|USFDC Home | USF Electronic Theses and Dissertations||| RSS|
This item is only available as the following downloads:
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 001931275
007 cr mnu|||uuuuu
008 080324s2006 flu sbm 000 0 eng d
datafield ind1 8 ind2 024
subfield code a E14-SFE0001777
Monitoring water quality in Tampa Bay :
b coupling in situ and remote sensing
h [electronic resource] /
by Zhiqian Chen.
[Tampa, Fla] :
University of South Florida,
ABSTRACT: Water quality in Tampa Bay was examined using concurrent in situ and satellite remote sensing observations. Chlorophyll and suspended sediment concentrations showed large short-term variability, primarily driven by tide and wind forcing. Superimposed on these high frequency variations were recurrent phytoplankton blooms stimulated by decreases in turbidity 1-2 days after wind-induced bottom sediment resuspension events; the blooms were particularly strong if neap tides occurred after the wind events. The in situ data show that observations once per month are inadequate to sample short-term variability and that therefore the current monthly water quality surveys may have uncertainties of -50 to 200% if they are used to represent the monthly mean concentrations of chlorophyll or suspended sediment. Such uncertainties make it difficult to identify trends and interannual variability based on the in situ monitoring program.^ Colored dissolved organic matter (CDOM) generally showed good relationship with salinity and primarily delivered by riverine inputs but showed conservative and non-conservative mixing behaviors for the dry and wet seasons, respectively. CDOM in Old Tampa Bay (OTB), however, showed properties that were different from those in other Bay segments, and the non-conservative CDOM mixing behavior may be simply due to a three-end-member mixing scenario in which Hillsborough Bay and Middle Tampa Bay also receive water from Old Tampa Bay. A turbidity algorithm was successfully developed for application of MODIS/Aqua 250 m imagery. The MODIS turbidity images showed distinct spatial and temporal patterns related to river runoff in the upper bay and wind-induced sediment resuspension events in the middle and lower portions of the Bay.^ Similarly, light attenuation from SeaWiFS estimated using a new semi-analytical algorithm confirmed that water clarity was related to river runoff and to wind-induced sediment resuspension events. Wind is shown repeatedly to be another important factor controlling water quality in the Bay. The study shows that remote sensing products have the potential to be an important tool to help resource managers assess conditions in a large estuary like Tampa Bay synoptically, frequently and repeatedly.
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 120 pages.
Adviser: Frank Muller-Karger, Ph.D.
x Marine Science
t USF Electronic Theses and Dissertations.
Monitoring Water Quality in Tampa Bay: Coupling in Situ and Remote Sensing by Zhiqiang Chen 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 E. Mller-Karger, Ph.D. Kendall L. Carder, Ph.D. Chuanmin Hu, Ph.D. Mark E. Luther, Ph.D. Peter W. Swarzenski, Ph.D. Date of Approval: November 13, 2006 Keywords: SeaWiFS, MODIS, bio-optical sensors, turbidity, water clarity Copyright 2006, Zhiqiang Chen
To my parents
Acknowledgments I thank my major professor, Dr. Frank Muller-Karger, for providing me with a wonderful opportunity to advance my research interests and for his support. I am grateful to Dr. Chuanmin Hu for his help with lear ning remote sensing. I also thank my other committee members, Drs. Kendall Carder, Mark Luther and Peter Swarzenski for their time and effort in providing constructive comments and advice. This work was supported by the University of South Florida (USF) and the U.S. Geological Survey (USGS) Cooperative Gr aduate Assistantship Program, and the National Aeronautics and Space Administra tion (NASA). Field work was supported by the Tampa Bay study of the USGS Coasta l and Marine Geology Program, and was conducted aboard the USGS R/V Mako with the capable assistance of Jordan Stanford and Dana Nelson. Dr. Mark E. Luther allowed the installa tion of bio-optical sensors on key PORTS (Physical Oceanographic Real-Time System) co astal towers in Tampa Bay, and he and Sheryl Gilbert helped install sensors and re trieve data from the PORTS sensors. The Environmental Protection Commission of Hills borough County also shared data that was critical to this study. I must thank my parents for their unc onditional support and my beloved wife for giving birth to and taking care of three love ly boys, Sam, Tim, and Angelo, who fill my heart with joy. Finally I thank all my friends at the Ins titute of Marine Remote Sensing (IMaRS) at the USF College of Marine Science, for their friendly atmosphere and kind help and encouragement throughout the past 5 years. This made studying at IMaRS always an enjoyable experience.
i Table of Contents List of Tables iii List of Figures iv Abstract viii Chapter 1. Introduction 1 1.1. The quality of the water of Tampa Bay 1 1.2. Remote sensing of water quality in estuarine waters 4 1.3. Objectives and outline 6 Chapter 2. Physical forcing of short-term bio-optical variabil ity in Tampa Bay: Observations from a coastal tower 7 2.1. Abstract 7 2.2. Introduction 8 2.3. Methods and Materials 9 2.3.1. Deployment of sensors 9 2.3.2. Ancillary data 11 2.3.3. Data analysis 12 2.4. Results and Discussion 13 2.4.1. Physical variations 13 2.4.2. Bio-optical variations 14 2.4.3. Sediment resuspension versus phytoplankton blooms 20 2.4.4. Implications for wa ter quality monitoring 21 2.5. Conclusions 24 Chapter 3. Colored Dissolved Or ganic Matter in Tampa Bay 26 3.1. Abstract 26 3.2. Introduction 27 3.3. Methods 28 3.3.1. Temperature, salinity, chloro phyll and total suspended solids 28 3.3.2. Dissolved Organic carbon 31 3.3.3. CDOM absorption coefficients and fluorescence 31 3.3.4. Phytoplankton pigment absorption coefficients 32 3.3.4. Statistic test 32 3.4. Results 33 3.4.1. Distribution of TSS and Chl 33 3.4.2. Light absorption by CDOM 33 3.4.3. DOC 42 3.4.4. CDOM fluorescence efficiency 44 3.5. Discussion 45
ii 3.5.1. CDOM sources and sinks 45 3.5.2. Seasonal variation in CDOM absorption 48 3.5.3. Implication of water clarify and remote sensing 49 3.6. Conclusions 49 Chapter 4. Monitoring turbidity in Tampa Bay using MODIS/Aqua 250 m imagery 51 4.1. Abstract 51 4.2. Introduction 52 4.3. Materials and Methods 54 4.3.1. Field data 54 4.3.2. Cross calibration of the MODIS 250 m with 1 km bands 56 4.3.3. Satellite image processing 59 4.3.4. Satellite in situ comparison 60 4.4. Results 61 4.4.1. Cross calibration and data validation 61 4.4.2. Turbidity from satellite R rs 63 4.4.3. Image series of turbidity 64 4.5. Discussions 67 4.5.1. Calibration and algorithm issues 67 4.5.2. Effects of colored dissolved organic matter (CDOM) 74 4.6. Conclusions 75 Chapter 5. Remote Sensing of water clarity in Tampa Bay 76 5.1. Abstract 76 5.2. Introduction 77 5.3. Materials and methods 80 5.3.1. Satellite data 80 5.3.2. Field and ancillary data 80 5.3.3. Satellite In situ comparison 81 5.4. Results 81 5.4.1. Comparison in sutu SDD and satellite Kd(490) 81 5.4.2. SDD image series 82 5.5. Discussion 96 5.5.1. Algorithm issues 96 5.5.2. Implication of Tampa Bay Ecosys tem restoration and water quality monitoring 97 5.6. Conclusion 98 Chapter 6. Summary and future work 99 List of References 102 Appendices 116 About the Author End Page
iii List of Tables Table 3-1 Correlation coefficients (r) between surface salinity and TSS and between surface salinity and Chl in Tampa Bay and the AR in the dry and wet seasons 35 Table 3-2 Results of regression anal ysis between aCDOM(400) and surface salinity in Tampa Bay and the AR in the dry and wet seasons. 39 Table 3-3 Table 4 Correlation coeffi cients (r) between aCDOM(400) and TSS and between aCDOM(400) and Chl in Tampa Bay and the AR in the dry and wet seasons 40 Table 3-4 The averages and standard de viations (in parentheses) of CDOM absorption (aCDOM(443)) and phytoplankton pigment absorption (aph(443)) in different bay segments and seasons 41 Table 3-5 Results of regression anal ysis between CDOM fluorescence and a CDOM (400) Tampa Bay and the AR during the dry season. 45 Table 4-1 Symbols, de finitions and Units 58 Table 4-2 Comparison between the m easured at-sensor radiances (mW cm -2 m -1 sr -1 ) at the MODIS 250 m bands and those predicted from the 1 km ocean color bands 62 Table 4-3 Comparison between in situ versus MODIS remote sensing reflectance at 645 nm (R rs (645), sr -1 ), with the latter derived using 1) single scattering approximati on and the original at-sensor radiance (i.e., pre-launch calibration) and 2) multi-scattering method and the cross-calibrated at -sensor radiance, respectively 62
List of Figures Figure 1-1. MODIS/Aqua 250 m resolution image of Tampa Bay showing the four sub-basins of the Bay and four major tributaries 2 Figure 1-2. NOAA/USGS merged bathymetric/topographic digital elevation model (DEM) of the Tampa Bay estuary at 30 m spatial resolution (Gesch & Wilson, 2001) 3 Figure 2-1. (Left): Schematic of in situ monitoring station 10 Figure 2-2. Hourly averaged time-series of (A) winds, (B) water levels, (C) currents, (D) salinity, and (E) temperature 15 Figure 2-3. Hourly averaged time-series of (A) significant wave period, (B) significant wave height, (C) near bottom wave orbital velocity, and (D) near-bottom wave shear stress 16 Figure 2-4. (A) Hourly averaged time series of backscattering coefficient (bbp(532)) in the surface (thin solid line) and bottom layers (thin dashed line) 19 Figure 2-5. Spectral density of time series of (A) water levels, (B) bbp(532), and (C) Chl 20 Figure 2-6. Variations of bbp(532) and Chl with relation to salinity, water levels, and currents from 11-19 December 2004 22 Figure 2-7. The relative errors (= %100 mmmmmi ) of Chl (thin blue) and bbp(532) (thick red) incurred by representing a monthly mean by a single snapshot measurements (mi) 22 Figure 2-8. The monthly time series of (A) Chl (mg m -3 ) and (B) turbidity (NTU) at one of the EPCHC stations (See Fig. 1 for station location) from 1997 to 2003 24 Figure 3-1. Map of Tampa Bay with sampling locations overlaid 29 iv
v Figure 3-2. (A) Percentage of average (1973-2003) mont hly mean flow rates of Hillsborough River (HR), Alafia Ri ver (AR), Little Manatee River (LMR), Manatee River (MR) relative to the total monthly mean flow rate from these four major rivers of each month 30 Figure 3-3. Surface salinity versus (A) total suspended solids concentration (TSS) and (B) chlorophyll conc entration (Chl). 34 Figure 3-4. Surface salinity versus CDOM a CDOM (400) for Tampa Bay and the AR in the dry and wet seasons (see also Table 3). 36 Figure 3-5. a CDOM (400) versus (A) TSS and (B) Chl and (C) aCDOM(443) versus aph(443) for Tampa Bay and the AR in the dry and wet seasons (also see Table3-3) 38 Figure 3-6. CDOM absorption spectral sl ope (S) versus surface salinity for Tampa Bay and the AR in the dry and wet seasons. 41 Figure 3-7. Surface salinity versus DOC concentration for Tampa Bay and the AR in the dry season in June, 2004 42 Figure 3-8. aCDOM(400) versus DOC concentration for Tampa Bay and the AR in the dry season in June, 2004 43 Figure 3-9. CDOM fluorescence (QSE) ve rsus aCDOM(400) (m-1) for Tampa Bay and the AR during the dry season in June, 2004 44 Figure 4-1. MODIS 250 m image of Tampa Bay showing the four segments of the bay and major tributaries 55 Figure 4-2. (a) MODIS at-sensor radiance at 667 nm (1 km) on 13 December 2004, 18:43 GMT over the study area 59 Figure 4-3. Scatter plots of the at-sensor radiance (mW cm -2 m -1 sr -1 ) in the two 250 m bands (645 nm in (a) and 859 nm in (b)) as measured by the sensor and predicted by the calibrated 1 km ocean color bands 63
vi Figure 4-4. Remote sensing reflectance in Tampa Bay at band 1 ( R rs (645), sr -1 ) from in situ measurements and MODIS es timates derived using 1) multiple-scattering atmospheric correction of the cross-calibrated atsensor radiance (filled circles); a nd 2) single-scattering atmospheric correction of the original (i.e., pre-launch calibration) at-sensor radiance (open circles) 65 Figure 4-5. The relationship between in situ turbidity (NTU) and MODIS remote sensing reflectance at band 1 (R rs (645), sr -1 ) in a log-log scale 65 Figure 4-6. The relationship between in situ turbidity from the 2004 measurements and that estimated fr om concurrent MODIS data using the relationship derived from the 2005 data 66 Figure 4-7. Climatological (May 2003 to April 2006) monthly means of turbidity derived from MODIS 250 m data 68 Figure 4-8. The climatological monthly m eans of turbidity derived from MODIS between May 2003 and April 2006 at several stations in Tampa Bay (see Fig. 1 for station locations) 70 Figure 4-9. Monthly turbidity images from MODIS 250 m data (top panels) and from in situ measurements (lower pa nels) for April of 2004, 2005, and 2006 (no in situ measurement was available for April 2006) 71 Figure 4-10.Daily averaged wind speed m easured at Port of Manatee (27.64, 82.56) in April 2004, 2005, and 2006 (data courtesy of NOAA) 72 Figure 4-11 Time-series of monthly turbidity estimates derived from in situ and MODIS measurements at selected stations shown in Fig. 1 73 Figure 5-1. A SeaWiFS quasi-true-color RGB image of Tampa Bay showing the four sub-segments,namely, Old Tampa Bay (OTB), Hillsborough Bay (HB), Middle Tampa Bay (MTB), and Lower Tampa Bay (LTB) 79 Figure 5-2. In situ secchi disk depth (SDD, m) vers us light attenuation coefficient at 490 nm ( K d (490), m -1 ) derived from SeaWiFS using (1) an empirical band-ratio algorithm (EA) (triangles, updated Muellers (2000) algorithm) and (2) a semi-a nalytical algorithm(SA) (circles, Lee et al., 2005a) 83
vii Figure 5-3. Climatological monthly composites (September 1997 December 2005) of SeaWiFS SDD (m) 84 Figure 5-4. Climatological monthly means of SeaWiFS (filled circles) and in situ (open circles) secchi disk depth (SDD, m) at several stations within Tampa Bay (Fig. 1) 87 Figure 5-5. Climatological monthly means of (A) in situ chlorophyll (mg m 3 ), (B) color (Pt-unit), and (C) turbidity (n ephelometric turbidity units, NTU) measurements from the EPCHC monitoring program at several stations from the various bay-segm ents (see Fig. 1 for the station locations) collected between September 1997 and December 2005 88 Figure 5-6. (a) Climatological monthly means of river flow from the Hillsborough River (open circles) and the Alafia River (filled circles) from 1997 to 2005 92 Figure 5-7. Monthly means of SeaWiFS and in situ SDD at selected stations (Fig. 1) from September 1997 to December 2005 93 Figure 5-8. SeaWiFS SDD anomaly between September 1997 and December 2005 at several stations in Tampa Bay (Fig. 1) 94 Figure 5-9. Monthly means of river flow from the Hillsborough River (open circles) and the Alafia River (fi lled circles) from September 1997 to December 2005 95 Figure 5-10. The number of days when th e daily averaged wind speed was > 4.0 m s -1 of each month from September 1997 to December 2005 at the PORT station near Saint Pete rsburg (27.6'N, 82.6'W) 95
viii Monitoring Water quality in Tampa Bay: Coupling in Situ and Remote Sensing Zhiqiang Chen ABSTRACT Water quality in Tampa Bay was examined using concurrent in situ and satellite remote sensing observations. Chlorophyll an d suspended sediment concentrations showed large short-term variability, prim arily driven by tide and wind forcing. Superimposed on these high frequency variat ions were recurrent phytoplankton blooms stimulated by decreases in turbidity 1-2 days after wi nd-induced bottom sediment resuspension events; the blooms were particular ly strong if neap tides occurred after the wind events. The in situ data show that observations once per month are inadequate to sample short-term variability and that theref ore the current monthly water quality surveys may have uncertainties of -50 to 200% if they are used to represent the monthly mean concentrations of chlorophyll or suspended se diment. Such uncertainties make it difficult to identify trends and interannual variability based on the in situ monitoring program. Colored dissolved organic matter (CDOM) generally showed good relationship with salinity and primarily delivered by riveri ne inputs but showed conservative and nonconservative mixing behaviors for the dry and wet seasons, respectively. CDOM in Old Tampa Bay (OTB), however, showed properties that were different from those in other Bay segments, and the non-conservative CDOM mixing behavior may be simply due to a three-end-member mixing scenario in which Hillsborough Bay and Middle Tampa Bay also receive water from Old Tampa Bay. A turbidity algorithm was successfully developed for application of MODIS/Aqua 250 m imagery. The MODIS turbidity images showed distinct spatial and temporal patterns related to river runo ff in the upper bay and
ix wind-induced sediment resuspension events in the middle and lower portions of the Bay. Similarly, light attenuation from SeaWiFS estimated using a new semi-analytical algorithm confirmed that water clarity was re lated to river runoff and to wind-induced sediment resuspension events. Wind is shown repeatedly to be another important factor controlling water quality in the Bay. The study shows that remote sensing products have the potential to be an important tool to help resource managers assess conditions in a large estuary like Tampa Bay synoptic ally, frequently and repeatedly.
1 Chapter 1. Introduction 1.1. The quality of the water of Tampa Bay Tampa Bay, with a surface area of ~ 1000 km 2 is Floridas largest open-water estuary. For practical purposes, it is typically divided into four sub-basins, namely: Old Tampa Bay (OTB), Hillsborough Bay (HB), Middle Tampa Bay (MTB), and Lower Tampa Bay (LTM) (Fig.1-1). The estuary has an average depth of ~ 4.0 m, but it has an extensive navigation system w ith a >10.0 m deep shipping ch annel running along its axis from the mouth to HB (Fig.1-2). This channe l plays an important role in determining estuarine circulation (Weisberg & Zheng, 2005) and water quality characteristics (Bendis, 1999). Tampa Bay is a diverse and productive natura l system that provides a vital habitat for crustaceans, fish, shellfish and a variety of marine mammals, reptiles and birds (Harwell et al., 1995), and cont ributes over $5 billion annually to the local economy from trade, tourism, development, and fishing (FDC A, 1996). It is therefor e critical that the development of the Bay be conducted in an environmentally sound way to sustain a clean and healthy system. Tampa Bay has undergone substantial anthropogenic alterations. In the decades prior to the 1980s, Tampa Bay was heavily po lluted by nutrient load ings from sources such as sewage and wastewater, causing severe eutrophication a nd phytoplankton blooms (Johansson, 2000). The decline in water quality due to pollution led to substantial losses of seagrass coverage (Lewis et al., 1998; To masko et al., 2005). Si nce then, significant ecosystem restoration efforts have been under way. In 1990, the Tampa Bay National Estuary Program (TBNEP) was established to in tegrate efforts to restore and protect the Bay. In 1996, TBNEP developed a Comprehens ive Conservation and Management Plan (CCMP), which focused on restoration of seagra ss to levels similar to those observed in the 1950s by improving water quality through co nstruction of better sewage treatment
facilities in the neighboring cities and controlling the amount of nitrogen reaching the bay (Janicki & Wade, 1996). These efforts gradually improved the water quality of the Bay, and some of the seagrass has recovered (Johansson, 2000; Tomasko et al., 2005). OTB HB MTB LTB A R M R LMR H R Figure 1-1. MODIS/Aqua 250 m re solution im ag e of Ta m p a Bay showing the four subbasins of the Bay and four m a jor tributarie s. The sub-basins are Old Tam p a Bay (OTB), Hillsboroug h Bay (HB), Middle Ta mpa Bay (MTB), and Lower Tam p a Bay (LTB). The river s ar e th e Hillsbo rou gh River (H B), the Al af ia River (AR), the Little Manate e River (LMR), and the Manatee River (MR). The Environm ental Protection Comm ission of Hillsboroug h Countys (EPCHC) water quality monitoring stations are overlaid with various sym bols to indicate different m onthly sampling tim es: diam ond (OTB) generally sam p led the first week, triangles (HB) the second week, and squares (MTB and LTB) the last week. T h e inse t sho w s the loca tion of Ta m p a Bay in the State of Florida. 2
Figure 1-2. NOAA/USGS merged bathymetric/topographic digital elevation model (DEM) of the Tampa Bay estuary at 30 m spatial resolution (Gesch & Wilson, 2001 ) The resolution of the gridded dataset is 250 m. Grey and white represent land and missing data, respectively. Previous studies have suggested that the primary factors regulating water quality seem local precipitation or river runoff and associated nutrients and sediments (Janicki et al., 2001). For example, HB receives the discharge of the two largest rivers emptying into Tampa Bay (the Hillsborough River and the Alafia River, Fig. 1-1). HB also generally has higher chlorophyll concentrations, turbidity and light attenuation coefficients than MTB and LTB (Janicki et al., 2001). The seasonal and inter-annual variability of water quality is closely related to hydrologic conditions (e.g., precipitation and river flow) (Lipp et al., 2001; Schmidt and Luther, 2002). Deteriorating water quality and the discontinuous recovery of seagrass in 1998 was widely attributed to increased rain fall associated with the stronger 1997-1998 El Nino event (Tomasko et al., 2005). Our current understanding of the causes of variability in water quality is based on in situ field surveys. As for most coastal environmental monitoring programs, at present water quality in Tampa Bay is measured on a monthly basis at established stations. Fig. 1-1 shows the location of sampling stations used in the Tampa Bay water quality 3
4 monitoring program conducted by the Environmental Protection Commission of Hillsborough County (EPCHC). Due to the size of Tampa Bay and logistical support, the EPCHC usually takes 3 weeks to sample the Bay, with approximately one segment sampled per week sequentially (Fig. 1-1, and Boler et al., 1991). These observations are used to represent the monthly conditions of the Bay, and to infer seasonal and interannual trends (Janicki et al ., 2001; Johansson, 2000). Clearly, the sampling is intensive, yet not synoptic. It would be ideal to sample the Bay synopti cally and more frequently, as routinely done in open ocean waters using satellite remote sensing technologies, to provide resource managers and decision-makers with a more realistic view of the variability of water quality Tampa Bay. 1.2. Remote sensing of water quality in estuarine waters Satellite remote sensing is a powerful tool to observe physic al and biological processes in the open ocean (e.g., Behrenfe ld et al., 2001). However, its routine application in estuarine and coastal waters has been limited (IOCCG, 2000). Retrieved products such as chlorophyll lack of sufficient accuracy in coastal wa ters due to problems with atmospheric correction and bio-optic al inversion algori thms. The operational atmospheric correction algorithms were desi gned primarily for the open oceans (Gordon & Wang, 1994; Gordon 1997). In coastal wate rs, these algorithms often result in erroneous masks of coastal waters and underestimates of water-leaving radiance, including negative values in the blue 412 nm band (Hu et al., 2000; Harding et al., 2005). The assumption that near infrared (NIR) wa velengths provide a black pixel breaks down due to high suspended sediment loads in coastal waters (Wang & Shi, 2005). In addition, absorbing aerosols over coastal wate rs can further degrad e the performance of the general atmospheric correction, especially for the blue bands. Various modifications have been proposed (Stumpf et al., 2000), but no fully satisfactory solution has been achieved yet. The generalized bio-optical inversion algorithms also contain large uncertainties when applied to coastal waters (Lee et al., 2005b). Therefore, a regional bio-optical algorithm has of ten been developed for speci fic coastal water (Dzwonkowski and Yan, 2005).
