JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 106, NO. C5, PAGES 9499-9507, MAY 15, 2001 Spatial and temporal variability of absorption by dissolved material at a continental shelf Emmanuel Boss,W. Scott Pegau,J. Ron V. Zaneveld, and Andrew H. Barnard Collegeof Ocean and AtmosphericSciences,Oregon State University,Corvallis Abstract. Opticalpropertiesof dissolved(coloreddissolvedorganicmaterial (CDOM)) and particulatematter and hydrographicmeasurementswere obtainedat the Mid-Atlantic Bight during the fall of 1996 and the springof 1997 as part of the CoastalMixing and Opticsexperiment.To assess the temporal and spatialvariability,time serieswere obtained at one location and cross-shelftransectswere carried out. On short timescales, variabilityin the vertical distributionof the dissolvedfractionwas mostlydue to highfrequencyinternalwaves.This variabilitywas conservative, resultingin no changeson isopycnals. Over longerperiodsand episodically,CDOM variabilitywas dominatedby storms.The stormswere associated with sedimentresuspension eventsand were accompaniedby an increasein the absorptionby the dissolvedmaterials.Data from spatialtransectsshowthat near the bottom, over the shelf, and in both springand fall, increasedparticulateabsorptionand increasedCDOM absorptionco-occur.These data supportthe hypothesisthat bottom sedimentscan act as a sourceof dissolvedorganic carbonduring sedimentresuspension events. cleshave also been suggestedas sinksfor CDOM [Thurman, 1985; M. Moran, personalcommunication,1999]. Colored dissolvedorganic material (CDOM) absorption The observeddistributionof CDOM is complicatedby physstronglyimpactsoceancolor [Bricaudet al., 1981;Carderet al., ical advectionand local mixing, which act to transport and 1989],especiallyin coastalregionswhere it is a major absorb- erase the signaturesassociatedwith the above sourcesand ing substancein the blue portion of the visible spectrum sinks,respectively.The objectiveof thispaper is to evaluatethe [Bukataet al., 1995]. In order to invert reliably the remotely contributionof these processesto the observedvariability of sensedocean color to obtain the concentrationof optically CDOM on a continental shelf. significantmaterial, it is of major importance to determine The data analyzedhere were collectedon the continental CDOM distributionand its effect on oceancolor.The spectral shelfat the Mid-Atlantic Bight.Hydrographicpropertiesat the 1. Introduction absorption by CDOM (here denotedby aa) at the visible Mid-Atlantic Bight are stronglyaffectedby estuarinesources wavelengthscan be described,to a high degreeof accuracy,by a singleexponentialor a sumof two exponentialsthat decrease with increasingwavelength[Carderet al., 1989;Jerlov,1968; Roesleret al., 1989]. As in previousstudies,we use the term CDOM to denote coloredmaterial that passesthrough a 0.2 /•m filter, regardlessof its origin and chemicalcomposition. Terrestrial and oceanic sources and sinks contribute to the at the land-seaboundaryand by the Gulf Stream, its eastern boundary.The yearly mean circulationis predominantlyalong the shelf (westward),oppositeto the direction of the Gulf Stream[Beardsley and Boicourt,1981].A densityfront is often found at the shelf break separatingthe cold and fresh shelf water from the salineand warm slopewater offshore.Associatedwith thisfront is a geostrophicwestwardjet, with observed observeddistributionof CDOM (alsoreferredto as Gelbstoff, velocities of the orderof 0.5 m s-• [Barthet al., 1998].A Gilvin,andyellowsubstance), in theworld'soceans[Bricaudet weaker across-shoresecondarycirculationwith vertical velocal., 1981;Carderet al., 1989;Jerlov,1968].Terrestrialrunoff is ities of the order of 10 m d -• associated with this front has consideredto be the main source of CDOM in coastal waters, beenobserved[Barthet al., 1998].This circulationis associated resultingin a tight salinityCDOM relationshipthere. An ad- with the upwelling of sediment-richbottom waters on the ditional sourceof CDOM comesfrom benthic flux; poorly shoreside of the front and downwellingchlorophyll-richnearoxygenatedsedimentsin highlyproductivecoastalareas have surface waters on the offshore side of the front. beenobservedto flux dissolved organiccarbon(DOC) into the Severalstudieshave addressedthe seasonaland spatialvariwatercolumn[Burdige andHomestead,1994;Chenet al., 1997]. abilityof surfaceCDOM in the Mid-Atlantic [DeGranpreet al., Decompositionof particulateorganiccarbon is the primary 1996; Vodaceket al., 1997] and South Atlantic Bight [Nelson source of the DOC released. Other potential sourcesfor and Guarda, 1995]. These studieshave concludedthat the CDOM in the water columnare its releaseas a byproductof major sourcefor CDOM absorptionis riverine runoff, resultprimary[Bricaudet al., 1981;Twardowski, 1998]and secondary ing in a strongseasonalvariability and a stronglinear relationproductivity[Nelsonet al., 1998]. The main known sink for ship between increasingsalinity and decreasingCDOM for CDOM is photooxidation[Miller, 1994; Vodaceket al., 1997]. eachseason(thoughthe relationshipvariedbetweenseasons). Bacterialmetabolismand CDOM adsorptionto sinkingpatti- Photo-oxidation of CDOM resulted in lower levels of CDOM Copyright2001 by the American GeophysicalUnion. near the surfacein the summer[Vodaceket al., 1997]. These studieswere mostly confined to the surface and used instrumentation that could not detect the low CDOM absorption Paper number 2000JC900008. 