging behind, riverflow peaks (March 1998, September 1998, November 1999, Septemb er 2000, August 2001, and January 2003) (Figure 2.11). Peaks in both occurre d during the 1997-98 and 2002-03 El Nios. Of the five Red Tides that developed during 1997-20 03, only two (December 1998, February 2003) started after periods of eleva ted CDOM concentrations and riverflow (Figure 2.11). While the data are limite d, with only 40% of the Red Tides occurring after peaks in riverflow and CDOM, this w ould suggest that there are other factors besides riverflow and CDOM involved in the development of Red Tides. Riverflow vs CDOM vs Red Tide vs El Nio (1997-2003 )0 50 100 150 200 250 300 350 400 450 500Sep-97 Feb-98 Jul-98 Dec-98 May-99 Oct-99 Mar-00 Aug-00 Jan-01Jun-01 Nov-01 Apr-02 Sep-02 Feb-03 Jul-03 Riverflow (m3/s)0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08CDOM (m-1) CENTRAL RIVERS CDOM Red Tide El Nino Figure 2.11. Total riverflow for central region al ong the WFS versus CDOM, El Nino, and red tide events from September 1997December 2 003. Rivers included are described in text. Arrows mark Red Tides that occu rred at, or immediately after, the onset of peaks in riverflow and CDOM. 2.4 Conclusions Satellite data permit the study of temporal and spa tial variability in CDOM concentrations along the WFS. We observed a large increase in CDOM concentrations close to the coast of the WFS relative to offshore waters. These coastal waters are adjacent to numerous river sources which would sugg est the CDOM source was riverine.
72 In addition, in the high-river discharge region, su ch as the Apalachicola/Suwannee area, we found a stronger correlation between riverflow a nd CDOM concentrations than in low-river discharge regions. While phytoplankton d ecay is another potential source of CDOM, the large river influence along the WFS would suggest rivers are the predominant source. Analysis of nutrient data for the ECOHAB region fro m 1999-2000 provides additional evidence that freshwater was a likely fa ctor underlying periods of increased CDOM concentrations during this period. During Mar ch 1999, November 1999, and January 2000, when CDOM concentrations were high, s urface phosphate, nitrate+nitrite, and silicic acid concentrations were elevated above normal levels at multiple coastal locations (3-29km off coast). In particular, silic ic acid concentrations averaged between 6-10 m m which is indicative of typical riverine concentra tions (~10 m m). In addition, examination of nutrient and current profiles did no t indicate upwelling during these periods (Vargo et al submitted; USF OCG, http://ocgweb.marine.usf.edu/ ). During April 2002 the CDOM concentrations along the coast were high, while the overall CDOM concentrations for the entire ECOHAB r egion were low. The elevated concentrations at the coast may be partially due to the elevated riverflow from the Suwannee and Apalachicola Rivers during March-April 2002. While these rivers are on the northern portion of the WFS shelf, during this period there was an area of low Sea Surface Height (SSH) (-30cm) located at 29-30 N, 84-85 W, and an area of higher SSH (0-2cm) to the west, at 29-30 N, 86-87 W (CCAR, http://ccar.colorado.edu ). The resulting pressure gradient, combined with the Cori olis effect, would have created a southeastward geostrophic current towards the ECOHA B region.
73 These conditions could have led to the introduction of riverine CDOM from these northern rivers into the ECOHAB region, producing t he elevated CDOM concentrations observed strictly along the coast within this regio n. Further analytical and visual analysis of the ECOHA B region within the WFS shelf, provides evidence as to the role of increase d rainfall and riverflow, indicative of El Nio conditions, in the spatial and temporal patter ns of CDOM concentrations observed. The largest concentrations of riverine CDOM (ag_443 0.05m-1) occurred during El Nio periods Â– winter 1997, spring 1998, and winter 2002 (Figure 2.4a). Increased riverflow and rainfall during these periods would e nable the transport of larger concentrations of CDOM offshore than during non-El Nio periods as evidenced by the greater surface area covered by coastal waters with ag >0.05 m-1(Figure 2.8). Cross correlations between rainfall and CDOM, and r iverflow and CDOM, reveal a 3 month lag period for the overall study period. While during the El Nio periods there was a direct correlation between the factors above, a lag period was observed during the non-El Nio periods. During the overall study peri od only 16% (12 of 76 months) existed under El Nio conditions, while 84% (64 of 76 months) existed under non-El Nio conditions which would explain the lag period observed for the overall study period. Studies within the ECOHAB region have shown that du ring normal non-El Nio years CDOM is temporally integrated from multiple s ources into Tampa Bay before being released into the coastal waters of the Gulf. The result is a lag effect in which peak riverflow occurs during the summer months and incre ased CDOM concentrations during
74 the winter months. Multiple linear regression ana lysis indicates that rainfall and riverflow combined explain 87% of the observed vari ability in CDOM concentrations. The seasonal and regional variability ag maps will help identify areas where and periods when CDOM may be problematic in the SeaWiFS bio-optical algorithms, influencing the chlorophyll concentrations. Increa sed CDOM concentrations are most prevalent along the WFS during the winter months an d in areas of high river influence, suggesting caution in the use of the OC-4 algorithm to estimate chlorophyll at these times and in these locations. In addition, during future El Nio periods when riv erflow is increased managers for the WFS shelf waters can use these maps as refe rences for identifying areas where potential conflicts with the satellite data exist, enabling more accurate evaluations and decisions affecting the coastal biology. Use of th e Carder semi-algorithm during periods of elevated CDOM may reduce the overestimations in chlorophyll observed with the standard NASA OC-4 algorithm. More accurate estima tes of satellite-derived chlorophyll within coastal waters will lead to an i mproved understanding of the ecological variability in these productive waters a nd a better estimate of their role in global productivity. Identification of areas and p eriods of increased CDOM may also contribute to our understanding of Red Tides and th e underlying mechanisms. Analysis of riverflow patterns, El Nios, and Red T ides over a 57-yr period reveals a slightly higher incidence of Red Tides fo llowing increased riverflow (62%). While El Nios are periods of increased rainfall an d riverflow, only 11% of the Red Tides occurred after El Nios. Additional factors in the generation and maintenance of Red Tides include such additional nutrient sources as benthic fluxes of remineralized
75 nutrients, zooplankton excretion, and the decay and remineralization of dead fish killed by K. brevis (Vargo et al submitted, Walsh et al press). Groundwater discharge also has been suggested as another nutrient source in the ge neration of Red Tides (Hu et al 2006). Additional studies have proposed a connection betwe en Saharan dust and Red Tides (Walsh et al 2005). Lau et al (2006) examined the Atlantic Oscillation and revea led that during positive oscillations rainfall is blocked. The result is draught conditions and increased dust concentrations. During the 1970s th ere was a major Sahel draught and the subsequent development of a large Red Tide off the WFL between Tampa Bay and Fort Meyers. Walsh et al (2001a, 2001b, 2002, 2005) hypothesized that Red T ides are the result of a sequence of physical and ecological events whi ch include a phosphorus-rich nutrient supply, aeolian iron from the Saharan Desert, and a subsequent Trichodesmium bloom with the release of Dissolved Organic Nitrogen (DON ). The DON initiates toxic dinoflagellates at the bottom of the euphotic zone, onshore currents then upwell these populations to CDOM-rich surface waters in coastal waters, resulting in the small initial blooms of Red Tides. The additional nutrient sourc e from dead fish results in the development of a large Red Tide (Walsh et al 2003) Our analysis of the 1997 to 2003 period reveals on ly 40% of the Red Tides occurred immediately after the onset of peak riverf low and peak CDOM concentrations, further suggesting that while riverflow and CDOM ma y play a role in Red Tides, there are other factors as well. Dixon et al (2003, 2004) did show that riverflow and rainfall were correlated with Karenia brevis concentrations along the WFS, particularly the
76 central region. Walsh et al (in press) have shown that CDOM serves as a sunscr een, alleviating light inhibition on K. brevis growth. While our study has shown that riverflow appears to be a possible factor underlying Red Tide development, the mechanism is v ery complicated. The mechanism for intitiating blooms also may be different than t he mechanism for sustaining them. Future research is needed to fully understand all t he factors and the potential role of riverflow and CDOM in Red Tides. Knowledge of the temporal and regional patterns of CDOM as revealed in this study, though, may be an i mportant link in unraveling the mystery of the Red Tide, as well as gaining a bette r estimate of coastal chlorophyll concentrations and thus a better understanding of c hlorophyll patterns and the overall ecology within coastal communities.