5 The optical properties of estuarine and coastal waters are highly variable and complex, posing a great challenge to de-c onvolution of the mixed contributions of different optical components. For example, colored dissolved organic matter (CDOM) and detritus or sediments often exhibit larger absorption than phytoplan kton (Babin et al., 2003; Maguson et al., 2004). Therefore, small er rors in sediment concentration or CDOM light absorption estimates cause substantial e rrors in chlorophyll estimates (Wozniak & Stramski, 2004). Indeed, to develop better bi o-optical inversion algorithms, a detailed characterization of optical properties for specifi c coastal waters is a necessary first step. Another problem is that current satellit e ocean color sensors have limited spatial and spectral resolution applications in estu arine and coastal waters. The standard Seaviewing Wide Field-of-View Sensor (SeaWiFS) and the Moderate Resolution Imaging Spectroradiometer (MODIS) ocean color data have 1 km resolution at nadir, while estuarine and coastal waters are highly patchy, with variables having spatial scales ranging from 100 to 1000 m (Sanderson and Taylor, 2003). This precludes wide application of ocean color observations in quantitative st udies of coastal waters. Recent advances in both algorithm devel opments and sensor technology promises to overcome some of the difficulties. First, an improved atmospheric correction algorithm has been developed using short-wave infrar ed wavelengths (SWIR) to account for the non-zero water-leaving radiance at NIR (W ang & Shi, 2005). This may improve the performance of atmospheric corrections in coastal waters with high sediment loads. Second, various semi-analytical bio-optical al gorithms have been proposed to estimate water constituents (Carder et al., 1999; Lee et al., 1999; Lee et al., 2002 Lee et al., 2005a; Maritorena et al., 2002). Thes e algorithms can be further improved using locally-derived parameterizations (Cota et al., 2004; Maguson et al., 2004). MODIS sensors are equipped with 250m and 500m sharpening bands, originally designed for land and aerosol ma pping. These medium resolution bands also have the serve to potential to map water quality indices due to their finer spatial resolution and adequate sensitivity (Hu et al ., 2004; Miller & Mckee, 2004). MODIS also has spectral bands designed to retrieve solar-induced chlorophyll fluorescence. Preliminary studies show that satellite-retrieved sun-indu ced chlorophyll fluorescence is related to phytoplankton concentr ations in coastal waters (Hoge et al., 2003; Hu et al.,
6 2005). These advances provide unprecedented opportunities of usi ng satellite remote sensing to monitor water quality in estuaries like Tampa Bay. 1.3. Objectives and outline The overall objective of this thesis is to characterize temporal and spatial variability of water quality in Tampa Ba y and understand the governing processes and mechanisms. For this purpose, data collected from discrete samples, continuous underway measurements, high frequency biooptical sensors, and synoptic remote sensing imagery were examined. The combination of in situ and satellite remote sensing observations provides better temporal and spatia l coverage of water quality and is expect to reveal more realistic and complete vari ation patterns and tre nds of water quality. Among a wide variety of indices, several key water quality indices were selected (Boler et al., 1991), namely: colore d dissolved organic matter (C DOM), turbidity or total suspended substance (TSS), and secchi disk depth (SDD, a surrogate of water clarity). This thesis is structured as follows. Chap ter 2 focuses on the short-term (tidal and subtidal) variability in chlorophyll and su spended sediment indices in Tampa Bay. Chapter 3 examines CDOM distribution in Tampa Bay in the dry and wet seasons. Chapter 4 investigates turb idity variability in the Ba y using MODIS 250 m imagery, which required cross-calibration with the 1 km resolution ocean color bands, and development of simple atmospheric correc tion and bio-optical al gorithms. Chapter 5 studies water clarity using SeaWiFS imag ery. Chapter 6 summarizes the findings and provides suggestions to improve monitoring of water quality in Tampa Bay.
7 Chapter 2. Physical forcing of short-term bio-optical variabil ity in Tampa Bay: Observations from a coastal tower 2.1. Abstract Short-term phytoplankton and sediment dynamics of Tampa Bay were examined with data collected from bio-optical, physical, and meteorol ogical sensors mounted on a coastal tower from 8 December 2004 to 17 Ja nuary 2005. Both chlorophyll concentration (Chl, 2.0 to 12.0 mg m -3 ) and the backscattering coeffi cient at 532 nm (bbp(532), an indicator of sediment concentration, 0.03 to 0.20 m -1 ) showed large fluctuations. These variations were dominated by the semidiurnal and diurnal frequencies of tides. In addition, three major sediment resuspension events were observed, and these were followed by increases in Chl. The sediment resuspension events were caused primarily by wind-induced mixing (e.g., hourly averaged wind speed >8.0 m s -1 ). Minimum critical bottom shear stress for sediment resuspension was ~0.2 N m -2 at the study site. Sediments remained in the water column for 2-3 days after the wind event. Moderate increases in Chl were often observed about 1-2 days af ter sediments settled, possibly due to the combined effects of increased nutrients and light regime. One such event developed into an intensive (Chl >10 mg m -3 ), transient (2 days) phytoplankton bloom, which was uniquely coincident with a neap tide. This suggested that phytoplankton blooming was related not only to the magnitudes of winds a nd tides but also to their relative timing. The results suggest that single monthly observations are not good estimat ors of monthly mean conditions, as they have uncertainties ranging from -50% to 200% relative to actual monthly mean conditions.
8 2.2. Introduction Estuaries are highly dynamic environments where rivers, winds, and tides interact to determine hydrological, biological, and geological variability. These forcings lead to a wide range of temporal scales of variation in phytoplankton and suspended sediment concentrations in estuaries. Riverine input s can determine the magnitude and extent of phytoplankton blooms by delivering nutrients an d influencing stratification of the water column over seasonal and interannual scales (Cloern, 1991; Harding, 1994). Semidiurnal or diurnal tides lead to vari ations in phytoplankton biomass and sediment concentrations through tidal advection and tidal mixing (Cloern et al., 1989; Li & Smayda, 2001; Roegner et al., 2002). Superimposed on thes e periodic variations are wind events. Pulsed winds modify estuarine circulation a nd water level, and gene rate waves that can suspend sediments (e.g., Schoellhamer, 1995) an d mix nutrients and benthic algae into overlying waters (Lawrence et al ., 2004; Yeager et al., 2005). Many studies have addressed high frequency variations of phytoplankton in estuarine ecosystems (Cloern et al., 1989; Cloern, 1991; Desmit et al., 2005; Li & Smayda, 2001; Roegner et al., 2002). However, one of the main obstacles has been the lack of reliable means for high frequency sampling of phytoplankton in larger estuaries over synoptic scales (e.g., Roegner et al., 2002) High frequency variability of sediments has been well documented using optical a nd/or acoustic sensors (Li & Amos, 2001; references therein). This variability results primarily from sediment resuspension, driven by highly variable winds, waves, tides, and in teractions among them (Jing & Ridd, 1996; Li & Amos, 2001; Schoellhamer, 1995). Few stud ies have been able to address the relationship between sediment and phytoplankton dynamics, even though turbidity is a crucial factor affecting phytoplankton bl ooms (Desmit et al. 2005; May et al., 2003). The Tampa Bay water quality monitori ng program has been conducted since 1974 by the Environmental Protection Commission of Hillsborough County (EPCHC). This program assesses water quality, but there has been little information available on shortterm scales (such as tidal and subtidal) of phytoplankton and sediment concentrations. The EPCHC conducts monthly field surveys which typically span 3 weeks, with approximately one in one week(F igure 2-1). The single monthly in situ measurements
9 conducted at each station are often used to ch aracterize variation and trends of Chl and sediments in the bay over seas onal and interannual scales (Janick et al., 2001; Johansson, 2000). But as observed in other estuarie s (e.g., Johnson et al., 2000), the monthly observations inherently suffer from aliasi ng. The extent to which the high frequency variability may influence a long term trend analysis has not been evaluated. In this chapter, we studied the shor t-term variability of phytoplankton and suspended sediments in Tampa Bay using data collected from bio-optical, physical, and meteorological sensors mounted on a coastal tower station. Our object ives were to (1) characterize high frequency variations in phytoplankton and suspended sediments in Tampa Bay (variability); (2) understand governin g processes responsible for the observed variability (forcing); and (3) discuss implicat ions of these findings on long-term trend analyses of water quality in the Bay. 2.3. Methods and Materials 2.3.1. Deployment of sensors Bio-optical sensors were de ployed on a coastal tower located in Manatee Channel (27. 661 o N, 82.5834 o W, Fig.2-1) on 14-27 July 2004 and from 8 December 2004 to 17 January 2005, respectively. The station is loca ted near the middle of the Bay at a bottom depth of ~ 4.6 m ( http://comps.marine.usf.edu/BRACE/ ). Two types of bio-optical se nsors were deployed, a WETLabs ECO-BBSB sensor measuring backscattering coeffi cient at 532 nm (bbp(532)) and a WETLabs ECO-FLNTUSB sensor measuring LED stimulated chlorophyll fluorescence near 685 nm. Both sensors had an internal battery a nd memory for data logging. To determine if there were differences in chlorophyll and se diment between the surface and bottom, one set of scattering and chlorophyll sensors was in stalled at 1m depth below the surface with sensors facing down, and another set at 1.5 m above the bottom with sensors facing up (Fig. 2-1). Sample frequency was set to once per hour. The recorded raw data (voltages) were processed with the WETLabs ECOView software and converted to Chl (mg m -3 ) and bbp(532) (m -1 ) using calibration coefficients provided by the manufacturer. Chl
LT OT H MT B Figure 2-1. (LEFT): Schem a tic of in situ m onitoring station. (Right): Moderate Resolution I m aging Spectror adiom e ter (MODI S/Aqua, 250m) im age of Tam p a Bay on 20 Decem be r 2004 showing four bay segm ents, nam e ly: Old Tam p a Bay (OLB), Hillsboroug h Bay (HB), Middle Ta mpa Bay (M TB), and Lower Tam p a Bay (LTB), and various sampling stations: Manatee Channel station ( ) where chlorophyll fluorom eters, backscattering m e ters and S ea-Bird S eaGauge were m ounted; Port of Manatee ( ) whe r e wind data were collected; Sunshine S kyway Bridge station ( ) where Acoustic Doppler Current Prof ile (ADCP) curre n t data near surface were obtained; One station ( ) from the EPCHC Ta mpa Bay wa ter quality monitoring pr ogram wher e 7-year m o nthly tim e series of Chl ( m g m -3 ) and turbidity (NTU) from 1997 to 2003 were obtained. Schem a tic diagram also shows how the sensors were m ounted onto the Manatee Channel station. MSL in the schem a tic sta nds for Mean Sea Level. 10
11 values were further calibrated using concurrent discrete wa ter samples collected at the station location and analyzed using the fl uorometric method (Strickland and Parsons, 1972). A Sea Bird Electronics SeaGauge wave and tide recorder was also deployed on the station about 2.2 m below the sea surface(Fi g. 2-1), measuring salinity, temperature, pressure, significant wave height and wave period ( ftp://comps.marine.usf.edu/pub/BRACE/seagauge_data ). The WETLab sensors were equipped with an anti-biofouling mechanism: before each measurement, a copper wiper cleaned the optical window. Unfortunately, during the two week summer deployment, barn acles grew and covered the sensors after only about one week, preventing further data co llection. In contrast, the sensors deployed during the winter successfully collected data for 40 days. Therefore, only winter data were used in this study. 2.3.2. Ancillary data Wind (both speed and direction) and currents from the Tampa Bay Physical Oceanographic Real Time Sy stem (PORTS) located at the Port of Manatee (27.637 o N, 82.5633 o W, ~ 3.3 km to the tower) and at the Sunshine Skyway Bridge (27.620 o N, 82.6558 o W, ~8.5 km to the tower), respectively, were obtained to represent general wind and current conditions. All data were binned into hourly data for this study. Pressure (pounds per square inch, PSI) co llected from the seagauge was converted to water level (m) using the following equation: Water Level = (pressure 14.7) 0.689476 2.05 (1) where 14.7 is the standard atmospheric pres sure (PSI), 0.689476 is the conversion factor for pressure (PSI) to depth (m), and 2.05 is the offset of the sea gauge relative to the mean lower low water (MLLW) (Fig. 2-1). Th e absolute accuracy of water level is not critical for this study and only relative os cillations are releva nt in this study. Monthly time series of Chl (mg m -3 ) and turbidity (Nephelometric Turbidity Unit, NTU, a practical indicator of sediment concentration) between 1997 and 2003 were obtained from a nearby EPCHC wate r quality station (station 23, 27.666 o N and 82 5992 o W, ~ 1.7 km from our tower) (Fig. 2-1).
2.3.3. Data analysis All time series data collected by the sensors were passed through a low-pass filter with cutoff time of 36 hours to separate tidal and subtidal variability. A spectral power analysis was also applied to extract dominant frequencies associated with the data. In principle, bottom shear stress ( b N m -2 ) includes shear stress from waves, currents, and interactions between waves and currents near the bottom boundary layer (Li & Amos, 2001). Due to the lack of current data at the tower station, only wave induced bottom shear stress ( w ) was estimated. Maximum wave bottom shear stress was estimated using the model proposed by Li and Amos (2001). Briefly, linear wave theory was used to calculate wave properties and estimate wave bottom shear stress as follows: )tanh(22hkTgL (2) ))sinh((hkTHub (3) TubAb 2 (4) w =0.5fwub 2 (5) fw=exp(5.213 )997.5)(194.0Abkb KbAb >1.7 (6a) fw=0.28, KbAb 1.7 (6b) where L, T, H, h, k are wave length (m), wave period (s), wave height (~0.707 times of wave significant height, m), water depth (m), and wave number (2/L, m -1 ), respectively; Ub, Ab, fw, are wave orbital velocity (m s -1 ), near-bed wave orbital amplitude (m), wave friction factor (dimensionless), and seawater density (kg m -3 ), respectively. The bottom roughness height (Kb) was taken as 0.003 m as obtained in Old Tampa Bay by Schoellhamer (1995) based on the assumption that within a limited geographic region this parameter is uniform (Peter Howd/USGS, personal communication). 12
13 2.4. Results and Discussion 2.4.1. Physical variations Five distinct wind events were documen ted during the 40 day winter deployment period. These events occurred around 11, 14, 19, 26 December 2004 and 14 January 2005, respectively (arrows in Figs. 2-2A and 23B). Wind speeds associated with these events were typically > 8.0 m s -1 for ~ 1-2 days (>18.0 m s -1 on 26 December 2004). Dominant directions of these wind events mos tly alternated from s outherly/southeasterly (from south or southeast toward north or northwest) wind to northeasterly/ northwesterly (from northeast or northwest toward southwest or southeast) wind. Th is transition pattern is typical of Southern Florida during passages of cold fronts from the north or northwest (Wang, 1998). Water levels exhibited distinctive oscill ations in mixed semidiurnal and diurnal forms with the highest water level of about 1.0 meter (Fig. 2-2B). Superimposed were fluctuations of water levels resulting from spring (e.g., ar ound 10 January 2005) and neap (e.g., around 18 December 2004) tides. Low-pass f iltered water levels revealed that local winds also played an important role in water levels. Northerly (from the north toward the south) winds led to decreased water levels (e.g. on 14 December 2004), while southerly (from the south toward the north) winds incr eased water levels (e.g., 25 December 2004). Current speeds ranged from 0.1 to 1.2 m s -1 with similar fluctuations as water levels, indicating that currents were driven primarily by tides. For example, despite similar wind conditions (< 3.0 m s -1 ), daily average of current sp eed on 10 January 2005 (under a spring tide) was approximately 0.60 m s -1 about two-fold of that on 18 December 2004 (,about 0.30 m s -1 during neap tide). The directions of currents mostly reversed from about 62 to 242 true within one tidal cycle, primarily along the orientation of the deep channel at the Sunshine Skyway Bridge station (Fig. 2-2C). Salinity showed similar fluctuations as water levels, ranging from 26 to 31 with higher salinity during flood tid es and lower salinity duri ng ebb tides (Fig. 2-2D), suggesting that water masses are horizonta lly transported by tidal advection. Water temperature ranged from 14 to 22 C, and it mi rrored passages of cold fronts depressing
14 temperature, otherwise water would gra dually warm up under calm conditions (Fig. 22E). Wave periods varied from 1 to 9 s with most waves having a period <2 s, showing that short waves prevailed in Tampa Bay in winter (Fig. 2-3A) due to the limited wind fetch inside Tampa Bay. Higher periods (e. g., >4 s) appeared independent of wind speed. For example there were higher periods on 4-10 January 2005, but wind speed was consistently lower. These longer waves (or sw ells) were likely transmitted from the West Florida Shelf (WFS). Significant wave height generally ranged from 25 to 200 cm with 5 peaks >100 cm (Fig. 2-3B), coincident with each of the five strong wind events. Wave bottom orbital velocities and bottom shear stre sses exhibited similar patterns as wave height and wind speed except for some m oderate peaks occurring on January 4-10 2005. The relatively higher wave periods during th at period may explain the increased bottom shear stresses (Figs. 2-3C and 2-3D), indicating that bottom shear stress can be influenced by both local wind-induced wa ves and remotely transported swells. 2.4.2. Bio-optical variations The backscattering coefficient, bbp(532), exhibited similar ch anges in the bottom and surface layers with the bottom values (mean and standard deviation were 0.092 and 0.044, m -1 respectively) slightly higher than the surface values (mean and standard deviation were 0.072 and 0.040 m -1 respectively) (Fig. 2-4A ). The bottom chlorophyll fluorometer malfunctioned after about 9 days, bu t prior to that, similar Chl patterns were observed between the surface and bottom layers (Fig. 2-4B). These results indicate
Dec.8 12 16 20 24 28 Jan.1 5 9 13 17 12.0 14.0 16.0 18.0 20.0 22.0 Temperature ( ) E Temp Flt_Temp 24.0 26.0 28.0 30.0 32.0 SalinityDSalFlt_Sal .0 .0 0.0 5.0 10.0 Wind (m s)A .05 0.00 0.50 1.00 1.50 Water Level (m)BObs.Flt_Obs. .6 .3 0.0 0.3 0.6 0.9 Current (m s ) C C h an nel ori e nt at i on Sp ri ng tid es Neap tides Figure 2-2. Hourly averaged tim e-series of (A) winds, (B) water levels (C) currents (D) salinity and (E) temperature. To show the su btida l var i ations, winds, water levels, salin ity, an d tem p eratur e data wer e f ilter e d usin g a low-pass f ilter with c u t o ff tim e of 36 hours and shown in wind stickplot (A) and with red solid lines in B, D, E panels. Current data (C) were not filtered because there were too m a ny m i ssing data poin t s 15
0.0 2.0 4.0 6.0 8.0 10.0 Wave Period (Second)AObs.Filt_Obs. 0 0.2 0.4 0.6 W ave orbital velocity (m/s ) C 0.0 50.0 100.0 150.0 200.0 250.0 Wave Height (cm)BSigHtFlt_SigHt Dec.8 12 16 20 24 28 Jan.1 5 9 13 17 100 bottom shear Stress (N/m2 ) D Figure 2-3. Hourly averaged tim e-series of (A) significant wave period, (B) significa nt wave height, (C) near bottom wave orbital velocity, and (D) near-b ottom wave shear stress. As in Figure 2-2, significant wave heig ht and wave period were also f ilter e d usin g the low-pa ss f ilte r and shown with red solid lin e s in A, B panels. To sh ow details o f bottom shear stress, y a x i s of panel D is presented with a logarithm scale and the black thick line stands for the m i ni m u m critica l shear stress of 0.2 N m -2 which was identified by visual inspection of tim e series of sedim e nt resuspension. The points of zero values are not shown in this pan e l du e to the loga rithm scale. 16
17 that the water column of Tampa Bay is well-mixed in winter. To simplify the comparison, we focused on examining the surf ace Chl and bbp (532) data in this study. The green backscattering coefficient, bbp(532), showed large variations ranging from 0.03-0.20 m -1 (Fig. 2-4A). A spectral analysis revealed that the highest energy of bbp(532) variation is associated with semidiurnal and diurnal frequencies, similar to the water level spectrum (Fig. 2-5). This cohere nce suggests that sedi ment variability was primarily driven by tidal mixing. A closer look into the time series of bbp(532), salinity, water level and currents revealed that highe r bbp(532) typically occurred in low salinity waters during ebb or early flood tides, sometimes displaying a doublepeak feature (Fig. 2-6A). Similar features were also found in other studies, and can be at tributed to the tidal effects on deposition and resuspension of sediments (e.g., Jing & Ridd, 1996). The lowpass filtered bbp(532) showed a baseline bbp(532) appeared to be about 0.06 m -1 (the maximum value under calm conditions at wind speed <3.5 m s -1 ). Compared to the baseline value, three sediment resuspension events can be identifie d, specifically on 8-16 and 25-30 December 2004 and 8-17 January 2005 (F ig. 2-4A). Apparently, the first two events were caused by winds or wind-indu ced waves. Sediments were typically suspended when hourly averaged wind speeds were > 8.0 m s -1 (or daily averaged wind speeds >5.0 m s -1 not shown) and bottom shear stress > 0.2 N m -2 (Fig. 2-3D). These suspended sediments remained in the water co lumn for ~ 2-3 days after a wind event. The above calculation of bottom shear stress (~0.2 N m -2 ) is, however, conservative given that the current contributions to bo ttom shear stress were not includ ed in our wave model, and therefore represents a minimum estimate of critical bottom shear stress for a pronounced sediment resuspension event. However, winds alone do not account for all sediment resuspension. For example, the wind event on 19 December 2004 with speed > 8.0 m s -1 did not result in a clear resuspension event. Likewise, the early stage of the third moderate sediment resuspension event could not be attributed to winds but was coincident wi th higher wave period at that period (Fig. 2-3A), suggesting the remotely transported wave may also contribute to sediment resuspension. Another possible reason for this resuspension is stronger currents observed at the same time (coincident with a spring tide). Unfortunately, we could not quantify the relative importance of currents versus waves fo r bottom shear stress in this
18 study, but previous studies suggested that cu rrents played a minor role in sediment resuspension at some sites in Tampa Bay (Schoellhamer, 1995; 1996). The observed currents in this study are occasionally > 1.0 m s -1 significantly higher than 0.2 m s -1 reported by Schoellhamer (1995, 1996) for Old Tampa Bay and Hillsborough Bay. Thus further studies are necessary to evaluate the importance of stronger currents in sediment resuspension in the Lower Tampa Bay. Chl showed high frequency variations, w ith concentrations ranging from ~ 2.0 to 12 mg m -3 (Fig. 2-4B). A spectral analysis suggeste d that Chl variation was coherent with water level variations and, therefore, dr iven by tides (Fig.2-5). These results are consistent with observations in other estuar ies (Cloern et al., 1989; Li & Smayda, 2001; Roegner et al., 2002). Within a tidal cycle, high er Chl values frequently occurred in lower salinity waters and/or during early flood tides indicating that tidal advection transports water masses with different chlorophyll concen trations and salinity back and forth along the Bay, in agreement with overall bay-wide chlorophyll and salinity distributions with higher chlorophyll and lower sa linity in the upper bay a nd vice versa (Bendis, 1999; Weisberg & Zheng, 2005). The low-pass filtered Chl time series show ed that the Chl ba seline is ~3.5 mg m -3 (the mode value of Chl data, representi ng the most common Chl values during the deployment period). Compared to this baselin e concentration, three Chl increases were also identified. These increases in Chl di d not exactly coincide with sediment resuspensions but occurred later, with a lag ti me of 1-2 days (Figs.2-4 and 2-6). One such an increase developed into an inte nsive phytoplankton bloom (Chl> 12.0 mg m -3 ) on 1718 December 2004, ~1 day after suspended sediments returned to the normal condition.