0148-0227/01/2000JC900008509.00 9499 9500 BOSS ET AL.: COLORED DISSOLVED ORGANIC ABSORPTION VARIABILITY than 5 m were not usedbecauseof contaminationby bubbles generatedby the shipmotion. The data analyzedhere include 162profiles(41 bursts)from CMOI beforehurricaneEdouard, 41 ø 30'N 40 profiles(9 bursts)from CMOI after Edouard,and 136 (40 bursts)from CMOII. During the analysisof the data the near-surfacevalues of •"• • 40 ø 30'N MATTER • • ,•transects line ag(676) werefoundtobeoftheorderof -0.01-0.02m-• and decayingto ---0 at 10 m depth and below. Sincewe expect CMO time-series site • negligible absorption by CDOM at the infraredag(676) was subtractedfrom the dissolvedabsorptionat all wavelengths. This baseline offset removal procedure assumesthat this depth-dependent error wasnot spectrallydependent.We have not been able to identify fully the sourceof this effect, but it appearsto be related to changesin flow rate causedby the 39 ø 30'N 72 ø 30'W 71 ø 30'W 70 ø 30'W 69 ø 30'W addition of the filter. A noticeable difference in CDOM concentration was found Figure 1. Studysite of the CoastalMixing and Opticsexperbetween days3 and 4 of CMOII, when the instrumentmeaiment (CMO). The locationof the time seriesstationis indicatedby an open circle,and the transectcourseis denotedby suringCDOM was changed.A vicariouscorrectionwas comthe bold line. Bottom depthis basedon the ETOPO5 database. puted by matchingthe CDOM absorptionalong the 6ø isotherms(---40m depth)beforeand after the instrumentchange. The 6 ø isotherm signalin the visiblerangeover the midshelfregion(detection of UV absorptionwas possiblesinceit is higher than in the visible).In thisstudywe analyzeCDOM absorptiondataat 440 nm collectedat the Mid-Atlantic Bight that spansthe whole water columnand a period of a few weeksin the late summer and spring. was chosen because at this location neither surface nor bottom water mass intrusions were observed. The offsetwasfoundto be +0.0095m-• for the first3 days.Similarly, a differencein salinityand densitywasobservedbetween days6 and 7 of CMOII when the CTDs were changed.The offsetswere computedsimilarlyto CDOM (salinitywas +0.05 practical salinity units(psu),anddensity was+0.03kgm-3 for the first 6 days).These offsetswere added to the data. The adjustments were madeto the shortertime recordin eachcase. 2. Methods The offsetsto the salinityand densitymake variationsin the Data were collectedduring two cruisesto the Mid-Atlantic ancillary data more clear. Since these data are not used in Bight as part of the CoastalMixing and Optics experiment calculations, it doesnot matterwhichperiodwasadjusted.The (CMO) duringAugust 17 to September7, 1996 (denotedby offsetto aa(440) wasaddedto the datacollected duringthe CMOI), andApril 26 to May 13, 1997(CMOII). A time series first3 daysbecausethe spectralslopeof the CDOM absorption station was occupiednear 40.5øN, 70.5øW,and three across- during that time was inconsistentwith the rest of the data that the calibrationerror most likely shelftransectswere performed,two in CMOI, before (August collected.This suggests 17-18, 1996) and after (September5-7, 1996) the time series occurredduringthe first 3 days.This temporaladjustmentto and one after the time seriesfor CMOII (May 12-13, 1997, the CDOM absorptionwasfound to be spectrallydependent, Figure 1). Sosiket al. [thisissue]and Gardneret al. [thisissue] but only the value at 440 nm is necessaryfor the discussion in summarizethe variabilityin opticaland hydrographic proper- this text. ties duringthe time series. The spectralslopesof aa were foundto be on average Two WET Labs ac-9swere usedto measureCDOM absorp- ---0.018 nm-• (excluding thefirst3 daysof CMOII), withinthe tion, total absorption,and the absorptiondue to particles(by range of publishedvalues[e.g.,Roederet al., 1989] and obdifference).To determinethe contributionof dissolvedmate- servedby DeGranpreet al. [1996]and Vodaceket al. [1997]in rialsto the total absorptioncoefficient,a 0.2 p.mfilter (Gelman the Mid-Atlantic Bight, increasingour confidencein our caliSuporcap100) was attached to the inlet of one