ARTICLE IN PRESS

0278-4343/$ - see

doi:10.1016/j.csr

�Correspondifax: +1401 782

E-mail addre

jay.oreilly@noa

(C.A. Oviatt).

Continental Shelf Research 27 (2007) 1677–1691

www.elsevier.com/locate/csr

Validation of SeaWiFS chlorophyll a in Massachusetts Bay

Kimberly J.W. Hydea,�, John E. O’Reillya, Candace A. Oviattb

aNOAA, National Marine Fisheries Service, 28 Tarzwell Drive, Narragansett, RI 02886, USAbGraduate School of Oceanography, University of Rhode Island, South Ferry Road, Narragansett, RI 02886, USA

Received 10 August 2006; received in revised form 25 January 2007; accepted 5 February 2007

Available online 28 February 2007

Abstract

Massachusetts Bay, a semi-enclosed embayment (50� 100 km) in the Northwest Atlantic, is the focus of a monitoring

program designed to measure the effects of relocating the Boston Harbor sewage outfall to a site 15 km offshore in

Massachusetts Bay. The Massachusetts Water Resources Authority (MWRA) in situ monitoring program samples selected

stations up to 17 times per year to observe seasonal changes in phytoplankton biomass and other water quality variables.

We investigated the feasibility of augmenting the monitoring data with satellite ocean color data to increase the spatial and

temporal resolution of quantitative phytoplankton measurements. In coastal regions such as Massachusetts Bay, ocean

color remote sensing can be complicated by in-water constituents whose concentrations vary independently of

phytoplankton and by inaccurate modeling of absorbing aerosols that tend to be concentrated near the coast. An

evaluation of in situ and sea-viewing wide field-of-view sensor (SeaWiFS) measurements from 1998 to 2005 demonstrated

that SeaWiFS overestimated chlorophyll a mainly due to atmospheric correction errors that were amplified by absorption

from elevated concentrations of chlorophyll a and colored dissolved organic matter. Negative water-leaving radiances in

the 412 nm band, an obvious artifact of inadequate atmospheric correction, were recorded in approximately 60–80% of the

cloud-free images along the coast, while the remaining portions of the Bay only experience negative radiances 35–55% of

the time with a clear nearshore to offshore decrease in frequency. Seasonally, the greatest occurrences of negative 412 nm

radiances were in November and December and the lowest were recorded during the summer months. Concentrations of

suspended solids in Massachusetts Bay were low compared with other coastal regions and did not have a significant impact

on SeaWiFS chlorophyll a measurements. A regional empirical algorithm was developed to correct the SeaWiFS data to

agree with in situ observations. Monthly SeaWiFS composites illustrated the spatial extent of a bimodal seasonal pattern,

including prominent spring and fall phytoplankton blooms; and the approximate 115 cloud-free scenes per year revealed

interannual variations in the timing, magnitude and duration of phytoplankton blooms. Despite known artifacts of

SeaWiFS in coastal regions, this study provided a viable chlorophyll a product in Massachusetts Bay that significantly

increased the spatial and temporal synoptic coverage of phytoplankton biomass, which can be used to gain a

comprehensive ecosystem-wide understanding of phytoplankton dynamics at event, seasonal and interannual timescales.

r 2007 Elsevier Ltd. All rights reserved.

Keywords: Ocean color remote sensing; Phytoplankton; Chlorophyll algorithm; SeaWiFS; Coastal remote sensing; Massachusetts Bay

front matter r 2007 Elsevier Ltd. All rights reserved

.2007.02.002

ng author. Tel.: +1401 782 3226;

3201.

sses: kimberly.hyde@noaa.gov (K.J.W. Hyde),

a.gov (J.E. O’Reilly), coviatt@gso.uri.edu

1. Introduction

Coastal and estuarine ecosystems typicallyexhibit high temporal and spatial variability in

.

ARTICLE IN PRESSK.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–16911678

phytoplankton biomass that is often too difficult tocharacterize with a limited set of in situ shipboardmeasurements. Continuous monitoring of spatialand seasonal patterns of near-surface phytoplank-ton concentrations can only realistically be achievedby ocean color remote sensing which providessynoptic spatial and temporal chlorophyll a datanot attainable by in situ sampling (Morel andBerthon, 1989; O’Reilly et al., 1998; Thomaset al., 2003; Harding et al., 2005). However,ocean color remote sensing in continental shelfand coastal regions such as Massachusetts Bay canbe complicated by constituents in the water whoseconcentrations vary independently of phytoplank-ton abundance and by absorbing aerosols that tendto be concentrated near the coast (IOCCG, 2000;O’Reilly and Yoder, 2003; Schollaert et al., 2003).

The sea-viewing wide field-of-view sensor (SeaWiFS)on board the OrbView-2 spacecraft was specificallydesigned to estimate the phytoplankton pigmentconcentration in the oceans. It has been operationalsince September, 1997 and has provided globalestimates of oceanic chlorophyll a and other bio-optical data that have increased the understandingof temporal variability of marine ecosystems andthe role of oceanic photosynthesis and primaryproductivity in the Earth’s carbon budget andclimate (Falkowski et al., 1998). SeaWiFS wasdesigned to measure chlorophyll a concentrationsin clear-water (Case 1) regions in the range of0.01–64mgm�3 to within 35% (Hooker andMcClain, 2000; Lavender et al., 2004). Coastalregions and lakes often have more complex bio-optical properties compared to the open ocean andare thus classified as ‘‘Case 2’’ waters. Bio-opticalcomplexity and inaccurate atmospheric modelinghave created three challenges to ocean color remotesensing in coastal regions: (1) interference of thechlorophyll signal due to increased concentrationsof optical constituents in the water that varyindependently of phytoplankton biomass (Moreland Prieur, 1977; IOCCG, 2000); (2) failure of theatmospheric correction model due to increasedturbidity in the water (Wang, 2000); and (3) failureof the atmospheric correction model due toabsorbing aerosols in the atmosphere (O’Reillyand Yoder, 2003; Schollaert et al., 2003; Stumpfet al., 2003).

The overall impact of absorbing aerosols andadditional in-water bio-optical constituents, such ascolored dissolved organic matter (CDOM, alsocalled gelbstoff, yellow substances, or gilvin) and

total suspended solids or particulate matter (TSS), isto decrease the reflectance band ratio (Rmax) used inthe current SeaWiFS ocean chlorophyll 4 algorithm(OC4). This decrease of Rmax ultimately causesan overestimation of chlorophyll a concentra-tions. OC4 is a modified cubic polynomial relatingthe ratio of the reflectance in the blue (wave-length ¼ 443, 490 or 510 nm) and green (wave-length ¼ 555 nm) bands to the chlorophyll a

concentration:

Chl a ¼ 10:0ð0:366�3:067Rmaxþ1:930R2maxþ0:649R3

max�1:532R4maxÞ,

(1a)

where Chl a is the chlorophyll a concentration andRmax is the maximum band ratio determined inEq. (1b) from the remote sensing reflectance (RRS)ratios at the given wavelengths (O’Reilly et al., 1998,2000).