Dec.8 12 16 20 24 28 Jan.1 5 9 13 17 0.0 3.0 6.0 9.0 12.0 Chlorophyll (mg m)BSurfaceBottomFlt_SfcFlt_Btm 0.00 0.05 0.10 0.15 0.20 0.25 bbp(532) (m)A SurfaceBottomFlt_SfcFlt_Btm Figure 2-4. (A) Hourly averaged time series of backscattering coefficient (bbp(532)) in the surface (thin solid line) and bottom layers (thin dashed line). The horizontal black thick solid line represents bbp(532) of 0.06 m -1 as a baseline value, which is the maximum bbp(532) under calm conditions (wind speed < 3.5 m s -1 ); (B) Hourly averaged time series of Chl data in the surface (thin solid line) and bottom layers (thin dashed line). The horizontal black thick solid line represents Chl of 3.5 mg m -3 as a baseline concentration, which is the mode of surface Chl time series. The thick red lines are the low-pass filtered data with cutoff time of 36 hours to show subtidal variability of surface bbp(532) and Chl. 19
20Figure 2-5. Spectral density of time series of (A) water levels, (B) bbp(532), and (C ) Chl. The two horizontal dashed lines in each panel are the lower and upper 90% confidence limits, respectively. Significant diurnal and semidiurnal frequencies are synchronous among these spectra, indicating dominant tidal effects on sediments and chlorophyll variability. 0 1 2 3 Water Levels 0 0.1 0.2 0.3 0.4 0. 5 0 1 2 3 4 Frequency (hr)Chl 0 1 2 3 4 Spectral Densitybbp Diurnal Semidiurnal 2.4.3. Sediment resuspension versus phytoplankton blooms Given that there was no simultaneous Chl increase with sediment resuspension events, we conclude that there were minimal concentrations of benthic algae in the bottom sediment near the tower station. Therefore, increases in Chl after sediment settlement must have resulted mainly from phytoplankton growth in the water column adjacent to the station due to enhanced nutrient levels and optimal light availability. Indeed, both direct and indirect experiments suggested that phytoplankton growth in Tampa Bay is limited by nutrient availability, specifically by nitrogen concentration (Johansson, 2000; Vargo et al., 1991). Sediment resuspension has been widely suggested to increase nutrients in the water column and therefore is closely coupled to
phytoplankton dynamics (Lawrence et al., 2004; Stephen et al., 2004). Increased nutrients, however, can't be used until sediment concentrations decline and sufficient light become available for phytoplankton growth. Sediment resuspension alone can't account for the intensive phytoplankton bloom observed December 17-18, 2004. In addition to winds, tides appeared to play critically important role in stimulate phytoplankton blooms. The sediment resuspension event of December 8-16, 2004 was triggered by high bottom shear stress (2.0 N m -2 Fig. 2-3D), and was followed by a neap tide. Observations in other estuaries have shown that rapid increase in phytoplankton growth often occurs during periods of low tidal energy (such as during a neap tide) (Cloern, 1991; Huisman et al., 1999; Monbet, 1992; May et al., 2003). Our data for October 31 to December 13, 2005 (not shown) showed a similar phytoplankton bloom, which also occurred after a sediment resuspension event and during a neap tide. 2.4.4. Implications for water quality monitoring The EPCHC water quality monitoring program performs monthly surveys over the entire Tampa Bay with each survey spanning ~ 3 weeks. Each week samples one segment of Tampa Bay (Boler et al., 1991). Because of the extensive sampling across the Bay, the program is not able to characterize the high frequency variations shown above. The question is whether this program can provide adequate information on variability over seasonal or interannual scales. Average conditions for a specific month in a time series are ideally represented by a monthly mean value (mm) derived by averaging all data collected over the month. If a single observation (mi) is used to represent monthly conditions, such as with the EPCHC water quality monitoring program, the relative error can be estimated as follows: Relative Error (RE) = %100 mmmmmi (7) 21
Figure 2-6. Variations of bbp(532) and Chl with relation to salinity, water levels, and currents from 11-19 December 2004. A double-peak feature of bbp(532) and higher Chl associated with tides are highlighted with red circles. Dec 11 12 13 14 15 16 17 18 19 .50 0.00 0.50 1.00 Current (m s)Current .50 0.00 0.50 1.00 Water Level (m)Water Level 0 3 6 9 12 Chlorophyll (mg m)ChlorophyllSalinity 24 26 28 30 32 Salinity 0.00 0.05 0.10 0.15 0.20 Bbp(532) (m)bbp(532)Salinity 24 26 28 30 32 Salinity Dec.8 12 16 20 24 28 Jan.1 5 9 0.00 50 100 150 200 250 Relative Error(x100%) Date Chl Bbp(532) Figure 2-7. The relative errors (= % 100 mm mm mi ) of Chl (thin blue) and bbp(532) (thick red) incurred by representing a m onthly m ean by a single snapshot m easurem ents (m i ) 22
23 The error analysis is based on a comparis on of a monthly means (mm) computed with data collected continuously from 8 Decembe r 2004 to 7 January 2005. The potential error ranges from -50% to 200% for both Chl and bbp(532) if a single snapshot is used to represent the monthly mean value. The in situ time series data collected with automated sensors between 8 December 2004 and 7 January 2005 suggest that the potential errors introduced by using a single observation as the monthly mean range from 50% to 200% for both chlorophyll and sediment concentrations (Fig.2-7). These errors are larger than, or at least comparable to, the seasonal and interannual variations in Chl and turb idity observed in the longer EPCHC records (Fig.2-8). Therefore for a part icular site, single m onthly observations are inadequate to infer either seasonal or inte rannual variability. Only after the multi-year data are averaged to reduce aliasing effect s does the seasonal cycle become apparent (Fig.2-8). Similarly, when data from multiple locations during one survey are averaged, the aliasing errors are also reduced. Hence, a trend analysis fr om the EPCHC monthly data can only be conducted either over a la rge region for any particular month or over multi-year averages for a partic ular station. In addition, the 3-week difference in sample time for different bay segments makes it more difficult to assess spatial biases or differences because different sites may have experienced different wind and/or tidal effects during the stratified sampling peri ods. This difference may introduce further errors in characterizing spatial and temporal variability in phytoplankton and sediment. Thus, it is highly desirable to combine the m onthly surveys with other data, such as buoy or mooring observations as well as remote sensing imagery, to obtain a more reliable assessment of water quality variations in estuaries like Tampa Bay. The ongoing implementation of the integrated ocean obs ervation system (IOOS) would provide an unprecedented opportunity to advance our unders tanding of variability of water quality and their driving mechanisms.
Figure 2-8. The monthly time series of (A) Chl (mg m -3 ) and (B) turbidity (NTU) at one of the EPCHC stations (See Fig.2-1 for station location) from 1997 to 2003. The thicker lines represent the 7-year averages. Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0.00 2.00 4.00 6.00 8.00 10.00 12.00 14.00 MonthChl (mg m)1997199819992000200120022003Average 0.00 2.00 4.00 6.00 8.00 Turbidity (NTU)1997199819992000200120022003Average A B 2.5. Conclusions Bio-optical, physical, and meteorological observations showed that short-term variability of phytoplankton and suspended sediment concentrations in Tampa Bay was driven primarily by winds and tides. Sediment resuspension events were caused by wind-induced waves, swell, and currents, including tidal currents. The minimum critical shear stress for sediment resuspension at the study site was ~ 0.2 N m -2 Once suspended, sediments remained in the water column for ~ 2-3 days after winds decreased. No increase in Chl was observed in the water column during resuspension events, implying that benthic algae contribute little to the phytoplankton biomass in the water column. Instead, phytoplankton increased 1-2 days after sediments settled. This may be due to enhanced nutrient concentrations released during sediment resuspension events and optimal light availability subsequent to sediment deposition. Intensive phytoplankton 24
25 blooms (Chl >10 mg m -3 from a background level of about 3.5 mg m -3 ) occurred when a sediment resuspension event was followed by low winds and a neap tide. This suggests that minimal mixing following strong mixing an d resuspension events leads to the largest blooms, since these conditions allow turbidity to decrease and phytoplankton to utilize new nutrients delivered either by sedime nt resuspension or riverine inputs. The tidal and subtidal variability highlights some limitations in the current water quality monitoring program. These surveys are useful to characterize seasonal trends for any particular site if multi-year data are av eraged, or characterize inter-annual trends if Tampa Bay is treated as seve ral bay segments or as a whole. For smaller regions or specific sites, -50% to 200% errors could be introduced if those snapshot measurements are used to represent monthly mean condi tions, making it impossible to derive interannual trends for those specific sites. Although the results presented here were obtained only from 40-day continuous measurements (one sample per hour) at one station, the observed variability and responsible processes should be representative for variability across the bay in winter. In shallow regions, benthic algae may play a more important role in influencing phytoplankton biomass in the water column. Furt her studies at multiple stations and in different seasons are necessary to fully char acterize the short-term spatial and temporal variability of phytoplankton and sediments across the bay.
26 Chapter 3. Colored Dissolved Or ganic Matter in Tampa Bay 3.1. Abstract Absorption and fluorescence of colore d dissolved organic matter (CDOM) and concentrations of dissolved organic car bon (DOC), chlorophyll and total suspended solids in Tampa Bay and its adjacent rivers were examined in June and October of 2004. Except in Old Tampa Bay (OTB), in June, 2004, CDOM properties showed a conservative relationship with salinity (a CDOM (400)= 0.19 salinity+6.78, R 2 =0.98, n=17, salinity range = 1.1 32.5) with little variation in absorption spectral slope and fluorescence efficiency. CDOM distribution was therefore co ntrolled by mixing between various end members, including rivers entering Tampa Bay and marine waters of the Gulf of Mexico. In October, 2004, the average CDOM absorption coefficients measured around the Bay (a CDOM (400), ~7.76 m -1 ) were about seven times larger than values observed in June (~1.11 m -1 ) and appeared to behave non-conservatively. The nonconservative behavior may have been caused by mixing between various end-members, specifically rivers, marine waters, and mixing of waters from Hillsborough Bay, Middle Tampa Bay and Old Tampa Bay. Alternatively, removal of CDOM may have occurred at intermediate salinities (e.g., a CDOM (400) removal >15% at salinity ~ 13.0) due to photobleaching due under stratified conditions. Th e spatial and seasona l distributions of CDOM showed that the two largest rivers, the Alafia River (AR) and Hillsborough River (HR) were dominant CDOM sources. In OTB, CDOM showed lower absorption coefficients, higher absorption spectral slopes, lower ratios of CDOM absorption to DOC, and higher fluorescence efficien cy. These differences may be due to (1) changes in CDOM composition by photobleaching due to the longer residence time in OTB; (2) other sources of CDOM such as local creeks, streams, groundwater, bottom re-suspension or primary production from either phytoplankt on or seagrass. Compar ed to phytoplankton pigment absorption, a ph (443), average a CDOM (443) was about five times higher in June
27 and about ten times higher in October, showing that blue light attenuation was dominated by CDOM rather than by phytoplankt on absorption throughout the year. 3.2. Introduction Colored Dissolved Organic Matter (CDOM ) is one of the most important constituents affecting water quality in estuarie s. It has strong eff ects on light attenuation (Branco & Kremer, 2005; Clementson et al., 2004; Magnuson et al., 2004), affecting the light available to phytopla nkton and submerged aquatic ve getation (Anastasiou et al., 2005). Further, CDOM may also serve as a trac er of dissolved organic carbon (DOC), if the ratio between the colored and uncolored fractions of DOC is known. Since DOC affects the transport and bio-availability of trace metals and organic pollutants (Santschi et al., 1997; Guo et al., 2001), it is of interest in the assessment of CDOM of estuarine and coastal waters. CDOM distribution and varia tion patterns, however, have not been well studied in Tampa Bay. The current water quality mo nitoring program conducts only visual comparison of water samples with color standa rds. Thus these observations lack detailed information about the optical properties to track sources, sinks and changes in the composition and relative abundance of constituents. Furthermore, these color measurements are not reliable when used to estimate CDOM concentrations, primarily due to subjective assessment of co lors by each user (Gallegos, 2005). In this study, absorption and fluorescence of CDOM were examined using observations collected during two cruise surv eys, specifically in June and October, 2004. I examined how CDOM optical properties va ried relative to salinity, Chlorophyll concentration (Chl), total suspended solids concentration (TSS), and DOC. The objectives of the study were to (1) characteri ze the distribution and variability of CDOM in Tampa Bay; (2) understand processes that co ntrol its origins, disp ersal and sinks; and (3) help understand the ecologi cal significance of high light attenuation due to CDOM in Tampa Bay.
28 3.3. Methods Fig. 3-1 shows sample locations for the two surveys, conducted on June 1 3 and October 12 14, 2004, respectively. Sevent een stations were collected from three of the four sub-basins of the Bay during each survey. Six samples were also collected from the Alafia River (AR) in June, 2004. The AR was not sampled in October, 2004, because the rivers effect was readily observable in th e bay due to high discharge at that time. River flow rates (1973 to 2003) were obtained from US Geological Survey National Water Information System (USGS NW IS). While four major rivers discharge fresh water to the Bay, the Alafia Ri ver (AR) and the Hillsborough River (HR) contribute, on average, >80% of the total freshwater deli vered to the Bay (Robinson, 2004; and Fig. 3-2a). Fig. 3-2 also shows the ge neral trend of river flow rates (the wet season from June to October and the dry se ason for the rest months of a year) and interannal variability of this typical seasonal ity for year of 2004. It can be clearly seen that the sampling dates of June 1-3 a nd October 12-14, 2004 are in the dry and wet seasons, respectively. Therefore the data from June and October are used to represent general CDOM concentrations in the dr y and wet seasons, respectively although interannual variation in rive r flow rates may impact CDOM seasonal contrasts. 3.3.1. Temperature, salinity, chlorophyl l, and total suspended solids Surface water samples for temperature, sa linity, chlorophyll, and total suspended solids (TSS) were collected using a bucket fr om a small boat. Temperature and salinity were measured immediately using a WTW TM Multi 340i Meter (Aqua tic eco-system, FL, USA). Samples were stored in brown plastic bottles in a cooler with ice. They were filtered the same day after returning to the lab through GF/F Whatman filters (pump pressure < 120 mmHg) to determine chloroph yll concentration (Chl) using a TurnerDesigns fluorometer (Muell er et al., 2003). Whatman TM Nylon membrane filters (0.2 m pore size) were used to obtain CDOM filtr ates. TSS estimates were based on 500~800 ml water samples filtered through pre-weighed 47mm Millipore GN filters (0.45 m pore size). Filters were dried in a desiccator and weighed again. The difference of filter
weights between withand without particulates, together with the volume of filtered seawater, was used to calculate TSS (Mueller et al., 2003). HR HB OTB AR MT LMR MR LTB Figure 3-1. Map of Ta mpa Bay with sam p ling lo cations overlaid. The Bay is divided into four segm ents: Old Tam p a Bay (OTB), Hillsborough Bay (HB), Middle Tam p a Bay (MTB), Lower Tam p a Bay (LTB). Major rive rs are th e Alafia River (AR), Hillsborough River (HR), Little Manatee Ri ver (LMR), and Manatee River (MR). AR station num bers increase tow a rd the m outh of the river (num bers discussed in the text but not shown here). The inset shows the loca tion of Ta m p a Bay in Florida. 29
020406080100123456789101112Month of a YearRelative Percentage 100%) MR LMR AR HR 2004 Daily FlowsClimatological Daily Flows 1101000.1 100150200250Day of the Year (1-365)300350400050 1101000.1 Alafia River Hillsborough River (A) (B) (C) Daily Flow Rate (m 3 s -1 ) Figure 3-2. (A) Percentage of average ( 1973-2003) m onthly m ean flow rates of Hillsboroug h River (HR), Alaf ia Riv e r (AR) Little Mana tee River (LMR ), Manatee River (MR) rela tive to th e tota l m onthly m ean flow rate from these four major rivers of each m onth. (B) The lon g -term average (open ci rcle) and 200 4 daily flow s (filled circle) of the AR a nd (C) of the HR. Highlighted areas indicate the sam p ling dates in this study: day 153-155 (1 3 June, 2004, dry season) and day 286-288 (12 14 October 2004, wet season). 30
31 3.3.2. Dissolved organic carbon Dissolved organic carbon samples (DOC) we re collected directly from surface waters of the Bay using 500 ml brown glass bo ttles previously rinsed three times with Milli-Q water. Samples were filt ered through pre-combusted (500 o C overnight) 47-mm Whatman GF/F filters. The initial 250 ml of filtrate was discarded, and 50 ml of subsequent filtrate was preserved by poisoning with 30 L of 1.1 M high purity hydrochloric acid, then stored at C. A Shimadzu TOC-5000A total DOC analyzer (using high temperature catalytic oxidation) equipped with the ASI-5000A accessory was used to analyze DOC samples. Milli-Q wate r was used as a reference blank and to prepare a standard solution of potassium hydr ogen phthalate (KHP). The accuracy and precision of the measurements were better than 5%. DOC was not measured in the wet season. 3.3.3. CDOM absorption coe fficient and fluorescence CDOM abundance based on its light absorption and fluorescence properties. A Hitachi U3310 double-beam spectrophotometer (300-850 nm, 2 nm resolution) was used to measure CDOM spectral absorbance (A( ), dimensionless) in 10 cm quartz cells with Milli-Q water as the blank reference. A( ) was measured three times and the mean was used to calculate the CDOM absorption coefficient, a CDOM ( ) (m -1 ) as follows (Hu et al., 2002): a CDOM ( ) = ln 10 A( ) /L, (1) where L is the cuvette pathlength (0.1 meter). A nonlinear least square regression was used to derive spectral slope (S, nm -1 ) over the wavelength range 350-550 nm: a CDOM ( ) = a CDOM (400) exp[ S( 400)]+K, (2) where a CDOM (400) is the CDOM absorption coefficient at a reference wavelength of 400 nm, and K is an offset to account fo r residual scattering and/or noise. CDOM fluorescence spectroscopy was perf ormed according to the method of Coble (1996) using a SPEX I ndustries Fluoromax-2 spectrofluorometer. Samples with A(300) > 0.02 were diluted to avoid self-shading (Green, 199 2). The fluorescence signal,
32 after corrections for Raman scattering and instrument configuration, was normalized to units of quinine sulfate equivalents (QSE) in ppb using the fluorescence of a diluted series of quinine sulfate di hydrate in 0.05 M sulfuric acid at excitation/emission (Ex/Em) of 350/450 nm. CDOM fluorescence intensity was reported as fluorescence at Ex/Em of 300/420 nm, located in Humic Peak C region (Coble, 1996). 3.3.4. Phytoplankton pigment absorption coefficient In the wet season, the spectral absorption coefficient due to particulate matter, a p ( ), was determined using the technique of Yentsch (1962). I used a custom-made, 512channel spectroradiometer (~350-850nm) to measure absorption. Pathlength amplification was corrected using the factor of Carder et al (1999). Detrital absorption spectra, a d ( ), were obtained after chemical extr action of phytoplankton pigments from the sample using hot methanol (Kishino et al., 1985). Phytoplankton pigment absorption spectra, a ph ( ), were then calculated by difference: a ph ( )= a p ( ) a d ( ). (3) Because the maximum absorption of phytoplankton occurs near 443 nm, a ph (443) was chosen to compare with a CDOM (443). In the dry season, when no particulate absorption was measured, a ph (443) was modeled using followi ng equation (Bricaud et al., 1995; Babin et al., 2003): a ph (443)=0.04 Chl 0.668 (4) 3.3.5. Statistical tests Pearsons product moment correlation and linear and non-linear regression analyses were used to evaluate the statistica l relationships between variables. Statistical significance was reported as either not signi ficant (NS) (p>0.05), weak (*, 0.01
33 3.4. Results Appendix 3-1 lists the results for all sta tions by segments and seasonal surveys. These results are discussed below. 3.4.1. Distribution of TSS and Chl TSS concentrations ranged from 1.6 mg/l in Middle Tampa Bay (MTB) to 18.2 mg/l in the Alafia River (AR). In general, there was no apparent relationship between TSS and salinity (Fig. 3-3A, Table 3-1) within individual sampling locations except in Hillsborough Bay (HB), where TSS was higher in the wet season than in the dry season (Fig. 3-3A, Appendix 3-1), indicative of higher riverine sediment inputs directly into this area in the wet season. MTB generally showed the lowest TSS of the three bay segments in both seasons. This suggests that TSS in Tampa Bay was not directly controlled by riverine inputs. Chl showed significant seas onal variations. The wet season Chl exceeded those of the dry season by factors ranging from 2 to 10 (Appendix 3-1). The highest Chl were observed in the wet season in HB (79.0 mg/m 3 ) and MTB (>40.0 mg/m 3 ). In the AR, highest Chl (18.0 mg/m 3 ) in June was observed in the intermediate salinity range 10.0~20.0, and Chl decreased with increasing salinity (Fig. 3-3B), consistent with earlier observations (Vargo et al., 1991). Looking at individual stations, there was a moderate inverse correlation between Chl and salinity in the wet season (Fig. 3-3B, Table 3-1). This correlation was attributed to nitrogen supply through rivers in the wet season, as Tampa Bay phytoplankton are generally nitrogen limited (Vargo et al., 1991). 3.4.2. Light absorption by CDOM CDOM absorption coefficients generally decreased with increasing salinity. The lowest CDOM absorption values occurred in MTB in the dry season (Fig.3-4, Appendix 3-1). The relationships between a CDOM (400) and salinity, howev er, varied with bay segments and seasons. In the dry season, except in Old Tampa Bay (OTB), a CDOM (400) was strongly inversely correlated to salinity, indicating that riverine CDOM was dominant. CDOM
from the AR was conservatively mixed into the MTB. In the wet season, however, a CDOM (400) in HB and MTB showed an exponential decrease with increasing salinity (Fig.3-4, Table 3-2). This may indicate that CDOM was not conservatively mixed or that there were multiple end-members mixing different proportions of CDOM and salinity in the central regions of the Bay. If the behavior is non-conservative, then >15% of a CDOM (400) was removed around salinities of about 13.0, which was the median salinity in October, 2004. Alternatively, CDOM absorption in OTB was significantly lower than in HB and MTB in both seasons,, indicative of possible differences in CDOM sources in OTB (Fig. 3-4). Mixing of OTB water with water from HB and MTB would also lead to an exponential decrease in the salinity-CDOM curve, which would give the appearance of non-conservative behavior. It is difficult to determine which effect caused the observed patterns without additional detailed experiments and observations. 34(B) Wet Season HBDry Season MTBWet Season MTBDry Season OTBWet Season OTBDry Season ARDry Season HB 5.010.020.025.030.035.015.0Salinity0.0 10Chl (mg/m3)1001(A) 5.015.020.0m)10.00.0 g/l S ( TS Figure 3-3. Surface salinity versus (A) total suspended solids concentration (TSS) and (B) chlorophyll concentration (Chl). Open symbols show dry season observations and filled symbols show wet season observations.
35 Table 3-1 Correlation coefficients (r) be tween surface salinity and TSS and between surface salinity and Chl in Tampa Bay a nd the AR in the dry and wet seasons Season Parameter AR HB+MTB HB+MTB+O TB TSS vs salinity 0.58 NS (n=6) 0.43 NS (n=11) 0.46 NS (n=17) Dry Season (June, 2004) Chl vs salinity 0.77 NS (n=6) 0.41 NS (n=11) 0.58 (n=17) TSS vs salinity N.D. 0.82 (n=9) 0.43 NS (n=14) Wet Season (October, 2004) Chl vs salinity N.D. 0.79 (n=9) 0.73 ** (n=14) Note: (1) Statistical significance is reported as either NS (p> 0.5), (0.01
aCDOM(400)=43.70 x exp( 0.11xsalinity) (R2=0.98, n=9 without OTB points)aCDOM(400)= 0.19 x salinity+6.78 (R2=0.98, n=17,without OTB points) Wet Season HBDry Season MTBWet Season MTBDry Season OTBWet Season OTBDry Season ARDry Season HB 5.0010.0020.0015.00aCDOM(400) (m-1)0.00 5.010.020.025.030.035.0Salinity15.00.0 Figure 3-4. Surface salinity versus CDOM a CDOM (400) for Tampa Bay and the AR in the dry and wet seasons (see also Table 3-2). The broken line is a theoretical conservative mixing line in the dry season in June, 2004, while the dashed line is the hypothesized mixing line for the wet season in October, 2004 (a CDOM (400)=16.81 1.10(Salinity-8.8)). CDOM absorption exhibited significant seasonal variations. Average a CDOM (400) in the wet season was about four, seven and ten times higher than in the dry season for OTB, MTB and HB, respectively (Appendix 3-1). For example, at TB17, a CDOM (400) in the dry season was 1.50 m -1 (salinity=27.3), but increased to 16.80 m -1 in the wet season (salinity=8.8). The average a CDOM (400) over the entire bay increased from ~1.11 m -1 in the dry season (salinity =27.9) to ~7.76 m -1 in the wet season (salinity =14.3), or about sevenfold. Furthermore, extrapolations from regression analysis between a CDOM (400) and surface salinity in the dry season and from the hypothesized conservative mixing line in the wet season indicate that a CDOM (400) of the river end member in the wet season (> 26.5 m -1 at salinity =0.0) was significantly higher than that in the dry season (about 6.78 m -1 ). The difference suggests a seasonal difference in the amount f land-based CDOM delivered to the Bay. CDOM absorption showed no significant correlation (p>0.05) with TSS or Chl in the dry season (Figs. 3-5A and 3-5B, Table 3-3). In the wet season, however, CDOM was 36
37 moderately correlated with Chl over most of Tampa Bay (Fig. 3-5B, Table 3-3), likely as a result of the coincidence that both high lo ads of nutrients and CDOM were delivered into the bay via increased riveri ne discharge in the wet season. As a result of the high CDOM concentra tions, CDOM absorption of blue light was significantly higher than that due to phytoplankton pigment absorption. Average a CDOM (443) was about 5 and 10 times higher than a ph (443) in the dry and wet seasons, respectively (Fig. 3-5C). The ratios of a CDOM (443) and a ph (443) ranged 3-11 for the different bay segments and seasons, with maxima in HB in the dry season and in MTB in the wet season. The minima were observed in OTB in both seasons (Table 3-4). CDOM absorption spectral slope (S) with in the AR, HB, and MTB stations were similar in both seasons (p<0.05). Higher sp ectral slopes were, however, found in OTB in both seasons, with much higher slopes in th e dry season (Fig.3-6). Chemical composition of the CDOM seems to be different in OT B, possibly due to phot obleaching or due to different CDOM sources in OTB, as discussed below.