77 CHAPTER 3: TEMPORAL AND SPATIAL IMPACTS OF BOTTOM REFLECTANCE ON BIOOPTICAL ALGORITHMS IN THE WEST FLORIDA SHELF 3.1 Introduction Clear, shallow coastal areas frequently fall into bio-optical case-II waters due to the contribution of bottom reflectance to satellite radiance measurements. This contribution "contaminates" bio-optical algorithms that assume all optically-active constituents reside within the water column (Marito rena et al 1994; Morel and Prieur 1997; OÂ’reilly et al 1998; Hu et al 2000; Lee et al 1998; Lee et al 1999; Lee et al 2002). Indeed, bottom reflectance may be a major but under studied factor in optically-shallow conditions of the West Florida Shelf (WFS) (Canniza ro et al 2006; Lee et al 1999; Carder et al 1993). Seasonal variation in factors affecting wa ter clarity, the depth of the water, and substrate-type all determine the amount of bott om-reflectance and the degree of contamination of the satellite signal. Satellite-se nsors such as the SeaWiFS and MODIS seek to provide good temporal and spatial resolutio n of parameters such as chlorophyll and colored dissolved organic matter (CDOM), yet sh allow waters can be problematic and affect such time series. The purpose of this study is to examine the effect of bottom reflectance on satellite-derived chlorophyll concentrations. I wi ll examine the temporal and spatial variability in bottom reflectance along the WFS and assess the impact on the chlorophyll
78 concentrations. Identification of areas and times of bottom reflectance will allow better interpretation and analysis of chlorophyll estimate s from space. The core hypothesis of the study is that contaminat ion due to bottom reflectance is greatest in shallow waters of the WFS during cal m periods, particularly during winter months when chlorophyll concentrations in these are as and river discharge are at a minimum. 3.2 Methods 3.2.1 Remote Sensing Reflectance and Semi-Analytical Mode ls A technique has been developed by Cannizzaro et al. (2006) for identifying satellite image pixels contaminated by bottom refle ctance. Cannizzaro et al (2006) collected in-situ data from the West Florida Shelf (WFS) and around the Bahamas in 1998-2001. Above-water measurements of upwelling r adiance (Lu) and sky radiance were used to calculate water-leaving radiances (Lw()). Downwelling irradiance (Ed()) was then used to derive above-water remote-sensing reflectance (Rrs()) from: Rrs() = Lw () / Ed() (Equation 1) Considering refraction and reflection effects at t he water-air interface it is possible to relate these above-surface measurements to below-surface measurements (rrs()) via the following Rrs() ~ 0.5 rrs (,0+) / (1 -1.5 rrs (,0-) (Equation 2) Using the semi-analytic model developed by Lee et al (1999), it is possible to determine water column and bottom components of the rrs values. Through inversion of the model, values for phytoplankton absorption, CDO M absorption, particulate
79 backscattering, bottom albedo at =550 ((555), as well as water depth are estimated. These values are then re-inserted into the model to determine rrs(555). The percentage of rrs(555) due to bottom, %bt_555, is determined by calc ulating the ratio rrs(555)bot / rrs(550)tot. Optically shallow waters are defined here as tho se with %bt_555 25%, while optically deep waters have %bt_555 < 25%. 3.2.2 Algorithm Parameters Band ratios involving Rrs(412), Rrs(490), Rrs(555), and Rrs(670) were evaluated for accuracy in deep and shallow water environments Results are consistent with the expectation that Rrs(670) is not impacted as severely by bottom influen ces as Rrs(555). Cubic polynomial regressions were determined betwee n the Rrs band ratios and chlorophyll concentrations to find the lowest error (Cannizzaro et al 2006). While in optically deep waters Rrs()/Rrs(555) produced lower errors. For the complete data set with both optically deep and shallow waters, Rrs(412)/Rrs(670) produced the best results. Therefore in optically shallow case-I waters where CDOM concentrations are low to moderate, or co-vary with chlorophyll, chlorophy ll concentrations can be determined more accurately using Rrs(412)/Rrs(670). The traditional ratio, Rrs(490)/Rrs(555) (or other algorithms such as the OC4v4, O'Reilly et al. 2000), can be used to derive chlorophyll concentrations in optically deep waters. The bestfit cubic polynomial functions were determined between these ratios and chlorophyll con centrations for the existing data sets and used to calculate chlorophyll in either deep or shallow waters (Table 1). For transitional waters between the optically deep and optically shallow waters, chlorophyll concentrations are determined using a combination o f the above ratios (Cannizzaro et al 2006).
80 Table 3.1. Coefficients were determined empiricall y from cubic polynomial functions (Cannizzaro et al 2006). To determine whether a pixel is optically deep, sha llow, or transitional, the spectral curvature about Rrs(555) was used to indicate the amount of bottom inf luence. This curvature value (CURVE) was calculated accordi ng to: [Rrs(412)*Rrs(670)] / Rrs(555)2 (Equation 3) As the bottom influence increases, the Rrs(555) increases, therefore decreasing the CURVE value. So, in shallow waters where bottom in fluence is high, the CURVE values will be low. Upper and lower thresholds were devel oped based on the CURVE values and a weighting factor (w) was calculated based on those thresholds (Figure 3.1). Figure 3.1. Upper and lower threshold CURVE (y-axi s) values (dotted lines) were established based on the best-fit quadratic polynom ial function for optically deep data (%bt_555 < 25%). These CURVE values were used to determine t he weighting factor, w (w=(CURVEmeas-CURVElower) / (CURVEupper-CURVElower)). Figure courtesy of Cannizaro et al (2006). Water-type Chlorophyll Algorithm R Opt Shallow Log(chl)=0.8840-2.0837log(R)+1.3061(R)20.3906(R)3 Rrs(412)/ Rrs(670) Opt Deep Log(chl)=0.0597-2.2291log(R)+2.6691(R)23.4144(R)3 Rrs(490)/ Rrs(555) Transitional Chl=w(Chldeep)+(1-w)(Chlshallow)
81 Optically deep, shallow, or transitional waters we re then classified based on a combination of values for w (weighting factor), Rrs(412)/Rrs(670), Rrs(490)/Rrs(555), and Rrs(412): Optically Shallow: w<0 and Rrs(412)/Rrs(670) Â£ 100 or Rrs(412) 0.015 Transitional: 0 Â£ w Â£ 1, Rrs(412)/Rrs(670) Â£ 100 Optically deep: All other conditions 3.2.3 Applying algorithms Seasonal climatologies were developed for the WFS u sing SeaWiFS data for the period September 1997-December 2003. Water leavin g radiances for the six SeaWiFS wavelengths (412, 443, 490, 510, 555, 670) were ext racted and the following remote sensing reflectance values were calculated: Rrs(412) = nLw(412)/170.79 (Equations 4a-d)) Rrs(490) = nLw(490)/193.66 Rrs(555) = nLw(555)/185.33 Rrs(670) = nLw(670)/153.41 where the denominators are values of extraterrestri al normalized solar irradiance values, E0 d ( l ). Band ratios of these Rrs values were then calculated for use in the differe nt chlorophyll algorithms as described in Table 3.1. Using the conditions described earlier, individual pixels within an image were classified a s optically deep, shallow, or transitional. The appropriate algorithm was then a pplied to that particular pixel to calculate the chlorophyll concentration. Chlorophy ll in the resulting image was
82 calculated using a combination of all three algorit hms, or what is termed the Â“blend algorithmÂ”. 3.2.4 Satellite Algorithm Comparison: OC-4 versus Blend Seasonal chlorophyll concentration climatologies (1 997-2003) were developed for the WFS using the OC-4 v 4 bio-optical algorithm de veloped by OÂ’Reilly et al (2000). Chlorophyll concentrations calculated with the blen d algorithm (Table 3.1) were compared with those concentrations derived using th e OC-4 for temporal and spatial correlation. The differences between the two algorithms were det ermined by subtracting the blend concentrations from the OC-4 concentrations. Positive differences indicated higher concentrations determined by the OC-4 algorithm rel ative to the blend algorithm and negative differences indicated higher concentration s by the blend relative to the OC-4. 3.2.5 In-situ versus satellite-derived chlorophyll In-situ chlorophylla concentration data were obtained from the ECOHAB (Ecology and Oceanography of Harmful Algal Blooms) program at the University of South Florida (USF) College of Marine Science, led by Dr. Cynthia Heil (Florida Marine Research Institute) and Dr. Gabriel Vargo (USF). Daily surface chlorophyll data from June 1998 Â– Nov ember 2000 were averaged on a monthly basis. These results were compared to satellite chlorophyll concentrations derived from both the OC-4 algorithm and the blend algorithm to determine the effectiveness of each algorithm. Root mean square errors (RMS) were calculated between each algorithm and the in-situ data as follows: RMS = 1/n( x )2 x 100 (Equation 5)
83 where x = (S-I)/I (S=satellite data; I=in-situ data) 3.2.6 CURVE values CURVE values were determined for the WFS as a measu re of bottom reflectance, with low CURVE values indicating high bottom reflectance Values below the 25th percentile (0.21) were considered periods of large bottom infl uence. These values were plotted over the time series to indicate periods when bottom ref lectance was high and therefore a possible source of contamination in the satellite c hlorophyll estimates. 3.3 Results 3.3.1 OC4 versus Blend-Derived Seasonal Chlorophyll conce ntrations Seasonal chlorophyll concentrations were calculated for the WFS region using both the OC-4 algorithm and the blend algorithm as described above. Throughout the time series, the OC-4 algorithm results were consis tently higher (averaging 28%) than the blend chlorophyll estimates (Figure 3.2). This wo uld suggest that either the OC-4 algorithm was overestimating chlorophyll due to bot tom reflectance, or that the blend algorithm was underestimating chlorophyll during th ese periods. Average Seasonal Chlorophyll OC4 and Blend 0 1 2 3 4Spr98 Sum98 Fall98 Win98 Spr99 Sum99 Fall99 Win99 Spr00 Sum00 Fall00 Win00 Spr01 Sum01 Fall01 Win01 Spr02 Sum02 Fall02 Win02 Spr03 Sum03 Fall03 Win03 OC4_Chlor Blend_Chlor AveragesOC-4 =1.80 0.63 mg m-3 Blend =1.300.48 mg m-3 Figure 3.2. Average seasonal chlorophyll concentra tions for the WFS determined using the OC-4 algorithm and the Blend algorithm. Error bars represent the standard deviation for each time series.