ac-9. Both bration procedure.In order to estimatethe effect of the subinstrumentswere calibrateddailywith opticallypure water as tractionof aa(676) on the verticalstructure of aa(440) the a reference(BarnsteadNANOpure) and correctedfor tem- correctedabsorptiondata were comparedto CDOM fluoresperature and salinityfollowingPegauet al. [1997].The partic- cencemeasurementson the samplingplatform (WET Labs ulate absorptionwas correctedfor scatteringusingmethod 3 SpectralAbsorptionand Fluorescence Instrument(SAFIRE), (CMOI) and method 1 (CMOII) of ZaneveMet al. [1994]. excitation/emission, 265 nm/460 nm) for randomly selected Physicalpropertieswere measuredon the samepackagewith a measurementprofileswhere both were available.The two inSeaBirdElectronicsSBE-911 conductivity-temperature-depthdependentmeasurements of CDOM concentrationwere found (CTD) (a SBE-25 CTD was usedafter April 31, 1997). The to behighlycorrelated (average (R2) -- 0.95), increasing our data from the CTD and ac-9 were mergedand binned to 1 m confidencein our processing procedure. intervals. For comparisonwith data collectedpreviouslyin the South The data were collectedduring daylightin samplingbursts Atlantic and Mid-Atlantic Bights we include in our analysis (on averagethree per day) that lasted an averageof 40 min data from Nelsonand Guarda [1995],DeGranpreet al. [1996], (+ 10 min) with an averageof four castsper burst.In order to and Vodaceket al. [1997].When absorption near440 nm was quantifythe short-termvariabilityin opticalandphysicalprop- not measured(first andlaststudy),we extrapolatedto 440 nm, erties the mean standard deviation and maximum standard assuming ag(440)= aa(h.)e -ø'ø18(44ø-x), onthebasis ofthe deviationof burstswere computedfor eachdepth.Daily means valueof the spectralslope(0.018nm-•) observed in these were computedby averagingthe burstmeans.Data shallower studies. BOSS ET AL.' CMO-I COLORED 10 DISSOLVED ORGANIC ag(440) [m-1] MATTER ABSORPTION VARIABILITY 9501 o [kg/m3] %(440)[m-1] ß ( 20 30 40 50 60 CMO-II 10 20 30 40 50 60 10-3 10-2 10-3 10-2 10-1 10-2 100 Figure2. Mean(bold)andmaximum of thestandard deviation (over40 minsampling bursts)of aa(440), ap(440), anddensities for CMOI (priorto hurricane Edouard)andCMOII. Theminimums werelowerthan the resolution 3. Results of the instruments. ap correlates lesswith density(r = 0.23 for CMOI andr = 0.31 for CMOII). The short-termvariabilityis higher, on avThe resultsare presentedfor differenttemporaland spatial erage,in all propertiesin CMOI relativeto CMOII (Figure2). samplingscales.We presentthe particulatefractionabsorption 3.1.2. Mesoscalevariability. Over scalesof 4-5 daysthe at 440 nm, ap(440), for comparison, sincetogetherwith andaa exhibit wave-like variation (Plate1) in aa(440) and the (almost)constantabsorption of water,it CMOIdensity constitutesthe total absorption.Sosiket al. [thisissue]discuss the thermocline, indicating the potential presenceof lowfrequencyRossbyor Kelvin waves associatedwith the shelf the mainsources of variabilityin ap(440). Physical forcing, break front or the bottom topography.These waves cause sinks,and sourcesnear boundariesare expectedto result in a vertical displacements of 0(5 m) at the pycnoclineandcanalso variation of optical propertiesthat depend on the physical be observed in the more evenlyspacedCTD data setcollected distancefrom the boundary(z). In the interior, on the other hand,opticalpropertiestend to stayconstantalongisopycnals. during the samecruiseGardneret al. [this issue]and in the In order to separatethe variability of optical properties at currentrecord of Boydet al. [1997]. Intrusionsof oceanicwarm saltywaterswere observeddurconstantdepth from that on constantisopycnalsurfacesthe ing severaloccasionsof both cruises.During CMOI, intrusions variabilityof the opticalpropertiesaswell asdensityare quantified at constantdepths,and the covarianceof densityand occurredon August22, 25, and 27 andAugust31 to September opticalpropertiesis computed.Salinityis presentedin order to 1, 1999(Plate 1). W. S. Pegau(Mixing of opticalpropertiesas compare it withtheaa(440) distribution; in thesimplest case, evidencedin salinityintrusionsobservedover the continental shelf in the Middle Atlantic Bight, submittedto Journalof where the only sourceis rivefine input of constantCDOM Research,2000,hereinafterreferredto asPegauet concentration,the two shouldinverselycovary.Last, temper- Geophysical ature is presentedin order to assess changesin opticalprop- al., submittedmanuscript,2000) discussintrusivefeaturesobwith meandering ertieswithinwater massesof constanthydrographicproperties. servedbetweenAugust25 and 27 associated of the shelfbreak front intrudingat the baseof the mixedlayer 3.1. Temporal Variability and near the bottom. Within these