(A) 100TSS (mg/l)101aCDOM(400) (m-1)(B) 10100Chl (mg/m3)1 1101000.11:10 line Wet Season Average1:1 line (C)1:5 line Dry Season Average Wet Season HBDry Season MTBWet Season MTBDry Season OTBWet Season OTBDry Season ARDry Season HBaCDOM(443) (m-1) 1010010.1 110aph(443) (m-1)0.10.01 Figure 3-5. a CDOM (400) versus (A) TSS and (B) Chl and (C) a CDOM (443) versus a ph (443) for Tampa Bay and the AR in the dry and wet seasons (also see Table 3-3). 38
39 Table 3-2 Results of regr ession analysis between a CDOM (400) and surface salinity in Tampa Bay and the AR in the dry and wet seasons. Region Regression equation (R 2 ) Number of samples Dry Season (June, 2004) AR a CDOM (400) = 0.19 salinity+6.77 0.90 ** 6 HB and MTB a CDOM (400) = 0.19 salinity+6.72 0.90 ** 11 AR,HB + MTB a CDOM (400) = 0.19 salinity+6.78 0.98 *** 17 HB+MTB+OTB a CDOM (400) = 0.077 salinity+3.27 0.29 NS 16 Wet Season (October, 2004) HB and MTB (without OTB) a CDOM (400) =43.70 exp( 0.11 Salinity) 0.98 *** 9 Note: (1) Statistical significance is reported as either NS (p>0.5), (0.01
40 Table 3-3 Correlation coefficients (r) between a CDOM (400) and TSS and between a CDOM (400) and Chl in Tampa Bay and the AR in the dry and wet seasons. Season Parameter AR HB+MTB HB+MTB +OTB TSS vs a CDOM (400) 0.43 NS (n=6) 0.60 NS (n=11) 0.36 NS (n=17) Dry Season (June, 2004) Chl vs a CDOM (400) 0.74 NS (n=6) 0.55 NS (n=11) 0.24 NS (n=17) TSS vs a CDOM (400) N.D. 0.79 (n=9) 0.22 NS (n=14) Wet Season (October, 2004) Chl vs a CDOM (400) N.D. 0.77 (n=9) 0.73 ** (n=14) Note: (1) Statistical significance is repo rted as either NS (p>0.5), (0.01
Table 3-4 The averages and standard deviations (in parentheses) of CDOM absorption (a CDOM (443)) and phytoplankton pigment absorption (a ph (443)) in different bay segments and seasons. Season Parameter Hillsborough Bay Middle Tampa Bay Old Tampa Bay a CDOM (443) 0.71(0.04) 0.50 (0.22) 0.42 (0.09) Dry Season (June, 2004) a ph (443) 0.11 (0.05) 0.09 (0.03) 0.13 (0.01) a CDOM (443) 7.75 (1.87) 5.09 (1.79) 1.80 (0.34) Wet Season (October, 2004) a ph (443) 1.14 (0.42) 0.46 (0.32) 0.30 (0.12) Wet Season HBDry Season MTBWet Season MTBDry Season OTBWet Season OTBDry Season ARDry Season HB 15.020.025.030.0CDOM Spectral Slope (x10-3, nm-1)10.0 5.010.015.020.025.030.035.0Salinity0.0 Figure 3-6. CDOM absorption spectral slope (S) versus surface salinity for Tampa Bay and the AR in the dry and wet seasons. The dashed and solid lines represent the average spectral slope without OTB samples in the dry and wet seasons, respectively. Students t-test showed that these average values were significantly smaller than those from OTB in both seasons (p<0.01). 41
3.4.3. DOC During the dry season, DOC varied from 200 to 500 M, with higher concentrations in the AR and OTB and lower in the HB and MTB. Unlike CDOM (Fig. 3-4), however, DOC in the AR did not decrease with increasing salinity but remained almost constant along the AR salinity gradient (Fig. 3-7). When comparing CDOM absorption with DOC, OTB showed consistently lower ratios of CDOM absorption to DOC (Fig. 3-8), suggesting that DOC in OTB may have originated from different sources or that the rate of bleaching of CDOM was higher than in other parts of the bay. Dry Season HB Dry Season OTBDry Season ARDry Season MTB 200300400500600DOC (100 5.010.015.020.025.030.0Salinity35.00.0 Figure 3-7. Surface salinity versus DOC concentration for Tampa Bay and the AR in the dry season in June, 2004. 42
Dry Season HBDry Season OTBDry Season ARDry Season MTB 200300400500600DOC(100 2.004.006.008.00aCDOM(400) (m-1)0.00 Figure 3-8. a CDOM (400) versus DOC concentration for Tampa Bay and the AR in the dry season in June, 2004. 3.4.4. CDOM fluorescence efficiency The ratio of CDOM fluorescence to absorption, or "equivalent fluorescence efficiency", was relatively constant throughout the bay and the AR during the dry season (Fig.3-9). This is consistent with the inference that most CDOM was derived from river inputs, and that CDOM was conservatively mixed in most of Tampa Bay. In contrast, the OTB CDOM showed slightly higher fluorescence efficiency than those in other areas (Fig.3-9, Table 3-5), along with other differences in the other CDOM optical properties in the OTB: lower CDOM absorption, higher spectral slopes, and lower ratios of CDOM absorption to DOC. 43
Fluorescence=11 x aCDOM(400)+9R2=0.95, n=22, all TB and AR Data Dry season HRDry season MTBDry Season ARDry season OTB 2.004.006.008.00aCDOM(400) (m-1)0.00 6090120CDOM Fluorescence QSE (300/420)300 Figure 3-9. CDOM fluorescence (QSE) versus a CDOM (400) (m -1 ) for Tampa Bay and the AR during the dry season in June, 2004. 44
45 Table 3-5 Results of regression an alysis between CDOM fluorescence and a CDOM (400) Tampa Bay and the AR during the dry season. Region Regression equation (R 2 ) Number of samples AR Fluorescence=12.02 a CDOM (400) +5.95 0.91 ** 6 HB and MTB (Without OTB) Fluorescence=13.23 a CDOM (400) +4.23 0.89 *** 11 HB+MTB+OTB Fluorescence=1095 a CDOM (400) +10.10 0.34NS 16 HB and MTB+AR Fluorescence=12.12 a CDOM (400) +5.53 0.99 *** 17 All data Fluorescence=12.12 a CDOM (400) +9.53 0.95 *** 22 Note: Statistical significan ce is reported as either NS (p>0.5), *(0.01
46 that CDOM properties from the HR and AR might be optically similar so that a single conservative mixing behavior (or effectiv ely the same riverine end-member) was observed. Indeed, Hastings et al. (2004) s uggested that CDOM from the HR was not optically distinct from that derived from th e AR. These rivers have similar watersheds (Chen et al., 2004), while rivers with differe nt watersheds show different CDOM loads (Chen and Gardner, 2004). Similarly, in Ta mpa Bay, CDOM from the Little Manatee River (LMR) and Manatee River (MR) s howed higher CDOM fluorescence/absorption ratios than those from the HR and AR. The AR and HR are associated with similar urban watersheds (Estevez et al., 1991), but the LMR and MR are relate d to more rural, agriculture-intensive waters heds (Hastings et al., 2004). Conservative mixing of CDOM is common in estuarine systems because high CDOM concentrations mask small variati ons caused by other processes, such as photochemical and biological processes (N ieke et al., 1997; Blough and Del Vecchio, 2002; Rochelle-Newall and Fisher, 2002; Kowalczuk et al., 2003; Chen et al., 2004). CDOM absorption in Tampa Bay is indeed rela tively high compared to those reported for other estuaries. For instance, at salinity of 25.0, a CDOM (400) in Tampa Bay in the dry season was about 1.50 m -1 while it was about 0.50 m -1 in the Chesapeake Bay (RochelleNewall and Fisher, 2002) and about 0.70 m -1 in the Mississippi Ri ver plume (Hu et al., 2003). In various ecological setti ngs, phytoplankton (Carder et al., 1989) or submerged aquatic vegetation such as seagrasses (Otis et al;, 2004) may release CDOM into the water. The poor correlation between a CDOM (400) and Chl in the dry season, and particularly the lack of co rresponding CDOM increase in Ch l peaks in the AR, however, indicate that phytoplankton played only a mi nor role in regulating CDOM abundance in the dry season in Tampa Bay. While a CDOM (400) showed a moderate correlation with Chl in the wet season, this correlation likely resulted from a coin cidence between high concentrations of nutrients (therefore Chl) and CDOM, rather than an inherent cause-andeffect relationship between Chl and C DOM. Little contribution of CDOM from phytoplankton has also observed in other estuaries, such as in the Chesapeake Bay (Rochelle-Newall et al., 1999; Rochelle-N ewall and Fisher, 2002). Similarly, the poor correlation between a CDOM (400) and TSS suggests that CDOM contribution from
47 sediment resuspension was also negligible. However, further study is required to provide more direct evidence for the relationshi p between CDOM variation and sediment resuspension events in Tampa Bay, since previous studies in other coastal and estuarine waters have suggested that CDOM could be derived from bottom sediments (e.g., Boss et al., 2001; Burdige et al., 2004). Several processes have been reported to be responsible for CDOM removal, such as flocculation and precipitation of sediment s in estuaries (Uher et al., 2001), microbial transformation and photobleaching (Blough and Del Vecchio, 2002; and references therein). We found no significant correlation between CDOM and TSS in either season, which suggested that TSS in Tampa Bay ( 1.6 ~ 18.2 mg/l) may not be actively involved in CDOM removal by sediment adsorption (Uher et al., 2001). CDOM removal by photobleaching in Tampa Ba y might also be negligible in the dry season. Previous studies have found that photobleaching usually takes weeks to months to effect a noticeab le removal of CDOM, depending on insolation and water column stability (e.g., Vodacek et al., 1997). Numerical mode ling of Tampa Bay, however, suggest that the e-folding time (the time required for the number of particles in a grid cell to decrease by 65%) in Lower Tampa Bay (LTB) and in the adjacent deep channel is about 10 days or less (Burwell et al., 2000). Furthermore, the water column was generally well mixed in the dry season, which would further limit photobleaching. This hypothesis is consistent with the small variati on observed in the spectral slope (Fig. 3-6) and fluorescence efficiency (Fig.3-9) along the salinity gradient from the AR to MTB. In the wet season, we observed an appa rent non-conservative behavior along the salinity gradient going from HB to MTB, suggesting the pres ence of a sink. If a sink was present, at least 15% of a CDOM (400) might have been removed at salinity ~13.0, assuming no CDOM was added from othe r processes (Fig.3-4). Indee d, if we hypothesize that conservative mixing yields a CDOM (400) =0.00 m -1 at salinity=36.0, then CDOM removal at salinity ~13.0 would be >50%. If CDOM re moval occurred in the wet season, it may have been due to photobleaching under increased stratification conditions in the wet season Tampa Bay due to buoyancy input from freshwater and heat (Burwell et al.,
48 2000). Also, CDOM delivered via rivers in the wet season may be younger and therefore susceptible to degradation (Z anardi-Lamardo et al., 2004). An alternate hypothesis is that the appa rent non-conservative mixing was due to mixing of water from more than two en dmembers. About 20% of the freshwater delivered to the Bay is derived from the LMR and MR (Fig. 3-2) (see also Hu et al., 2004). Yet, the contribution of OTB water can cause lower CDOM concentrations than those from mixing between HB and MTB becaus e CDOM in OTB is about half of that observed in HB at similar salinities. As mentioned above, CDOM in OTB showed distinctive propert ies relative to other bay segments. The higher spectral slope s might be an indicator of photobleaching due to loss of the CDOM fraction with higher molecular wei ght (Twardowski and Donaghay, 2002), in agreement with the longer water residence time in OTB (> 140 days) (Burwell et al., 2000) allowing for sufficient photobleaching (Vodacek et al., 1997). The longer water residence time also implies that the CDOM exchange through water circulation is weaker in OTB compar ed with in the HB and MTB, making local processes (phytoplankton and benthic production/degradation, CD OM released by sediment resuspension, photobleaching) more effective in determining its optical properties. The differences in CDOM propert ies in OTB may also simply arise from different sources of CDOM, including local cr eeks, streams, rivers, groundwater, or from autochthonous production from phytoplankton (C arder et al., 1989) or submerged aquatic vegetation (Otis et al;, 2004). 3.5.2. Seasonal variation in CDOM absorption Average a CDOM (400) in Tampa Bay during the wet season was ~7.76 m -1 or about 7-fold higher than in the dry season (~1.11 m -1 ). These values are near the upper limit of published ranges of CDOM absorption coefficients in coastal and estuarine waters (Kowalczuk et al., 2003). Indeed, a CDOM (400) extrapolated to a riverine endmember (salinity = 0.0) are ~7.0 m -1 and 26.5 m -1 in the dry and wet seasons, respectively. Thus, about 4-fold higher CDOM was pr esent in rivers in the wet s eason than in the dry season. These changes are linked to seasonal variat ion in river flow (Fig.3-2). Hurricanes or tropical storms may transport a dditional CDOM from watersheds into rivers and estuaries
49 (e.g., Avery et al., 2004), such as in 2004, when four major hurricanes (category three or greater) affected Tampa Bay. The variations in freshwater inputs may also alter CDOM composition, as observed in other southern Fl orida estuarine and coastal waters (e. g., Zanardi-Lamardo et al., 2004). 3.5.3. Implications for water clarity monitoring and remote sensing The absorption of blue light was dominated by CDOM rather than by phytoplankton pigments (Fig. 3-5C). Our re sults show that CDOM and ph ytoplankton pigments covary in the wet season, which is likely the primar y reason why earlier studies concluded that chlorophyll could account for some of the vari ation in light attenu ation (e.g., Janicki et al., 2001). However, this correlation is s easonal and is uncommon in the dry season. A better index of water clarity is CDOM absorp tion or fluorescence, since this serves as a proxy for CDOM concentration and light absorp tion. This is a simple measurement that would benefit water quality monitoring program s. The high ratio of CDOM to pigment absorption in Tampa Bay makes it difficult to estimate Chl reliably using the sea spectral reflectance band-ratio algorithms th at are applied in Case I waters (O'Reilly et al., 2000). Instead, in complex CDOM-rich coastal e nvironments, chlorophyll fluorescence line height (FLH) observations are better suited to assess synoptic patterns of Chl distribution (Hu et al., 2005). Both MODIS (Moderate Resolution Imaging Spectroradiometer Sensors) and the European MERIS (Medium Resolution Imaging Spectrometer) collect remotely-sensed fluorescence data, and these im ages should be evaluated for application in Tampa Bay. 3.6. Conclusions Two surveys, one each in June and October of 2004, were conducted in Tampa Bay to study the applicability of optical observations to assess water quality indices. The results show that colored dissolved organic matter (CDOM) dominates the absorption of blue light relative to phyt oplankton pigments. Average a CDOM (443) was five and ten times higher than phytoplankton pigment absorption, a ph (443), in the dry and wet seasons, respectively. The Alafia River and the Hillsborough River were the main CDOM
50 sources to Tampa Bay. These rivers showed a CDOM (400) of about 7.00 m -1 and 26.50 m -1 at zero salinity in the dry and wet seasons, respectively. CDOM absorption coefficient within Tampa Bay showed significant seasonal variations with average a CDOM (400) of ~1.11 m -1 and ~7.76 m -1 in the dry and wet seasons, respectively. In the dry season, ex cept for Old Tampa Bay, CDOM distribution was primarily controlled by conservative mi xing between riverine inputs (Alafia and Hillsborough Rivers) and marine coastal waters ( a CDOM (400)= 0.19 salinity+6.78, R 2 =0.98. N=17, salinity=1.1 ~ 32.5). Other proc esses such as phyt oplankton production, sediment resuspension, and photobleaching seemed to have little impact on CDOM abundance. In the wet season, a CDOM (400) showed an exponential decrease with increasing salinity along a gradient from Hillsborough Bay to Middle Tampa Bay. Two possibilities are proposed to explain this. On e is that CDOM was removed during mixing. The other is that there is a significant c ontribution of CDOM from Old Tampa Bay with lower CDOM concentrations. Old Ta mpa Bay showed relatively lower a CDOM (400), higher CDOM spectral slope, lower ratios of CDOM absorption to DOC and higher fluorescence efficiency than other parts of Tampa Bay. These differences might suggest more intensive photobleaching in shallower waters, or different CDOM sources, such as local rivers, streams, ground water, bottom resuspension or even phytoplankton or submerged aquatic vegetation. Further studies may help clarify these differences. The predominance of CDOM absorption over phytoplankton pigment absorption precludes reliable estimates of Chl using conventional empirical or semi-analytical biooptical algorithms. Chlorophyll fluorescence lin e height from MODI S or MERIS may be a good alternative to these absorp tion-based chlor ophyll algorithms. The study provided first results of seasonal va riations of CDOM distribution as well as general inferences of CDOM sources, distribution and sinks in Tampa Bay. However, these were based on two quasi-synoptic survey s in two short time periods of two typical seasons. For a better understa nding of the interactions between CDOM and rivers, sediment and phytoplankton, more intensive sampling is required.
51 Chapter 4. Monitoring turbidity in Tampa Bay using MODIS/Aqua 250 m imagery 4.1. Abstract An approach to map turbidity in Tampa Ba y is developed for application with the 250 m imagery collected with the Moderate Resolution Imaging Spectroradiometer Sensor (MODIS). The approach includes cr oss-calibration of M ODIS 250 m data with well-calibrated MODIS 1 km data, a simple atmospheric correction, and development of a bio-optical inversion algorithm based on in situ reflectance and turbidity measurements. Results show that the MODIS pre-launch radi ometric calibration of the 250 m bands was adequate for this application. A simple atmospheric correction provided reliable retrievals of remote sensing reflectance at 645 nm (0.002 < R rs (645) < 0.015 sr -1 median bias = 7%, slope = 0.95, intercept = 0.00, r 2 =0.97, n=15). A more rigorous approach, using a multiple-scattering atmospheric correction of the cross-calibrated at-sensor radiance, retrieved similar R rs (645). R rs (645) estimates, after rigorous quality control, showed a close correlation with in situ turbidity (turbidity =1203.9 R rs (645) 1.087 0.9 < turbidity < 8.0 NTU, r 2 =0.73, n=43). The MODIS turbidity ma ps derived using this algorithm showed distinct spatial and te mporal patterns related to ri ver runoff in Upper Tampa Bay, and to wind-induced sediment resuspension ev ents in the middle and lower portions of the bay. The monthly mean turbidity patterns estimated from MODIS were different from those determined from single monthly in situ observations, which I at tribute to aliasing in a fast-changing estuary. Synoptic and frequent sampling facilitated by satellite remote sensing helps improve estimates of the mean patterns of turbidity in estuaries like Tampa Bay and are a valuable tool that should be used in monitoring of water quality of estuarine and coastal waters.
52 4.2. Introduction The "turbidity" of water is a common i ndex used to assess coastal and estuarine water quality. It is used to help understand factors that control light attenuation and therefore the productivity of planktonic and benthic algae (Cloern, 1987; Cole & Cloern, 1987; Fisher et al., 1999; Pennock & Sharp, 1994) seagrass, and coral reefs (Anthony et al., 2004; Moore et al., 1997). In coastal and estuarine waters turbidity is frequently directly associated to concentrations of tota l suspended solids or sediments (TSS) in the water column. Knowledge of the distribution of suspended or resuspended particles is important to understand processes like coasta l erosion and mobilization of chemicals or pollutants (Heyes et al., 2004). Therefore, unders tanding how turbidity vari es is of critical interest to coastal resource managers and researchers. Turbidity in coastal and estuarine wate rs shows a wide range in spatial and temporal variability. Rivers deliver terrestrial materials to estuaries following seasonal patterns, but also in events that may trigger marked interannual vari ability (Pribble et al., 2001). Currents and waves lead to suspension of bottom sediments, changing turbidity in response to storms and other wind events at tid al and subtidal freque ncies (Cloern et al., 1989; Schoellhamer, 1995). In addition to these natural forcings, human activities such as transportation and dredging also influence th e magnitude and distribution of turbidity (Schoellhamer, 1996).As a result of these pr ocesses, turbidity in estuaries is highly variable. Conventional sampli ng methods often fail to characterize turbid ity patterns because of their limitations in temporal and spatial sampling (Chen et al., submitted). Satellite remote sensing has been used successfully to map sediment concentrations and turbidity in coastal and estuarine waters (Doxara n, et al., 2002; Miller et al., 2005; Ruddick et al., 2003; Stumpf & Pennock, 1989). However, due to the inherent problems associated with sensor s (e.g., spatial resolutions, revisit times and accessibility of images), routine application of satellite remote sensing to monitor turbidity and sediments in coastal waters has been limited. For example, although Landsat ETM+ provides high-resolution (30-m) imagery, the revisit time is ~ 16 days. This long revisit time, cloud cover, and high co st make it inadequate to resolve turbidity dynamics in coastal and estuarine waters. Sensors with nearly daily overpasses and designed to observe marine waters (e.g., the Advanced Very High Resolution Radiometer
53 (AVHRR), the Sea-viewing Wide Field-of-Vie w Sensor (SeaWiFS), and the ocean bands of the Moderate Resolution Imaging Spectrora diometer Sensor (MODIS)) typically have coarse spatial resolution, such as ~1 km per pixel, and are of limited value to examine processes in most estuarine waters. The MODIS flown aboard the Aqua spacecraft, launched in July 2002, provides near-daily coverage of the subtropical ocean and has two bands that observe the Earth at 250 m resolution (band 1: 645 nm, from 620-670 nm; band 2: 859 nm, from 841-876 nm). These bands have sufficient sensitivity to detect a wide range of changes in the color of estuarine waters (Hu et al ., 2004). Several studies have de monstrated the potential of these bands to monitor water quality in coas tal and estuarine wate rs (Hu et al., 2004; Miller & Mckee, 2004). Two key issues, however, must be reso lved before MODIS 250 m data can be routinely applied in coastal studi es (Hu et al., 2004). First, the reliability of at-sensor (i.e., the top of atmosphere, TOA) radiance obser vations has to be a ssessed, and a method has to be available to estimate remote sensing reflectance (R rs ) of aquatic environments. The MODIS 250 m bands were originally designed to serve as "sharpening bands" to detect land, aerosol, and cloud features. Therefore, ther e has been no effort to apply an oceanic vicarious calibration such as done for the MODIS ocean 1 km bands. Remote sensing of aquatic environments demands rigor and accu racy in sensor calibration because a 5% error in at-sensor radiance may result in a 50% R rs error or worse. Also, bio-optical algorithm (an algorithm to convert R rs to water-quality parameters such as turbidity or TSS) typically derived from limited in situ data frequently is inadequate for application elsewhere or at other times (Hu et al., 2004; Miller & Mckee, 2004). Hu et al. (2004) discussed desired improvements to the MODIS 250 m products, including vicarious calib ration and coupling with in situ measurements. Here I present results of a vicarious calibration effect ed by cross-calibration of MODIS 250 m bands with the MODIS 1 km ocean color bands whic h have been calibrated using the Marine Optical Buoy (MOBY) located near Hawaii. I then develop and validate an empirical algorithm to convert the atmos pherically corrected data (R rs ) to turbidity. Finally, I derive monthly mean patterns of turbidity using a time series of MODIS/Aqua 250 m images collected between May 2003 and Ap ril 2006, and compare them with in situ observations
54 from the Tampa Bay water quality monitori ng program conducted by the Environmental Protection Commission of Hillsborough County's (EPCHC). 4.3. Methods and Materials 4.3.1. Field data Turbidity data (reported in nephelometric turbidity units or NTU) were obtained from the Environmental Protection Comm ission of Hillsborough County's (EPCHC) Tampa Bay water quality monitoring progr am (Bolder et al., 1991). The EPCHC conducts monthly surveys which span 3 weeks. Each week covers approximately one segment of the bay (Fig. 4-1). Water sample s were collected approximately about mid10depth between the surface and the bottom if the depth was greater than 3 m, otherwise only surface samples were collected. The EPCHC estimates turbidity of water samples in the lab with a Hach Model 2100N Turbidimeter. This device measures light intensity (peak spectral response is ~ 570 nm) at 3 different angles (90 scattered, forward scattered, and transmitted light). The normalized scattered light at 90 relative to the total scattered light at 3 angles was calibrated with four standard solutions ranging from 2 to 2000 NTU. This ratio mode was constantly run to remove the possible effects of colors on turbidity measurement (Hach Model 2100N Turbidimeter manual, 1999). Thus turbidity values generally re present the bulk scattering ca used by particles in water samples, primarily by suspended sediments. Th erefore, turbidity is frequently used as a good indicator of sediment con centration, although th e relationship between turbidity and sediment concentration can be highly vari able, and depends on se diment properties (e.g., size, shape, composition, and refractive i ndex) and measurement uncertainties. For example, when TSS concentration is low, relative uncertainty in the TSS concentration determined by the dry weight method can be larger (Christian & Sheng, 2003).