Rmax ¼ log10RRS443

RRS5554

RRS490

RRS5554

RRS510

RRS555

� �.

(1b)

The maximum reflectance band ratio changes as afunction of chlorophyll a concentration, whereRRS443/RRS555 is maximal from 0 to �0.3mgm�3;RRS490/RRS555 commonly dominates at chloro-phyll a concentrations between 0.3 and�1.5mgm�3; and when concentrations exceed�1.5mgm�3, the RRS510/RRS555 ratio is greatest(O’Reilly et al., 1998). Dissolved and suspendedmatter in productive and turbid coastal watersincrease both absorption and scattering of light andinfluences the optical signal received by the satellitesensor (Morel and Prieur, 1977; IOCCG, 2000).Inaccurate modeling of aerosols during the atmo-spheric correction process will lower the remotelysensed reflectance ratio because more light at thetop of the atmosphere (TOA) is attributed to theatmosphere, resulting in decreased water-leavingradiances with increasing severity in the lowerwavelengths (Gordon et al., 1997; O’Reilly andYoder, 2003; Schollaert et al., 2003; Stumpf et al.,2003). In coastal waters this may cause negativewater-leaving radiances in the blue bands and resultin errors in the calculation of chlorophyll a (Stumpfet al., 2003). One advantage of using the OC4algorithm is that at the higher concentrations ofchlorophyll a expected in Massachusetts Bay thepredominant Rmax ratio of 510/555 is less affectedby atmospheric correction errors due to absorbingaerosols (O’Reilly et al., 1998; Harding et al., 2005).

ARTICLE IN PRESSK.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–1691 1679

Despite these known limitations in coastal remotesensing, numerous researchers have developedregional algorithms and models to estimate phyto-plankton pigments as well as suspended particulateand color dissolved organic matter concentrationsin coastal regions such as: the Bay of Biscay(Froidefond et al., 2002; Lavender et al., 2004),the Baltic Sea (Ohde et al., 2002; Darecki et al.,2003; Kowalczuk et al., 2005), coastal Ireland(Mitchelson et al., 1986; Darecki et al., 2003), theWhite Sea (Pozdnyakov et al., 2003), the Gulf ofNaples (Tassan, 1994) and the Eastern Arabian Sea(Desa et al., 2001). In addition, refinements andimprovements to the atmospheric correction processin the fourth and fifth SeaWiFS reprocessing yieldeda reduction in the frequency of pixels with negativewater-leaving radiances in the blue wavelengthbands and have increased the accuracy of SeaWiFSin coastal regions (O’Reilly and Yoder, 2003).

The SeaWiFS sensor can be expected to provideup to 100 relatively cloud-free scenes per year in theNorthwest Atlantic (Harding et al., 2005) andsupplement the Massachusetts Water ResourcesAuthority (MWRA) efforts to capture the seasonalfluctuations of phytoplankton biomass. The currentSeaWiFS algorithm, OC4, is presumed to over-estimate chlorophyll a concentrations in Massachu-setts Bay; however, a thorough evaluation with anextensive dataset has not been performed. Theobjective of this study is to validate SeaWiFSchlorophyll a in Massachusetts Bay and to evaluatethe impact of suspended solids and atmosphericcorrection errors to determine under what condi-tions SeaWiFS measurements are less accurate forthe Massachusetts Bay region. The overall goal is toassess whether SeaWiFS can be used in conjunctionwith in situ monitoring data to develop a morecomprehensive characterization and understandingof spatial, seasonal and interannual variations inphytoplankton biomass in Massachusetts Bay.

2. Methods

Massachusetts and Cape Cod Bays and theadjacent Boston Harbor are the locations of anintensive study conducted by the MWRA designedto assess the effects of relocating the Deer IslandWater Treatment Plant sewage effluent dischargefrom the original discharge site in Boston Harbor toa location 15 km offshore in the Massachusetts Bay.Assessment of the existing conditions began in 1992,and continued monitoring is designed to study the

impacts of the new outfall, which began dischargingon September 6, 2000 (Albro et al., 1998; Libbyet al., 2002). Monitoring data at 15 stationscollected between 1998 and 2005 coincides withthe first eight full years of SeaWiFS and were usedto validate SeaWiFS products in Massachusetts Bay(Fig. 1).

2.1. In situ measurements

Duplicate 25–400mL chlorophyll a samples werevacuum-filtered on glass fiber filters (Whatman47mm-diameter GF/F filters). Two drops of asaturated solution of magnesium carbonate(MgCO3) was added during filtration and thesamples were frozen until analysis. Pigment sampleswere extracted by mechanical grinding followed bya 2–24 h steep in 90% acetone at �20 1C. Samplescollected in 2003 were extracted by soaking wholefilters in 90% buffered acetone for 24 h prior toanalysis. The extract was centrifuged and analyzedusing a Turner Designs Fluorometer (Yentsch andMenzel, 1963; Arar and Collins, 1997). Laboratorycalibration errors and degraded standards causedinaccurate chlorophyll a calculations of samplescollected from 1998 to 2000 and in February andMarch 2003. Samples were accordingly correctedand qualified ‘‘use with caution’’ (Libby et al.,2001), however no significant difference was foundif the data were eliminated from the SeaWiFScomparison analysis.

Total suspended solids were measured by vacuumfiltering duplicate 300mL aliquots of whole watersamples through tared 0.4 mm pore size Nuclepore47mm-diameter membrane filters. Filters were driedin a desiccator for a minimum of 7 days, and thenweighed using an ATI CAHN Model C-44 Toploading microbalance (Albro et al., 1998; Libbyet al., 2002).

2.2. Optically-weighted chlorophyll a concentration

For remote sensing purposes, the penetrationdepth of light in the sea is defined as the depthabove which 90% of diffusely reflected irradianceoriginates (Gordon and McCluney, 1975). Thepenetration depth (Z90) is less than 25% ofthe thickness of the euphotic depth (Zeu), thus thepigment concentration estimated by the satellitesensor is not the concentration within the euphoticdepth, but rather the concentration within thepenetration depth. During the summer stratified

ARTICLE IN PRESS

Fig. 1. Location of the MWRA stations where chlorophyll a samples were collected. Data obtained from the Office of Geographic and

Environmental Information (MassGIS), Commonwealth of Massachusetts Executive Office of Environmental Affairs (http://

www.mass.gov/mgis/massgis.htm).