84 The greatest differences between the two algorithms occurred during spring 1998 (El Nio period), winter 1998, summer/fall 2001, an d summer 2003 (Figure 3.3). During fall 2003 the blend chlorophyll estimate was actual ly higher than the OC-4 estimate resulting in a slightly negative difference value. This would suggest that all pixels recovered were optically deep (i.e. high chlorophyl l or CDOM and/or no retrievals for shallow water). It should be noted, though, that i n optically deep waters the blend algorithm uses a modified OC-2 algorithm, not OC-4, which may account for the slightly negative difference value. The average RMS differe nce between the two algorithms was 13%. -0.50 0.00 0.50 1.00 1.50Spr98 Sum98 Fall98 Win98 Spr99 Sum99 Fall99 Win99 Spr00 Sum00 Fall00 Win00 Spr01 Sum01 Fall01 Win01 Spr02 Sum02 Fall02 Win02 Spr03 Sum03 Fall03 Win03Difference0 0.1 0.2 0.3 0.4 0.5Avg_Curve Difference Avg_Curve Figure 3.3. Differences between seasonal chlorophy ll for the WFS as determined by the OC-4 and the Blend algorithms (OC-4 minus Blend). The average seasonal curvature about Rrs(555) is shown as a measure of bottom infl uence. CURVE = [Rrs(412)*Rrs(670)] / Rrs(555)2 Periods of large differences and increased bottom influence are circled. Horizontal line represents the 25th percentile for CURVE (0.21). 3.3.2 CURVE Values Analysis of the CURVE estimates, which are an indic ator of bottom reflectance, suggests that bottom influence typically occurs dur ing the winter months (low CURVE values) (Figure 2). In 1998, though, the bottom in fluence is also large during the spring months, which was the end of the strong 1997-98 El Nino. Throughout the study period
85 bottom influence was large (CURVE0.21) during spring 1998, winter 1998, fall 1999, winter 2001, winter 2002, fall 2003, and winter 200 3 (Figure 3.3). In addition, the greatest differences between the O C-4 and the blend chlorophyll estimates occur when the bottom influence is greate st as indicated by low curve values (spring and winter 1998, fall 2001; Figure 3.3). T he large differences during periods of increased bottom reflectance are due to the removal of the bottom influence by the blend algorithm, suggesting that the blend algorithm chlo rophyll estimates are more accurate for locations with bottom influence. 3.3.3 In-situ versus satellite-derived chlorophyll Analysis of in-situ chlorophyll and satellite-deriv ed chlorophyll reveals good overall agreement (Figure 3.4). Throughout the tim e series, though, the blend algorithm is in better agreement than the OC-4 algorithm. Th e average RMS error between the OC4 algorithm and the ECOHAB in-situ chlorophyll was 20% while the average RMS between the blend algorithm and ECOHAB in-situ chlo rophyll was 11%. This could be explained by the fact that the blend algorithm was regionally tuned for the area. During October 1999 there was a large peak in the i n-situ chlorophyll that is not reflected in either the OC-4 or blend measurements. This was the period of a large red tide bloom (up to 5,270,000 cells/l ) between Tampa Bay and Charlotte Harbor (FWRI 2006). These blooms are notoriously patchy, so try ing to match a local high concentration with a 1 km pixel is not possible whi ch may account for the disparity between the satellite and in-situ measurements.
86 ECOHAB Region-1 0 1 2 3 4 5Jun98 Jul98 Aug98 Sep98 Oct98 Nov98 Dec98 Jan99 Feb99 Mar99 Apr99 May99 Jun99 Jul99 Aug99 Sep99 Oct99 Nov99 Dec99 Jan00 Feb00 Mar00 Apr00 May00 Jun00 Jul00 Aug00 Sep00 Oct00 Nov00Chlor_a Blend ECOHAB OC-4 Averages OC-4: 0.910.19Blend: 0.43 0.12 Ecohab : 0.630.67 Figure 3.4. ECOHAB in-situ chlorophyll values vers us OC-4 derived chlorophyll and blend-derived chlorophyll. 3.3.4 Imagery Analysis Seasonal images of the CURVE values reveal specific times and locations where the bottom effect is present. The shallow coastal waters off the Suwannee River were consistently areas of large bottom influence (winte rs of 1999, 2000, 2002, 2003; summer 1998; spring 2000 and 2001; fall 2001; spring and s ummer 2002) (Figure 3.5b). Increased bottom effect was also present between Ta mpa Bay and Charlotte Harbor (summer 1998; spring and summer 2002, fall and wint er 2003) (Figure 3.5b). In addition, chlorophyll concentrations as determined via the OC-4 and the blend algorithms were compared for spatial and temporal pattern diff erences (Figures 3.5a & c). Summer 1998
87 Spring 2000 Winter 1999 Winter 2000 Summer 1999 Spring 1999 Winter 1998
88 Fall 2001 Spring 2002 Summer 2002 Spring 2001 Spring 2003 Winter 2002
89 (a) (b) (c) Figure 3.5. Average seasonal chlorophyll estimate s as determined by the OC-4 algorithm (a) and the blend algorithm (c). CURVE e stimates are shown in (b) images. Areas outlined in red in the CURVE images are areas of increased bottom effect. Only seasons where areas of excessive bottom influence ( CURVE0.06) are present are shown. Bathymetry images developed using SeaDAS indicate that depths average less than 8m along the WFS with some areas less than 4m (Figure 3.6). These images reveal two pockets of waters along the coast with depths g reater than 40m. These occur between 28.2-28.5 N and south of 27 N. Analysis of NOAA depth charts indicate that actual depths in these regions average between 4-5m with no waters deeper than 6m. SeaDAS uses the ETOPO2 dataset to generate bathymet ry images, which is a combination of echo-sounder data and altimeter data from GEOSAT and ERS-1. The Winter 2003 Summer 2003 Fall 2003
90 lower resolution and accuracy of satellite altimetr y in shallow waters such as the WFS may account for the errors observed in the bathymet ry images. R. Figure 3.6. Bathymetric image for the central WFS generated from SeaWiFS imagery using SeaDAS software. Depths are based on the ETO PO2 2-minute bathymetry grid from NGDC. 3.3.5 Regional Analyses 22.214.171.124 Suwannee River Region In the coastal waters off the Suwannee River (29.029.4 N, 83.24-83.6 W) chlorophyll concentrations calculated via the OC-4 algorithm averaged 5.19 mg m-3 and ranged between 1.9 and 12.5 mg m-3, while concentrations calculated via the blend algorithm averaged 3.4 mg m-3 and ranged between 1.2 and 7.5 mg m-3 (34% difference in averages) (Figure 3.7). The lower chlorophyll concentrations estimated with the blend algorithm are presumed to be the result of the remo val of bottom influence and consequently are more accurate estimations of chlor ophyll concentration. The RMS difference between the two algorithms was 18%. Suwannee R. Waccasassa R.
91 Avg Seasonal Chlorophyll Suwannee Region OC4 and Blend 0 5 10 15 20Spr98Sum9 Fall98 Win9 Spr99Sum9 Fall99 Win9 Spr00Sum0 Fall00 Win0 Spr01Sum0 Fall01 Win0 Spr02Sum0 Fall02 Win0 Spr03Sum0 Fall03 Chlor (mg m3) Blend OC-4 Averages OC-4 =5.19 2.9 mg m-3 Blend =3.431.7mg m-3 Figure 3.7. Average seasonal chlorophyll estimate s for the Suwannee River region (defined in text) as determined by the OC-4 and ble nd algorithms. The greatest differences between the two algorithms (Figure 3.8) for the Suwannee region, though, did not always occur when bottom influence was the greatest (winters of 1999, 2000, 2003; summers of 1998 and 1 999, spring 1999 and 2000, falls of 2001 and 2003; spring and summer 2002 (Figure 3.5b) ). During winter 1998 the difference was 6.6 mg m-3, although the bottom influence was identified as l ow (curve 0.6). Vanderbloemen (2006) has shown that chloroph yll concentrations were increased (10 mg m-3) in this area as well as CDOM concentrations (>0.0 5 m -1) during this time which may explain the large difference between the algorithms. As CDOM concentrations increase, absorption at the shorter wavelengths ( l =412, 490nm) will increase at a greater rate than at the longer wavelengths ( l =555, 670nm). Rrs(412) and Rrs(490) will decrease, thereby increasing the chlorop hyll estimates when either the Rrs(412)/ Rrs(670) or the Rrs(490)/ Rrs(555) ratio is used with the blend algorithm. Chlorophyll estimates with the blend al gorithm would therefore be increased
92 in either shallow or deep waters due to the impact of CDOM absorption at the blue wavelengths. This would explain the large differen ces observed during this time. During winter 2002 the blend algorithm chlorophyll estimates were larger than the OC-4 estimates (difference= 4) (Figure 3.8) i ndicating that the blend algorithm may have undercorrected for the bottom effect or that t he atmosphere was over-corrected by the algorithm, lowering values at both 412 and 670n m. In high CDOM regions, the effect on Rrs(412) would be greater than Rrs(670), increasing the derived chlorophyll values. This wa s the beginning of the 2002-2003 El Nino when elevated rainfall and riverflow led to in creased chlorophyll and CDOM concentrations, particularly along the coast adjace nt to the mouth of the Suwannee River (Figures 3.5a, c). The increased chlorophyll and C DOM concentrations would have increased the absorption at l =412 and 443nm, reducing the Rrs(412) and Rrs(443). High CDOM concentrations would also reduce Rrs(555) (Cannizzaro et al 2006), and so the overall curve value would not be impacted and the w aters would be correctly identified as optically deep. The resulting chlorophyll concentr ations in such optically deep waters may be inaccurate, though, since the blend algorith m has been regionally tuned. This may explain the negative difference between the OC4 and blend algorithms observed in winter 2002 (Figure 3.8).