intrusions,Pegau et al. 3.1.1. Short-term variability (40 min bursts sampling). (submittedmanuscript,2000) found CDOM mixed conservaWhen strongstratificationis present(CMOI), CDOM variabil- tively,while particulateabsorptiondid not. In the spring,beity (its standarddeviationas a function of depth) correlates tween May 4 and 6, a salty bottom intrusion is observed,and with the variabilityin density(Figure 2, correlationcoefficient after May 7 a freshwater massis observedin the upper 20 m. r of the standarddeviationof all bursts,r = 0.8). This implies The salty bottom intrusion is associatedwith relatively low that on this timescale,CDOM behavesto a large extent con- CDOM absorption,while the freshsurfaceintrusionis associservativelyand its variabilityis mostlydue to isopycnaladvec- ated with higher valuesof CDOM. tion. The maximumin variabilityoccursat the pycnocline(s) 3.1.3. Variability due to storms. Intense episodicstorms >0.4 N m-2) occurred duringbothcruises; during duringtimes of intensehigh-frequencyinternal wave activity (windstress associated with solitarywavepackets[Boydet al., 1997;Chang CMOI (September2, 1996) the eye of hurricane Edouard and Dickey,thisissue].During CMOII, variabilityin CDOM is passed110 km from the time-seriesstation while a seriesof smallerthantheinstrument accuracy (0.005m-•, precision is springstormstookplaceduringCMOII (April 27, 1997,MaY 4, -0.002m-•), andthecorrelation between density andaa is 1997,and May 7, 1997). These stormshave contributedto the weak (r = 0.18). Higher variabilitynear the bottom, where variabilityin opticalpropertiesby verticallymixinggradients, sediments densityisnearlyconstant,is suggestive of nonconservative pro- reducingthe physicalstratificationandresuspending cesses and/oradvection there.Particulateabsorption ap in (Plate 1). Edouard reduced the near-surfacegradients in bothcruises is morevariablethanaa' In CMOI, variabilityin CDOM while increasingthe near-bottomgradientsin partic- 9502 BOSS ET AL.: COLORED DISSOLVED ORGANIC ulate absorption(Plate 1). The springstormscausedmixingof the opticalpropertiesdownto 20 m (Plate 1). 3.1.4. Seasonal variability. The vertically averaged CDOM value at the CMO site is decreasedby 0.01 _+0.005 m- • fromfall to spring.Thisdifference is associated withthe MATTER ABSORPTION VARIABILITY 1.155, computedwithout usingthe absorptionvaluescloseto the surface,where photo-oxidationmay be present). The data of Nelsonand Guarda, [1995, Figure 8] from the SouthAtlanticBighthavevaluesof CDOM absorption at least a factor of 2 higher than those observedin the Mid-Atlantic Bightin the late summerandspringfor the samesalinityvalues mean salinityduringCMOI being,on average,0.25 _+0.06 psu fresherthan during CMOII. During the spring,CDOM is less (approximate linearrelationship: aa(440) = -0.0727s + stratifiedthan in the summer(Plate 1). The hydrographicand 2.618). This may be due to (1) higherCDOM concentration CDOM absorptionpropertiesof the waters denserthan the in riverine inputsfrom southof the Mid-Atlantic Bight (e.g., shelf break front (tr > 25.5) and below the pycnoclineare from the ChesapeakeBay). (2) Accumulationof CDOM derived from in-water productivity and/or sediment-released nearlyconstantthroughoutthe year (Plate 2 and Figure 3). CDOM into the southwardflowingwaters originatingin the 3.2. Spatial Variability Mid-Atlantic Bight. 3.2.1. Vertical structure of CDOM during the time series. The correlationbetweensalinityand CDOM absorptionin as observedin the refThroughout bothcruises, valuesof aa monotonically increase the surfacelayer is season-dependent with depthexceptfor the last4 daysin CMOII where,near the erencedstudiesas well as here.At eachdepthhorizon,aa surface,a low-salinity high-aa watermassis present. decreasesand s increaseswith distance from shore. However, 3.2.2. Variability in transects. The optical and hydro- the observed relationship betweens and aa in the vertical graphicpropertiesin the first transectare stronglystratified duringthe time seriesstationis, mostfrequently,oppositeto with increasing s (Plate2). Valuesof aa abovethe shelfexhibita subsurfacethe horizontalrelation,with aa increasing maximum at depths of 40-60 m and a minimum near the surface. The bottommaximum in aa nearthe 60 m isobathis associated withthe20-30% elevated particulate absorption ap values. (Plate 1). 