K.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–16911680

season, the subsurface chlorophyll maximum typi-cally occurs within the euphotic zone, but below thepenetration depth and may be partially or comple-tely invisible to a remote sensor (Sathyendranathand Platt, 1989; Ballestero, 1999). To estimate the in

situ chlorophyll a concentration observed by thesatellite, in situ chlorophyll a concentrations wereoptically-weighted to the penetration depth (Eq. (2))(Gordon and Clark, 1980)

Chlwt ¼

R z900 ChlðzÞf ðzÞdzR z90

0 f ðzÞdz, (2a)

Where Chlwt is the optically-weighted chlorophyll a

concentration (mgm�3), Chl (z) is the chlorophyll a

concentration at depth z and f (z) is given by

f ðzÞ ¼ exp �2

Z z

0

kdðzÞdz0� �

, (2b)

where kd (z) is the attenuation coefficient for thepenetration depth.

2.3. Satellite-derived observations

Individual daily merged local area coverage(MLAC) SeaWiFS images of the Northwest

ARTICLE IN PRESS

0

60

120

180

240

300

360

420

480

540

600

Fre

qu

ency

0

10

20

30

40

50

60

70

80

90

100

Cu

mu

lati

ve

%

0.01 0.1 1 10 100

Chlorophyll a (mg m-3)

Fig. 2. Log-normal distribution of in situ chlorophyll a samples

collected at all depths and stations (n ¼ 6773 , geometric

mean ¼ 1.53mgm�3).

K.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–1691 1681

Atlantic were mapped at 1 km pixel resolution forthe Massachusetts Bay coast using a Lambert conicconformal map projection. SeaWiFS MLAC consoli-dates all available LAC from a single orbit andoverlapping scenes from multiple HRPT (high resolu-tion picture transmission) stations are evaluated toselect a single ‘‘best quality’’ scan and increase thesignal to noise ratio. Chlorophyll a was derived usingSeaWiFS version 5.1 standard processing. Using thecoordinates for each in situ sampling station, a 5� 5array of pixels was extracted from the images taken onthe same day of the in situ chlorophyll a sampling. 64%of the images were collected within 3h of in situ

sampling and 95% were within 5h. The geometricmeans of chlorophyll a arrays with eleven or more(40%) cloud-free pixels were calculated and comparedto in situ chlorophyll a samples. The geometric mean isthe arithmetic average of the log-transformed data thatis then untransformed. The geometric mean wasselected rather than the arithmetic mean becausechlorophyll a concentrations are commonly log-nor-mally distributed and the geometric mean has theadvantage of reducing the influence of a small numberof high or low values (outliers) on the mean (Campbell,1995; Bricaud et al., 2002; Yoder et al., 2002; Yoderand Kennelly, 2003).

2.4. Environmental data

The Charles and Merrimack Rivers are the twoprincipal freshwater sources influencing Massachu-setts Bay. Daily stream flow records and long-termmeans were obtained from the United StatesGeological Survey for both rivers (http://waterdata.usgs.gov/nwis). The water gage for theCharles River is located on the Moody Street Bridgein Waltham, Massachusetts and the Merrimackgage is 335m downstream of the Concord River inLowell, Massachusetts. Daily precipitation is mea-sured at locations in Middlesex County (42.3839—latitude, �71.2147—longitude) and BarnstableCounty (41.9758—latitude, �70.0247—longitude)by the National Atmospheric Deposition Program/National Trends Network (http://nadp.sws.uiuc.edu/).

3. Results and discussion

3.1. Chlorophyll a

In situ chlorophyll a concentrations in Massachu-setts Bay between January 1998 and December 2005

were log-normally distributed. The geometric meanof the in situ concentration for all depths sampled(n ¼ 6773) was 1.5mgm�3. The minimum andmaximum concentrations were 0.003–44mgm�3,respectively; however 98% of the chlorophylla measurements ranged from 0.11 to 13mgm�3

(Fig. 2). The shallowest chlorophyll a sample foreach profile was, on average, collected at 2m belowthe surface and never greater than 4m. In compar-ison, the penetration depth, or rather the maximumdepth observed by the satellite sensor, ranged from1 to greater than 15m (mean ¼ 4.3m), indicatingthat the shallowest measurement may not always bethe best representation of the in situ chlorophyll a

concentration observed by the satellite sensor due tothe possible presence of a sub-surface chlorophyllmaximum.

Approximately three-quarters (73%) of the in situ

profiles had a subsurface maximum chlorophyll a

concentration at an average depth of 15m. How-ever, the ratio between the maximum and thesurface concentration was less than 1.5 in morethan 50% of the subsurface chlorophyll maximumprofiles. Furthermore, the mean concentrations ofboth the surface and optically-weighted penetrationdepth chlorophyll a were similar (1.9mgm�3), yetslightly higher than the mean for all depths.Similarly, there was a strong relationship betweenthe surface and optically-weighted chlorophyll a

concentrations (intercept ¼ 0.009, slope ¼ 0.986,r2 ¼ 0.999), suggesting that the weighted estimatewas rarely influenced by a subsurface chlorophyllmaximum. Despite the close relationship betweenthe surface and optically-weighted chlorophyll a

concentrations, the optically-weighted value was

ARTICLE IN PRESSK.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–16911682

used for all comparisons to SeaWiFS chlorophyll a

in order to represent best the concentrationsmeasured by the satellite sensor. Further referencesto in situ chlorophyll a refer to the optically-weighted chlorophyll a concentration.

There were over 400 match-ups between in situ

samples and cloud-free SeaWiFS images takenwithin 12 h of the sample time from 1998 to 2005.The geometric mean SeaWiFS chlorophyll a con-centration of the match-ups was 3.11mgm�3 andwas greater than the corresponding in situ meanconcentration of 2.02mgm�3. The SeaWiFS match-ups were compared to the optically-weighted in situ

chlorophyll a concentration and statistically eval-uated using a Type II reduced major axis linearregression (Laws and Archie, 1981; Press andTeukolsky, 1992) of the log10-transformed data(Fig. 3). The match-up analysis showed a positiverelationship between the SeaWiFS and in situ

chlorophyll a, however SeaWiFS clearly overesti-mated chlorophyll a in Massachusetts Bay at lowerconcentrations (Fig. 3).

3.2. Possible sources of error

The SeaWiFS chlorophyll a match-up resultswere similar to those observed by Harding et al.(2005) who demonstrated that SeaWiFS overesti-mates chlorophyll a in Chesapeake Bay and the MidAtlantic Bight due to in-water optical constituentsand atmospheric correction errors. Overestimationerrors in Chesapeake Bay decreased along thesalinity gradient as the distance increased from the

0.1 1 10 100

0.1

1

10

100

Sea

WiF

S C

hlo

rophyll

a

Int: 0.248Slope: 0.832Rsq: 0.387

Fig. 3. Type II linear regression analysis of the optically-

weighted chlorophyll a and the mean of the SeaWiFS extract

with 11 or more cloud-free pixels from the 5� 5 array from

images collected on the same day as the in situ sampling

(n ¼ 426). The dashed line represents the one to one line.