93 -4.00 -2.00 0.00 2.00 4.00 6.00 8.00Spr98 Sum98 Fall98 Win98 Spr99 Sum99 Fall99 Win99 Spr00 Sum00 Fall00 Win00 Spr01 Sum01 Fall01 Win01 Spr02 Sum02 Fall02 Win02 Spr03 Sum03 Fall03Difference Figure 3.8. Differences between seasonal chlorophy ll for the Suwannee Region as determined by the OC-4 and the blend algorithms (OC -4 minus Blend). Negative values indicate larger estimates by the blend algorithm re lative to the OC-4. 126.96.36.199 Waccasassa River Region In the coastal waters off the Waccasassa River (28. 0-28.5 N, 82.6-83.2 W) chlorophyll concentrations calculated via the OC-4 algorithm averaged 4.3 mg m-3 and ranged between 2.1 and 9.1 mg m-3, while concentrations calculated via the blend algorithm averaged 3.5 mg m-3 and ranged between 1.3 and 12.9 mg m-3 (20% difference in averages) (Figure 3.9). These lower chlorophyl l concentrations estimated with the blend algorithm are due to the removal of bottom in fluence and consequently are more accurate estimations of chlorophyll concentration. The RMS difference between the two algorithms was 20%. Avg Seasonal Chlorophyll Waccassassa Region OC4 and Blend0 5 10 15 20Spr98Sum9 Fall98 Win9 Spr99Sum9 Fall99 Win9 Spr00Sum0 Fall00 Win0 Spr01Sum0 Fall01 Win0 Spr02Sum0 Fall02 Win0 Spr03Sum0 Fall03 Chlor (mg m3) Blend OC-4 Averages OC-4 = 4.34 1.9 mg m-3 Blend = 3.482.5 mg m-3
94 Figure 3.9. Average seasonal chlorophyll estimate s for the Waccasassa River region (defined in text) as determined by the OC-4 and ble nd algorithms. Bottom influence was observed in this region during the spring/summer of all years, fall 2000-2003, and winter 2002-2003. The g reatest differences between the two algorithms (Figure 3.10) occurred during the summer s of 1998, 2002, and 2003 when bottom influence was great in this area (Figure 3.5 b). While the differences were generally positive due to the larger OC-4 concentra tions relative to the blend concentrations, during summer 1998 and winter 2002 the differences were negative. During these times, chlorophyll concentrations dete rmined with the blend algorithm were larger than those determined by the OC-4 algorithm most likely due to the misidentification of shallow waters as deep waters and therefore the use of the wrong chlorophyll algorithm within the blend algorithm. Presence of seagrass beds may be a contributing factor. -6 -4 -2 0 2 4Spr98 Sum9 Fall98 Win9 Spr99 Sum9 Fall99 Win9 Spr00 Sum0 Fall00 Win0 Spr01 Sum0 Fall01 Win0 Spr02 Sum0 Fall02 Win0 Spr03 Sum0 Fall03 Difference Figure 3.10. Differences between seasonal chloroph yll for the Waccasassa Region as determined by the OC-4 and the blend algorithms. 3.4 Discussion In shallow waters such as the WFS, contamination due to bottom influence impacts satellite estimates such as chlorophyll con centration. By using a combination of remote sensing reflectances (Rrs) at different wavelengths I have been able to iden tify
95 bottom influence. Since bottom influence is greate r at =555, Rrs(555) was used to identify when bottom influence is high (i.e. optica lly shallow waters), low (optically deep waters), or intermediate (transitional) relative to values at 412 and 670nm. Different chlorophyll algorithms were used for each condition resulting in the Â“blend algorithmÂ”. Bottom influence along the entire WFS was found to be greatest during the winter months when CURVE values were low. Chlorophyll con centrations typically were low during these times which would indicate the waters were clearer, allowing the bottom to be seen more readily. Reflectance from the bottom in these shallow coastal waters would interfere with the water-leaving radiances thus aff ecting the chlorophyll estimates. Comparisons of in-situ chlorophyll concentrations with chlorophyll estimates from the blend algorithm versus the OC-4 algorithm revealed better agreement with the blend algorithm (RMS=11%) than the OC-4 (RMS=20%). Comparisons of chlorophyll estimates from both algorithms indicate that the OC -4 estimates were, on average, 28% higher than the blend estimates. The periods when the differences were largest were often when bottom influence was large, suggesting b ottom influence was removed by the blend algorithm. One should be able to Â“tuneÂ” the Cannizzaro algorithm to better perform when using satellite data. Regional analysis of bottom influence revealed var iability in the ability of the blend algorithm to remove bottom effects. The diff erence between the OC-4 chlorophyll estimates and the blend estimates for the Suwannee River region was 34%, while the difference for the Waccasassa River region was 20%. The average slope in the Suwannee River region is 72 while the average slope in the Waccasassa River r egion is 40 A slope of 20 can introduce up to 16% errors in estimates of bot tom albedo (Carder
96 et al 2003) which would mean that a 72 slope would introduce much larger errors in bottom albedo. Bottom albedo is used in the classi fication of optically deep and shallow waters and so errors in its estimation may explain the differences observed in the chlorophyll estimates derived from the blend algori thm and the OC-4 algorithm. In extremely shallow coastal waters (z 5m) where bottom reflectance exceeds the water column reflectance, it is difficult to ob tain accurate chlorophyll estimates with the blend algorithm. Reflectances at 412 and 670nm are impacted more by the bottom in these waters resulting in their misclassification a s deep waters (Cannizzaro et al 2006). This may explain why in winter 2002 the chlorophyll estimates using the blend algorithm were actually higher for the Suwannee River region than the OC-4 estimates. Chlorophyll concentrations were relatively low for this area at this time as estimated by the OC-4 (2.4 mg m-3), which would also suggest waters were clearer and more subject to bottom influence. So bottom influence was increased due t o both the shallow water depth and the increased water clarity resulting from the redu ced chlorophyll within the water column. During high runoff periods with excessive CDOM to chlorophyll ratios, the Cannizzaro algorithm may need Â“retuningÂ” since it w as based on the mean ratio during the ECOHAB cruises for waters less rich in CDOM. While the intent of this study was to examine tempo ral and spatial patterns in bottom influence along the WFS, the use of seasonal averages introduces potential biases. Environmental conditions such as cloudiness may res ult in one season being undersampled relative to another, thus introducing biases within the seasonal averages. As a result, shallow waters may be over-sampled, re sulting in indications of greater bottom influence than is what is true, or vice-vers a.