4. Discussion The secondtransect,performedafter the hurricanepassage, 4.1. Role of AdvectionVersus Local Change showslittle correlationbetween physicaland optical properThe short-term variability is higher in all properties in ties.While the water columnhasrestratified,intenseresuspenCMOI relativeto CMOII. Someof thisvariabilityis associated sionof particleshastakenplacenear the 60 m isobath.Vertical with isopycnaladvectiondue to higher stratificationand assogradients of aa are weakwith bottomandsurfacemaximum ciatedinternalwaveactivity(Plates1 and 2). Other sourcesof observed towardshorecoincident witha maximum in ap. Horshort-termvariabilitymay be horizontaladvectionof horizon- izontalgradients in aa followingthe hurricaneare stronger tal gradients in absorption (uOaa/Ox)aswellasvariability due than the gradientsin the first transect. to local processes. The mean horizontal velocities during The spring(CMOII) sectionexhibitslow verticalstratifica- CMOI wereof the orderof 0.2 m s-• (the upper95% was 0.4 m s-•) [Boydet al., 1997].For a localchangein CDOM are higher both at the surface and bottom above the shelf absorption of theorderof 0.01m-• overa 40 minintervalto (midwaterminimum). be totallydueto horizontaladvection by a 0.2 m s-• current, the spatialgradienthasto be of the orderof 0.2 m-• over10 3.3. Salinity-ag Relationship tionin bothhydrography andopticalproperties. Valuesof aa In the casewhere the only CDOM sourceis riverineinput, onewouldexpectaa to increase with decreasing salinitys as km. The maximumalong-isopycnal CDOM gradientsobserved in the transects were of the order of 0.005 m -• over 20 km. oceanicwatersdilute the rivefine input. Indeed, at the surface This is consistentwith the vertical advection of isopycnals in the SouthAtlantic and Mid-Atlantic Bightsa linear inverse beingthe maincausefor CDOM variabilityin CMOI, resulting in a high correlationbetweenthe variabilityin densityand that relationship of salinityandaa hasbeenobserved in previous studies[DeGranpreet al., 1996;Nelsonand Guarda, 1995;Vo- in CDOM absorptionduring this time. daceket al., 1997];thisis alsothe generaltrend observedin our spatialsections(Figure 3). By dividingthe data into two seasonsand addingthe data of DeGranpreet al. [1996] and Vodaceket al. [1997] (collected3 years earlier and downstream from the CMO site)we observethe followingtrends:(1) Variability in CDOM concentrationwithin watersof givensalinity is much higher in the late summer; twofold differencesin CDOM absorptionare found in waterswith salinityfrom 31.5 to 32.3psuin that time. (2) There is little variabilityin the deep watersbeyondthe shelfbreak front (s > 32.3) comparedto the near-surfacewaters. The data of DeGranpreet al. [1996] and Vodaceket al. [1997] fit the trends observedin our data well, suggestinga consistencyin the range of CDOM values overa 3 yeartimescale.(3) The CDOM absorptionvaluesfrom the fall, measuredafter the hurricanepassage,obey a similar relationshipwith salinityasthe springdata (approximatelinear 4.2. Storms and SeasonalVariability The effect of a singlestorm can causeas muchvariabilityin CDOM absorptionas is observedbetweenseasons.Over the shelfthedifference in valuesof aa at thebeginning andendof CMOI (At • 3 weeks)is largerthan the differencein aa betweenthe end of CMOI and CMOII (At • 8 months,Plate 2 and Figure 3). Seasonalvariability,in addition to winter verticalmixing, includes potential changes in aa - s in riverine sourcesand effectsof seasonallymodulatedprimary and secondary productivity.Our data are by no means adequateto resolvetheseprocesses. 4.3. Variability of CDOM in the Water off the Shelf Break Front CDOM propertiesbeyondthe shelfslopefront are foundto that prorelationship aa(440) = -0.0157s + 0.58). Thisrelationship varylittle throughoutthe year (Figure3), suggesting is different from that computedwith the data taken in the fall prior to the hurricanepassage(aa(440) = -0.032s + cessesaffectingCDOM in thesewatershave timescaleslonger than 0(6 months).It is interestingthat a nearly linear rela- BOSSET AL.: COLORED DISSOLVED ORGANIC MATTER ABSORPTION VARIABILITY 9503 Data fromAugustand September (a) 0.2 ...... 0.15 (b) **•. 20 . E • 15 t3.1 0.05 . . , I ...... Data fromApril andMay (C)0.2 (d) 20 0.15 t3.1 10 0.05 0 30 i 31 i 32 i 33 I 34 i 35 i 1 53 32 Salinity(psu) 33 34 35 36 Salinity(psu) Figure3. (a) and(b) Theaa(440)-salinity andtemperature-salinity diagrams compiled withthedata collected during thesummer-fall transect (August 17-18,1996,soliddots)andthepost-hurricane Edouard transect (September 5-7,1996, shaded dots)anddatacollected 3years earlier intheMid-Atlantic Bightfrom Vodacek etal. [1997, Table1] (extrapolating to 440nmassuming s - 0.018onthebasis of Vodacek etal. [1997, Figure 7],denoted bystars) andbyDeGranpre etal.[1996, Table1](and442nmabsorption given by Vodacek etal.