O T B H LTB AR M R LM H MT Figure 4-1. MODIS 250 m i m age of Tam p a Bay showing the f our segm ents of the bay and m a jor tr ibuta r ies. Th e segm ents are Old Tam p a Bay (OTB), Hillsbo rou gh Bay (HB), Middle Tam p a Bay (MTB), and Lower Tam p a Bay (LTB). The rivers are the Hillsboroug h River (HB), the Alaf ia River (A R), the Little M a natee Riv e r (LMR), and the Mana tee River (MR) The Enviro nm ental Pro t ection Comm ission of Hillsboroug h Countys (E PCHC) wate r quality m o nitoring st ations are overlaid with various sym bols to indicate d i fferent sampling tim e s in each m onth: diam ond (OTB) generally in the first week, triangle (HB) in the second week, and square (MTB and LTB) in the last week. Five stations m a rked with white crosses and labeled with numbers are also overlaid (stations 92, 23, 14, 40, 55 from the EPCHC program ). 55
56 Remote sensing reflectance ( R rs sr -1 ) data were collected in situ during several field surveys on 21-22 October 2003, 1-3 June 2004, 12-14 October 2004, 13 December 2005, and 27 February 2006, respectively, with an handheld PR650 (Photo Research Inc.) or Analytical Spectral Device (ASD Inc. ) spectroradiometer, following the method described in Hu et al. (2004). Briefly, upwelling radiance from the water surface and from a standard reflectance plaque (as a re ference), and the sky radiance were measured. R rs was derived by dividing the water spectra, co rrected for sky radiance reflected off the surface, by downwelling irradiance es timated from the plaque spectra. These in situ hyperspectral R rs data were then integrated over the relative spectral response (RSR) function of MODIS band 1 to obtain R rs (645) for MODIS validations. 4.3.2. Cross-calibration of the MODIS 250 m with 1 km bands Vicarious calibration of the satellite at-s ensor radiance is the procedure used to adjust the measured radiance to a pred icted radiance, based on well calibrated measurements and a radiative transfer model. It is critical for ocean color sensors to conduct periodic vicarious calibrations, becau se the pre-launch laboratory calibration typically has uncertainties of 2-5%, which can translate to 20-50% relative errors or worse in the retrieved R rs after atmospheric correction. Th e vicarious calibration and the subsequent atmospheric correc tion using consistent radiat ive transfer codes can be considered to be as a self-tuning process. Vicarious calibration of ocean color sens ors is typically performed at a welldefined site (i.e., spatially homogeneous water away from land) with a well-calibrated instrument (e.g., the Marine Optical Buoy or MOBY at Hawaii). When such an option is not available, an alternative is to calibrate one sensor against a nother well-calibrated sensor, for example calibrating MOS usi ng SeaWiFS (Wang & Franz, 2000) or calibrating Landsat/ETM+ using SeaWiFS (Hu et al., 2001b). Calibra tion against another calibrated satellite sensor is hereafter referred to as cross-calibration in this paper. Here I take the approach described in Hu et al (2001b) to calibrate the 250 m bands using MODIS/Aqua 1 km ocean bands. Because all MODIS bands have identical solar/viewing geometry, the procedure is simpler than when cross-calibrating two sensors are on
57 separate satellites as in Hu et al. (2001b). In short, the 250 m data are adjusted according to the at-sensor radiances predicted by the 1 km data. For clarity I briefly describe this cross-calibration procedure below. After removal of the ozone effect (H u et al., 2004), at-sensor radiance, L t can be predicted as (for brevity the waveleng th dependency is suppressed here): L t = L r + L a + t v L w (1) where definitions of these terms are given in Table 4-1. Here the effects of sunglint and whitecaps are omitted because during the quality control step, data affected by these artifacts can be discarded. The effects of water vapor, oxygen absorption, or light polarization are also negligible for the 250 m bands (e.g., Meister et al., 2005). The terms on the right-hand side of Eq. (1) were estimated from the MODIS 1 km ocean bands in the following way: (1) Eleven cloud-free MODIS images from May 2003 to April 2006 were randomly chosen. From each image, a clear-water ocean area adjacent to Tampa Bay was chosen as the calibratio n site (Fig. 4-2). (2) MODIS 1 km Level 1 data were proce ssed using SeaDAS 4.8 to estimate L r L a t v and L w for each of the 1 km wavelengths ( =412, 443, 488, 531, 551, 667, 674, 748, 859 nm), and further adjusted to the nominal center wavelengths of MODIS/Aqua similar to the procedure fo r SeaWiFS described by Hu et al.(2001b). (3) The above multispectral data were used to construct an artificial hyperspectral dataset covering from 400 to 900 nm by interpolation. L w (859) for the clear-water site was assumed 0, and L w (645) was approximated as 1.30 L w (667) according to the extensive in situ hyper-spectral data collected from the West Florida shelf (Cannizzaro et al., accepted). Sensitivity an alyses indicated that the results were not sensitive to using a constant (i.e. 1.30) if a substitute d value is in the range of 1.0 1.4 (results not shown here). (4) The simulated hyperspectral data (except the L w term) were integrated over the bandpass of the 250 m bands, while modulated by the relative spectral response (RSR) function, to provide the corresponding radiance for the 250 m bands:
L x (band) = dSdSLx)()()( (2) where S() is the RSR of the 250 m bands, x can be r or a, and band is 645 or 859 nm (see Table 4-1 for nomenclature). Table 4-1 Symbols, definitions and Units Symbols Definitions Units t 0 Diffuse transmittance from the sun to the ground due to Rayleigh and aerosol scattering Dimensionless t v Diffuse transmittance from the ground to a sensor due to Rayleigh and aerosol scattering Dimensionless t oz Diffuse transmittance from the sun to the ground and from the ground to a sensor due to absorption by ozone Dimensionless L a Radiance from aerosol scattering and aerosol-Rayleigh interaction mW cm -2 m -1 sr -1 L r Radiance from Rayleigh scattering in the absence of aerosols mW cm -2 m -1 sr -1 L t Total radiance at the top of atmosphere (TOA) mW cm -2 m -1 sr -1 L w Water-leaving radiance at the sea surface mW cm -2 m -1 sr -1 Lgn Normalized sun glint radiance if there were no atmosphere and solar irradiance F 0 =1 sr -1 0 Solar zenith angle Degree v Sensor zenith angle Degree F 0 Adjusted extraterrestrial solar irradiance mW cm -2 m -1 For simplicity, the wavelength dependency of all terms (except solar and sensor angles) is suppressed. 58
(b) (a) Figure 4-2. (a) MODIS at-senso r rad i ance at 667 nm (1 km ) on 13 December 2004, 18 :43 GMT over the study area; (b) MODIS at-senso r radiance at 645 nm (250 m ) from the sam e satellite pass. Data in the rectan gular box w e re used for cross-calibration. 4.3.3. Satellite im age process i n g M ODI S L1 B di r e c t br oa dc a s t da t a we r e captured in real-tim e by an X-band antenna loca ted at the U n ivers ity of South Fl orid a (USF) in Saint Pe tersb u rg, Florid a. For regions without a local antenna, historical data can be obtained from the NASA Goddard Space Flight Center's Distribute d Active Archive Center (DAAC). Clouds were m a sked according to the TOA reflectance (sr -1 ) at 859 nm 0 0 cos F L R t t (3) with a threshold of 0.018 sr -1 This step filters out m o st of the clouds and severe sun glint contam inations. Then, an atm o spheric correc tion step was carried out using the m e thods outlin ed in Hu et al. (20 04), and su mmarized as the followin g equation. 59
R rs (645)= 00000cos)645()645()645()645()859()645())859()859()859(()645()645((FttLFFLtLtLvrroztozt (4) Both single scattering approximation and multi-scattering exact calculations (the latter used with the cross-calibrated at-sensor radiance) were used to estimate L r The latter method was used to ensure consistency with the cross-calibrated data. The details of these two methods can be found in Hu et al. (2004). 4.3.4. Satellite in situ comparison Comparison between satellite and in situ observations is complicated in coastal and estuarine waters, where high spatial and temporal variations in both water and atmosphere properties make it difficult to have observations congruent in space and time. Bailey & Werdell (2006) made several key recommendations for such comparisons. Here we used the following criteria to select the satellite and in situ matching pairs: First, the satellite data must be collected within 2 hours of the in situ measurements. For comparison of remote sensing reflectance, the time window was relaxed to 4 hours to increase the matching points. Second, to ensure relative homogeneity and to remove sensor noise (Hu et al., 2001a), a 33 pixel box centered at the in situ measurement site was used. Within this box, there must be at least 4 valid (see below) pixels, and the coefficient of variance (standard deviation divided by mean) of the valid pixels must be <0.4. Then, the median value of valid pixels in the 33 box is compared with the in situ data. By trial and error, any pixel that meets one of the following criteria is considered invalid: 1) water depth <2.8 m (bottom contamination); 2) satellite zenith angle >50 (scan edge); 3) (R t (859) R r (859))>0.019 sr -1 (large aerosol); 4) normalized sun glint radiance (L gn ) > 0.0001 sr -1 (sun glint contamination, Wang & Bailey, 2001). 60
61 4.4. Results 4.4.1. Cross calibration and data validation Fig. 4-3 shows the comparison between the ozone-corrected at-s ensor radiances in the 250 m bands and those predicted from the 1 km bands, with sta tistics presented in Table 4-2. Both MODIS 250 m bands show ex cellent linear rela tionships with the estimates derived from the 1 km bands, with median ratios (indicati ng overall biases) of 0.96 and 1.01 for the 645 nm and 859 nm bands, respectively. The median absolute percentage differences (MPD, Bailey & Werd ell, 2006, a measure of uncertainty) were 3.61% and 1.30%, respectively. These uncertain ties correspond to ~2 counts for the 645 nm band (L t (645) is normally > 60 counts with TSS ranging between 2-15 mg L -1 in Tampa Bay) and <1 count for the 859-nm ba nd. From the slopes and intercepts of the linear regressions, the vicarious gains and o ffsets for the two 250 m bands are [1.0904, 0.067] and [1.0303, -0.0124], respectively. The at-sensor radiances were adjusted by the above vicarious gains and offsets, and the rigorous atmospheric correction (mu lti-scattering) was applied to process the MODIS imagery for those dates when concurrent in situ R rs data were collected. Note that the 11 cloud-free scenes we re randomly chosen between May 2003 and April 2006 and therefore these days with in situ R rs measurements were different than those used in the cross-calibration. The derived R rs (645) is shown in Fig. 4-4 to compare with in situ R rs (645), with statistics listed in Table 4-3. To show how much improvement may be gained from this extra effort (cross-ca libration and rigorous at mospheric correction), R rs (645) derived from the simple atmospheri c correction and the original at-sensor radiance (i.e., without cross-ca libration) is also shown in Fig.4-4 and Table 4-3. The improvement in terms of bias (5% versus -7 %) or other measures is not significant (paired Students t-test, p<0.001). Therefore, the pre-launch calibrati on (i.e., original atsensor radiance) and the singl e-scattering atmospheric correc tion were deemed sufficient to process the entire time -series of MODIS imagery be tween May 2003 and April 2006.
62 Table 4-2 Comparison between the m easured at-sensor radiances (mW cm -2 m -1 sr -1 ) at the MODIS 250 m bands and those predicted from the 1 km ocean color bands. Band Ratio a % Difference b Slope Intercept r 2 RMSE c N L t (645) 0.96 3.61 0.91 0.07 0.997 0.04 11 L t (859) 1.01 1.30 0.97 0.01 0.995 0.02 11 a Median ratio between the measured and pr edicted radiances, indi cating overall biases. b Median absolute percentage difference (MPD) between the measured and predicted atsensor radiance, indicating typical uncertainties (B ailey and Werdell, 2006). c RMSE represents root mean square errors of the linear regression fitting, in units of mW cm -2 m -1 sr -1 Table 4-3 Comparison between in situ versus MODIS remote sensing reflectance at 645 nm (R rs (645), sr -1 ), with the latter derived using 1) single scattering approximation and the original at-sensor radiance (i.e., pre-launch calibration) and 2) multi-scattering method and the cross-calibrated at -sensor radiance, respectively. Calibration Ratio % Difference Slope Intercept r 2 RMSE N Pre-launch 0.93 8 0.95 0.00 0.97 0.0007 15 Cross-calibration 1.05 10 1.04 0.00 0.98 0.0007 15 The ratio, % difference, and RMSE are defined in the same way as in Table 4-2.
0.200.300.400.50Predicted Lt(859) 0.200.300.400.50Measured Lt (859) 1.001.201.401.601.80Measured Lt (645) 1.001.201.401.601.80Predicted Lt(645) Figure 4-3. Scatter plots of the at-sensor radiance (mW cm -2 m -1 sr -1 ) in the two 250 m bands (645 nm in (a) and 859 nm in (b)) as measured by the sensor and predicted by the calibrated 1 km ocean color bands. The dashed lines are 1:1 lines. (Results of the comparison are shown in Table 4-2). 4.4.2. Turbidity from satellite R rs After rigorous quality control of the satellite data (see Methods), a total of 43 matching pairs were found for concurrent satellite and in situ measurements. Fig. 4-5 shows that MODIS R rs (645) is closely related to in situ turbidity for values ranging from 0.9 to 8.0 NTU and R rs (645) from 0.001 to 0.008 sr -1 respectively (r 2 =0.73, n=43). Further, the relationship appeared to be stable over time, specifically for 2004 and 2005 (Fig. 4-5), indicating that a time-independent regression relationship could be obtained. Indeed, if the relationship derived from the 2005 data is used to predict the 2004 turbidity, the results agree well with the in situ turbidity values (median predicted versus in situ ratio of ~ 0.98 and median absolute percentage difference/MPD of 10%; Fig. 4-6). Hence, the regression relationship of Fig. 4-5 and the rigorous data quality control criteria were applied to the MODIS R rs (645) imagery to obtain the turbidity time-series maps. 63
64 4.4.3. Image series of turbidity The 3-year (May 2003 to April 2006) MODIS tu rbidity maps were used to generate monthly and climatological monthly means. The climatological monthly means showed distinctive spatial and temporal patterns in Tampa Bay (Figs. 4-7). Turbidity in Hillsborough Bay (HB) was consistently higher than in other sub-regions except in August and September; at this time, however, turbidity in HB may be underestimated due to high input of colored dissolved organi c matter (CDOM) from major rivers (see below). Turbidity in the upper Middle Tamp a Bay (MTB, around Station 14) was lower relative to that in HB and OTB throughout the year. In contrast, both Old Tampa Bay (OTB) and Lower Tampa Bay (LTB) showed cl ear seasonal variations LTB, particularly near the bay mouth, showed high turbidity between November-March and lower turbidity between May-October, while opposite seasonality existed in OTB. Bay-wide high turbidity was found in April. Such seasonality can also be visualized by extracting timeseries data from several selected stations from the various sub-regions (Fig. 4-8). For example, the upper bay (Sta. 55 in HB and Sta. 40 in OTB; see Fig. 4-1 for location) typically showed higher turbidity than MT B and LTB, particularly in the wet season (from May to October, turbidity > 3.0 NTU). MTB (Sta. 14) showed the lowest turbidity and seasonality (typically ~2.0 NTU). LTB s howed moderate turbidity but the largest seasonal variations. Turbidity also showed significant interannual variations. For example, Fig.4-9 shows the contrast between the turbidity patterns from April of 2004, 2005, and 2006. Such interannual variations can be attributed to different wi nd forcing, as shown in Fig.410. We may assume that 6.0 m s -1 wind speed is a threshold above which sediment resuspension occurs. Then, there are 7, 8, and 2 days for April of 2004, 2005, and 2006, respectively, which can explain the observed spatial patterns variations.
Single-scattering and pre-launch calibrationMultiple-scattering and cross calibration MODIS Rrs(645) (sr-1) 0.0000.0050.0100.0150.020 0.0000.0050.0100.0150.020In situ Rrs(645) (sr-1) Figure 4-4. Remote sensing reflectance in Tampa Bay at band 1 (R rs (645), sr -1 ) from in situ measurements and MODIS estimates derived using 1) multiple-scattering atmospheric correction of the cross-calibrated at-sensor radiance (filled circles); and 2) single-scattering atmospheric correction of the original (i.e., pre-launch calibration) at-sensor radiance (open circles). The dashed and dotted line represents the 1:1 ratio (Comparison results listed in Table 4-3). 2004 fitted line2004 data2005 data2005 fitted line2004+2005 data Turbidity=1203.9xRrs(645)1.087R2=0.73, n=430.0080.003 110 In Situ turbidity (NTU) 0.001MODIS Rrs(645) (sr-1) Figure 4-5. Relationship between in situ turbidity (NTU) and MODIS remote sensing reflectance at band 1 (R rs (645), sr -1 ) in a log-log scale. The 2004 and 2005 data are represented with open and filled circles, respectively. The fitted lines from 2004, 2005, and 2004+2005 data are represented with dashed line, thin solid line, and thick solid line, respectively. 65
Slope=0.99Intercept=0.2R2=0.83n=13Median Ratio=0.98MPD=10% 0.02.04.06.0 In Situ Turbidity (NTU) 0.02.04.06.0MODIS Turbidiity (NTU) 66elationship between in situ turbidity from the 2004 m Figure 4-6. Reasurements and that estimated from concurrent MODIS data using the relationship derived from the 2005 data. The dashed line is the 1:1 line. Fig. 4-9 shows an example of the contrast between the monthly turbidity maps derived from satellite observations and from in situ data. While the monthly MODIS turbidity estimates were generally consistent with those determined from in situ observations (Fig. 4-9), there were some important differences. Some fine-scale features are different due to either the different temporal and spatial sampling frequencies or the artificial effects from the interpolation method for the in situ data. In addition, the single monthly in situ turbidity observations are frequently within 1 standard deviation of the multiple satellite estimates collected during the same month (Fig. 4-11). However, MODIS mean turbidity showed less short-term temporal variability, but much more pronounced seasonal and interannual variability than in situ observations. For example, abnormally high in situ turbidity values (> 10.0 NTU) were observed at Sta. 55 in October and December 2003, respectively, but the monthly MODIS means showed low turbidity, comparable to that seen in other months. The high turbidity events sampled in situ through single monthly collections unfortunately masked the relative increases in turbidity observed in the wet season, thus obscuring seasonal changes (Fig. 4-8). Similarly, the high in situ turbidity observed at Sta. 23 in late 2003 and at Sta. 92 in early 2004 obscured the higher overall turbidity inferred from MODIS data for late 2004 and early 2005 (Fig. 4-11).
67 These differences between these data sets are likely due, in great part, to the mismatch in sampling frequencies between the in situ program and the satellite collections. MODIS typically has >4 high-qua lity observations per month (about once per week), and even >10 observations were coll ected frequently betw een October and March (Fig. 4-11). This helps minimize the aliasing of temporal variations and helps construct more reliable monthly means than using the single in situ monthly measurements. I also find that the satellite observations provide higher detail of the spatial heterogeneity in turbidity patterns. Because of these increased temporal and spatial sampling frequency, MODIS provides a useful synoptic produc t that helps improve assess ments of the monthly mean turbidity. Clearly, MODIS obser vations require cloudless skie s and minimal sun glint. These factors limit the number of MODIS obser vations in the wet season. For example, in the upper bay (e.g., Sta. 55) I had no valid observations during some months. 4.5. Discussion 4.5.1. Calibration and algorithm issues: The results I obtained using the complex cross-calibration scheme and the exact atmospheric correction Ire similar to those obtained with the simp ler, single-scattering atmospheric correction of data calibrated with pre-launch coefficients. On the other hand, there may be some small calibration uncertain ties in the 1 km MODIS ocean band data that can erroneously propagate to the 250 m bands through the cross-calibration. This may be the cause for the residual errors observed in the retrieved R rs data obtained from this rigorous approach. Overall, I find that MODIS 250 m imagery can be used with the pre-launch calibration and simple atmospheric correction to assess long-term trends in water turbidity. Other error s ources are discussed below.
Figure 4-7. Climatological (May 2003 to April 2006) monthly means of turbidity derived from MODIS 250 m data. White color inside Tampa Bay (not offshore) represents the mask of shallow water (bottom depth < 2.8 m) and grey color represents land. 68
Figure 4-7. (Continued) 69
Sta 40Sta 92Sta 23Sta 55Sta 14 1.02.03.04.05.0Turbidity (NTU) 1234567Month of a Year89101112 Figure 4-8. The climatological monthly means of turbidity derived from MODIS between May 2003 and April 2006 at several stations in Tampa Bay (see Fig.4-1 for station locations). Sta. 55 was affected by high CDOM and cloud cover during later summer; consequently, turbidity was possibly underestimated and there was no valid data in August. 70
Figure 4-9. Monthly turbidity images from MODIS 250 m data (top panels) and from in situ measurements (lower panels) for April of 2004, 2005, and 2006 (no in situ measurement was available for April 2006). Crosses represent sample stations with bottom depths > 2.8 m, and the values shown are in situ turbidity measurements. The in situ monthly map is a graphical composite of single observations taken at different times in different parts of Tampa Bay and is not a true monthly mean. The satellite-derived mean is based on multiple MODIS observations during the month. 71
0.03.06.09.012.0051015202530Day of the Month (April)Wind Speed (m s-1) 2004 2005 2006 Figure 4-10. Daily averaged wind speed measured at Port of Manatee (27.64, -82.56) in April 2004, 2005, and 2006 (data courtesy of NOAA). The dashed line represents 6 m s -1 I use a white aerosol assumption, which considers aerosol reflectance at 645 nm equal to that at 859 nm. Yet this assumption is simplistic and it works because these bands are close spectrally, making extrapolation from 859 to 645 nm feasible even if the aerosol is not white. Indeed, for non white-aerosols the errors can be several counts (e.g., Hu et al., 2001b), and of the order of 5-10% in the retrieved R rs (645). However, the black-pixel assumption (L w (859) =0) may not be accurate in very turbid waters (Hu et al. 2000; Siegel et al., 2000; Wang & Shi, 2005). In this case an alternative may be the nearest neighboring approach to apply nearby aerosol reflectance from clear waters to the turbid pixels (e.g., Hu et al., 2000; Miller & Mckee, 2004). In coastal areas where aerosol reflectance is often spatially heterogeneous (Hu et al., 2004), this technique is thus limiting. Using longer wavelengths may circumvent the non black-pixel problem (Wang & Shi, 2005), but the larger spectral distance between the wavelength of interest (645 nm) and the reference wavelength (1640 nm) may yield some errors in the aerosol reflectance extrapolation. Therefore, at present there is no perfect atmospheric correction method for MODIS 250 m bands in coastal waters. However, future sensors may add one 72
more band in the short wave infrared (SWIR) wavelength, in addition to the present 1640 nm MODIS band. This will help determine aerosol type as well as aerosol reflectance in 73 Figure 4-11. Time-series of monthly turbidity estimates derived from in situ and MODIS measurements at selected stations shown in Fig 4-1. The period covered is May 2003 to December 2005 (no in situ data available for 2006). The number of MODIS observations during each month is shown on the right-hand side. Where number of samples was >1, the standard deviation of the monthly mean is also shown on the MODIS data. There is only one in situ observation in each month. The circles highlight the relatively high turbidity observed from late 2004 to early 2005 in the Lower Tampa Bay, which are not observed with the in situ measurements. The y-axis scale for station 55 is different from that in the other panels. 5 7 9 11 1 3 5 7 9 11 1 3 5 7 9 11 0.0 3.0 6.0 9.0 12.0 MonthSta. 92 0 3 6 9 12 0.0 3.0 6.0 9.0 12.0 Sta. 23 0 3 6 9 12 0.0 3.0 6.0 9.0 12.0 Turbidity (NTU)Sta. 14 0 3 6 9 12 Number of MODIS observations 0.0 3.0 6.0 9.0 12.0 Sta. 40 0 3 6 9 12 Number of MODIS observations 0.0 5.0 10.0 15.0 20.0 Sta. 55MODIS TurbidityIn Situ Turbidity 0 3 6 9 12 2003 2 0 0 5 2 0 0 4
74 the SWIR region, making extrapolation of the atmospheric correction of shorter wavelengths more accurate. Hu et al. (2004) showed differences be tween MODIS/Terra a nd MODIS/Aqua data and inconsistency in the time-dependent R rs retrievals, especially for the 500 m bands (469 and 555 nm). The more extensive MODIS and in situ data set examined in the present turbidity study allowed for a rigorous data quality cont rol. These data showed that using a single sensor, specifically MODI S/Aqua 250 m data and rigorous data quality control, helps improve accuracy and time-i ndependent consistency. Indeed, when only the 2-hour time window and homogeneity ch eck (see the Methods) were allowed for match-ups between MODIS and in situ data, 751 pairs were found. After several qualitycontrol criteria (bathymetry, Rayleigh corrected R t (859), sensor viewing angles and normalized sun glint radiance) were applied, on ly 43 matching pairs were left and used to develop the R rs (645) => turbidity algorithm. The th reshold values were found by trialand-error, and therefore may need to be adjusted if the same approach is applied to other estuaries. 4.5.2. Effects of colored dissolv ed organic matter (CDOM) The apparent low turbidity in the wet season (e.g, August and September) in HB is likely an artifact due to the interference by CDOM, whose ab sorption coefficient at 645 nm (a cdom (645)) could be ~ 0.40 m -1 in the wet season (Chen et al., in press). This is greater than the water absorpti on at this wavelength (~0.33 m -1 Pope & Fry, 1997). Because R rs (645) for most waters is not only a func tion of turbidity (a proxy for TSS) but also inversely proportional to total absorption coeffi cient, presence of high and variable CDOM will make R rs (645) smaller, leading to low, unrea listic turbidity re trievals. Similar effects have also been observed elsewher e (e.g., Woodruff et al ., 1998). This effect, however, is limited to HB in the wet season only and generally neg ligible for the other bay segments and seasons when a cdom (645) is typically < 0.03 m -1 Indeed, MODIS turbidity showed comparable variation patterns to those from in situ measurements in most time (Fig. 4-11).