Susquehanna River, which is a large source ofsuspended solids and CDOM (Harding et al., 2005).Suspended solids, as well as terrigenous CDOM, aretransported to coastal regions by riverine inputs.Unlike Chesapeake Bay, there are no major riversdirectly discharging into Massachusetts Bay, butrather two indirect freshwater sources. The Gulf ofMaine Rivers, dominated by the Merrimack River,discharge roughly 1100m3 s�1 and merge with theGulf of Maine Coastal Current. The current flowssouth along the Maine and New Hampshire coastsand branches at Cape Ann, north of MassachusettsBay. Much of the flow follows the topographysouthward past Stellwagen Bank and to the east ofCape Cod, however a weaker branch does enterMassachusetts Bay (Signell et al., 2000). BostonHarbor, fed by the Charles and other rivers, is amore localized freshwater source discharging ap-proximately 41m3 s�1, half of which was suppliedby sewage outfalls prior to the relocation of theMWRA outfall on September 6, 2000 (Signell et al.,2000). Boston Harbor also acts as a long-term sinkfor fine grain sediments because of its restrictedflushing and low-wave energy environment (Butmanet al., 2002). As a result of the indirect nature of thefreshwater inputs, surface suspended solid concen-trations in Massachusetts Bay were low (mean-0.98mgL�1, max ¼ 4.8mgL�1, Fig. 4) compared toother coastal regions such as, the Santa BarbaraChannel (0–200mgL�1) (Warrick et al., 2004), SanFrancisco Bay (5–30mgL�1) (Cloern et al., 1989),East China Sea (100–1000mgL�1) (Deng and Li,2003), and the northern Gulf of Mexico along theLouisiana coast (100–400mgL�1) (Myint and

0 1 2 3 4 5

Total Suspended Solids (mg L-1)

0

10

20

30

40

50

60

70

80

90

100

Fre

quen

cy

0

10

20

30

40

50

60

70

80

90

100C

um

ula

tive

%

Fig. 4. Frequency diagram of surface total suspended solid

concentrations observed in Massachusetts Bay (n ¼ 959, med-

ian ¼ 0.84mgL�1, and mean ¼ 0.98mgL�1). The maximum

concentration measured was 4.8mgL�1.

ARTICLE IN PRESS

-1.0 -0.5 0.0 0.5 1.0 1.5 2.0

LWN (412) (µW cm-2 sr-1 nm-1)

0.0

0.2

0.4

0.6

0.8

1.0

Rel

ativ

e F

requen

cy

0.0

0.2

0.4

0.6

0.8

1.0

Cum

ula

tive

Fre

quen

cy

Fig. 6. The relative and cumulative frequency of 412 nm water-

leaving radiances (n ¼ 8488 images, mean ¼ 0.153mWcm�2

sr�1 nm�1). The gray solid line represents the relative frequency

and the dashed black line is the cumulative frequency. 23% of the

measured radiances values were less than or equal to

0mWcm�2 sr�1 nm�1 (the vertical dashed line).

K.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–1691 1683

Walker, 2002). There were only 32 SeaWiFS in situ

match-ups when suspended solid concentrationswere greater than 2mgL�1 (Fig. 5). A Student’st-statistic (Zar, 1999) of the slope and interceptshowed no significant difference in the SeaWiFS toin situ chlorophyll a regression when the high TSSsamples (42mgL�1) were removed.

When concentrations of chlorophyll a andCDOM are high, water-leaving radiances (Lwn) inthe 412 nm band approach zero and may even beslightly negative because of greater absorption ofblue wavelengths by the in-water constituents. Themean Lwn (412) for the Massachusetts Bay regionwas 0.153 (n ¼ 8488 images), however 23% of themeasured radiances were p0 (Fig. 6). Negative Lwn

in the 412 nm band are also an indication ofinaccurate SeaWiFS atmospheric correction be-cause they occur when too much of the TOA signalis attributed to the atmosphere, due to inaccurateaerosol modeling or failure of the black pixelassumption (O’Reilly and Yoder, 2003; Schollaertet al., 2003). Thus, when the atmospheric portion issubtracted from the TOA signal, the result isreduced or negative values from the water and anoverestimation of chlorophyll a. Along the Massa-chusetts Bay coast (depth less than 25m and within7 km of the shore), negative water-leaving radianceswere recorded in approximately 60–80% of thecloud-free images from 1998 to 2005 (Fig. 7). Theremaining portions of the Bay only experiencednegative radiances 35–55% of the time with a clearnearshore to offshore decrease in the frequency of

0.1 1 10 100

in situ Chlorophyll a (mg /m3)

0.1

1

10

100

Sea

WiF

S C

hlo

rop

hy

ll a

(m

g m

-3) Int: 0.238

Slope: 0.828Rsq: 0.410

TSS < 2 mg L-1

TSS > 2 mg L-1

Fig. 5. SeaWiFS versus in situ chlorophyll a comparison of when

total suspended solids (TSS) were measured (n ¼ 310, no TSS

data were available prior to 2000). The black circles (n ¼ 32)

represent suspended solid concentrations greater than 2mgL�1.

There was no significant difference in the regression when the

high TSS samples were eliminated from the analysis.

negative radiances. Monthly composites of the 8-year study period revealed that the highest occur-rences of negative 412 nm radiances were inNovember and December, while the lowest frequen-cies were during the summer months (Fig. 7).

This seasonality of the negative radiances andatmospheric correction errors is most likely to beexplained by a combination of inaccurate atmo-spheric correction modeling and the presence ofblue wavelength absorbers in the water, such asphytoplankton and CDOM. Measurements ofCDOM absorption were not available for thisstudy, however it can be assumed that CDOMconcentrations fluctuate as a function of chlorophylla concentration and/or river runoff (IOCCG, 2000;Twardowski and Donaghay, 2001). The absence ofa direct freshwater source into Massachusetts Baysuggests that concentrations of the allochthonousCDOM, which is conservative and covaries withfreshwater input and salinity, will be lower thanother coastal ecosystems with direct freshwater flowsuch as Chesapeake Bay. The nonconservativeportion of CDOM is a by-product of phytoplank-ton production and for remote sensing purposes isassumed to linearly vary with chlorophyll a

(Twardowski and Donaghay, 2001).A joint probability distribution of the monthly

SeaWiFS chlorophyll a and percent frequency ofnegative 412 nm radiances explains a majorityof the observed seasonality. All months demon-strated a positive relationship between the percentfrequency of negative radiances and chlorophyll a

ARTICLE IN PRESS

1998-2005

0 20 40 60 80 100Percent (%)

Fig. 7. The percent frequency of 8488 images collected from 1998 to 2005, and the monthly percent frequency when the water-leaving

radiance in the 412 nm band is negative. A negative 412 nm radiance is an indication of a failure of the atmospheric correction algorithm.

