Jul 15, 1998 - of 45-50/â¢atm occurred after Hurricane Felix passed near Bermuda in August ...... Hurricanes Luis and Marilyn passed to the west of Bermuda.
JOURNAL
OF GEOPHYSICAL
RESEARCH,
VOL. 103, NO. C8, PAGES 15,567-15,585, JULY 15, 1998
Variability ofpCOz on diel to seasonal timescales in the Sargasso Sea near Bermuda Nicholas
R. Bates
BermudaBiologicalStationfor Research,Inc., Ferry Reach, Bermuda
Taro Takahashiand David W. Chipman Lamont-DohertyEarth Observatory,Columbia University,Palisades,New York
Anthony H. Knap BermudaBiologicalStationfor Research,Inc., Ferry Reach, Bermuda
Abstract. Continuousunderwaymeasurementsof atmosphericand surfaceseawater pCO2 were collectedon numerouscruisesin the SargassoSea (32øN,64øW) near Bermuda from June 1994 to November 1995. We observedthat seawaterpCO2 was highly variable on different timescales,rangingfrom diet to seasonal.On diet timescales,p CO: changesof 5-25/•atm occurredin responseto diurnal warming and coolingassociated with solar heat fluxes.Over longer timescales,p CO: was influencedby atmospheric forcing and tropical cyclones.For example,a surfacecoolingof 3øC and decreasein p CO: of 45-50/•atm occurredafter Hurricane Felix passednear Bermuda in August 1995. The decreasein pCO: was significantconsideringthe annual changewas 90-100/•atm. Over all timescales,temperaturewas the dominantcontrol on pCO2 variability.We found that surfacepCO 2 conditionswere accuratelypredictedfrom temperatureswith small errors (4-9/•atm) if seasonalpCO2-temperaturerelationshipswere established.In future synthesisof regionalpCO: data it shouldbe feasibleto use surfacetemperature,remotely sensedfrom space,as a tool for extrapolationover wider spatial scalesin the North Atlantic subtropicalgyre.Net annualfluxesof CO: for 1994 and 1995 (-0.25 to -0.6 mot
CO: m-: yr-•) weredirected fromatmosphere to oceanandweresimilarto values reportedfor 1989-1993 by Bateset al. [1996b].We found that short-termvariabilityof pCO2 (diet warmingand coolingor atmosphericforcing),frequencyof sampling(every 3-4 daysor monthly),or use of temperature-derived p CO2 did not affect estimatesof net yearly CO: fluxesby more than 10-20%. However, strongwinds associatedwith hurricanesdecreasedthe net annual flux of CO2 into the ocean by 19-28% in 1995. The major sourcesof error for air-sea gasexchangewas uncertaintyassociatedwith gas transfer-windspeedrelationshipsand differencesin the typesof wind speeddata used (daily averagedversusclimatological).Suchuncertaintiesmake it difficultto quantifythe contributionof gas exchangeto the carboncycleand the balanceof carbonimport and export terms in the upper ocean of the SargassoSea. 1.
Introduction
of the inorganiccarbon cyclefrom monthly observationsat a few oceanictime series[Keeling,1993; Winn et al., 1994;Bates et al., 1996a, b] and repeat transect occupationsof certain oceanicregions[e.g.,Feelyet al., 1987;Penget al., 1987; Takahashi et al., 1993; Metzl et al., 1995]. However, only a few studieshave investigatedthe in situ variability of surfaceseawaterpCO 2 on lessthan seasonaltimescales[e.g.,Robertsonet al., 1993; Friederich et al., 1995; McNeil and Merlivat, 1996; DeGrandpreet al., 1997;Goyetand Peltzer,1997].For example, Robertsonet al. [1993] followed diel changesof pCO2 in the northeastAtlantic over a 5-day period, while Friederichet al. [1995] continuously monitored surface pCO 2 for several months to investigatecoastalupwellingoff the coastof Cali-
Biogeochemicaland physicalprocessesinfluenceboth the distributionof seawaterCO2 and the strengthand directionof air-seaexchangeof CO2 over a range of timescalesand space scales.In temperate and subpolarseas,seasonaldecreasesof the partial pressureof CO2 (pCO2), total carbon dioxide (TCO2), and nutrients are typically associatedwith phytoplanktonblooms[e.g.,Codispotiet al., 1982,1986;Takahashiet al., 1983;Watsonet al., 1991; Wongand Chan, 1991;Chipman et al., 1993; Takahashi et al., 1993; Poissonet al., 1993, 1994; Robertsonand Watson,1995], while in oligotrophicregions, surfaceseawaterpCO 2 is typicallycorrelatedwith temperature and salinity[Weisset al., 1982]. Over the last decadewe have fornia. From June 1994 to November 1995 we collected a gainedinsightsabout the seasonaland interannualvariability detailed time seriesof sea surfacetemperature,seawater,and atmosphericpCO 2 in the SargassoSea near two oceanictime Copyright1998 by the American GeophysicalUnion. series:HydrostationS (32ø10'N,64ø30'W) and the U.S. Joint Global Ocean Flux Study (JGOFS) Bermuda Atlantic Time Paper number 98JC00247. 0148-0227/98/98JC-00247509.00 SeriesStudy(BATS) site(31ø50'N,64ø10'W).In thispaperwe 15,567
15,568
BATES ET AL.: DIEL/SEASONAL VARIABILITY OF SEAWATER pCO 2 I
64ow
l 2000m J • - 32ON
32øN-
¸ Altomoor mooring site
ß BATS
65ow
64ow
I
I
Figure 1. Locationmapof the islandof Bermuda,Hydrostation S (32ø10'N,64ø30'W),andthe U.S. Joint GlobalOceanFluxStudy(JGOFS)BermudaAtlanticTimeSeriesStudy(BATS)site(31ø50'N,64ø10'W).The OceanFlux Programtime seriesof deepocean-moored sedimenttrapsare locatedjust to the southof the BATS site[Deuser etal., 1995].For clarity,cruisetracksbetweenBermuda,Hydrostation S,andBATSarenot shown.