[1997, Figure 7],denoted bysolidcircles]. (c)and(d)Thedatafromthespring transect (May 12-13,1997).Squares denotethe 5 m dataof eachsampling station. tionshipstillexistsbetweensalinityandCDOM withinthese 4.5. watersdespitetheir distancefromriverineinput. A back of the envelopecalculationbasedon observed coastalbenthicfluxesof DOC in Chesapeakebay (2 mmol 4.4. Primary ProductivityAs Sourcefor CDOM Primaryproductionover GeorgesBank and the MidAtlanticBight,wherethe CMO siteis located,is nearlyfive timeshigherthanthemeanoceanic productivity [O'Reilly etal., 1987].Primaryproductivity is an important sourceof DOC at Georges BankandtheMid-Atlantic Bight;onaverage, 14%of the primaryproduction is released as DOC [O'Reillyet al., 1987].In our datathereis no correlation betweenparticulate andCDOM absorption, thoughon the basisof DOC budgets, phytoplankton isanimportant source. Evenin theopenocean, Benthic Source DOG m-2 d-•) [e.g.,Burdige andHomestead, 1994]suggests a 440nmabsorption fluxof theorderof 0.01m-• m-2 d-• (assuming thathalfofDOC isCDOCanda specific absorption at440of0.7m-• (mgL-•) onthebasisofBlough etal. [1993]). O'Reillyet al. [1987]foundthe annualprimaryproduction in thevicinityof theCMO siteto be of theorderof 210-450g C m-2 [O'Reillyet al., 1987,Table21.3].Walshet al. [1987] estimated that between 5 and 10% of the annual primary production is releasedas organicDOC from the sediment et al., 1987,Table22.2].Thesenumbers suggest an where CDOM is believedto be mostlythe result of primary [O'Reilly fluxof the sameorderof magnitude asthoseobproduction, no correlation hasbeenfoundbetweenbiomass absorption [1994].In orderfor the and CDOM [Bricaudet al., 1981],suggesting that the other servedby Burdigeand Homestead benthic flux to have a significant effect relativeto the backprocesses controlling CDOM distribution havea highertemground values, near-bottom CDOM has to accumulate overa poralfrequency of variability, masking the contribution from this source. periodof the orderof a few weeks. 9504 BOSSET AL.: COLOREDDISSOLVEDORGANICMATTER ABSORPTIONVARIABILITY Hurricane Edouard • ag(440)[m-•] Spring•.4N/m2) 0.08 20 0.07 0.06 0.05 a•(440) [m-•] 0.2 0.15 0.1 0.05 60 cr[kg/m3] 26 [ 20 ß 25 24 40 23 ' ß i •! i • 22 33 20 32.5 40 32 31.5 60 31 Temperature[øC] 18 16 14 12 10 6.5 8 60 6 19 21 23 25 27 29 31 August Date in 1996 2 4 6 September 26 28 30 2 4 April 6 8 10 May Date in 1997 Plate1. Contour plotof dailymeanofag(440),ap(440),salinity, anddensity at thetimeseries station. Contour density isdifferent forthetwocruises because ofa weaker stratification inCMOII.In bothcruises, ag(440) increases nearthebottom associated withbottom resuspension (higher ap(440)) following storm passage. BOSS ET AL.: COLORED DISSOLVED 8/17-18/96(CMO-I) ORGANIC MATTER ABSORPTION 9/5-7/96 (CMO-I) VARIABILITY 9505 5/12-13/97(CMO-1I) 20 0.08 40 0.06 60 0.04 I 80 I 100 ag(4-40) (44o) l, 20 0.02 i ,• It••[o. ! I-" I 40 0.1 --- >>0.12 60 0.05 80 %(440) %(440) lOO { • yP %(440) 27 I 20 '' 40 I I I I 26 I I ! 25 60 24 80 I [• lOO ø't 23 35 20 I I I I I I 34 i I ! I 33 32 lOO • 31 20 t8.5 6.5 60 15 ß 10 80 • 100 40 • 40.2 • T 40.4 40.6 degreeslatitudeN 40.8 40 T 40.5 degreeslatitudeN 41 [: T ß 40 40.4 40.8 degreeslatitudeN Plate2. Contourplotof aa(440), ap(440), salinity,anddensityof threespatialtransects. Theboldblack line indicatesbottom depth at each samplingstation.Dashed lines denote the location of the sampling stations.Notice the decreasein stratificationfollowingthe hurricanepassage(secondtransect)and spring transects compared to thelatesummer transect. Noticeanelevated valueof aa(440) nearthebottomin all transects to the north of 40.5øN near the 60 m isobath. 9506 BOSS ET AL.' COLORED DISSOLVED ORGANIC While diffusive fluxes may be low, suddenstirring of the sedimentsmay releasewater with very high concentrationof CDOM. CDOM concentrationsin the upper sediment of coastalwatershavebeen found to be 5 timeshigherthan those abovewaters(e.g., D. J. Burdigeet al., Fluorescentdissolved organicmatter in marine sedimentpore waters,submittedto Marine Chemistry,1999)while CDOM fluorescence in the upper meter of anoxicsedimentshasbeen observedto be 10-500 timeshigherthan the overlyingwaters[Chenand Bada, 1994]. 5. MATTER ABSORPTION VARIABILITY Conclusion Our data set is unique in that it resolvesthe whole water column and includesboth temporal and spatialinformation. These data, together with previousstudies,suggestthat the distribution of aa is determined bya coastal rivefinesource, a shelfbottomsource,and a photo-oxidativesinkat the surface. Advectionand diffusionby the backgroundflow further modulatesthe distributionof CDOM absorption.There may also be an additional in-water source of CDOM associated with In all thetransects weobserve highervaluesofaa ontopof the phytoplankton[Twardowski,1998],whichwe did not observein shelf,in correlationwith higher particulatevalues.In the last our data,thoughit maycontribute to the differentaa - s two transectsa midwater minimum in CDOM absorptionis observedon top of the shelf, uncorrelatedwith the hydrographicproperties(Plate 2). A sedimentarysourceis consistent with sucha