97 Factors not addressed in the current study include differences in optical properties amongst phytoplankton species. Numerous studies ha ve demonstrated variability in the spectral absorption amongst species, particularly c hlorophyll-specific absorption, due to variability in pigment composition and the package effect (Bricaud et al 1988, Bricaud et al 1981). Changes in the spectral absorption per uni t chlorophyll impact the reflectance spectra, thereby impacting derived chlorophyll and the functioning of the blend algorithm that is based on remote sensing reflectances. Typical phytoplankton along the western edge of the WFS include prokaryotes, prymnesiophytes, and pelagophytes (Qian et al 2003). These phytoplankton are well adapted to these oligotrophic, low-nutrient waters. In the high-nutrient waters along the coast, though, diatoms predominate due to their hig h nutrient-uptake efficiencies (Qian et al 2003, Smayda 1997). Absorption spectra for phyto plankton demonstrate major absorption peaks around 440nm and 675nm due to chlo rophyll a, but the overall absorption spectra will vary based on pigment varia bility and the degree of pigment packaging amongst species. This variability in abs orption will affect the remote sensing reflectances and ultimately the chlorophyll estimat es derived from the blend algorithm. Reflectance spectra for typical phytoplankton along the WFS demonstrate a shift in peaks from the blue wavelengths in low-chlorophy ll waters to the green wavelengths in high-chlorophyll waters. In waters with high co ncentrations of Karenia brevis a toxic dinoflagellate that is responsible for the red tide prominent along the WFS, the peak at the green wavelengths (~570nm) is much lower than i n nonK. brevis waters (Cannizzaro 2004). In addition, the overall amount of reflecta nce across the spectrum is
98 approximately 3-4 times lower in K. brevis waters (>104 cells/l) than in nonK. brevis waters (<104 cells/l), resulting in darker colored waters (Canni zzaro 2004). Trichodesmium is a genus of cyanobacteria that is present within warm oceanic waters such as the WFS often forming bloom conditio ns (Lewis et al 1988). These phytoplankton contain phycourobilin and phycoerythr obilin, phycobilipigments that absorb at 495, 545, and 565nm (Subramaniam et al 1999). They also produce a larger chlorophyll-specific backscattering coefficient acr oss the spectrum (400-700nm) than other WFS phytoplankton (Subramanium et al 1999). Since remote sensing reflectances are derived from absorption and backscattering coef ficients, chlorophyll estimates derived from the blend algorithm in Trichodesmium -rich waters may be affected, especially if they are concentrated in stratified l ayers or at the surface. All algorithms used here assume vertical homogeneity. Shallow wat ers may be misclassified as deep, or vice-versa, resulting in the use of the incorrect c hlorophyll algorithm and inaccurate chlorophyll estimates. Bottom albedo ( r ) is another factor not addressed in this study. A reas with seagrasses have lower albedos (0.02-0.07) across th e spectrum (400-700nm) than sandy bottom areas (0.1-0.2) thereby affecting bottom fla gs (Lee et al 2001; Voss et al 2003; Carder et al 2003; Werdell et al 2003). Variability in the sand type will also impa ct the albedo, i.e. bottoms covered with pristine quartz s and will have higher albedos than muddy quartz sand. The geology of the seafloor of the WFS is nearly in finitely complex and enormously variable (personal communication, Dr. Hi ne, USF). The typical distribution along the central portion of the WFS is a predomina nce of quartz along the coast, shifting
99 to a mixture of quartz and carbonate just offshore, and then an increase in carbonate as you move onto the shelf and then the slope. The ca rbonate constituents also vary as you move from the shelf to the slope, shifting from pri marily mollusks, algae, and ooids, to forams (Hine et al 2001). This variability in bottom type, and there fore bottom albedo, has an impact on the overall reflectances observed for the WFS. Oyster reefs are also prevalent along portions of the WFS and may affect the effectiveness of the chlorophyll algorithms. In th e Florida Springs Coast, the region between the Pithlachascotee and Waccasassa Rivers, oyster reefs are present adjacent to many river mouths ((Wolfe 1990), including the Suwa nnee River. The primary species in this area is the Eastern or American oyster, Crassostrea virginica The appearance of these oyster reefs changes over time. During ebb t ide when exposed to the atmosphere, the reefs appear gray, but when submerged the color becomes greenish-brown due to the presence of the algae. The reefs actually have la yers beneath the greenish-brown layer which vary in color due to the presence of either d etritus (reddish-brown) or shells from ferrous sulfide-rich anaerobic environments (silver -black) (Wolfe 1990). This variability in reef color will impact the remote sensing reflec tances for these areas and thus the chlorophyll estimates. Additional studies of the WFS have revealed the eff ect changes in bottom composition have on bottom albedo. From 1985 to 19 97 there was a decline in seagrass within Florida Bay. This decline led to increased sediment resuspension and turbidity, reflectance, and bottom albedo (Stumpf et al 1999). Such changes are possible in other regions along the WFS where seagrasses predominate, thus potentially impacting the
100 reflectance measurements and the ability to accurat ely estimate chlorophyll concentrations. Seagrass beds may confound all algorithms. Approxi mately 9100 km2 of seagrasses exist along the WFS, with approximately 3000 km2 present in the Big Bend area and 5500 km2 in the Florida Bay region (Iverson et al 1986). The three prominent species found along the Florida Springs Coast/Big B end area are Thalassia testudinum (turtlegrass), Syringodium filiforme (manateegrass), and Halodule wrightii (shoalgrass) (Wolfe 1990). Each of these species is different i n its leaf structure, overall height, and depth distribution. Werdell et al (2003) has shown that there is variability in spect ral albedos for different substrates, including differe nt plant species, particularly between 500-550nm. All of these factors will affect the up ward light field and therefore the satellite-derived chlorophyll estimates. Modelling studies have implicated the importance of seagrass canopies (canopy architectur e and leaf structure) in the ability to accurately determine the irradiance distribution (Z immerman 2003). There is also seasonality for all species with greatest productiv ity typically occurring between April and November, resulting in seasonal impact on chlor ophyll estimates. The ability of the blend algorithm to successfully identify times and locations where bottom influence is large along the WFS will be affected by the factors described above, and is ineffective for waters shallower than 4 or 5m. Here, a depth flag should eliminate all pixels from consideration unless extr emely high chlorophyll shields the bottom (Cannizzaro et al 2006). Such a flag has yet to be developed. In addition, application of the blend algorithm to satellite data has its limitations. The current atmospheric correction algorithms emplo yed with SeaWiFS data provide
101 barely adequate estimates of water-leaving radiance s at the shorter wavelengths (412nm) (Gordon et al 1994). Measurement of water-leaving radiances at the longer wavelengths (670nm) is also difficult due to the large absorpti on by pure water at these wavelengths and the low signal-to-noise ratio. Since these are two primary wavelengths employed by the blend algorithm, the ability to retrieve accura te reflectances using SeaWiFS data may be affected by inaccurate atmospheric corrections. While the results of this study are promising, the ability of the blend algorithm to accurately identify bottom influence along the WFS using SeaWiFS data for all times and locations has its limitations. Modifications in th e algorithm may need to be made to account for the low signal-to-noise ratio at the lo nger wavelengths and the problems at the shorter wavelengths due to the atmospheric corr ection techniques. Variations in bottom albedo, as well as phytoplankton populations present, need to be addressed. In addition, separate analysis of shallow versus deep sites along the WFS would be instrumental in comparisons of the blend and OC-4 a lgorithm. These are areas for future studies since accurate retrieval of chlorophyll con centrations along the WFS requires accurate accountability for bottom effects.
102 REFERENCES Ault, D. 2006. Temporal and Spatial Distribution o f Chlorophyll on the West Florida Shelf. MS thesis, Dept. of Marine Science, University of S outh Florida, St Petersburg, FL. 126 pp. Biggs, D.C., and P.H. Ressler. 2001. Distribution a nd abundance of phytoplankton, zooplankton, ichthyoplankton, and micronekton in the deepwater G ulf of Mexico. Gulf Mex. Sci. 19: 735. Biggs, D.C., A.E. Jochens, M.K Howard, S.F. DiMarco K.D. Mullin, F.E. Mller-Karger, and C. Hu. in press. Eddy-forced variations in on-marg in and off-margin summertime circulation along the 1000 m isobath of the norther n Gulf of Mexico, 2000-2003, and links with sperm whale distributions along the midd le slope. in: New Developments in the Circulation of the Gulf of Mexico, Amer. Geophy s. Union monograph in press. Bricaud, A., A. Morel, and L. Prieur. 1981. Absorpt ion by dissolved organic matter in the sea (yellow substance) in the UV and visible domain s. Limnol. Oceanogr., 26:4353.Brown, O. B., J. W. Brown, and R. H. Evans. 1985 Calibration of advanced very high resolution radiometer infrared observatio ns. J. Geophysical Research 90. 11667-11678. Bricaud, A., A. L. Bedhomme and A. Morel. 1988. Op tical properties of diverse phytoplanktonic species: Experimental results and t heoretical interpretation. J. Plankton Res. 10:851-873. Cannizzaro, J.P. 2004. Detection and Quantificati on of Karenia brevis blooms on the West Florida Shelf from Remotely Sensed Ocean Color Imagery. MS thesis, Dept. of Marine Science, University of South Florid a, St Petersburg, FL. 71 pp. Cannizzaro, J.P., K.L. Carder. 2006. Estimating C hlorophyll a Concentrations from Remote-Sensing Reflectance in Optically Shallow Wat ers. Remote Sensing of Environment. 101:13-24. Carder, K. L., R. G. Steward, G. R. Harvey, and P. B. Ortner. 1989. Marine humic and fulvic acids: Their effects on remote sensing of oc ean chlorophyll. Limnol. Oceanogr. 34:68-81.