4.6. Conclusions Two issues towards operational application of MODIS 250 m data for monitoring of estuarine turbidity were addressed: calibration of the at-sensor radiance and methods to convert such radiance to turbidity. Cross-calibration of the MODIS 250 m bands with the 1 km ocean color bands, and use of a rigorous, multiple-scattering atmospheric correction led to no substantial improvements over using the pre-launch calibration and a simple, single-scattering atmospheric correction approach in terms of the retrieved surface remote sensing reflectance (R rs ) for a large dynamic range. Further, the retrieved MODIS R rs after rigorous quality control, showed a high correlation with concurrent in situ turbidity observations. Therefore, the simple approach presented here seems adequate to generate accurate and consistent time-series of turbidity maps over estuaries of moderate size, such as Tampa Bay. Although similar features were observed both in the MODIS turbidity maps and those based on in situ observations, of particular importance are their differences. MODIS provides multiple cloud-free measurements at high spatial resolution, and these provide an improved record of the mean state of the Bay each month, as well as of seasonal and inter-annual variability. The distinctive spatial and temporal viability revealed from MODIS imagery can be explained well by wind-driven bottom-resuspension events during the winter and river inputs during the rainy season. There still remain several issues to be addressed in the future, for example the interference of high CDOM and shallow (< 2.8 m) but bright bottom in the conversion from R rs to turbidity. However, because of the relative ease, robustness, low cost, and simple methods for data processing, as well as the repeated global coverage, MODIS has important advantages over traditional in situ sampling methods. Thus we strongly recommend implementing a similar scheme for all estuaries of moderate size and coastal regions to help better manage estuarine and coastal resources. 75
76 Chapter 5. Remote Sensing of water clarity in Tampa Bay 5.1. Abstract Understanding the distribution and variabil ity of water clarity in Tampa Bay is critical for ecosystem restoration and protec tion efforts focused on this major estuary. Traditionally, water clarity in Tampa Bay has been measured with a Secchi disk (Secchi Disk Depth or SDD) during monthly surveys at established stations. Here I estimate SDD and the light attenuation coefficient at 490 nm, K d (490) (m -1 ), from Sea-viewing Wide Field-of-View Sensor (SeaWiFS) imagery us ing a new semi-analytical light attenuation algorithm (SDD=1.04K d (490) -0.82 ). K d (490) derived with the traditional band-ratio algorithm showed no signifi cant relationship with in situ SDD (r 2 =0.14, n=80), but the new estimates were closely correlated with in situ SDD (0.9
77 5.2. Introduction Tampa Bay is the largest open-water estuary in Florida, featuring a surface area of ~1,000 km 2 It has been traditionally divided in to 4 sub-segments, namely: Old Tampa Bay (OTB), Hillsborough Bay (HB), Middle Tampa Bay (MTB), and Lower Tampa Bay (LTM) (Fig. 5-1). Contributing more than $5 billion annually from trade, tourism, development, and fishing (FDCA, 1996), Tamp a Bay is a diverse and productive natural system that provides a vital habi tat for crustaceans, fish, shellfish and a variety of marine mammals, reptiles and birds (H arwell et al., 1995). It is therefore critical that the development of the Bay be conducted in an environmentally sound way to sustain a clean and healthy system. Tampa Bay has undergone substantial anthropogenic alterations. In the decades prior to the 1980s, Tampa Bay was heavily po lluted by nutrient load ings from sources like sewage and wastewater. The decline in wa ter quality due to pollution led to severe eutrophication and increases in light attenuation, therefore ca using substantial losses of seagrass coverage (Lewis et al., 1998; To masko et al., 2005). Since then, significant ecosystem restoration efforts have been under the way. In 1990, Tampa Bay National Estuary Program (TBNEP) was established to in tegrate efforts to restore and protect the Bay. In 1996, TBNEP developed a Comprehens ive Conservation and Management Plan (CCMP), which focused on restoration of seagra ss to levels similar to those observed in the 1950s (Janicki & Wade, 1996). These effort s gradually improved the water quality of the Bay, and some of the seagrass has recovered (Johansson, 2000; Tomasko et al., 2005). Water clarity is an index of water qual ity, taken to reflect th e impact of nutrient load on phytoplankton concentration. It is also a critical factor controlling seagrass extent in coastal and estuarine wa ters (Gallegos, 2001; Kirk, 1994). It therefore has been selected as a key parameter that is rou tinely monitored in Tampa Bay, and ultimately serves to direct riverine nutrient input management plans (Janicki et al., 2001). Water clarity in Tampa Bay is measured with a Secchi disk once a month at established stations (Janicki & Wade, 1996; Fig. 5-1), similar as in most coastal environmental monitoring programs. A white or white-black disk (usually ~20 cm in diameter) is lowered into the wa ter, and the depth at which th e disk is no longer visible is
78 recorded as the Secchi Disk Depth (SDD, m). SDD provides an inexpensive index of the rate at which light is attenua ted with depth. Due to the si ze of Tampa Bay and logistic support, it usually takes three weeks to co llect the field SDD obs ervations across the estuary (see Fig. 5-1). While these observati ons were used to characterize the monthly conditions of the Bay, clearly the sampling program is not synoptic and also inadequate to reflect the mean status of water clarity in a fast changing estuarine system like Tampa Bay (Chen et al., submitted, and Chapter 2). Satellite remote sensing has been used to derive light attenuation coefficients in coastal and open waters using site-specifi c (and frequently time-specific) empirical algorithms (Austin & Petzold, 1981; Muller, 2000; Prasad et al., 1998; Stumpf & Pennock, 1991). Further, the standard empirical algorithm used to estimate the diffuse light attenuation coefficient at 490 nm ( K d (490), m -1 ) from Sea-viewing Wide Field-ofView Sensor (SeaWiFS) satellite data was developed based largely on open ocean observations where most K d (490) < 0.15 m -1 (Muller, 2000). Therefore, these historical empirical methods are prone to generate large errors in coastal and estuarine waters (Lee et al., 2005b). More recently, improved light attenuation coefficient estimates for coastal waters have been possible by application of a semi-analytical algorithm (Lee et al., 2005a; Lee et al., 2005b). However, the previous al gorithm development and evaluation relied entirely on in situ measurements of above-water remote sensing reflectance ( R rs ), and it remains unknown that how well the semi-analyti cal model retrieves the light attenuation from satellite observations because satellit e derived remote sensing reflectance often contains uncertainties in coas tal and estuarine waters (Harding et al., 2005; Hu et al., 2000). Here I evaluate the semi-analytical me thod of Lee et al. (2005a) to estimate K d (490) using in situ SDD measurements collected in Tampa Bay by the EPCHC. I then estimate SDD using SeaWiFS imagery between September 1997 and December 2005 and characterize the temporal and spatial variability of water clarity in Tampa Bay. These analyses are compared to patterns inferred from the in situ surveys. Finally, some recommendations are provided for improving the water quality m onitoring program of Tampa Bay and similar programs.
OTB HB LTB AR MR LMR HR MTB Figure 5-1. A Sea W iFS quasi-t rue-color RGB im age of Ta mpa Bay showing the four sub-segm ents, nam e ly, Old Tam p a Bay (O TB), Hillsbo rough Bay (HB), Middle Tam p a Bay (MTB), and Lower Tam p a Bay (LTB). Major rivers sho w n are Hills borough River (HR), Alafia River (AR), Little Manat ee River (L MR), and Manatee River (MR). Environm ental Pro t ectio n Comm ission of Hillsborough Countys (EPC HC) water quality monitoring stations are shown using differe nt symbols to indicate m onthly sam p ling tim es: diamond (OTB) is generally sam p led in th e first week of the m onth, triangle (HB) generally in the second week, and squares (MTB and LTB) in the third or fourth weeks. Four stations (92, 23, 14, and 40, crosses) wher e tim e-series data were extracted for this study are also shown. The inset represents the relative position of Ta m p a Bay estuary in the state of Florida. 79
80 5.3. Materials and methods 5.3.1. Satellite data SeaWiFS merged local area coverage (M LAC) Level-1A data were downloaded from the NASA Goddard Space Flight Center (GSFC, http://oceancolor.gsfc.nasa.gov/ ) and processed using the SeaWiFS Data Anal ysis System software (SeaDAS, Version 4.9). Two methods were used to estimate K d (490). One was the empirical band-ratio algorithm (the default algorithm in SeaDAS; http://oceancolor.gsfc.nasa.gov/REPROCESSING/SeaWiFS/R5.1 /k490_update.html updated Muellers (2000) algorithm). The second used the semi-analytical algorithm of Lee et al. (2005a). The standard method is an empirical algorithm based on the relationship between K d (490) and the blue-to-green rati o of water leaving radiance, L w (Mueller, 2000). The second method is a semi-analytical approach, which first derives absorption and backscattering coefficien ts from remote sensing reflectance, R rs and then uses these coefficients to estimate K d (Lee et al., 2005a). Standard SeaWiFS processing assessments for atmospheric correction failure, land, clouds, sun glint, and large solar/sensor angle were used to mask invalid pixels. Flags for stray light, shallow water, negative water-leaving radiance, and turbid case 2 water were disabled to increase the data coverage in time and space. 5.3.2. Field and ancillary data In situ K d (490) observations were not available, so in situ SDD was used instead as a surrogate of water clarity. Previous studies ha ve shown that SDD is empirically related to K d at certain wavelengths (Giesen et al. 1990; Jean-Franc & Giuseppe, 2004; Kirk, 1994; Kratzer et al., 2003 ). SDD was obtained from the Environmental Protection Commission of Hillsborough County's (EPCHC) Tampa Bay water quality monitoring program (Boler et al., 1991). Chlorophyll concentrations, turb idity (reported in nephelometric turbidity units or NTU), and color (Ptunits) were also collected along with SDD observations. Daily averaged river flow rates of the Al afia River and the Hillsborough River, the two largest tributaries discharging fresh wa ter into Tampa Bay, were obtained from the United State Geological Survey National Water Information System (USGS NWIS)
81 between 1997 and 2005. Monthly and climatologi cal monthly means of river flow were derived. Wind data were obtained from one of the National Oceanic Atmospheric Administration (NOAA) Tampa Bay physical oceanographic real-time (PORT) stations at Saint Petersburg (27.6'N, 82.6'W) from 1997 to 2005. Wind data were binned into daily means of wind speed and the numbe r of days when daily wind speed was > 4.0 m s -1 was calculated with the intent of understanding wind-induced sediment resuspension events. The cutoff wind speed of 4.0 m s -1 was chosen based on Chen et al.s (submitted) observations relating wind speed and possible sediment resusupension events in Tampa Bay. 5.3.3. Satellitein situ comparison A narrow window of 2 hours between in situ SDD measurements and the SeaWiFS overpass was used to find matching in situ and satellite observations. A median value from a 3 3 pixel box centered at the measurement site was used to filter out sensor and algorithm noise (Hu et al., 2001). To ensure spatial homogeneity in the satellite data, the median value was used only when the number of the valid pixels within the 3x3 box was >4 and the coefficient of variation (CV) of the valid pixels was <0.4. To remove pixels where the bottom contaminated satelli te radiance observations pixels with water depths <2 m ( Gesch & Wilson, 2001) were masked and excluded from the match-up comparisons. 5.4. Results 5.4.1. Comparison between in situ SDD and satellite K d (490) A total of 80 matching pairs were found after applying the various data quality restrictions described above Old Tampa Bay (OTB) had seven matching station pairs, Middle Tampa Bay (MTB) had eight, and Lowe r Tampa Bay (LTB) had five stations. Only one match-up station was found for Hills borough Bay (HB) due to the small area of that segment of the Bay, where there are several islands. The matching pairs were
82 relatively evenly distributed in time across four seasons (24 in spring, 20 in summer, 9 in fall, and 27 in winter). The statistical relationships between in situ SDD and K d (490) derived empirically (EA) and semi-analytically (SA) were significantly different (Fig. 5-2). K d EA (490) showed no significant relationship with in situ SDD (SDD=0.99K d EA (490) -068 r 2 =0.14, n=80), while K d SA (490) showed a high correlation with in situ SDD (SDD=1.04K d SA (490) -0.84 0.9
months (see below). Due to these artifacts and insufficient validating point (e.g., only one station available in HB) for the relationship shown in Fig 5-2, we omitted HB from subsequent analyses. 1.02.01.010.00.50.1SDD=1.04xKd(490)(-0.82) r2=0.67,n=80 SeaWiFS Kd(490) (m-1) In Situ Secchi Disk Depth (m) Points from SAPoints from EA Figure 5-2. In situ secchi disk depth (SDD, m) versus light attenuation coefficient at 490 nm (K d (490), m -1 ) derived from SeaWiFS using (1) an empirical band-ratio algorithm (EA) (triangles, updated Muellers (2000) algorithm) and (2) a semi-analytical algorithm(SA) (circles, Lee et al., 2005a). The spatial and seasonal variations revealed in the images can be better visualized by extracting time-series data at selected stations (Fig.5-4). In general, chlorophyll was highest in the upper Bay and decreased systematically toward the Lower Bay, while turbidity was lowest in the intermediate reaches of the Bay and higher in both the upper and lower Bay. Yet, SDD was relatively uniform across the Bay in November-March, but increased from the upper (Sta. 40 and Sta. 14) to lower bay portions (Sta. 23 and Sta. 92) in late spring (May to June). These patterns are visible both in the satellite and in situ data (e.g., within 1 standard deviation; Fig. 5-4). 83
Figure 5-3. Climatological monthly composites (September 1997 December 2005) of SeaWiFS SDD (m). Color legend also shows the corresponding K d (490) values (m -1 ). White color within Tampa Bay represents shallow water (bottom depth < 2.0 m). Grey color represents land. 84
Figure 5-3. (Continued) 85
86 The seasonal patterns are cons istent with the time series of chlorophyll, color, and turbidity measurements (Fig. 5-5). The largest SDD coincided with the lowest chlorophyll concentration, color, and turbid ity in May or June. Then, SDD gradually decreased with increasing ch lorophyll and color while turb idity was slightly lower, suggesting that during the summer (correspond ing to the rainy season from mid-June to mid-September; Fig.5-6) SDD was primarily controlled by phytoplankton concentration. Chlorophyll reached maxima between July and October depending on location, and SDD showed lower values during th is time, corresponding to a wi nter turbidity maximum. In the Middle and Lower Bay, minimal chlorophy ll concentrations o ccurred starting in January through late spring around May, but turbidity did not reach minima until May, suggesting that SDD in the spring is controlle d by factors such as sediment resuspension events. The seasonality in water clarity in differe nt parts of the Bay can therefore be explained by variations of river runoff and wind-driven sediment resuspension events (Fig. 5-6). Larger river runo ff in August-October (Fig 56a) delivers a higher nutrient flux into the Bay, contributing to phytoplankt on growth. In comparison, stronger winds in March-April (Fig. 5-6b) lead to higher turbidity due to sediment resuspension. SDD showed significant inte rannual variation. Fig. 5-7 shows the time series of monthly mean SDD extracted at several stat ions from the various bay segments. This interannual variations are bett er observed in SDD anomalies (the difference between the monthly means and the climatological monthl y means; Fig. 5-8). Relatively lower SDD occurred in winter-spring of 1998 and su mmer-fall of 2001, 2003, and 2004 than in other years for all four stations. For stations 23 and 92 (middle to lower bay), SDD appeared higher between late 2002 and ear ly 2003 and in late 2005. In part icular, at station 92 there appeared to be a trend toward higher SDD after 2002. A Students t-test indicated that mean SDD (2.80 m) from July 2002 to December 2005 is significantly larger than that (2.42 m) between January 1999 and June 2002 (p<0.05, n=42, data before 1999 were not used due to El Nio effects on water quality in 1998, see below). Whether or not this is an indication of sustained improved water quali ty needs to be further investigated by extending the series beyond 2005.
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0.0 2.0 4.0 6.0 Months of a YearSta. 92 SeaWiFS 20 40 60 70 Number of observation days 0.0 2.0 4.0 6.0 SDD (m) Sta. 23 In Situ 10 20 30 40 0.0 2.0 4.0 6.0 Sta. 14 10 20 30 40 number of observation days 0.0 2.0 4.0 6.0 Sta. 40 10 20 30 40 Figure 5-4. Climatological monthly means of SeaWiFS (filled circles) and in situ (open circles) secchi disk depth (SDD, m) at several stations within Tampa Bay (Fig. 5-1). The number of SeaWiFS observation days during each month is shown on the right-hand side, note that the y-axis scale for Sta. 92 is different from those for other stations and there are only 8 in situ observations in each month. The vertical error bars indicate the standard deviation of satellite or in situ observations. No station from HB is shown here due to the uncertainties and insufficient points for the relationship shown in Fig. 5-2. 87
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0.0 3.0 6.0 9.0 12.0 Months of a YearSta. 92 0.0 5.0 10.0 15.0 In Situ Chl (mg m3)Sta. 23 0.0 5.0 10.0 15.0 20.0 Sta. 14 0.0 10.0 20.0 30.0 40.0 50.0 Sta. 40 Figure 5-5. Climatological monthly means of (A) in situ chlorophyll (mg m 3 ), (B) color (Pt-unit), and (C) turbidity (nephelometric turbidity units, NTU) measurements from the EPCHC monitoring program at several stations from the various bay-segments (see Fig. 5-1 for the station locations) collected between September 1997 and December 2005. The horizontal lines indicate the means of all measurements during 8 years at each station 88
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0.0 2.0 4.0 6.0 8.0 10.0 Months of a YearSta. 92 0.0 5.0 10.0 15.0 20.0 In Situ Color (Pt Unit)Sta. 23 0.0 5.0 10.0 15.0 20.0 25.0 Sta. 14 0.0 5.0 10.0 15.0 20.0 Sta. 40 Figure 5-5. (Continued) 89
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 0.0 3.0 6.0 9.0 12.0 Months of a YearSta. 92 0.0 2.0 4.0 6.0 In Situ Turbidity (NTU)Sta. 23 0.0 2.0 4.0 6.0 Sta. 14 .0 0.0 2.0 4.0 6.0 8.0 10.0 Sta. 40 Figure 5-5. (Continued) 90
91 The observed interannual SDD variations are generally in agreement with the variability in river discharg e patterns (Fig.5-9). Low SDD coincided with high river flow and vice-versa. For example, abnormally higher river flow during winter of 1997-1998, linked to the 1997-1998 El Nio event (Schmidt & Luther, 2002), led to the lowest SDD in early 1998. The hurricane-induced high river flow in August and September 2004 (Hu et al., 2006) led to the lower SDD in the corr esponding months, particularly in the middle and lower Bay. There are, however, occasional exceptions. No corresponding high SDD values were found for the low river flow during 2000. Relatively higher SDD between 2002 and early 2003 occurred during a period of larger river flow, although wind speed during 2002 was slightly below the climatol ogical means (Fig.5-10). Clearly, although river flow and wind are the two major factors that affect SDD, other factors such as tides and estuarine circulation patterns (Weisber g & Zheng, 2005) may also contribute to modulate water quality in different parts of the Bay. Overall, SeaWiFS SDD estimates are consistent with in situ SDD (Figs.5-4 and 57, e.g., within 1 standard deviation). However, SeaWiFS SDD showed more pronounced seasonal and interannual variability that is not captured by the in situ SDD. For example, abnormally high in situ SDD values (> 5.0 m) were observe d at Sta. 92 in February and March 2003, but the monthly SeaWiFS means (> 4 observations in each month) showed lower SDD values, comparable to those seen in other years. Although the high in situ SDD values in those months may capture actu al events, they mask the seasonal patterns revealed in satellite observations (Fig. 5-7). These large differences (e.g, > 1-2 m) between two data sets are likely due, in great part, to the mismatch in sampling frequencies between the in situ program and the satellite collections because the uncertainties in the SeaWiFS SDD estimates, as measured by the RMS error (0.55 m in Fig. 5-2), are much smaller than the observed differences. Indeed SeaWiFS typically has > 4 quality-controlled observations per mont h (about once per week), and sometimes >10 observations per month between January and May (Fig. 5-7). This helps reduce aliasing of temporal variations and helps construct more reliable monthly means than using the single in situ monthly measurements.