describediel to seasonalvariabilityofpCO 2 and evaluatethe potentialcausesof suchvariability.Over shorttimescales we observedvariabilityof p COg due to severaltypesof physical forcing,suchas the diurnal thermalcycle,small-scale(3600 atmosphericpCO 2 readings.(A summaryof surfacehydrographicandp CO2 data is available from N. Bates and the following web address: http:// www.bbsr.edu on BATS and HydrostationS data page.) We used TCO 2 data in our analysisthat were collected monthly at BATS. TCO 2 was determinedby using a highly
precise(-0.4 p•molkg-•) andaccurate coulometric method, detailsof whichwere reportedby Bateset al. [1996b].Surface TCO 2 was calculatedfrom our underwaypCO 2 data, and alkalinity was calculatedfrom underwaysalinitydata usingappropriatealkalinity-salinity regressions [Bateset al., 1996a,b]. In addition, surfacepCO 2 was calculatedfrom monthly measurementsof surfaceTCO 2 and alkalinity using relevant dissociationconstants[i.e., Goyetand Poisson,1989;Roy et al., 1993;Millero, 1995] and thermodynamicconsiderations[Dickson and Goyet, 1994]. 2.4.
Gas Exchange Calculations
The exchangeof CO2 between ocean and atmosphereis driven by concentrationdifferencesin CO2 at the air-sea interface.The net flux of CO2 (F) can be expressedas
F = ks(Ap CO2)
(1)
where k is the transfervelocity,s is the solubilityof CO2, and ApCO2 is the concentrationdifferencebetween atmosphere and ocean.The transfervelocityk is a functionof wind speed. The relationshipbetweenk and wind speedhas considerable
15,570
BATES ET AL.' DIEL/SEASONAL VARIABILITY
OF SEAWATER pCO 2
29
28-
+
27-
262524-
23-
19
++ +
a
BATES ET AL.: DIEL/$EA$ONAL VARIABILITY
OF SEAWATER pCO 2
15,5'71
420
O
410 ß
400
Seawater pCO 2
a
........ •........ Atmospheric pCO 2
ß
390 ß
380 370 .
360 350 340 330 ß
320 ß
310 300
IOI N IDI jI FIM•A
j I F IM IA IM• J Ij
366
364362360358356354352350348346344
jI FIMI A IMI J Ij •A I S I OINID I jI F•MtA IMI j Ij. IAI SIO •N I D
2070
+ +
C
2060 -
•,,
+
,.• •m
+ 2050 -
+
'$
+
•
2040
•
2030
::.-:!.
+
+
+
2020
2010
J F M A M J J A S O N D J F M A • M• J IJ •A I S•O •NI D 1994
1995
Figure 4. Time seriesof surfaceandatmospheric propertiesin the regionaroundHydrostation S (32ø10'N, 64ø30'W)andBATS(31ø50'N,64ø10'W)fromJanuary1994to December1995.Eachdatapointdenotesthe mean value of all continuoussurface observationscollectedon each single day between June 1994 and
November1995.(a) Mean daily surfacepCO2 (/xatm) and atmospheric pCO2 (/xatm),(b) mean daily atmospheric xCO: (ppm),and(c) TCO: calculated frommeandailysurface pCO: andalkalinity,whichwas estimatedfrom the salinity-alkalinity regression observedat BATS [Bateset at., 1996a].
of seawater)was calculatedfrom the observedtemperature --•6.5 m and (2) daily averagedwind speedfrom Harbour and salinityusingthe equationsof Weiss[1974].Wanninkhof Radio on Bermuda, collected in !995, with the anemometer [1992]developed(5) for steadywinds(or short-termwinds), located at a height of --•25 m. These wind speedswere corwhereas (6) wasfittedto long-term 14Ctracerdataandthus rectedto 10 m usingthe equationsof Largeand Pond [1982] suitablefor climatologicallong-termwinds.We useddaily av- and Smith [1988]. eragedwindspeeddata(averaged fromhourlymeasurements) collected on the island of Bermuda
at the former U.S. Naval
Air Station.The anemometerwas locatedat a height of 10 m. Two additionalwind speeddata were used for comparative
purposes:(1) daily averagedwind speedfrom the ALTOMOOR (BermudaTestbedMooring) surfacemooringnear BATS [e.g.,Dickeyetat., 1998a,b], collectedfrom June1994to January1995,with the anemometerwaslocatedat a heightof
3.