The greatest percentage of negative radiances was recorded along the coast (up to 80%) and decreased offshore. The total number of

images for January ¼ 718, February ¼ 659, March ¼ 723, April ¼ 710, May ¼ 711, June ¼ 702, July ¼ 724, August ¼ 721,

September ¼ 714, October ¼ 730, November ¼ 665 and December ¼ 713.

K.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–16911684

ARTICLE IN PRESS

0

20

40

60

80

100

Per

cent

Neg

ativ

e L

wn 4

12

0.1 0.3 1 3 10

Int : 28.11Slope: 44.16Rsq: 0.45

JAN

0.1 0.3 1 3 10

Int: 24.32Slope: 52.19Rsq: 0.59

FEB

0.1 0.3 1 3 10

0

20

40

60

80

100Int: 16.53Slope: 52.23Rsq: 0.64

MAR

0

20

40

60

80

100

Per

cent

Neg

ativ

e L

wn 4

12

Per

cent

Neg

ativ

e L

wn 4

12

Per

cent

Neg

ativ

e L

wn 4

12

Int: -1.83Slope: 116.64Rsq: 0.74

APR

Int: 9.00Slope: 122.29Rsq: 0.81

MAY 0

20

40

60

80

100Int: 14.49Slope: 85.12Rsq: 0.82

JUN

0

20

40

60

80

100Int: 4.42Slope: 82.59Rsq: 0.72

JUL

Int: 6.89Slope: 74.36Rsq: 0.69

AUG 0

20

40

60

80

100Int: 12.03Slope: 81.21Rsq: 0.83

SEP

0.1 0.3 1 3 10

Chlorohyll a (mg/m3) Chlorohyll a (mg/m3)

0

20

40

60

80

100Int: 6.15Slope: 93.52Rsq: 0.87

OCT

Int: 30.20Slope: 77.75Rsq: 0.76

NOV

Percent Frequeny (%)

0.01 0.1 1

0.1 0.3 1 3 10

0

20

40

60

80

100Int: 46.01Slope: 62.76Rsq: 0.71

DEC

Fig. 8. Monthly joint probability distribution of the percent frequency of negative 412 nm water-leaving radiances and Log10 SeaWiFS

chlorophyll a.

K.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–1691 1685

concentrations. From April through December thecoefficient of determination (R2) was greater than�0.70 (Fig. 8). November and December had

moderate concentrations of chlorophyll a (meangreater than 1.6mgm�3), yet higher than averagenegative frequencies, indicating that a combination

ARTICLE IN PRESS

0.1 1 10 100

in situ Chlorophyll a (mg m-3)

0.1

1

10

100

Cor

rect

ed S

eaW

iFS

Chl

orop

hyll

a Int: 0.000Slope: 1.000Rsq: 0.387

Fig. 10. Type II linear regression of the corrected SeaWiFS

chlorophyll a versus the in situ chlorophyll a (n ¼ 426). The

correction algorithm used the slope and intercept from the

regression statistics in Fig. 3 to calculate the corrected values.

K.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–16911686

of in-water absorbers and high solar zenithangles (approximately 57–65.61) yielded negativefrequencies up to 70% in some portions ofMassachusetts Bay. Conversely, the probabilitydistribution was more dispersed in January,February and March, when chlorophyll a concen-trations and river flow were relatively low andvariable, and the solar zenith angle was near itsseasonal maximum.

The use of SeaWiFS’s OC4 maximum band ratioalgorithm does have an advantage in that itconsistently selected the 510:555 nm ratio becausethe average chlorophyll a concentration in Massa-chusetts Bay was greater than 1.5mgm�3, therebyreducing the overestimation by not using the lowerwavelength radiances that are more adverselyaffected by atmospheric correction errors. Forthe 8-year study period (n ¼ 8488 images), the510:555 nm ratio dominated the maximum bandratio for more than 75% of the time in a majorityof the Massachusetts Bay region (Fig. 9(a)).The dominant band ratio shifted to 490:555 nmfurther offshore (Fig. 9(b)) and the 443:555 nm ratiorarely dominated in the Massachusetts Bay region(Fig. 9(c)).

3.3. SeaWiFS chlorophyll a correction

There were no specific instances when externalparameters such as suspended solids were able toaccount for the overestimation of chlorophyll a bySeaWiFS. However, a localized empirical correction

a b

Fig. 9. The percent of 8488 images when the maximum band ratio selec

555, and (c) 443/555.

algorithm, based on the type II linear regression,was developed to correct regionally SeaWiFSchlorophyll a in Massachusetts Bay (Eq. (3)). Thestandard SeaWiFS chlorophyll a product (Sea-WiFS(chlorophyll)) was corrected (SeaWiFS(corrected)),using the slope (b ¼ 0.832) and intercept (a ¼ 0.248)from Fig. 3.

SeaWiFSðcorrectedÞ ¼ 10

Log10 SeaWiFSðchlorophyllÞ � 0:248

0:832

� �

(3)

The algorithm adjusts the slope and intercept ofthe regression, however the R2 associated with the

c

ted for the OC4 chlorophyll a algorithm was (a) 510/555, (b) 490/

ARTICLE IN PRESS

a b

Fig. 11. The 1998–2005 mean chlorophyll a concentration of (a) SeaWiFS and (b) corrected SeaWiFS (n ¼ 8490). The black contour line

represents 1mgm�3. Note the colorbar is a log10 scale.

K.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–1691 1687

scatter does not change (Fig. 10). The highuncertainty was partially due to the mismatch insampling size between the in situ (o1m2) andsatellite (�25 km2) measurements and the small-scale, high-frequency variability in phytoplanktonbiomass common in Massachusetts Bay. The overallimpact of applying the correction algorithm was toreduce the overestimated SeaWiFS chlorophyll a toconcentrations typically observed in MassachusettsBay (Figs. 10 and 11). The geometric mean of theSeaWiFS images for the 426 match-ups decreasedfrom 3.11 (original data) to 2.11mgm�3 (correcteddata), which is similar to the correspondingoptically-weighted in situ mean (2.02mgm�3).

3.4. Seasonal and interannual variations of

chlorophyll a in Massachusetts Bay

SeaWiFS increased the spatial and temporalresolution in Massachusetts Bay over the MWRAsampling regime and the combination of in situ andremotely sensed data allowed for a more accurateand comprehensive assessment of phytoplanktondynamics. Spatially, chlorophyll a concentrationswere greater along the coast and in Cape Cod Bayand decreased with distance offshore (Fig. 11). Inaddition, there was a high degree of interannualvariability. The lowest mean concentration mea-sured by SeaWiFS at the MWRA stations (Fig. 1)was observed in 2004 (1.14mgm�3), while the

highest annual concentration for the region was in2000 (2.10mgm�3), followed by 1999 (1.98mgm�3)and 2005 (1.81mgm�3). Seasonally, the 8-yearmonthly means of phytoplankton biomass exhibiteda bimodal pattern similar to the seasonal cycledescribed by Cura (1991), Thomas et al. (2003) andOviatt et al. (2007) with peak spring concentrationsin April and fall maxima in October and November(Fig. 12, Table 1).