pattern. Similarly,the stronggradientalong isopycnalsfollowingthe hurricanesuggeststhat CDOM and particulateabsorptionhave not mixed conservatively. Follow- relationshipobservedfarther downstreamat the SouthAtlantic Bight [Nelsonand Guarda, 1995]. This sink/sourcelayout resultsin a spatialdistributionof CDOM, whichis decreasing awayfrom shore(with increasein salinity)and increasing with depthat the shelf(alsowith increasein salinity). ingthehurricane, aa increased in thefreshest waters(s < 32 found to be mostly due to conservativeprocessesconsistent with the advectionof isopycnals.Over periodsof weeks and psu),whileit decreased in the saltierwaters(s > 32 psu).This decreasein CDOM absorptionmay be due to vertical mixing of saltywaterswith the relativelyCDOM-poor fresherwater. The increasein CDOM of the freshestwatersmaybe due to a benthicinjectionof CDOM into the waterson the shallowpart of the shelf. This is also consistentwith increasingbottom stressand elevated resuspensionactivity in shallowerwater causedby the wind and wavesassociatedwith the hurricane. Advectionof a freshwatermassof a different CDOM-salinity relationshipthan that previouslyfound on the shelfcannotbe ruled out. 4.6. Photooxidation Short-term variability in theverticaldistribution of aa was longer,sources andsinksof aa causea distribution thatdeviates from that of physicalproperties near the surface and abovethe shelf.Episodicstormsare found to havean impor- tantrolein the high-frequency variability in ag throughboth redistributionand resuspension of bottom sediment. Over the shelf, salinity and CDOM are found to have a different relationshipdependingon the seasons,aswas found previously[DeGranpreet al., 1996;Nelsonand Guarda, 1995; Vodaceket al., 1997].Beyondthe shelfbreak front, subsurface aa wasfoundto behavein a conservative fashionfor a period longerthan 8 months(Plate 2 and Figure 3). Comparisonof the (scarce)data obtainedfor this study and those of past studiesover the Mid-Atlantic Bight shelf suggeststhat the seasonalcyclerepeatsitself over the years. Though we do not observea direct reductionof surfacea duringthe time series,photo-oxidationmay havebeenpresent Thevariability inaa(440) isfoundtobesignificantly smaller andis probablythe main contributionto the trend of decreased than that in ap(440) on all timescales and space scales. HowCDOM in waters of decreasedsalinity observedabove the ever, the averagemagnitudeof the dissolvedabsorptionis shelf(Plate1).Themeanaa distribution asa function of depth similarto that of the particulatefractionat 440 nm, providing in the late summercouldbe the result of a balanceof photo- a similarcontributionto the total absorption.This impliesthat oxidationin the mixed layer (ML) and a diffusiveflux from high-frequencyvariabilityof ocean color in a given surface lower in the water column(integratingthe conservation equa- water massis most likely due to variationsin the particulate tion over the ML): fraction. 0 = -HMi•'yaa+ K daa/dz (at z = HM•), where HML is the ML depth, •/is the ML-averagedphotooxidationrate, and K is the eddy diffusivity.Taking HMi• • Further studyof the spectralshape(the exponentialslope) of aa isrequiredto distinguish betweenmarineandterrestrial sources and to determine whether the surface a has been photo-oxidized[e.g.,Carderet al., 1989;Vodaceket al., 1997]. 10 m, K • 2 x 10-s m-2 s-1 (on the basisof the in situ We do not havethe spectralresolutionto addressthisquestion microstructuremeasurementof MacKinnonand Gregg(per- fully with our data set. sonalcommunication, 1998)), and the relativeverticalgradient of CDOMacross theML baseaa ldaa/dz• 0.2 m-1 (Plate 1), the averagephoto-oxidation rate constant•/corresponds to a timescale (•/-1) of theorderof 1 month.GiventhatK varied Acknowledgments.Discussions with M. Twardowskyand P. Coble are gratefullyacknowledged. We thank JenniferMcKinnon and Mike Greggfor eddydiffusivities data,T. Boydand M. Levinefor providing current data, S. Lentz, A. Plueddernann, S. Anderson, J. Edson, and the WHOI Upper Ocean ProcessesGroup for providingthe wind stressdata. Supportwas providedby the EnvironmentalOpticsProgram of the Office of Naval Research. by a factor of 5 near the ML base,this value shouldbe interpreted as an order of magnitude estimate of the rate. This value is consistentwith the estimatedphoto-oxidationratesof Nelson et al. [1998] and Vodaceket al. [1997]. We cannot, however,rule out the possibilitythat the surfacereductionwas due to the advectionof lower CDOM waters, though this is References unlikely given the general trend of decreasingCDOM with decreasing salinityobservedin the time series(Plate 1). Photo- Barth, J. A., D. Bogucki,S. D. Pierce, and P.M. Kosro, Secondary oxidationof ag is a well-documented process and hasbeen previouslyobservedin the Mid-Atlantic Bight [Vodaceket al., 1997]. circulationassociated with a shelfbreakfront, Geophys. Res.Lett., 25, 2761-2764, 1998. Beardsley,R. C., and W. C. Boicourt, On estuarineand continentalshelfcirculationin the Middle Atlantic Bight, in Evolutionof Phys- BOSS ET AL.: COLORED DISSOLVED ORGANIC MATTER ABSORPTION VARIABILITY 9507 ical Oceanography,edited by B. A. Warren and C. Wunch, pp. sorptionon the continentalshelfof the southeasternUnited States, J. Geophys.Res., 100, 8715-8732, 1995. 