103 Carder, K. L., S. K. Hawes, K. A. Baker, R. C. Smit h, R. G. Steward and B. G. Mitchell. 1991. Reflectance model for quantifying chlorophyl l a in the presence of productivity degradation products. J. Geophys. Res. 96:20,599-20,611. Carder, K. L., R. G. Steward, R. F. Chen, S. Hawes, and Z. Lee. 1993. Aviris calibration and application in coastal oceanic environments:tra cers of soluble and particulate constituents of the Tampa Bay coastal plume. Phot ogrammetric Engineering & Remote Sensing, 59(3):339-344.Carder Kendall L., C. Liu, Z. Lee, D.C. English, J. Patten, F.R. Chen, J.E. Ivey, and C.O. Davis. 2 003. Illumination and turbidity effects on observing faceted bottom elements with u niform Lambertian albedos. Limnol. Oceanogr 48(1, part 2). 355-363. Carder, K. L., F.R. Chen, Z.P. Lee, and S.K.Hawes. 1999. Semianalytic ModerateResolution Imaging Spectrometer algorithms for chlo rophyll a and absorption with bio-optical domains based on nitrate-depletio temperatures J. Geophysical Research. 104: 5403-5421. Carder, K.L., C. Liu, Z. Lee, D.C. English, J. Patt on, F.R. Chen, and J.E. Ivey. 2003. Illumination and Turbidity Effects on Observing Fac eted Bottom Elements with Uniform Lambertian Albedos. Limnol. Oceanogr. 48: 355-363. Coble, P. G. 1996. Characterization of Marine and Terrestrial DOM in Seawater using Excitation-Emission Matrix Spectroscopy. Mar. Chem. 52: 325-346. Coble, P. G., C. E. Del Castillo and B. Avril. 199 8. Distribution and optical properties of CDOM in the Arabian Sea during the 1995 Southwes t Monsoon. Deep Sea Research Part II 45:2195Â–2223. Del Castillo, C.E., F. Gilbes, and P.G. Coble. 2000 On the dispersal of riverine colored dissolved organic matter over the West Florida Shel f. Limnol. Oceanogr 45: 1425-1432. Del Castillo, C. E., P. G. Coble, R. N. Conmy, F. E Muller-Karger, L. Vanderbloemen, and G. A. Vargo. 2001. Multispectral in situ measur ements of organic matter and chlorophyll fluorescence in seawater: Documenting t he intrusion of the Mississippi River plume in the West Florida Shelf. Limnol. Oceanogr. 46(7):1836-1843. Ding, K., and Gordon, H. R. 1995. Analysis of the i nfluence of O2 A -band absorption on atmospheric correction of ocean-color imagery. Appl. Opt 34:2068-2080.
104 Dixon, L. K. and K. A. Steidinger. 2004. Correlat ion of Karenia brevis presence in the eastern Gulf of Mexico with rainfall and riverine f low. In: Steidinger, K. A., J. H. Landsberg, C. R. Tomas and G. A. Vargo (eds.), Harm ful Algae 2002. Florida Fish and Wildlife Conservation Commission, Florida Institute of Oceanography, and Intergovernmental Oceanographic Commission of U NESCO, St. Petersburg, Florida, USA, pp. 29-31. Dixon, L.K. 2003. Red Tide Bloom Dynamics with r espect to Rainfall and Riverine Flow. Mote Marine Laboratory Technical Report No. 795. Sarasota, FL. Submitted to the Sarasota Bay National Estuary Prog ram, Sarasota, FL. DÂ’Sa, E. J., C. Hu, F. E. Muller-Karger, and K. L. Carder. 2002. Estimation of colored dissolved organic matter and salinity fields in cas e 2 waters using SeaWiFS: Examples from Florida Bay and Florida Shelf. Proc. Indian Acad. Sci. (Earth Planet Sci.), 111(3):197-207, Sept., 2002. El Sayed, S.Z. 1972. Primary productivity and sta nding crop of phytoplankton. in Chemistry, primary production, and benthic algae of the Gulf of Mexico. Ser. Atlas Mar. Environ, Am. Geogr. Soc. 22: 8-13. Elliott, B.A. 1982. Anticyclonic rings in the Gulf of Mexico. Journal of Physical Oceanography. 12: 1292-1309. Eppley, R.W., 1972. Temperature and phytoplankton g rowth in the sea. Fisheries Bulletin 7, 1063-1085. Feinstein, A. 1956. Correlations of various phenome na with red tide outbreaks on the Florida West coast. Bulletin of Marine Science of the Gulf and Caribbea n 6(3). 209-232. Fogg, F. E., and G. T. Boalch. 1958. Extracellula r products in pure cultures of brown alga. Nature 181:789-790. Fratantoni, P.S., T.N. Lee, G.P. Podesta, and F. Mu ller-Karger. 1998. The Influence of Loop Current Perturbations on the Formation and Evo lution of Tortugas Eddies in the Southern Strait of Florida. J. Geophysical Research-Oceans. 103: 2475924779. FWRI (Florida Wildlife Research Insititute), Mote M arine Laboratory, NOAA, USF College of Marine Science. 2006. Red Tides in Flor ida HAB Historical Database Version 2.0 (1954-2002). [CD-ROM].
105 Gilbes, Fernando, Carmelo Thomas, John J. Walsh, an d Frank E. Muller-Karger. 1996. An episodic chlorophyll plume on the West Florida S helf. Continental Shelf Research. 16: 1201-1224. Gilbes, Fernando, F.E. Muller-Karger, and C.E. Del Castillo. 2002. New evidence for the West Florida Shelf Plume. Continental Shelf Research 22: 2479-2496. Gordon, H. R., and Wang, M. (1994): Retrieval of wa ter-leaving radiance and aerosol optical thickness over the oceans with SeaWiFS: a p reliminary algorithm. Appl. Opt. 33:443-452. Gordon, L.L., J. Jennings, J.C., A.A. Ross, and J.M Krest. 1992. A suggested protocol for continuous flow automated analysis of seawater nutrients in the WOCE hydrographic program and the Joint Global Ocean Flu xes Study. Grp. Tech. Rpt. 92-1. OSU College of Oceanography. Grassl, H. 1976. The dependence of the measured coo l skin of the ocean on wind stress and total heat flux. Boundary Layer Met 10. 465-474. Gyory, J, A.J. Mariano, E.H. Ryan. The loop current OceanSurface Currents. (). http://oceancurrents.rsmas.miami.edu/atlantic/loopcurrent.html. Hagemeyer, Bart. (n.d.). El Nino & La Nina. Retrie ved October 29, 2004 from http://www.floridadisaster.org/hwa03/el_nino.htm Hanson, K., and G. A. Maul, 1991. Florida precipita tion and the Pacific El Nino, 1895Â– 1989. Florida Sci. 54, 160Â–68. He, R. and R. Weisberg. 1999. West Florida Shelf ci rculation and temperature budget for the 1999 spring transition. Continental Shelf Research. 22: 719-748. He, R. and R. H. Weisberg. 2002. West Florida she lf circulation and temperature budget for the 1999 spring transition. Continental Shelf Res., 22:719-748. He, R. and R. Weisberg. 2003. West Florida Shelf ci rculation and temperature budget for the 1998 fall transition. Continental Shelf Research. 23: 777-800. Hine, A.C., G.R. Brooks, R.A. Davis, L.J. Doyle, G. Gelfenbaum, S.D. Locker, D.C. Twitchell, and R.H. Wiesberg. 2001. A Summar y of Findings of the West-Central Florida Coastal Studies Project. USGC Open File Report 01303. 41 pp. Hofmann, E.E. and S.J. Worley. 1986. An investigat ion of the circulation of the Gulf of Mexico. Journal of Geophysical Research. 86(C5): 4186-4192.
106 Hu, C., K.L. Carder, and F.E. Muller-Karger 2000. Atmospheric correction of SeaWiFS imagery over turbid coastal waters: a pract ical method. Remote Sens. Environ. 74: 195-206. Hu, C., F.E. Muller-Karger, D.C. Biggs, K.L. Carder B. Nababan, D. Nadeau, and J. Vanderbloemen. 2003. Comparison of ship and sat ellite bio-optical measurements on the continental margin of the NE Gu lf of Mexico. Int.J.Rem.Sensing 24: 2597-2612. Hu, C., E. T. Montgomery, R. W. Schmitt, and F. E. Muller-Karger. 2004. The Amazon and Orinoco River plumes in the tropical Atl antic: Observation from space and S-Floats. Deep-Sea Research-II, 51:1151-1171. Hu, C., F.W. Muller-Karger, and P.W. Swarzenski. 2 006. Hurricanes, Submarine Groundwater Discharge, and FloridaÂ’s Red Tides. Geophysical Research Letters. 33: L11601. Iverson, R. L. andH.F. Bittaker. 1986. Seagrass di stribution and adundance in Eastern Gulf of Mexico. Estuarine Coastal and Shelf Science. 22: 577-602. Jochens, A.E., S.F. DiMarco, W.D. Nowlin, Jr., R.O. Reid, and M.C. Kennicutt II. 2002. Northeastern Gulf of Mexico Chemical Oceanography a nd Hydrography Study: Synthesis Report. U.S. Dept. of the Interior, Miner als Management Service, Gulf of Mexico OCS Region, New Orleans, LA. OCS Study MMS 2002-055. 586 pages. Jolliff, J.K., J.J. Walsh, R. He, R. Weisburg, A. S tovall-Leonard, P.G. Coble, R. Conmy, C. Heil, B. Nababan, H. Zhang, C. Hu, and F.E. Mull er-Karger. 2003. Dispersal of the Suwanee River plume over the West Florida Sh elf: Simulation and observation of the optical and biochemical conseque nces of a flushing event. Geophysical Research Letters. 30(13): 1709-1712. Kirk, J. T. O. 1994. Light and hotosynthesis in a quatic ecosystem. 2nd ed. Cambridge University Press. Cambridge, 509 pp. Lau K.M., S. Shen, K.-M. Kim and H. Wang, 2006: A Multi-model Study of the 20th Century Simulations of Sahel Drought from the 1970s to 1990s. J. of Geophysical Research 111, D0711, doi:10.1029/2005JD006281. Lee, Zhongping,, K.L. Carder, C.D. Mobley, R.G. Ste ward, and J.S. Patch. 1998. Hyperspectral remote sensing for shallow waters. I. A semianalytical model. Applied Optics. 37: 6329-6338. Lee, Zhongping,, K.L. Carder, C.D. Mobley, R.G. Ste ward, and J.S. Patch. 1999. Hyperspectral remote sensing for shallow waters. 2. Deriving Bottom Depths and Water Properties by Optimization. Applied Optics. 38: 3831-3843.