JanMarAprJunAugSepOctNovDecFebMayJul 20304050607080Number of Days Months of a Year Hillsborough RiverAlafia River 5.010.015.020.025.0Climatological river flow (m3/s)0.0 (a) (b) Figure 5-6. (a) Climatological monthly means of river flow from the Hillsborough River (open circles) and the Alafia River (filled circles) from 1997 to 2005. (b) The total number of days when the daily averaged wind speed was > 4.0 m s -1 from 1997 to 2005 at the PORT station near Saint Petersburg (27.6'N, 82.6'W). 92
S J M S J M S J M S J M S J M S J M S J M S J M S 0.0 2.0 4.0 6.0 8.0 Secchi Desk Depth (m) Sta. 92SeaWiFS Figure 5-7. Monthly means of SeaWiFS and in situ SDD at selected stations (Fig. 5-1) from September 1997 to December 2005. The number of SeaWiFS observation days during each month is shown on the right-hand side. When this number is >1, standard deviation is also shown. Note that there is only one in situ observation in each month. J, M, and S stand for the month of January, May, September, respectively, and are denoted in a same way in the subsequent figures. 0 4 8 12 16 Number of observation days 0.0 2.0 4.0 6.0 8.0 Sta. 23In Situ 0 2 4 6 8 0.0 2.0 4.0 6.0 8.0 Sta. 14 0 2 4 6 8 number of observation days 0.0 2.0 4.0 6.0 8.0 Sta. 40 0 2 4 6 8 2005 2004 2003 2002 2001 2000 1999 1998 93
1998 1999 2000 2001 2002 2003 2004 S J M S J M S J M S J M S J M S J M S J M S J M S .0 0.0 1.0 2.0 3.0 4.0 Sta. 92 .0 .0 0.0 1.0 2.0 3.0 4.0 Anomaly of Secchi Desk Depth (m)Sta. 23 .0 .0 0.0 1.0 2.0 3.0 4.0 Sta. 14 .0 0.0 1.0 2.0 3.0 4.0 Sta. 40 2005 Figure 5-8. SeaWiFS SDD anomaly between September 1997 and December 2005 at several stations in Tampa Bay (Fig.5-1). The anomaly is defined as the difference of SDD between the current month (Fig. 5-8) and the climatological month (red lines). Some gaps in anomaly lines at Sta. 14 and Sta. 40 are due to no valid values in those months. The box areas indicate 4 negative anomalies occurred around winter-spring of 1998, summer-fall of 2001, 2003, and 2004, respectively, while filled arrows in Sta. 92 indicate the positive anomalies in late 2002-early 2003 and in late 2005, respectively. 94
Alafia RiverHillsborough River 0.010.020.030.040.050.060.070.080.0Monthly mean of river flow rate (m3 s-1)JMSJMSJMSJMSJMSJMSJMSJMSS200220052004200319971998199920002001 95 igure 5-10. The number of days when the daily averaged wind speed was > 4.0 m s-1 of ach month from September 1997 to December 2005 at the PORT station near Saint etersburg (2745.6'N, 82.6'W). The overlaid red line indicates climatological means f number of the days of 12 months. Figure 5-9. Monthly means of river flow from the Hillsborough River (open circles) and the Alafia River (filled circles) from September 1997 to December 2005. SJMSJMSJMSJMSJMSJMSJMSJMS199919982004200520031997 200220012000 5.0 10.0 Dys 15.00.0Number ofa F e P o
96 .5. Discussion ith ut any fine tuning was applied. The improvement an be attributed to two unique features employed the semi-a nalytical algorithm. Firstly, the semi-analytical algorithm explicitly estimates absorption and backscattering ajor contributors to Kd) based on the vigorous radi ative transfer theory, water types. More importantly, various semi-analytical algorithms have demonstrated that the absorp tion and backscattering coefficients can be well r A detailed look into the satellite derived R found that after atm 5 5.5.1. Algorithm issues The semi-analytically derived K d (490) showed a much improved relationship w in situ SDD relative to the empirical band-ratio K d (490). The agreement is remarkably good given a general algorithm witho c coefficients (two m therefore it is less dependent of etrieved from ocean color remote sensing with ~15% in various waters (Lee et al., 2005a, and references therein). Secondly, semi-a nalytical algorithms use more than two bands in the algorithm to estimate the light attenuation coefficients; therefore retrievals contain more information than those from si mple two-band ratio algorithms (Lee et al., 2005b). These results suggest that although there may be relatively large uncertainties in the atmospheric correction over coastal and estuarine waters (e.g., Hu et al. 2000), water quality parameters derived from an improve d bio-optical algorithm may be sufficiently accurate for coastal water m onitoring programs. Indeed, Magnuson et al. (2005) also showed better performance in retrievals of chlorophyll concentration using an improved, locally tuned bio-opt ical algorithm. However, the apparent underestimates of light attenuation in Hillsborough Bay (HB) during the summer/rainy season are like ly due to the artifacts in the atmospheric correction. At this time of the year and in this region, remote sensing reflectance ( R rs ) in the blue-green wavelengths is very low (the maximum R rs at visible domain is < 0.002 sr 1 unpublished data) due to the high terrestrial input of colored dissolved organic matter (CDOM) (Chen et al., in press). A slight e rror in the atmospheric correction may cause much larger relative errors in R rs (Harding et al., 2005; Hu et al., 2001) and therefore create erroneous estimates in light attenuation. rs ospheric correction R rs (412) in those invalid pixels was often larger than R rs (443), an impossible feature for this type of water. The exact reason for this
97 precipitation and river runoff (Chen et al., in press; Lipp et al., 2001; Schmidt & Mark, 2002). This is likely the primary reason th at chlorophyll variation accounts for the on this Tampa Bay National Estuary Program (TBNEP) proposed a Nitrogen Manag gh y rainy arity) is an important factor impacting water clarity, as seen in previous stud For discrepancy was not clear. However, these e ffects appear to be limited to HB only, and are negligible for other bay-segments as shown in the re lationship between K d (490) and SDD (Fig., 5-2). Therefore, th e results presented for other bay segments are still valid. 5.5.2. Implications for Tampa Bay ecosystem restoration and water quality monitoring SeaWiFS SDD imagery clearly shows that the seasonal cycle is influenced by river flow through inputs of CDOM and nutrien ts. This observation is consistent with previous studies that found that Tampa Ba y water quality is closely related to majority of light attenuation variation in Ta mpa Bay (Janicki et al., 2001). Based finding, the ement Strategy that seeks to increase water clarity for seagrass restoration throu reducing phytoplankton concentrat ion by controlling total nitr ogen loadings into the ba (Janicki et al., 2001). However, our results suggest that SDD is al so controlled by other processes. Indeed, although increased phytoplankton and color lead to decreasin g SDD in the summer/ season, the SDD values are still la rger than those in the winter season when turbidity is at larger levels (Fig. 5-3). Part icularly SDD reaches a maximum (i.e., highest water cl in May only when turbidity turns into lowe r values, while phytoplankton and color show little change during this vari ation. This suggests that turb idity due to an increase in sediment resuspension ies in Tampa Bay (McPherson & Mill er, 1994) and in other estuaries (e.g., Christian & Sheng, 2003) Therefore, an optimal management plan may also include controls of sediment load or at least greater understanding of factors leading to sediment resuspension in the Bay. Because of more frequent and synoptic coverage, SeaWiFS provides improved estimates of the monthly mean patterns of SDD and monitoring long-term trends. example, it is difficult to determine trends from the in situ data because they are collected
98 ans should nsing to help interp ret the spatial-temporal patterns as well as the y a ed that one of the cri tical water quality parameters, the Secchi Disk Depth (SDD, an index for water clarity), can be accurately derived from SeaWiFS data for most of Tampa Bay waters and for al l seasons (standard e rror of 0.55 m for SDD ng .0 m). Although some artif acts exist due to erroneous atmospheric correct e se ter manage estuarine and coastal resources. only once per month and are easily biased by extreme events. Complementing the in situ data with satellite observat ions helps detect, for example, trends such as the water clarity improvement in the lower bay after 2002. Clearly, future monitoring pl include satellite remote se long-term trends. 5.6. Conclusions One major issue in application of satellite data to monitor estuarine water qualit has been the lack of sufficient accuracy due to uncertainties in the atmospheric correction and bio-optical inversion algor ithms. Using a semi-analytical algorithm and rigorous dat quality control, I show rangi from 0.9 to 8 ion, the frequent and s ynoptic coverage of the satel lite data provides better estimates of the spatial-temporal patterns as well as long-term trends of SDD. This can greatly benefit long-term monito ring efforts of estuarine waters, most or all of which are based only on field surveys. The distinctive spatial and temporal viability revealed in SeaWiFS SDD imagery highlights the importance of river runoff (phytoplankton and color) in the rainy season and of wind-induced sediment resuspensi on (turbidity) in th dry season in controlling the light attenua tion and therefore SDD across Tampa Bay, particularly in the middle a nd lower portions of the Bay. Therefore, optimal management plans should not only focus on total nitrogen control, but also on understanding and mitig ating total sediment load from various sources such as storm water, river runoff, dredging and transportati on activities. Becau of the synoptic, robust, and the repeated global coverage, satell ites like the SeaWiFS represent an important complement to traditional in situ sampling methods. Thus, I also strongly recommend incorporat ing satellite observations of light attenuation or water clarity to various coastal mon itoring programs to help bet
99 Chapter 6. Summary 6.1. Main conclusions This study addressed monitoring of wa ter quality in Tampa Bay using both in situ and remote sensing techniques. Four key wate r quality indices, specifically chlorophyll, dissolved colored organic matter, turbidity, a nd water clarity were examined. The results showed the following: Hourly observations collected with bio-opt ical sensors attached to a moored buoy showed large short-term variab ility at tidal and sub-tidal scales in both chlorophyll and suspended sediment concentrations in the dry season. The variability was primarily driven by tides and winds. During the wet season, the data wa s not useful due to severe bio-fouling of the instruments. Phytoplankton blooms occurred 1-2 days after high turbidity caused by winddrive sediment suspension events subsided, with the largest bloom s occurring if a neap tide coincided with the subsidence of turbidity. The hourly in situ observations collected with the automated sensors showed that the sampling strategy presently used by th e Environmental Protection Commission of Hillsborough County (EPCHC), which consists of single observations at a series of stations occupied over a period of three weeks every month, introduces biases (-50% to 200%) if they are used to represent the monthly mean condi tions in the chlorophyll and suspended sediment concentration of Tamp a Bay. A higher frequency of sampling is needed to characterize a fast changing estuary like Tampa Bay. The in situ spatial sampling of optical propertie s revealed showed that CDOM is delivered to Tampa Bay primarily by riverine inputs, and that it exhibited conservative mixing during the dry season and apparent non-conservative mixing during the wet season. The non-conservative behavior of CD OM in the gradient between Hillsborough
100 Bay and Middle Tampa Bay could be explai ned by photobleabching, but it could also be due to CDOM delivered by Old Tampa Bay in a scenario of multiple endmembers. Old Tampa Bay CDOM showed distinctively differe nt properties with hi gher spectral slopes and fluorescence efficiency and lower absorp tion coefficients relative to CDOM in the other basins. These differences may be rela ted to the shallow bathymetry in OTB and some local processes like photobleaching, i nputs from primary production in water column and/or seagreass. Satellite remote sensing show ed promising results as a tool to monitor Tampa Bay water quality frequently and synoptically. A turbidity algorithm for application of MODIS/Aqua 250 m imagery was develope d by systematically evaluating the radiometric observations, atmospheric correction algorithm, and developing a relationship between turbid ity and remote sensing reflectance at 645 nm. MODIS turbidity images showed distinct spatial and temporal patterns related to river runoff in the upper bay, and to wind-induced sediment re suspension events in the middle and lower portions of the bay. As compared to the si ngle station measurements from the monthly water quality surveys, MODIS de rived turbidity provided more realistic and consistent estimates of the mean and varying conditi ons of turbidity in Tampa Bay. MODIS provided as many as 10 views of the Bay per month. Light attenuation coefficients were derived using SeaWiFS images through application of a new semi-analytical algor ithm designed to estimate secchi disk depth (SDD). The SeaWiFS SDD image series confirmed that water clarity is related to temporal and spatial patterns of river runoff and also to wind-induced sediment resuspension events. 6.2. Future work (1) My study of short-term variability wa s limited to one sta tion and one (dry) season. To fully understand the spatial a nd temporal variabil ity in chlorophyll and sediment dynamics, bio-optical sensor s should be deployed in different bay segments and in different seasons. Howe ver, current bio-optical sensors suffer
101 from bio-fouling in the summer season. Therefore a new deployment method has to be developed. (2) More studies are needed to understand the cause for increased chlorophyll during neap tides after wind induced bottom sediment resuspension events. (3) The effects of bottom contamination on satellite remotely sensed signals have not been fully addressed in my thesis. I simply masked suspected areas of shallow water. Because Tampa Bay is on average only 3-4 m deep, quantitative evaluation of this effect is necessary; (4) More effort should be placed in deve loping a robust algorithm to estimate chlorophyll in the Bay. Similarly, near -IR band ratios are two promising methods that should be tested in the Bay. (5) Work with resource managers to develop a prototype water-quality monitoring tool based on information collected automatically from sensors deployed on buoys and satellite imagery of turbidity and water clarity would significantly help efforts to monitor Ta mpa Bay. These tools and applications need to be developed.
102 List of References Abbot, M.R.,& Letelier, R.M.(1999). Ch lorophyll fluorescence. In MODIS Ocean Science Team Algorithm Theoretical Ba sis Documents, 22, version 3, pp. 1 (Greenbelt :NASA). Anastasiou, C. J., Johansson, R., Aver y, W., Bishop, & D. F.(2005).Using percent transmission for calculating light attenuati on in clear shallow waters: A case study in Tampa Bay. USGS 4 th Annual Tampa Bay Scien ce Conference, Feb. 8&9, Gulfport, Florida. Anthony, K.R.N ., Ridd, P.V ., Orpin, A.R ., Larcombe, P & Lough, J .(2004).Temporal variation of light availab ility in coastal benthic habitats: Effects of clouds, turbidity, and tides. Limnology and Oceanography, 49(6), 2201-2211. Austin, R. W., & Petzold, T. J. (1981). The determination of the diffuse attenuation coefficients of sea water using the coastal zone colo r scanner. In oceanography from space. (J.F.R. Gower, Ed.), Plenum, New York, pp. 239-256. Avery, G. B., Kieber, R. J., Willey, J. D., Shank, G. C., Whitehead, R. F., 2004. Impact of hurricanes on the flux of rainwater and Cape Fear River water dissolved organic carbon to Long Bay, S outheastern United States. Global Biogeochemical Cycle, 18 (3), GB3015, doi:10.1029/2004GB002229. Babin, M., Morel, A.,& Genitili, B.(1996). Re mote sensing of sea surface Sun-induced fluorescence: consequences of natural variations in the optical characteristics of phytoplankton and the quantum yield of ch lorophyll a fluorescence. International Journal of Remote Sensing 17, pp. 2417. Babin, M., & Stramski,.D. (2002). Light abso rption by aquatic particles in the nearinfrared spectral region. Limnology and Oceanography 47, 911-915.
103 Babin, M., D., Stramski, Ferrari, G.M., Claustre, H., Bricaud, A., Obolensky, G., Hoepffner, N.(2003). Variations in th e light absorption coefficients of phytoplankton, nonalgal particle s, and dissolved organic matter in coastal waters around Europe. Journal of Geophysical Research, 108(C7) 3211, doi: 10. 1029/2001JC000882. Babin, M., & Stramski, D. (2005) Variations in the mass-sp ecific absorption coefficient of mineral particles suspended in water. Limnology and Oceanography 49 756767. Bailey, S.W., & Werdell, P.J.(2006). A multi-sensor approach for the on-orbit validation of ocean color satellite data products. Remote Sensing of Environment, 102(12), 12-23. Bendis, B.(1999). Water quality trends in Tampa Bay, Florida. University of South Florida, Master Thesis, 74p. Blough, N. V.& Del Veccico, R.(2002). Chromophoric DOM in the coastal environment. In: Hansell, D.S., Carlson, C.A., (Eds.), Biogeochemistry of Marine Dissolved Organic Matter, Elsevier Science, US, pp 509-546. Boler, R.N., Molloy, R.C.,& Lesnett, E. M. (1991). Surface water quality monitoring by the Environmental Protection Commission of Hillsborough Country. In: proceedings from BASIS II, Treat, S.F ., Clark, P.A. Editors. Tampa, FL., 528p. Boss, E., Pegau, W. S., Zaneveld, J.R.V.,& Barnard, A.H.(2001). Spatial and temporal variability of absorption by dissolved material at a continental shelf. Journal of Geophysical Research 106(C5), 9499. Branco, A. B.& Kreme, J. M..(2005). The Relative Importance of Chlorophyll and Colored Dissolved Organic Matter (CDOM) to the Prediction of the Diffuse Attenuation Coefficient in Shallow Estuaries. Estuaries: Vol. 28, No. 5, pp. 643 652. Bricaud, A., Basbin, M., Morel, A.,&Claustre H.(1995).Variability in the chlorophyllspecific absorption coefficients of natural phytoplankton: Analysis and parameterization. Journal of Geophysic al Research 100(C7), 13,321,332. Burdige, D. J, Kline S. W.,& Chen, W. H.(2004). Fluorescent di ssolved organic matter in marine sediment pore waters. Marine Chemistry 89 (1-4), 289-311.
104 Burwell, D., Vincent, M., Luther, M., & Ga lperin, B.(2000). Modeling Residence Times: Eulerian vs Lagrangian. In: M. L.,Butle r, H. L., (Eds.), Estuarine and Coastal Modeling, Spaulding, ASCE, Reston, VA, pp 995-1009. Cannizzaro, J. P., Carder, K. L., Chen, F. R., Heil, C.A., Vargo, G.A. (Accepted). A novel technique for detection of the toxic dinoflagellate Karenia brevis in the Gulf of Mexico from remotely sensed ocean color data. Continental Shelf Research. Carder, K.L., Steward, R.G., Harvery, G.R.,& Ortner, P.B. (1989) Marine Humis and Fulvic Acids: their effects on remote sensing of ocean chlorophyll. Limnology and Oceanography, 34(1), 68-81. Carder, K.L., Chen, F.R., Lee, Z.P., Hawes, S.K., Kamykowski, D.,(1999). Semi-analytic Moderate-Resolution Imaging Spectro meter algorithms for chlorophyll a and absorption with bio-optical domains based on nitrate-depletion temperatures. Journal of Geophysical Research, 104, 5403-5422. Chen, R. F.& Gardner, G. B.(2004). High -resolution measurements of chromophoric dissolved organic matter in the Mississippi and Atchafalaya River plume regions. Marine Chemistry, 89 (1-4), 103-125 Chen, Z., Li, Y., & Pan, J.(2004). Distributions of colored dissolved organic matter and dissolved organic carbon in the Pearl River Estuary, China. Continental Shelf Research, 24, 1845-1856. Chen, Z., Hu, C., Conmy, R.N., Swar zenski, P., & Mullerr-Karger, F. (In Press). Colored Dissolved Organic Matter in Tampa Bay, Florida. Marine Chemistry Chen, Z., Hu, C., Muller-Karger, F., & Luther, M. (Submitted). Physical forcing of short bio-optical variability in Tampa Bay: Observations from a coastal tower. Limnology and Oceanography. Chen, Z., Muller-Karger, F., & Hu,. C. (Submitt ed). Remote Sensing of water clarity in Tampa Bay. Remote Sensing of Environment Chen, Z., Hu,. C.,&Muller-Karger, F.(in revi sion). Monitoring turbidity in Tampa Bay using MODIS 250 M imagery. Remote Sensing of Environment
105 Christian, D., & Sheng, Y. P. (2003). Relative influence of various water quality parameters on light attenuation in Indian River Lagoon. Estuarine, coastal and shelf science, 57(5-6), 961-971. Clementson, L.A., Parslow, J. S., Turnbull, A. R., Bonham, P.I., .2004. Properties of light absorption in a highly colored estuarine sy stem in south-east Australia which is prone to blooms of the toxic dinof lagellate Gymnodinium catenatum. Estuarine Coastal and Shelf Science, 60 (1) 101-112. Cloern, J.E.(1987). Turbidity as a control on phytoplankton biomass and productivity in estuaries. Continental Shelf Research 7(11), 1367-1381. Cloern, J.E., Powell, T.M.,& Huzzey,L.M. (1989). Spatial and temporal variability in South San Francisco Bay. II. Temporal chan ges in salinity, suspended sediments, and phytoplankton biomass and produc itivity over tidal time scales. Estuarine, Coastal and Shelf Science, 28, 599-613. Cloern, J. E. 1991. Tidal s tirring and phytoplankton bloom dynamics in an estuary. Journal Marine Research, 49 203-221. Cloern, J. E., T.M. Powell, L. M. Huzzey. 1989. Spatial and temporal variability in South San Francisco Bay (USA). II temporal ch anges in Salinity, suspended sediments and phytoplankton biomass and productivity over tidal time scales. Estuarine, Coastal and shelf Science, 28, 599-613. Coble, P.G.(1996). Characterization of marine and terrestrial DOM in seawater using excitation-emission matrix spectroscopy. Marine Chemistry, 51 325-346. Cole, B.E., & Cloern, J.E. (1987). An em pirical model for estimating phytoplankton productivity in estuaries. Marine EcologyProgress Series, 36 299-305. DallOlmo, G., Gitelson, A.A., Rundquist, D.C., Leavitt, B.,Barrow, T., and Holz J. C.(2005). Assessing the potential of SeaWiFS and MODIS for estimating chlorophyll concentration in turbid productive waters using red and near-infrared bands. Remote Sensing of Environment, 96 176 187. Desmit,.X; J. P. Vanderborght, P. Regni er and R. Wollast 2005. Control of phytoplankton production by physical forc ing in a strong tidal, well-mixed estuary. Biogeoscicence 2, 205-218.
106 Doxaran, D., Froidefond, J. M., Lavender, S., & Castaing, P. (2002) Sp ectral signature of highly turbid waters. Application w ith SPOT data to quantify suspended particulate matter concentrations. Remote Sensing of Environment 81,149-161. Doxaran, D., Cherukuru, R.C.N. & Lavender, S. J. (2006). Inherent a nd apparent optical properties of turbid estuarine waters: m easurements, modelling and application to remote sensing. Applied Optics, 45, 2310-2324 Fisher, T. R ., Gustafson, A. B ., Sellner, K ., Lacouture, R ., Haas, L.W ., Wetzel, R.L ., Magnien, R ., Everitt, D ., Michaels, B .,& Karrh, R (1999). Spatial and temporal variation of resource li mitation in Chesapeake Bay. Marine Biology 133 (4), 763778. Florida Department of Community Affaires. ( 1996). State of the coastal report: preparing a sustainable future. Prepared by the Fl orida Center for Public Management. Florida State University, Tallahassee, FL. Gallegos, Charles L. (2001). Calculating optical water qua lity targets to restore and protect submerged aquatic vegetation: overcoming problems in partitioning the diffuse attenuation coefficient for photosynthetically active radiation Estuaries, 24(3), 381-397. Gallegos, C.L.(2005) Optical water quality of a blackwater river es tuary:the Lower St. Johns River, Florida, USA. Estuarine, Coastal and Shelf Science, 63 57. Gesch, D ., & Wilson, R (2001). Development of a seamless multisource topographic/ bathymetric elevation model of Tampa Bay. Marine Technology Society Journal 35(4) 58-64. Giesen, W. B. J. T., Van Katwijk, M. M ., & Den Hartog, C. (1990). Eelgrass condition and turbidity in the Dutch Wadden Sea. Aquatic Botany, 37, 71-85. Gower, J.F.R., Doerffer, R.,& Borstad, G.A. (1999). Interpretation of the 685nm peak in water-leaving spectra in terms of fluores cence, absorption and scattering, and its observation by MERIS. International Journal of Remote Sensing, 20 1771. Gower, J.F.R., Brown,L,& Borstad, G.A.(2004). Observation of chlorophyll fluorescence in west coast waters of Canada using the MODIS satellite sensor. Canadian Journal of Remote Sensing, 30 17.
107 Green, S. A. (1992). Applications of fl uorescence spectrosco py to environmental chemistry. Ph. D. Thesis, Woods Hole Oceanographic Institute. Woods Hole, MA, USA. Guo, L. D., Hunt, B. J., & Santschi, P. H.(2001). Effect of dissolved organic matter on the uptake of trace metals by American oysters. Environmental Science and Technology 35 (5) 885-893. Hach Model 2100N laboratory turbidimeter inst ruction manual for use with software version 1. (1999). Hach Company. pp. 1-80. Harding, L.W. Jr.(1994). Long-term trends in the distribution of phytoplankton in Chesapeake Bay: Roles of light, nutrients, and stream flow. Marine Ecology Progress Series,104 : 267-291. Harding, L.W., Kramer, J.G., Phinney, J.( 2002). Estuarine and watershed monitoring using remote sensing technology. Present Status and Future Trends. A Workshop report. Scientific and Technical Advisory Committee and Maryland Sea Grant College. pp54. Harding, L.W., Magnusona A., Mallonee M.E. (2005). SeaWiFS retrievals of chlorophyll in Chesapeake Bay and the mid-Atlantic bight. Estuarine, Coastal and Shelf Science, 62 75. Harwell, M.A., Ault, J.S., Gentile, J. H., 1995. Comparison of the Ecol ogical Risks to the Tampa Bay Ecosystem from Spills of Fuel #6 and Orimulsion. Comparative Ecological Risk Assessment, Volume 1. Center for Marine and Environmental Analyses. University of Miami, Miami, Florida. Hastings, R. H., Conmy, R. N., Coble, P. G.,(2004).CDOM characterization in riverdominated Environments. Paper presen ted at AGU December 2004 meeting, San Francisco. Heyes, A, Miller, C ., & Mason, R.P (2004). Mercury and methylmercury in Hudson River sediment: impact of tidal resu spension on partitioning and methylation. Marine Chemistry 90 (1-4), 75-89. Hoge F.E., Lyon, P.E., Swift, R.N., Yungerl, J.K., Abbot, M.R., Letelier, R.M.,& Esaias, W.E.(2003). Validation of Terra-MODIS phytoplankton chlorophyll fluorescence line height. I. Initial airborne lidar results. Applied Optics, 42,
108 Hu, C., Carder, K. L., & Mulle r-Karger, F.E. (2000). Atmosp heric correction of SeaWiFS imagery over turbid coastal waters: A practical method. Remote Sensing of Environment 74, 195. Hu, C., Carder, K. L., & Muller-Karger, F. E. (2001a). How precise are SeaWiFS ocean color estimates? Implication of digitization-noise errors. Remote Sensing of Environment 76, 239-249. Hu, C. Muller-Karger F. E, Andrefouet, S,& Carder K.L.(2001b). Atmospheric correction and cross calibration of LANDSAT-7/ETM + imagery over aquatic environments: A multiplatform approach using SeaWiFS/MODIS. Remote Sensing of Environment 78(1-2) 99-107. Hu, C., Muller-Karger, F.E., & Zepp, R.G.( 2002). Absorbance, absorption coefficient, and quantum yield of photo production: Comment on common ambiguity in the use of these optical concepts. Limnology and Oceanography, 47 (4), 1261-1267. Hu, C., Muller-Karger, F., Biggs, D, Carder, K., Nababan, B., Nadeau, D.& Vanderbloemen, J.(2003). Comparison of ship and satellite bio-optical measurements on the continental margin of the NE Gulf of Mexico. International Journal of Remote Sensing, 24(13) 2597-2612. Hu, C.,Chen, Z.,Clayton, T. Swarzenski, P., Brock, J. and Muller-Karger, F.(2004) Assessment of estuarine water quality monitoring with MODIS mediumresolution bands: Initial re sults from Tampa Bay, FL. Remote Sensing of Environment, 93, 423-441. Hu, C., Muller-Karger, F., Ta ylor, J., Carder, K.L., Kelble C., Johns, E., & Heil, C.A. (2005). Red tide detection and tracing using MODIS fluorescence data: A regional example in SW Florida coastal waters. Remote Sensing of Environment, 97,311-321. Hu, C., Muller-Karger,F. E., & Swarzensk i, P. W. (2006). Hurricanes, submarine groundwater discharge, and Floridas red tides. Geophysical Research Letters, 33 L11601, doi:10.1029/2005GL025449. Huot, K.L.Y., Brown, C.A.,& Cullen, J.J.(2005). New algorithms for MODIS suninduced chlorophyll fluorescence and a co mparison with present data products. Limnology and Oceanography, Methods 3 108.