Results
Variability of surfacetemperatureand salinity,seawater, and atmospheric pCO: (and other CO2 species)in the SargassoSea was observedon different timescalesand spatial scales.This variabilitycan be viewedin contextof the seasonality of physicsand biogeochemical patternsobservedhistoff-
15,572
BATES ET AL.: DIEL/SEASONAL VARIABILITY OF SEAWATER pCO 2 3
2.5-
•
0
1.5-
o
0.5x
J
F
M
A
M
J
J
A
S
O
N
X
•
D
0.35
•
0.30
•
0.25
•
0.20 •
0.15
o•
"•
X
0.10
•
•
D
• •
0.05
0.00
I
J
I
F
I
M
I
A
o
I
I
M
:• I
J
J
x I
A
I
S
I
O
I
N
D
454035-
x•
E•
30-
D
o
25-
x
o o
20-
15-
og
X
lO-
X0 x• x
xX xo
x
5-
• I
J
X i
F
x o o
i
A
00OxX:•X o xXoOcD o• •Sx•X • c• XX•xOx xo• X
i
M
oX
x
o
x• •
^X
i
M
x
xx • X
C])X
I
J
X
I
J
I
A
i
S
I
O
I
N
D
Figure 5. Daily rangesof surfacepropertiesplottedagainsttime of year.Symbols are designated byprimary causeof observedvariabilityon individualdays.Open circlesdenotedielwarmingandcooling,crosses denote mesoscale heterogeneity, and starsdenotespatialvariabilitycloseto the Bermudaplatform. (a) Surface temperaturerange(øC),(b) surfacesalinityrange,and (c) surfacepCO 2 range(/xatm).
cally and from 1994 to 1995. The annual changeof surface temperaturewas ---10øC,with a minimum of---19øC in February and March and a maximumof---29øCin August(Figure 3a). Suchchanges in surfacetemperaturehavebeenpreviously observed[e.g.,Menzel and Ryther,1960; Talleyand Raymer, 1982; Jenkins and Goldman, 1985; Lohrenz et al., 1992;
Michaels etal.,1994a] andresulted fromseasonal changes in
period of strong thermal stratificationin summer and fall, when nutrientsare generallynot measurablein the euphotic
zone(NO3 concentrations 0.5øC was frequentlyobservedbetweenMay and October(Figure5), when surfaceheat fluxeswere higher,wind speedswere lower, and mixed layer depthswere shallower(---5-20 m). Occasionally, diet thermaleffects(---0.2ø-0.4øC) were observedduringcalm daysin winter (November-April)despitereducedsurfaceheat flux and deep mixedlayers(>100 m). Surfaceseawaterp CO2 varied coherenttywith temperature on daysthat exhibitedcharacteristic diurnalwarmingandcool-
ing patterns (Figure 6). Day-to-dayvariability of the diet rangesof pCO 2 (---5-35 txatm),shownin Figure 5, reflected variabilityin the processes controllingthe diurnal thermal cycle. On dayswhen diurnal warmingwas greatest(>1.5øC), pCO 2 varied by ---25-35 txatm.This variabilitywas nearly 3040% of the annualvariabilityobservedeachyear. Analysisof p CO2 and temperaturedata from cruisesshownin Figure 6 yieldedslopesfrom a 13.6-16.9 txatmchangeinpCO 2 per iøC (or 3.3-4.0%) temperaturechange(Figure7). Theseratesof p CO2 changeare consistentwith ratespredictedfor the thermodynamiceffectof temperatureonpCO 2 (variouslyreported in a rangefrom 4.0 to 4.3% per iøC change[e.g.,Takahashiet al., 1993;Goyetet al., 1993;Millero, 1995]). This findingindicatesthat diet changesin pCO 2 were primarily controlledby diurnalwarmingand coolingrather than other processes such asbiologicalproductionor gasexchange.The observed pCO 2temperatureslopeswere lower than predictedby thermodynamicsperhapsbecauseof a minor loss of CO2 from the surfacethrough productionand gas exchangeof CO2. Daily changesinpCO 2 due to productionwere minor (---0.5-3/xatm
d-•), sincesurfaceprimaryproduction rateswerelow (5-20 mg m3 d-• [Knapet al., 1991,1992,1993]).Similarly, pCO2 changes dueto gasexchange werealsominor(2.5-4 m s- • [Fairalletal., data from the U.S. Naval Air Station.Difference in CO2 fluxes 1996a],while maximumskineffectsoccurat light wind speeds for 1994and 1995betweenthe threewind speeddata setswere (---1.5rn s-•) [Coppinet al., 1991].Light wind conditions 60-70% and almost equivalentto the uncertaintyassociated (