107 Lee, Zhongping,, K.L. Carder, R.F. Chen, and T.G. P eacock. 2001. Properties of the Water Column and Bottom Derived from Aviris Visible Infrared Imaging Spectrometer (AVIRIS) Data. J. Geophys. Res. 106(C6): 11,639-11,651. Lee, Z., K.L. Carder, and R.A. Arnone. 2002. Derivi ng inherent optical properties from water color: a multiband quasi-analytical algorithm for optically deep waters. Appl. Opt. 41. 5755-5772. Lenes, J.M., B.P. Darrow, C. Cattrall, C.A. Heil, M Callahan, G.A. Vargo, R.H. Byrne, J.M. Prospero, D.E. Bates, K.A. Fanning, J.J. Walsh 2001. Iron Fertilization and the Trichodesmium Response on the West Florida Shelf. Limnol. Oceanogr. 46: 1261-1277. Lewis, M.R., O. Ulloa, and T. Platt. 1988. Photos ynthetic Action, Absorption, and Quantum Yield Spectra for a Natural Population of Oscillatoria in the North Atlantic. Limnol. Oceanogr 33: 92-98. Li, Z. and R. H. Weisberg. 1999a. West Florida con tinental shelf response to upwelling favorable wind forcing: Kinematics. J. Geophys. Res. 104(C6):13,507-13,527. Li, Z. and R. H. Weisberg. 1999b. West Florida con tinental shelf response to upwelling favorable wind forcing. 2. Dynamics. J. Geophys. Res. 104(C10):23,427-23,442. Livezey, R.E., M. Masutani, A. Leetmaa, H. Rui, M. Ji, and A. Kuma r. 1997. Teleconnective response of the PacificÂ–North Americ an region atmosphere to large central equatorial Pacific SST anomalies. J. Climate 10. 1787Â–1820 Maritorena, S., A. Morel, and B. Gentili. 1994. Dif fuse reflectance of oceanic shallow waters: Influence of water depth and bottom albedo. Limnol. Oceanogr. 39: 16891703.Maul, G.A. 1977. The annual cycle of the Gulf Loop Current, Part 1: Observations during a one-year time series. Journal of Marine Research. 35: 2947. Maul, G.A. and F. M. Vukovitch. 1993. The relatio nshiop between annual cycle of the Gulf Loop Current, Part 1: Observations during a o ne-year time series. Journal of Marine Research. 35: 29-47. Melo Gonzalez, N.; F.E. Mller-Karger; S. Cerdeira Estrada; R. Prez de los Reyes.; I. Victoria del Rio; P. Crdenas Perez; I. Mitrani Are nal. 2000: Near-surface phytoplankton distribution in the western Intra-Ame ricas Sea: The influence of El Nio and weather events. Journal of Geophysical Research. 105(C6). 1402914043.
108 Minnett, P. J. 1991. Consequences of sea surface te mperature variability on the validation and applications of satellite measurements. Journal of Geophysical Research. 96(C10). 18,475-18,489. Molinari, R.L. and J. Morrison. 1988. The separatio n of the Yucatan Current from the Campeche Bank and the intrusion of the Loop Current into the Gulf of Mexico. Journal of Geophysical Research. 93: 10.645-10,654. Momzikoff, A., S. Dallot, and G. Gondry. 1994. Dist ribution of seawater fluorescence and dissolved flavins in the Almeria-Orant front (A lboran Sea, Western Mediterranean Sea). J. Mar. Sys ., 5:361-376. Morel A. and L. Prieur. 1997. Analysis of variatio ns in ocean color. Limnol. Oceanogr., 22:709-722. Morey, S. L., W. W. Schroeder, J. J. OÂ’Brien, and J Zavala-Hidalgo. 2003. Annual cycle of riverine influence in the eastern Gulf of Mexico basin. Geophys. Res. Lett., 30(16):1867-1870.Morrison, J. and W.D. Nowli n. 1977. Repeated nutrient, oxygen, and density sections through the Loop Curre nt. Journal of Marine Research. 35(1): 105-128. Mote OrganizationÂ’s Red Tide Chronology. (n.d.). Re trieved February 2006, from http://isurus.mote.org/~mhenry/rtchrono.phtml Mueller, J.L., R.W. Austin. 1995. Ocean Optics Prot ocols for SeaWiFS Validation, Revision 1 NASA Tech. Memo. Muller-Karger, F.E., J.J. Walsh, R.H. Evans, and M. B. Meyers. 1991. On the seasonal phytoplankton concentration and sea surface tempera ture cycles of the Gulf of Mexico as determined by satellites. J. Geophysical Research. 96(C7). 1264512665. Muller-Karger, F. E. 2000. The spring 1998 Northe astern Gulf of Mexico (NEGOM) cold water event: Remote sensing evidence for upwel ling and for eastward advection of Mississippi water (or: How an errant L oop Current Anticyclone took the NEGOM for a spin). Gulf of Mex. Sci., 18(1):55-67. Murphy, E.B., K.A. Steidinger, B.S. Roberts, J. Wil liams, and J.W. Jolley. 1975. An explanation for the Florida east c oast Gymnodinium breve red tide of November, 1972. Limnol. Oceanogr. 20: 481-486. Nababan, B. 2005. Bio -optical Variability of Surface Waters in the North eastern Gulf of Mexico. PhD dissertation, Dept. of Marine Science, Universi ty of South Florida, St Petersburg, FL. 124 pp.