The sampling frequency increased from approxi-mately 125 in situ measurements to nearly 1000SeaWiFS measurements at the scale of a singlestation for the 8-year study period. The enhancedsampling revealed significant interannual variabilityin the magnitude, duration and timing of spring andfall phytoplankton blooms that were unresolvedwith in situ sampling (Fig. 13). Based on the in situ

data, the phytoplankton patterns in 1998 werecharacterized as atypical due to the absence of awinter–spring bloom (Keller et al., 2001; Oviattet al., 2007). However, our results based onSeaWiFS showed that a phytoplankton bloomoccurred in April 1998, but it was much lower inboth intensity and duration, relative to the winter–spring blooms observed in 1999, 2000, 2002, 2004and 2005 (Fig. 13). Several high intensity, shortduration blooms occurred between in situ measure-ments and were observed by SeaWiFS including afall bloom in September, 1998, and late spring/earlysummer blooms in June, 2000 and 2003 (Fig. 13).

ARTICLE IN PRESS

1998 1999 2000 2001

20052004 20032002

J F M A M J

DNOSAJ

0.1 0.3 1 3 10 30

Chlorophyll a (mg m-3)

Fig. 12. Annual and monthly corrected SeaWiFS chlorophyll a of Massachusetts Bay from 1998 to 2005. The black contour line

represents 1mgm�3. Note the colorbar is a log10 scale.

Table 1

Monthly mean and percentile (1, 25, 50, 75 and 99) SeaWiFS

chlorophyll a concentrations measured at station N04 from 1998

to 2005

Date N Mean 1% 25% 50% 75% 99%

January 91 0.85 0.26 0.67 0.86 1.05 2.86

February 84 0.96 0.33 0.60 0.91 1.34 9.81

March 84 1.61 0.38 0.98 1.34 2.85 12.36

April 70 2.58 0.56 1.47 2.25 4.28 16.76

May 57 1.89 0.54 1.25 1.84 2.84 6.14

June 60 1.40 0.32 0.98 1.35 2.01 9.83

July 78 1.27 0.41 0.89 1.28 1.79 4.71

August 94 1.52 0.45 0.99 1.34 2.07 8.23

September 110 2.18 0.30 1.01 2.36 4.52 20.37

October 89 3.16 0.42 1.57 3.45 6.47 20.59

November 59 2.11 0.44 1.12 1.92 3.33 25.83

December 64 1.17 0.37 0.84 1.13 1.64 7.97

Station N04 is located to the northwest of the outfall (Fig. 1).

K.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–16911688

In addition, the increased SeaWiFS resolutionunmasked rapidly varying chlorophyll a concentra-tions, which indicate that the spring and fall bloomperiods were not a single event as suggested by thein situ data, but rather a number of small pulses ofincreased phytoplankton biomass. These resultswere similar to findings of Flagg et al. (1994) whoreported that short-term, synoptic-scale phyto-plankton events were superimposed on theannual cycle and were often several times largerthan those seen on a seasonal scale in the Mid-Atlantic Bight. Moreover, the small pulses ofincreased phytoplankton biomass with concentra-tions greater than 5mgm�3 were also observedduring the summer months (June, August andSeptember; Table 1). These previously unreportedevent-scale pulses characterized by SeaWiFS are

ARTICLE IN PRESS

In situ Chlorophyll a (mg m-3)

1998

1999

2000

2001

2002

2003

2004

2005

Corrected SeaWiFS Chlorophyll a (mg m-3)

1998

1999

2000

2001

2002

2003

2004

J F M A M J J A S O N D J

J F M A M J J A S O N D J

2005

Fig. 13. (a) Annual integrated in situ chlorophyll a at station N04, (b) annual integrated corrected SeaWiFS chlorophyll a extracted at the

location of the in situ station N04. Tick marks indicate the sampling date. Station N04 is located to the northwest of the outfall (Fig. 1).

K.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–1691 1689

potentially significant to the ecology and waterquality of this region.

4. Conclusion

The purpose of this study was to assess theaccuracy of SeaWiFS chlorophyll a in Massachu-setts Bay with respect to in situ measurements andto evaluate possible error sources associated withthe overestimation of SeaWiFS in this coastalregion. Coastal regions are economically importantecosystems that typically express high temporal andspatial variability in phytoplankton biomass andproductivity that can easily be missed by in situ

sampling programs. Ocean color remote sensorsprovide unprecedented synoptic coverage of chlor-ophyll a that has increased the understanding of themagnitude and seasonal cycling of phytoplanktonbiomass and production, which is critical for thecomprehension of the general ecology of anecosystem. However, complex bio-optical waterproperties and inaccurate modeling of the nearshoreatmosphere often complicate the retrieval of accu-rate data from ocean color remote sensors in coastal

regions. This study demonstrated that SeaWiFSoverestimated chlorophyll a in the MassachusettsBay region and provided a simple empirical correc-tion scheme to adjust the SeaWiFS estimation. Thecorrected SeaWiFS accurately and reliably charac-terized the seasonal phytoplankton biomass inMassachusetts Bay revealing event, seasonal andinterannual variability and established that remo-tely sensed data can be an important component ofcoastal monitoring programs.

Acknowledgments

The Massachusetts Water Resources Authority(MWRA) supported in situ data collection andmanuscript preparation. Technical reports contain-ing water-quality data are available at http://www.mwra.state.ma.us/harbor/enquad/trlist.html.We thank personnel at Battelle Ocean Sciences,Duxbury, Massachusetts and the Marine EcosystemResearch Laboratory, University of Rhode Islandfor data collection and support. All in situ data wereused with permission from MWRA. SeaWiFS datasupport was also provided by Teresa Ducas,

ARTICLE IN PRESSK.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–16911690

NOAA—National Marine Fisheries Service, Nar-ragansett, Rhode Island, and James Yoder, Mau-reen Kennelly and Colleen Mouw, University ofRhode Island, Narragansett, Rhode Island. We alsothank Chuck McClain, Gene Feldman and JamesAcker, NASA—Goddard Space Flight Center, forSeaWiFS technical support and advice.

References

Albro, C.S., Trulli, H.K., Boyle, J.D., Sauchuk, S.A., Oviatt,

C.A., Keller, A.A., Zimmerman, C., Turner, J., Borkman, D.,

Tucker, J., 1998. Combined work/quality assurance plan for

baseline water quality monitoring: 1998–2000. Report EN-

QUAD ms-48, Massachusetts Water Resources Authority,

Boston.