198-233, MIT Press,Cambridge,Mass., 1981. Blough, N. V., O. C. Zafiriou, and J. Bonilla, Optical absorption Nelson,N. B., D. A. Siegel,and A. F. Michaels,Seasonaldynamicsof coloreddissolvedmaterial in the SargassoSea,DeepSeaRes.,Part I, spectraof watersfrom the OrinocoRiver outflow:Terrestrial input 45, 93!-957, 1998. of colored organicmatter to the Caribbean,J. Geophys.Res., 98, O'Reilly, J. E., C. Evans-Zetlin,and D. A. Busch,Primaryproduction, 2271-2278, 1993. in GeorgesBank, edited by R. H. Backus,pp. 220-233, MIT Press, Boyd, T., M.D. Levine, and S. R. Gard, Mooring observationsfrom Cambridge,Mass., 1987. the Mid-Atlantic Bight,Rep.97-2-164,226 pp., OregonState Univ., Pegau,W. S., D. Gray, and J. R. V. Zaneveld, Absorptionand attenCorvallis, 1997. uation of visibleand near-infraredlight in water: Dependenceon Bricaud,A., A. Morel, and L. Prieur, Absorptionby dissolvedorganic temperature and salinity,Appl. Opt., 36, 6035-6046, 1997. matterof the sea(yellowsubstance) in the UV andvisibledomains, Roesler,C. S., M. J. Perry, and K. L. Carder, Modeling in situ phytoLimnol. Oceanogr.,26, 43-53, 1981. plankton absorptionfrom total absorptionspectra in productive Bukata, R. P., J. H. Jerome,K. Y. Kondratyev,and D. V. Pozdnyakov, inland marine waters,Limnol. Oceanogr.,34, 1510-1523, 1989. OpticalProperties and RemoteSensing of Inland and CoastalWaters, Sosik,H. M., R. E. Green, W. S. Pegau,and C. S. Roesler,Temporal 362 pp., CRC press,Boca Raton, Fla., 1995. and vertical variability in optical propertiesof New England shelf Burdige,D. J., and J. Homestead,Fluxesof dissolvedorganiccarbon waters during late summerand spring,J. Geophys.Res.,this issue. from ChesapeakeBay sediments,Geochim.Cosmochim.Acta, 58, Thurman,E. M., OrganicGeochemistry of Natural Waters,497 pp., Dr. 3407-3424, 1994. Carder, K. L., R. G. Steward,G. R. Harvey, and P. B. Ortner, Marine humic and fulvic acids:Their effects on remote sensingof ocean chlorophyll,Limnol. Oceanogr.,34, 68-81, 1989. Chang, G. C., and T. D. Dickey, Optical and physicalvariability on timescalesfrom minutesto the seasonalcycleon the New England shelf:July 1996 to June 1997,J. Geophys.Res., this issue. Chen, R. F., and J. L. Bada, The fluorescenceof dissolvedorganic matter in porewatersof marine sediments,Mar. Chem., 45, 31-42, 1994. Chen, R. F., D. B. Chadwick,and S. H. Lieberman,The applicationof time-resolvedspectrofluorometry to measuringbenthicfluxesof organiccompounds,Org. Geochem.,26, 67-77, 1997. DeGranpre, M.D., A. Vodacek,R. K. Nelson,E. J. Bruce,and N. V. Blough, Seasonalseawateroptical properties of the U.S. Middle Atlantic Bight, J. Geophys.Res., 101, 22,727-22,736, 1996. Gardner,W. D., et al., Optics,particles,stratification,and stormson the New England continentalshelf,J. Geophys.Res., this issue. Jerlov,N. G., OpticalOceanography, ElsevierSci.,New York, 1968. Miller, W. L., Recent advancesin the photochemistryof natural dissolvedorganicmatter, in Aquaticand SurfacePhotochemistry, edited by G. R. Helz, R. G. Zepp, and D. G. Crosby,pp. 111-127, P. F. Lewis, New York, 1994. Nelson, J. R., and S. Guarda, Particulate and dissolvedspectral ab- W. Junk, Norwell, Mass., 1985. Twardowski,M. S.,The finescaledistribution,originandphotobleaching of the spectralabsorptionof dissolvedorganicmatter in coastal waters, Ph.D. thesis,Univ. of R. I., Kingston,1998. Vodacek,A., N. V. Blough,M.D. DeGranpre, E. T. Peltzer, and R. K. Nelson, Seasonal variations of CDOM and DOC in the Middle Atlantic Bight: Terrestrial inputs and photooxidation, Limnol. Oceanogr.,42, 674-686, 1997. Walsh, J. J., T. E. Whitledge, J. E. O'Reilly, W. C. Phoel, and A. F. Draxler, Nitrogencyclingon GeorgesBank and the New York Shelf: A comparisonbetweenwell-mixedand seasonallystratifiedwaters, in GeorgesBank, edited by R. H. Backus,pp. 234-246, MIT Press, Cambridge,Mass., 1987. Zaneveld,J. R. V., J. C. Kitchen, and C. C. Moore, Scatteringerror correctionof reflectingtube absorptionmeters,Proc. SPIE Int. Soc. Opt. Eng., 12, 44-55, 1994. A. H. Barnard,E. Boss,W. S. Pegau,and J. R. V. Zaneveld,College of Ocean and Atmospheric Sciences,Oregon State University, 104 Ocean Administration Bldg., Corvallis, OR 97331. (boss@oce. orst.edu) (ReceivedApril 2, 1999;revisedAugust11, 1999; acceptedDecember10, 1999.)