109 Janicki, A., Pribble, R., Ja nicki, S., & Winowitch, M. (2001) An analysis of long-term trends in Tampa Bay Water Quality. Tampa Bay Estuary Technical Report #0401, 48 pp. Janicki, A., & Wade, D. (1996). Estimating critical nitrogen loads for the Tampa Bay Estuary: an empirically based appro ach to setting management targets. Report to the Tampa Bay National Estuary Program # 94-444-14. Jean-FrancO. B., & Giuseppe, Z. (2004). Bi o-optical relationships for the northern Adriatic Sea. International Journal of Remote Sensing 25 (7), 1527. Jing, L.& Ridd, P. V (1996). Wave-current bottom shear stresses and sediment resuspension in Cleveland Bay, Australia. Coastal Engineering, 29, 169-186. Johansson, J. O. R. (2000). Historical overv iew of Tampa Bay water quality and seagrass: Issues and trends. In H. Greening (Ed.) Seagrass Management: Its Not Just Nutrients! Tampa Bay Estuary Program Symposium Proceedings, 22 August, St. Petersburg, FL (pp. 1 10). IOCCG (2000) Remote Sensing of Ocean Colo r in Coastal, and Other Optically-Complex Waters. Sathyendranath, S. (ed.), Reports of the International Ocean-Colour Coordinating Group, No. 3, IOCCG, Dartmouth, Canada. Kirk, J. T. O. (1994). Light and phot osynthesis in aquatic ecosystems. (2 nd ed.). New York, NY: Cambridge University Press, 599pp. Kishino, M., Takahashi, M., Okami, N., Ichimura, S., 1985. Estimation of the spectral absorption coefficients of phytoplankton in the sea. Bulletin of Ma rine Science, 37, 634-642. Kowalczuk, P., Cooper, W.J., Whitehe ad, R.F., Durako, M.J., Sheldon, W., 2003. Characterization of CDOM in an organic-rich river an d surrounding coastal ocean in the South Atlantic Bight. Aquatic Science, 65 384-401. Kratzer, S Hakansson, B & Sahlin, C (2003). Assessing Secchi and photic zone depth in the Baltic Sea from satellite data. Journal of the Human Environment, 32(8), 577-585. Ladika, S.(2002). Disciplines team up to take the pulse of Tampa Bay. Science, 295, 1455-1456.
110 Lawrence, D., M. J.Dagg, H. Lou, S. R. Cu mming, P. B. Ortner and C. Kelble. 2004. Wind events and benthic-pelagic coupling in a shallow subtropical bay in Florida. Marine Ecology Progress Series, 266 ,-13. Lee, Z., K.L. Carder, C. D. Mobley, R.G. Steward, and J. S. Patch (1999) Hyperspectral remote sensing for shallow waters: 2. Deriving bottom depths and water properties by optimization. Applie. Optics, 38, 3831-3843. Lee, Z.P.,Carder, K.L.,& Arnone, R.. (2002).Deriving inherent optical properties from watercolor: A multi-band quasi-analytical algorithm for optically deep waters. Applied Optics 41, 5755-5772, 2002. Lee, Z.,& Carder, K.L.(2004). Absorption spectrum of phytoplankton pigments derived from hyperspectral remote-sensing reflectance. Remote Sensing of Environment 89, 361. Lee, Z.-P., Du, K.P, & Arnone, R. (2005a ). A model for the diffuse attenuation coefficient of downw elling irradiance. Journal of Geophysical Research 110, C02016, doi:10.1029/2004JC002275.1 Lee, Z.P., Darecki, M., Carder, K. L, Davis, C. O., Stramski, D., & Rhea, W. J. (2005b). Diffuse attenuation coefficient of downwelling irradiance: An evaluation of remote sensing methods. Journal of Geophysical Research 110, C02017, doi:10.1029/2004JC002573. Lewis, R.R ., Clark, P.A ., Fehring, W.K ., Greening, H.S ., Johansson, R.O .,& Pau,l R.T (1998).The rehabilitation of the Tampa Bay Estuary, Florida, USA, as an example of successful integrated coastal management. Marine Pollution Bulletin 37(8-12), 468-473. Li, M. Z. and C. L. Amos. 2001. SEDTRANS 96: the upgraded and better calibrated sediment-transport model for continental shelves. Computers & Geoscience, 27 : 619. Li, Y., & Smayda, T.J.(2001). A chlorophy ll time series for Narragansett Bay: Assessment of the potential effect on tidal phase on measurement. Estuaries, 24(3), 328-336.
111 Lipp, E. K., Schmidt, N., Luther, M. & Rose, J. B. (2001). Determining the effects of El Nio-Southern Oscillation events on coastal water quality. Estuaries, 24 491497. Magnuson, A, Harding, L.W., Mallonee a E. M., & Adolf, J. E. (2004) Bio-optical model for Chesapeake Bay and th e Middle Atlantic Bight. Estuarine, Coastal and Shelf Science, 61, 403. Maritorena, S., Siegel, D.A.,& Peterson, A. R.(2002). Optimization of a semianalytical ocean color model for gl obal-scale applications. Applied Optics ,14, 2705-2714. May, C. L., J. R. Koseff, L.V. Lucas, J. E. Cloern and D.H. Schoellhamer. 2003. Effects of spatial and temporal variability of turbid ity on phytoplankton blooms. Marine Ecology Progress Series, 254, 111-128. McPherson, B. F., & Miller, R. L.(1994). Ca use of light attenuation in Tampa Bay and Charlotte Harbor, southwestern Florida. Water Research Bulletin 30, 43-53. Meister, G., Kwiatkowska, E.J., Franz, B.A., Patt, P.S., Feldman, G.C. & McClain, C.R.(2005). Moderate-Resolution Imaging Spectroradiometer ocean color polarization correction. Applied Optics 44(26), 5524-5535. Miller, R.L. & Mckee, B.A. (2004). Using MODIS Terra 250 m imagery to map concentrations of total suspended matter in coastal waters. Remote Sensing of Environment, 93, 259-266. Miller, R.L., Mckee, B.A.,& DSA, E. (2005) Monitoring bottom sediment resuspension and suspended sediments in shallow coasta l waters. (eds.) Miller, R.L., Remote sensing of coastal Aquatic environments, 259-276. Mobley, C.D., Hydrolight 3.0 Users' Guide, SRI International, Menlo Park, Calif., 1995. Monbet, Y. 1992. Control of phytoplankton bioma ss in estuaries: A comparative analysis of microtidal and macrotidal estuaries Estuaries, 15(4), 563-571. Moore, K. A., Wetzel, R. L., & Orth, R. J. (1997). Seasonal pulses of turbidity and their relations to eelgrass (Zostera marina L.) survival in an estuary. Journal of Experimental Marine Biology and Ecology 215 (1), 115-134.
112 Mueller, J. L. (2000). SeaWiFS algorithm for the diffuse attenuation coefficient, K(490), using water-leaving radiances at 490 and 555 nm. In O'Reilly, J.E. and coauthors: SeaWiFS Postlaunch Calibration and Validation Analyses, Part 3. S.B. Hooker and E.R. Firestone, Eds., NASA/TM-2000-206892, Vol. 11, NASA Goddard Space Flight Center, Greenbelt, Maryland, 24-27. Mueller, J. L., Bidigare, R. R., Trees, C., Dore, J., Karl, D., Van Heukelem, L., 2003. Biogeochemical and bio-optical measurements and data analysis protocols: ocean optics protocols for satellite ocean color sensor validation. Re vision 4, Vol. 2. NASA/TM-2003-21621. Nieke, B., Reuter, R., Heuermann, R., Wa ng, H., Babin, M., Therriault, J.C., 1997. Light absorption and fluorescence propertie s of chromophoric dissolved organic matter (CDOM), in the St. Lawrence Estuary (case 2 water). Continental Shelf Research 17, 235-252. Otis, D. B., Carder K. L., English D. C., & Ivey, J. E.(2004). CDOM transport from the Bahamas Banks. Coral Reef, 23(1), 152-160. OReilly, J. E., Maritorean, S., Siegel, D.A., O Brien, M. C., Toole, D.,Chavez, F.P.et al., 2000. (Eds), Ocean Chlorophyll-a Algorith ms for SeaWiFS, OC2 and OC4: version 4. SeaWiFS Post-launch Calibrati on and Validation Analyses, Part 3. NASA Tech. Memo. 2000-206892, Vol. 11, S.B. Hooker and E.R. Firestone, Eds., NASA Goddard Space Flight Center, Greenbelt, Maryland, 9-23. Pennock, J.R ., & Sharp, J. H (1994).Temporal alternation between light-limitation and nutrient-limitation of phytoplankton pr oduction in coastal-plain estuary. Marine Ecology-Progress Series, 111 (3), 275-288. Pope, R. M., & Fry, E. S.(1997). Absorption sp ectrum (380 -700 nm) of pure water. II. Integrating cavity measurements. Applied Optics 36, 8710-8723. Prasad, K. S., Bernstein, R. L., Kahru, M., & Mitchell, B.G. (1998) Ocean color algorithms for estimating water clarity (S ecchi depth) from SeaWiFS. Journal of Advance Marine Science T echnical Society, 4(2), 301-306. Pribble, R., Janicki, A., Zarbock, H., Janick, S., Winowitch, M., 2001. Estimates of total nitrogen, total phosphorus and total susp ended solids to Tampa Bay, Florida. Tampa Bay National Estuary Program Technical Publication, # 05-01.
113 Robinson, D., 2004. Habitat assessment in Tampa Bay tidal river. In: The Tampa Bay Estuary Program Tidal Rivers and St reams Workshop. June 23, 2004. Tampa Bay Regional Planning Council. Rochelle-Newall, E. J., Fish, T. R., Fan, C., & Glibert, P.M. (1999). Dynamics of choromophoric dissolved organic matte r and dissolved organic carbon in experimental mesocosms. International Journal of Remote Sensing, 20, 627-641. Rochelle-Newall, E. J., & Fish, T. R. ( 2002). Chromophoric dissolved organic matter and dissolved organic carbon in Chesapeake Bay. Marine Chemistry, 77, 23-34. Roegner, G. C.,B. M. Hickey, J. A. Newton, A. L. Shank, D.A. Armstrong. 2002. Windinduced plume and bloom intrusions into Willapa Bay, Washington. Limnology and Oceanography, 47(4), 1033-1042. Ruddick, K., Park, Y, & Nechad, B. (2003). MERIS imagery of Belgian coastal waters: mapping of suspended particulate ma tter and chlorophyll-a. MERIS user workshop, 10-13 th November, (Frascati), ESA, SP-549 Santschi, P.H., Lenhart, J. J.,& Honeym an, B. D.(1997). Heterogeneous processes affecting trace contaminant distribution in estuaries: The role of natural organic matter. Marine Chemistry, 58 (1-2), 99-125. Schoellhamer, D. H. (1995). Sediment Resuspension Mechanisms in Old Tampa Bay, Florida. Estuarine, Coastal and Shelf science, 40(6): 603-620. Schoellhamer, D.H. (1996). Anthropogenic sediment resuspension mechanisms in a shallow microtidal estuary. Estuarine, Coastal and Shelf science, 43(5), 533-548. Schmidt, N ., & Luther, M.E (2002). ENSO impacts on salinity in Tampa Bay, Florida. Estuaries, 25(5), 976-984. Siegel, D. A., Wang, M., Maritorena, S., & Robinson, W. (2000). Atmospheric correction of satellite ocean color imagery: The black pixel assumption. Applied Optics, 39, 3582-3591. Stephen, E. D.;Cable, J.E.; Childers, D. L. ;Coronado-Malina, C.;Da y, J.W.; Hittle, C.D.; Madden, C.J.;Reyes, E.; Rudnick, D.; & Sklar, F.(2004). Importance of storm events in controlling ecosystem structur e and function in a Florida Gulf Coast Estuary. Journal of Coastal Research, 20 (4), 198-1208.
114 Strickland, J.D.H., and Parsons, T.R. (1972). A Practical Handbook of Sea Water Analysis. Fisheries Research Board of Canada, 310 pp. Stumpf, R. P., & Pennock, J. R. (1989), Calib ration of a general optical equation for remote sensing of suspended sediment s in a moderately turbid estuary, Journal Geophysical Research 94(C10), 14,363-14,371. Stumpf, R.P.,& Pennock, J. R.(1991).Remote estimation of the diffuse attenuation coefficient in a moderately turbid estuary. Remote sensing of Environment, 38, 183-191. Tomasko, D.A., Corbett, C.A., Greening, H.S., & Raulerson, G.E. (2005). Spatial and temporal variation in seag rass coverage in Southwes t Florida: assessing the relative effects of anthropogenic nutrient load reductions and rainfall in four contiguous estuaries. Marine Pollution Bulletin, 50 797-805. Twardowski, M. S., Donaghay, P.L., 2002. Photobleaching of aquatic dissolved materials: Absorption removal, spectral alternation, and thei r interrelationship. Journal of Geophysical Research, 107 (C8), 3091,doi:10.1029/1999JC000281. Tzortziou,M., Herman, J.R.,Gallegos, C.L.,N eale, P.K., Subramaniam,A.,Harding, L.W. Jr.,& Ahmad, Z.(2006).. Bio-optics of th e Chesapeake Bay from measurements and radiative transfer closure. Estuarine, Coastal and Shelf Science, 68 348-362 Uher, G., Hughes, C., Henry, G., Upstill-G oddard, R.C., 2001. Non-conservative mixing behavior of colored dissolved organic matter in a humic-rich, turbid estuary. Geophysical Research Letters, 28 (17), 3309-3312. Vargo, G.A., Richardson, W. R., Howard, D., Paul, J.H., 1991. Phytoplankton production in Tampa Bay and two tidal rivers: The Alaf ia and Little manatee. In: S. Treat and P. Clark (Eds.), Proceedings of Tampa Bay Area Scientific Information Symposium 2, Tampa, FL, pp 49-66. Vodacek, A., Blough, N.V., Degrandpre, M. D., Peltzer, E.T., Nelson, R.K., 1997. Seasonal variation of CDOM and DOC in the Middle Atlantic Bight: terrestrial inputs and photo-oxidation Limnology and Oceanography, 42 674-686. Wang, J.D. (1998). Subtidal flow pa tterns in Western Florida Bay. Estuarine, Coastal Shelf Science, 46, 901-905.
115 Wang, M., and B. A. Franz (2000). Compari ng the ocean color measurements between MOS and SeaWiFS: A vicarious inte rcalibration approach for MOS. IEEE Trans. Geosci. Remote Sens. 38,184-197. Wang, M.,& Bailey, S.W.(2001). Correction of sun glint contamination on the SeaWiFS Ocean and atmosphere products. Applied Optics 40(27), 4790-4798. Wang, M., & Shi, W. (2005). Estimate of ocea n contribution at the MODIS near-infrared wavelengths along the east coast of the U.S.: Tw o case studies. Geophysical Research Letter, 32, L13605, doi: 10.1029/2005GL022917. Weisberg, R. H.,& Zheng, Y, L. (2005). Circulation of Tampa Bay driven by buoyancy, tides, and winds, as simulated using a finite volume coastal ocean model. Journal Of Geophysical Research 111, C01005, doi:10.1029/2005JC003067. Woodruff, D. L., Stumpf, R. P., Scope, J.A, & Paerl, H. W.(1999).Remote sensing of water clarity in optically complex estuarine waters. Remote sensing of Environment, 68, 41-52. Yeager, C.L.J., L.W. Harding, M. E. Mallo nee. 2005. Phytoplankt on production, biomass and community structure following a summer nutrient pulse in Chesapeake Bay. Aquatic Ecology, 39 (2), 135-149. Yentsch, C.S., 1962. Measurement of visible li ght absorption by partic ulate matter in the ocean. Limnology and Oceanography, 7 207-217. Zanardi-Lamardo, E., Moore, C. A., Zika, R. G., 2004. Seasonal variation in molecular mass and optical properties of chromophoric dissolved organic material in coastal waters of southwest Florida. Marine Chemistry, 89 34-57. Zibordi, G., Mlin, F.,& Berthon, J.F. (2006). Comparison of SeaWiFS, MODIS and MERIS radiometric products at a coastal site Geophysical Research Letter 33 L06617, doi:10.1029/2006GL025778.
Appendix 3-1. Surface salinity, CDOM absorption coefficient at 400 nm (a CDOM (400), m -1 ) and spectral slope (S, nm -1 ), total suspended solids concentration (TSS, mg/l), chlorophyll concentration (Chl, mg/m 3 ), dissolved organic carbon concentration (DOC, M) and normalized CDOM fluorescence (Quinine Sulfate equivalent, QSE) in Tampa Bay and the Alafia River (AR) in the 2004 dry (June, first row per station) and wet (October, second row per station) seasons. Station locations can be found in Fig. 1. Hillsborough Bay Stations Latitude (Degree) Longitude (Degree) Salinity a CDOM (400) (m -1 ) Spectral Slope (10 -3 nm -1 ) TSS (mg/l) Chl (mg/m 3 ) DOC (M) Fluorescence (QSE) TB01 27.896 82.464 26.8 1.41 18 1.8 2.4 280 26 9.2 16.06 14 8.1 30.8 TB02 27.883 82.421 26.8 1.53 18 2.8 3.2 263 27 12.6 10.99 14 11.8 45.2 TB17 27.860 82.453 27.3 1.50 16 5.0 7.9 N.D. 24 8.8 16.81 14 11.5 79.1 N.D. N.D. Dry Season Average 27.0 (0.3) 1.48 (0.06) 17 (1) 3.2 (1.6) 4.5 (3.0) 272 (11) 26 (12) Wet Season Average 10.2 (2.1) 14.62 (3.16) 13 (1) 10.5 (2.0) 51.7 (24.8) N.D N.D. 117
Middle Tampa Bay Stations Latitude (Degree) Longitude (Degree) Salinity a CDOM (400) (m -1 ) Spectral Slope (10 -3 nm -1 ) TSS (mg/l) Chl (mg/m 3 ) DOC (M) Fluorescence (QSE) TB03 27.667 82.565 31.5 0.61 18 2.0 2.3 232 13 N.D. N.D. N.D. N.D. N.D. N.D. N.D. TB04 27.760 82.576 28.5 1.27 13 3.6 2.3 291 20 17.2 6.48 16 1.9 4.7 TB05 27.695 82.544 30.9 0.72 16 2.0 3.4 227 14 N.D. N.D. N.D. N.D. N.D. N.D. N.D. TB06 27.723 82.515 30.3 N.D. N.D. 3.2 2.4 N.D. N.D N.D. N.D. N.D. N.D. N.D. N.D. N.D. TB12 27.784 82.445 27.7 1.58 16 8.4 6.5 260 24 13.7 9.10 15 4.7 12.4 TB13 27.778 82.505 28.5 1.47 15 7.0 5.4 235 21 13.4 9.03 15 8.0 32.5 N.D. N.D. TB14 27.805 82.555 28.3 1.28 15 2.8 4.0 234 20 14.2 8.12 14 4.8 41.4 N.D. N.D. TB15 27.723 82.594 30.6 0.69 17 1.6 2.1 215 15 18.5 5.81 15 3.2 7.7 N.D. N.D. TB16 27.683 82.633 32.5 0.60 16 1.8 2.7 175 11 17.7 5.97 15 3.3 9.7 N.D. N.D. Dry season average 29.8(1.8) 1.03 (0.41) 15 (1) 3.7 (2.6) 3.5 (1.8) 234 (33) 17 (5) Wet season Average 15.8 (2.3) 7.42 (1.52) 15 (1) 4.3(2.1) 18.0 (15.1) N.D N.D. Appendix 3-1 (Continued) 118
Old Tampa Bay Stations Latitude (Degree) Longitude (Degree) Salinity a CDOM (400) (m -1 ) Spectral Slope (10 -3 nm -1 ) TSS (mg/l) Chl (mg/m 3 ) DOC (M) Fluorescence (QSE) TB07 27.984 82.657 24.4 1.13 23.0 9.6 5.4 N.D. 33 13.0 5.00 17.5 7.1 25.1 N.D. N.D. TB08 27.942 82.670 24.3 1.19 20.5 5.2 5.9 447 32 14.2 3.39 17.5 3.5 6.7 N.D. N.D. TB09 27.952 82.599 25.2 1.16 18.6 6.6 6.6 374 28 15.3 3.57 18.4 8.1 10.4 N.D. N.D. TB10 27.891 82.598 26.4 0.75 24.0 6.8 7.8 356 26 15.2 3.65 17.8 15.1 19.6 N.D. N.D. TB11 27.879 82.562 27.1 0.92 21.7 6.0 7.3 326 24 16.8 4.65 17.8 10.4 26.9 N.D. N.D. Dry Season Average 25.5(1.2) 1.03 (0.19) 21.6 (2.1) 6.8(1.7) 6.5 (0.9) 375 (51) 28 (3.6) Wet Season Average 14.9(1.4) 4.05 (0.72) 17.8 (0.4) 8.8 (4.3) 17.7 (8.9) N.D. N.D. Appendix 3-1 (Continued) 119
Tampa Bay Average Stations Latitude Longitude Salinity a CDOM (400) (m -1 ) Spectral Slope (10 -3 nm -1 ) TSS (mg/l) Chl (mg/m 3 ) DOC (M) Fluorescence (QSE) Dry Season Average 27.9 (2.4) 1.11 (0.35) 18 (3) 4.6 (2.6) 4.6 (2.2) 230 (72) 22 (7) Wet season Average 14.3 (2.8) 7.76 (4.33) 16 (2) 7.2 (3.9) 25.1 (20.4) N.D. N.D. The Alafia River Stations Latitude Longitude Salinity a CDOM (400) (m -1 ) Spectral Slope (10 -3 ,nm -1 ) TSS (mg/l) Chl (mg/m 3 ) DOC (M) Fluorescence (QSE) AR 1 27.870 82.314 1.1 6.96 17 4.4 3.6 516 91 AR 2 27.861 82.320 1.7 5.88 16 3.8 5.0 390 83 AR 3 27.868 82.328 4.1 5.74 16 6.0 12.5 425 77 AR 4 27.858 82.348 7.4 5.72 14 6.0 16.1 366 68 AR 5 27.857 82.362 12.1 4.78 15 5.4 10.4 391 58 AR 6 27.859 82.389 18.1 3.01 17 18.2 18.3 398 45 Dry Season Average 7.4 (6.6) 5.35 (1.34) 16 (1) 7.3 (5.4) 11.0 (5.9) 414 (56) 70 (17) Appendix 3-1 (Continued) Notes: (1) CDOM absorption spectral slope (S) was determined over the wavelength range 350 440 nm. (2) N.D. = not determined. 120 (3) The numbers in parentheses in each average row represent standard deviations.
About the Author Zhiqiang Chen was enrolled in the Colle ge of Marine Science (CMS) at the University of South Florida (USF) in Augus t, 2001. Before coming to USF, he obtained his B.S. of marine chemistry at Xiamen Un iversity in Fujian, China and his M.S. of Chemical Oceanography at the Second In stitute of Oceanography (SIO) in Hangzhou, China. The study at Institute of Marine Remote Sensing (MaRS) at USF provides him opportunity of receiving extensive trai ning in radiative transfer theory in situ and lab experiments and other oceanographic topi cs. In 2003, he was awarded the U.S. Geological Study (USGS)-USF cooperative rese arch assistantship. Since then he has focused on monitoring water quality in Tamp a Bay using the lastest remote sensing techniques. During his study pe riod, he published 5 papers.