Arar, E.J., Collins, G.B., 1997. In vitro determination of

chlorophyll a and phaeophytin a in marine and freshwater

phytoplankton by fluorescence. Method 445.0, Version 1.2,

US Environmental Protection Agency, Environmental Mon-

itoring Systems Laboratory, Office of Research and Devel-

opment, Cincinnati, OH.

Ballestero, D., 1999. Remote sensing of vertically structured

phytoplankton pigments. Topicos Meteorologicos Y Oceano-

graficos 6 (1), 14–23.

Bricaud, A., Bosc, E., Antoine, D., 2002. Algal biomass and sea

surface temperature in the Mediterranean Basin: intercom-

parison of data from various satellite sensors, and implica-

tions for primary production estimates. Remote Sensing of

Environment 81 (2–3), 163–178.

Butman, B., Bothner, M.H., Lightsom, F.L., Gutierrex, B.T.,

Soupy, A.P., Martini, M.A., Strahle, W.S., 2002. Long-term

oceanographic observations in western Massachusetts Bay

offshore of Boston, Massachusetts; data report for

1989–2000. In Data Report for 1989–2000 US Geological

Survey Digital Data Series 74, US Geological Survey, Woods

Hole.

Campbell, J.W., 1995. The lognormal distribution as a model for

bio-optical variability in the sea. Journal of Geophysical

Research 100 (C7), 13237–13254.

Cloern, J.E., Powell, T.M., Huzzey, L., 1989. Spatial and

temporal variability in South San Francisco Bay (USA). II.

Temporal changes in salinity, suspended sediments, and

phytoplankton biomass and productivity over tidal time

scales. Estuarine, Coastal and Shelf Science 28, 599–613.

Cura, J.J., 1991. Review of phytoplankton data: Massachusetts

Bay. Technical Report 91–1, MWRA, Boston.

Darecki, M., Weeks, A., Sagan, S., Kowalczuk, P., Kaczmarek,

S., 2003. Optical characteristics of two contrasting Case 2

waters and their influence on remote sensing algorithms.

Continental Shelf Research 23 (3–4), 237–250.

Deng, M., Li, Y., 2003. Use of SeaWiFS imagery to detect three-

dimensional distribution of suspended sediment. Interna-

tional Journal of Remote Sensing 24 (3), 519–534.

Desa, E., Suresh, T., Matondkar, P.S.G., Desa, E., 2001. Sea

truth validation of SeaWiFS ocean colour sensor in the

coastal waters of the Eastern Arabian Sea. Current Science 80

(7), 854–860.

Falkowski, P.G., Barber, R.T., Smetacek, V., 1998. Biogeochem-

ical controls and feedbacks on ocean primary production.

Science 281 (5374), 200–206.

Flagg, C.N., Wirick, C.D., Smith, S.L., 1994. The interaction of

phytoplankton, zooplankton and currents from 15 months of

continuous data in the Mid-Atlantic Bight. Deep Sea

Research II 41 (2/3), 411–435.

Froidefond, J.M., Lavender, S.J., Labordes, P., Herbland, A.,

Lafon, V., 2002. SeaWiFS data interpretation in a coastal

area in the Bay of Biscay. International Journal of Remote

Sensing 23 (5), 881–904.

Gordon, H.R., Clark, D.K., 1980. Remote sensing optical

properties of a stratified ocean: an improved interpretation.

Applied Optics 19 (20), 3426–3428.

Gordon, H.R., McCluney, W.R., 1975. Estimation of the depth

of sunlight penetration in the sea for remote sensing. Applied

Optics 14 (2), 413–416.

Gordon, H.R., Du, T., Zhang, T., 1997. Remote sensing of ocean

color an aerosol properties: resolving the issue of aerosol

absorption. Applied Optics 36 (33), 8670–8684.

Harding, J., Lawrence, W., Magnuson, A., Mallonee, M.E.,

2005. SeaWiFS retrievals of chlorophyll in Chesapeake Bay

and the mid-Atlantic bight. Estuarine, Coastal and Shelf

Science 62 (1–2), 75–94.

Hooker, S.B., McClain, C.R., 2000. The calibration and

validation of SeaWiFS data. Progress in Oceanography 45,

427–465.

IOCCG, 2000. Remote sensing of ocean colour in coastal, and

other optically-complex waters. In: Sathyendranath, S. (Ed.),

Reports of the International Ocean-Colour Coordinating

Group, No. 3. IOCCG, Dartmouth, Canada.

Keller, A.A., Taylor, C., Oviatt, C.A., Dorrington, T., Hol-

combe, G., Reed, L.W., 2001. Phytoplankton production

patterns in Massachusetts Bay and the absence of the 1998

winter–spring bloom. Marine Biology 138, 1051–1062.

Kowalczuk, P., Olszewski, J., Darecki, M., Kaczmarek, S., 2005.

Empirical relationships between coloured dissolved organic

matter (CDOM) absorption and apparent optical properties

in Baltic Sea waters. International Journal of Remote Sensing

26 (2), 345–370.

Lavender, S.J., Pinkerton, M.H., Froidefond, J.M., Morales, J.,

Aiken, J., Moore, G.F., 2004. SeaWiFS validation in

European coastal waters using optical and bio-geochemical

measurements. International Journal of Remote Sensing 25

(7–8), 1481–1488.

Laws, E.A., Archie, J.W., 1981. Appropriate use of regression

analysis in marine biology. Marine Biology 65, 13–16.

Libby, P.S., Gagnon, C., Albro, C.S., Mickelson, M.J., Keller,

A.A., Borkman, D., Turner, J.T., Oviatt, C.A., 2002.

Combined work/quality assurance plan for baseline water

quality monitoring: 2002–2005. ENQUAD ms-074, Massa-

chusetts Water Resources Authority, Boston.

Libby, P.S., Hunt, C.D., McLeod, L.A., Geyer, W.R., Keller,

A.A., Oviatt, C.A., Borkman, D., Turner, J.T., 2001. 2000

annual water column monitoring report. Report ENQUAD

2001-17, Massachusetts Water Resources Authority, Boston.

Mitchelson, E.G., Jacob, N.J., Simpson, J.H., 1986. Ocean colour

algorithms from the Case 2 waters of the Irish Sea in

comparison to algorithms from Case 1 waters. Continental

Shelf Research 5 (3), 403–415.

Morel, A., Berthon, J.-F., 1989. Surface pigments, algal

biomass profiles, and potential production of the euphotic

ARTICLE IN PRESSK.J.W. Hyde et al. / Continental Shelf Research 27 (2007) 1677–1691 1691

layer: relationships reinvestigated in view of remote

sensing applications. Limnology and Oceanography 34 (8),

1545–1562.

Morel, A., Prieur, L., 1977. Analysis of variations in ocean color.

Limnology and Oceanography 22 (4), 709–722.