109 OÂ’Reilly, J. E., S. Maritorena, B.G. Mitchell, D.A. Siegal, K.L. Carder, S.A. Garver, M. Kahru, and C. McClain. 1998. Ocean Color Algorith ms for SeaWiFS. J. Geophysical Research. 103: 24,937-24,953. OÂ’Reilly, John E., et al. 2000. SeaWiFS Postlaunch Calibration and Validation Analyses, Part 3. NASA Tech. Memo. Vol. 11. 2000-206892. S.B Hooker and E.R. Firestone, eds., NASA Goddard Space Flight Center, 49 pp. Parsons, T.R., M. Takahashi, B. Hargrave. 1984. Bio logical Oceanographic Processes. 3rd ed. Pergamon Press. 330 pp. Qian, Y., A.E. Jochens, M.C. Kennicutt II, and D.C. Biggs. 2003. Spatial and temporal variability of phytoplankton blooms and community s tructure over the continental margin of the northeast Gulf of Mexico based on pigment an alysis. Contl. Shelf Res. 23: 1-17. Roberts, B.S. 1979. Occurrence of Gymnodinium brev e red tides along the west and east coasts of Florida during 1976 and 1977. pp. 199-202, in: T aylor, D.L, and H.H. Seliger (eds.), Toxic Dinoflagellate Blooms. Elsevier, NY. Schmidt, N., E.K. Lipp, J.B. Rose, and M. Luther. 2 001. ENSO Influences on Seasonal Rainfall and River Discharge in Florida. Journal of Climate. 14. 615-628. Sheinbaum, J., J. Candela, A. Badan., and J. Ochoa. 2002. Flow structure and transport in the Yucatan Channel. Geophysical Research Letters. 29(3). 10,1029. Siegel, D. A., S. Maritorena, N. B. Nelson, D. A. H ansell, and M. Lorenzi-Kayser. 2002. Global distribution and dynamics of colored dissolv ed and detrital organic materials. J. of Geophys. Res., 107(C12):3228-3242. Singh, L. 2005. Seasonal and Spatial Trends of Kare nia brevis Blooms and Associated Parameters Along the 10-Meter Isobath of the West F lorida Shelf. MS thesis, Dept. of Marine Science, University of South Florida, St Pet ersburg, FL. 78 pp. Sittel, M. C., 1994. Marginal probabilities of the extremes of ENSO events for temperature and precipitation in the southeastern U nited States. Tech. Rep. 94-1, Center for OceanÂ–Atmospheric Studies, The Florida S tate University, Tallahassee, FL, 155 pp. Smayda, T.J., 1997. Harmful Algal Blooms: Their Ec ophysiology and General Relevance to Phytoplankton Blooms in the Sea. Limnology and Oceanography 42(5), Part 2: The Ecology and Oceanography of Harm ful Algal Blooms. 11371153 South Florida Water Management District. 2006. The Everglades Water Flow Maps: Past, Present & Future. Retrieved August 2006, fro m http://www.sfwmd.gov/
110 Stumpf, R.P., M. L. Frayer, M. J. Durako, J. C. Bro ck. 1999. Variations in Water Clarity and Bottom Albedo in Florida Bay from 1985 to 1997. Estuaries, Vol. 22, No. 2, Part B: Dedicated Issue: Florida Bay: A Dynamic Sub tropical Estuary (Jun., 1999), pp. 431-444. Sturges, W. and J.C. Evans. 1983. On the variabilit y of the Loop Current in the Gulf of Mexico. Journal of Marine Research. 41: 639-653. Subramaniam A., E.J. Carpenter, D. Karentz, and P.G Falkowski. 1999. Bio-optical Properties of the Marine Diazotrophic Cyanobacterti a Trichodesmium spp. I. Absorption and Photosynthetic Action Spectra. Limnol. Oceanogr. 44: 608-617. Tester, P.A., R.P. Stumpf, F.M. Vukovich, P.K. Fowl er, and J.T. Turner.1991. An expatriot red tide bloom: transport, distribution, and persistence. Limnol. Oceanogr. 36: 1053-1061. Tester, P.A., K.A. Steidinger. 1997. Gymnodinium breve reed tide blooms: Initiation, transport, and consequences of surface circulation. Limnol. Oceanogr. 42: 10391051. Vargo, G.A., C.A. Heil, K.A. Fanning, L.K. Dixon, M .B. Neely, K. Lester, D. Ault, S. Murasko, J. Havens, J. Walsh, and S. Bell. submitt ed May 2004. Nutrient Availability in Support of Karenia brevis blooms on the central West Florida Shelf: What keeps Karenia blooming? submitted to Cont. Shelf Research Virmani, J. 2005. Ocean-Atmosphere Interactions on the West Florida Shelf. PhD dissertation, Dept. of Marine Science, University o f South Florida, St Petersburg, FL. 242 pp. Voss, K.J., C.D. Mobley, L.K. Sundman, J.E. Ivey, a nd C.H. Mazel. 2003. The spectral upwelling radiance distribution in optically shallo w waters. Limnol. Ocean. 48(1, part 2). 364-373. Vukovitch, F.M., B.M. Crissman, M. Bushnell, and W. J. King. 1979. Some aspects of the oceanography of the Gulf of Mexico u sing satellite and in-situ data. Journal of Geophysical Research. 84(C12): 7,749-7,768. Walsh, J.J, R.H. Weisberg, D.A. Dieterle, R. He, B. P. Darrow, J.K. Jolliff, K.M. Lester, G.A. Vargo, G.J. Kirkpatrick, K.A. Fanning, T.T. Su tton, A.E. Jochens, D.C. Biggs, B. Nababan, C. Hu, and F.E. Muller-Karger. 2 003. Phytoplankton response to intrusions of slope water on the West F lorida Shelf: Models and observations. Journal of Geophysical Research. 108(C6): 3190-3202.
111 Walsh, J.J., D.A. Dieterle, F.E. Muller-Karger, R. Bohrer, W.P. Bissett, R. Aparicio, R. J. Varela, H.T. Hochman, C. Schiller, R. Diaz, R. Thun ell, G.T. Taylor, M.I. Scranton, K.A. Fanning, and E.T. Peltzer. 1999. A n umerical simulation of carbon/nitrogen cycling during spring upwelling in the Cariaco Basin J. Geophys. Res. 104:7807-7825. Walsh, J.J., B. Penta, D.A. Dieterle, and W. P. Bis sett. 2001. Predictive ecological modeling of harmful algal blooms. Hum. Ecol. Risk Assess 7:1369-1383. Walsh, J.J. and K.A. Steidinger. 2001. Saharan dust and Florida red tides: the cyanophyte connection. J. Geophys. Res 106:11597-11612. Walsh, J. J., K.D. Haddad, D.A. Dieterle, R.H. Weis berg, Z. Li, H. Yang, F.E. MullerKarger, C.A. Heil, and W.P. Bissett. 2002. A numer ical analysis of landfall of the 1979 red tide of Karenia brevis along the west coast of Florida. Cont. Shelf Res 22:15-38. Walsh, J.J., R.H. Weisberg, D.A. Dieterle, R. He, B .P. Darrow, J.K. Jolliff, K.M. Lester, G.A.Vargo, G.J. Kirkpatrick, K.A. Fanning, T.T. Sut ton, A. E. Jochens, D.C. Biggs, B. Nababan, C. Hu, and F. E. Muller-Karger. 2003. The phytoplankton response to intrusions of slope water on the West F lorida shelf: models and observations J. Geophys. Res. 108:3190. Walsh, J.J., G.J. Kirkpatrick, B.P. Darrow, G.A. Va rgo, K.A. Fanning, E.B. Peebles, C.A. Heil, J. Havens, and K.A. Steidinger. 2005. Fish fa rming by Karenia brevis : a nutrient source for large red tides. Cont. Shelf Res Walsh, J.J., J.M. Lenes, J.K. Jolliff, B.A. Darrow, S. Milroy, D.A. Dieterle, K.A. Lester, L.A. Vanderbloemen1, R.H. Weisberg1, A. Barth1, L. Zheng, G.A. Vargo, K.A. Fanning, K.L. Carder, R. Chen, C.A. Heil, G.J. Kirk patrick, C. Hu, and F.E. Muller-Karger. in Press. A simulation analysis of nutrient supplies for maintenance of red tides in the eastern Gulf of Mex ico during 2001-2002: implications for future operational models. J. Geophys. Res. in Press. Weisberg, R.H. 1994. Transport of Mississippi Rive r water to the west shelf.. in Â“Coastal Oceanographic effects of 1993 Mississippi River floodingÂ”, ed. M.J. Dowgiallo, Special NOAA Report USDOC/NOAA/Coastal O cean Office/National Weather Service. Weisberg, R.H., B. Black, and Z. Li. 2000. An upwel ling case study on FloridaÂ’s west coast.. J. Geophys. Res. 105:11459-11469. Weisberg, R.H., Z. Li, and F.E. Muller-Karger. 2001 West Florida shelf response to local wind forcing: April 1998. J. Geophys. Res., 106, 31239-31262.
112 Weisberg, R.H. and R. He. 2003. Local and deep-oce an forcing contributions to anomalous water properties on the west Florida cont inental shelf. J. Geophys. Res 108, 3184, doi:10.1029/2002JC001407. Weisberg, R.H., R. He, G. Kirkpatrick, F. Muller-Ka rger, and J.J. Walsh. 2004. Coastal Ocean Circulation Influences on Remotely Sensed Opt ical Properties: A West Florida Shelf Case Study. Oceanography. 17:68-75. Weisberg, R.H., L.Y. Zheng. 2006. Circulation of Ta mpa Bay driven by Buoyancy, Tides, and Winds, as Simulated using a Finite Volum e Coastal Ocean Model. J. Geophysical Research-Oceans III. A rt. No. C01005. Werdell, P.J. and C.S. Roesler. 2003. Remote asse ssment of benthic substrate composition in shallow waters using multispectral r eflectance. Limnol. Ocean. 48(1, part 2). 557-567. Westberry, T.K. and D.A. Siegel. 2005. An Improve d Bio-optical Model for the Remote Sensing of Trichodesmium spp. Blooms. J. Geophys. Res. 110: C06012. Wolfe, Steven H. (ed). 1990. An ecological charact erization of the Florida Springs Coast: Pithlachascotee to Waccasassa Rovers. U.S. Dept. of the Interior, Fish and Wildlife Service and Minerals Mgmt Service. Bi ological Report 90(21). Yentsch, C. S. and C. A. Reichert. 1962. The inter relationship between water-soluble yellow substances and chloroplastic pigments in mar ine algae. Bot. Mar., 3:6574. Zhang, C., C. Hu, S. Shang, F.E. Muller-Karger, Y. Li, M. Dai, B. Huang, X. Ning, and H. Hong. In Press. Remote Sensing of Environment. In Press. Zimmerman, R.C. 2003. A biooptical model of irrad iance distribution and photosynthesis in seagrass canopies. Limnol and Ocean. 48(1, part 2). 568-585.
ABOUT THE AUTHOR Lisa Anne Vanderbloemen was born in Charlottesville Virginia and earned her Batchelor of Arts degree with honors in biology fro m the University of Virginia in 1981. She earned a Masters of Science in geographic and c artographic sciences from George Mason University in 1990 and a Masters of Science i n oceanography from the University of Maryland in 1998. In 1989 Ms. Vanderbloemen joi ned the Navy Reserves and is currently a Commander in the intelligence field. S he served during Operations Enduring and Iraqi Freedom, traveling to numerous countries within the Middle East. She is an intelligence analyst supporting the Global War on T errorism. Ms. Vanderbloemen also has extensive teaching experience, including teachi ng at the United States Naval Academy, the University of Maryland, and St. Peters burg College, and several community colleges.