Myint, S.W., Walker, N.D., 2002. Quantification of surface

suspended sediments along a river dominated coast with

NOAA AVHRR and SeaWiFS measurements: Louisiana,

USA. International Journal of Remote Sensing 23 (16),

3229–3249.

O’Reilly, J.E., Yoder, J.A., 2003. A comparison of SeaWiFS

LAC products from the third and fourth reprocessing:

Northeast US ecosystem. In: Patt, F.S., Barnes, R.A., Eplee,

Jr., R.E., Franz, B.A., Robinson, W.D., Feldman, G.C.,

Bailey, S.W., Gales, J., Werdell, P.J., Wang, M., Frouin, R.,

Stumpf, R.P., Arnone, R.A., Gould, Jr., R.W., Martinolich,

P.M., Ransibrahmanakul, V., O’Reilly, J.E., Yoder, J.A.

(Eds.), Algorithm Updates for the Fourth SeaWiFS Data

Reprocessing. NASA Goddard Space Flight Center, Green-

belt, MD, pp. 60–67.

O’Reilly, J.E., Maritorena, S., Mitchell, B.G., Siegel, D.A.,

Carder, K.L., Garver, S.A., Kahru, M., McClain, C.R., 1998.

Ocean color chlorophyll algorithms for SeaWiFS. Journal of

Geophysical Research 103 (C11), 24937–34953.

O’Reilly, J.E., Maritorena, S., Siegel, D.A., O’Brien, M.C.,

Toole, D.A., Mitchell, B.G., Kahru, M., Chavez, F., Strutton,

P.G., Cota, G.F., Hooker, S.B., McClain, C.R., Carder, K.L.,

Muller-Karger, F.E., Harding, J.L.W., Magnuson, A.,

Phinney, D., Moore, G.F., Aiken, J., Arrigo, K.R., Letellier,

R., Culver, M., 2000. Ocean color chlorophyll a algorithms

for SeaWiFS, OC2 and OC4: version 4. In: Hooker, S.B.,

Firestone, E.R. (Eds.), SeaWiFS Postlaunch Technical

Report Series, vol. 11. SeaWiFS Postlaunch Calibration and

Validation Analyses, Part 3. NASA Goddard Space Flight

Center, Greenbelt, MD, pp. 9–23.

Ohde, T., Sturm, B., Siegel, H., 2002. Derivation of SeaWiFS

vicarious calibration coefficients using in situ measurements

of Case 2 water of the Baltic Sea. Remote Sensing of

Environment 80, 248–255.

Oviatt, C.A., Hyde, K.J.W., Keller, A.A., Turner, J.T., 2007.

Production patterns in Massachusetts Bay with outfall

relocation. Estuaries and Coasts 30 (1), 35–46.

Pozdnyakov, D., Pettersson, L., Johannessen, O.M., Liaskovski,

A., Filatov, N., Bobylev, L., 2003. SeaWiFS maps water

quality parameters of the White Sea. International Journal of

Remote Sensing 24 (21), 4065–4071.

Press, W.H., Teukolsky, S.A., 1992. Fitting straight line data with

errors in both coordinates. Computers in Physics 6 (3),

274–276.

Sathyendranath, S., Platt, T., 1989. Remote sensing of ocean

chlorophyll: consequence of nonuniform pigment profile.

Applied Optics 28 (3), 490–495.

Schollaert, S.E., Yoder, J.A., O’Reilly, J.E., Westphal, D.L.,

2003. Influence of dust and sulfate aerosols on ocean color

spectra and chlorophyll a concentrations derived from

SeaWiFS off the US east coast. Journal of Geophysical

Research 108 (C6, 3191), 1–13.

Signell, R.P., Jenter, H.L., Blumberg, A.F., 2000. Predicting the

physical effects of relocating Boston’s sewage outfall.

Estuarine, Coastal and Shelf Science 50, 59–72.

Stumpf, R.P., Arnone, R.A., Gould Jr., R.W., Martinolich,

P.M., Ransibrahmanakul, V., 2003. A partially coupled

ocean-atmosphere model for retrieval of water-leaving

radiance from SeaWiFS in coastal waters. In: Patt, F.S.,

Barnes, R.A., Eplee, Jr., R.E., Franz, B.A., Robinson, W.D.,

Feldman, G.C., Bailey, S.W., Gales, J., Werdell, P.J., Wang,

M., Frouin, R., Stumpf, R.P., Arnone, R.A., Gould, Jr.,

R.W., Martinolich, P.M., Ransibrahmanakul, V., O’Reilly,

J.E., Yoder, J.A. (Eds.), Algorithm Updates for the Fourth

SeaWiFS Data Reprocessing. NASA Goddard Space Flight

Center, Greenbelt, MD, pp. 51–59.

Tassan, S., 1994. Local algorithms using SeaWiFS data for the

retrieval of phytoplankton, pigments, suspended sediment,

and yellow substance in coastal waters. Applied Optics 33

(12), 2369–2378.

Thomas, A.C., Townsend, D.W., Weatherbee, R., 2003. Satellite-

measured phytoplankton variability in the Gulf of Maine.

Continental Shelf Research 23, 971–989.

Twardowski, M.S., Donaghay, P.L., 2001. Separating in situ and

terrigenous sources of absorption by dissolved materials in coastal

waters. Journal of Geophysical Research 106 (C2), 2545–2560.

Wang, M., 2000. The SeaWiFS atmospheric correction algorithm

updates. In: McClain, C.R., Ainsworth, E.J., Barnes, R.A.,

Eplee, Jr., R.E., Patt, F.S., Robinson, W.D., Wang, M.,

Bailey, S.W. (Eds.), SeaWiFS Postlaunch Calibration and

Validation Analyses, Part 1. NASA Goddard Space Flight

Center, Greenbelt, MD, pp. 57–63.

Warrick, J.A., Mertes, L.A.K., Siegel, D.A., Mackenzie, C.,

2004. Estimating suspended sediment concentrations in turbid

coastal waters of the Santa Barbara Channel with SeaWiFS.

International Journal of Remote Sensing 25 (10), 1995–2002.

Yentsch, C.S., Menzel, D.W., 1963. A method for the determina-

tion of phytoplankton chlorophyll and phaeophytin by

fluorescence. Deep Sea Research 10, 221–231.

Yoder, J.A., Kennelly, M.A., 2003. Seasonal and ENSO

variability in global ocean phytoplankton chlorophyll derived

from 4 years of SeaWiFS measurements. Global Biogeo-

chemical Cycles 17 (4), 1112.

Yoder, J.A., Schollaert, S.E., O’Reilly, J.E., 2002. Climatological

phytoplankton chlorophyll and sea surface temperature

patterns in continental shelf and slope waters off the northeast

US coast. Limnology and Oceanography 47 (3), 672–682.

Zar, J.H., 1999. Biostatistical analysis. Prentice-Hall, Upper

Saddle River, NJ.