The relationship between synoptic weather ... - Wiley Online Library
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 104, NO. C8, PAGES 18,159-18,185, AUGUST 15, 1999
The relationship between synoptic weather systems and meteorological forcing on the North Carolina inner shelf JayA. Austin1 Massachusetts Institute of Technology/WoodsHole OceanographicInstitution Joint Program in Physical Oceanography,Woods Hole, Massachusetts Steven J. Lentz Department of Physical Oceanography,Woods Hole OceanographicInstitution, Woods Hole, Massachusetts
Abstract. A strong relationshipis observedbetween synopticweather systemsand atmosphericforcing of the oceanas estimated from buoy measurementsmade on the North Carolina inner shelfduring August and October-November1994 as part of the
CoastalOceanProcesses (COOP) Inner ShelfStudy. Synopticvariation (timescales of daysto weeks)in the meteorological time serieswas primarily associatedwith the passageof atmosphericfrontal systems. The most common synoptic weather pattern observedwas the passageof a low-pressurecenter to the north of the study site, which causedthe associatedcold front to pass over the study region. Before passageof the cold front, warm, moist northeastwardwinds increasedthe heat flux
intothe ocean,whereas afterthe coldfrontpassed, cold,dry southwestward winds decreasedthe heat flux into the ocean. In addition, in the presenceof oceanic stratification, northeastward winds drove coastal upwelling, bringing colder water to the surface, further increasingthe air-sea temperature contrast and hence the heat flux into the ocean inshore of the surface front between cool upwelled water and warmer water offshore. The decreasein surface heat flux during the passage
of a coldfront wasof order400W m-2, due primarilyto a decrease in latent heat flux. Although other synoptic patterns were observed,including one warm front passageand two tropical storm systems,the dominanceof cold fronts as a sourceof variability resultedin a strongpositivecorrelationbetweenthe along-shelf component of wind stressand the surfaceheat flux. To addressthe issueof spatial variation in the surface heat fluxes, data from several different sourceslocated along a cross-shelftransect were analyzed. This analysissuggeststhat the temperature of the atmospheric boundary layer undergoesadjustment when warm air blows over cold water but not when cold air blows over warm water. This producescross-shelf gradients in the bulk estimatesof turbulent heat fluxes during offshorewinds but not during onshorewinds. 1.
Introduction
estimates from observations taken off the coast of North
Carolina, north of Cape Hatteras during August and October/November1994 as part of the CoastalOcean Processes (COOP)InnerShelfStudy(ISS)fieldprogram [Burman,1994]. The primary purposeof this paper is
The surfaceheat flux and wind stressplay a crucial role in determining the behavior of the upper ocean, especially in the shallow coastal zone, where the entire water column can be directly influencedby atmospheric
forcing[WinantandBeardsley, 1979;Leeet al., 1989]. to describethe effectof synopticweathersystemson the
This paper presents surface heat flux and wind stress temporal and spatial variation in meteorologicalforcing. There have been few previousobservationalstudiesof the surface heat flux over U.S. continental shelves and even fewer which use direct in situ measurements of the
1Nowat theCollege ofOceanic andAtmospheric Sciences, OregonState University, Corvallis Copyright1999bytheAmericanGeophysical Union. Papernumber1999JC900016. 014g-O227/99/1999JC900016509.00
radiative fluxes. On the U.S. west coast, Beardsleyet
al. [1998]estimatedsurfacefluxesduringthe Coastal OceanDynamicsExperiment(CODE) and the Surface Mixed Layer Experiment (SMILE). They studiedseasonal and synoptic variation during both experiments. 18,159
18,160
AUSTINANDLENTZ:METEOROLOGICAL FORCINGONTHENCSHELF
They foundthat variationin shortwaveradiationand either sideof the front, they estimateda sharpdecline, ofthecomsea surfacetemperature(due to upwelling)were the oftheorderof400W m-2, in bulkestimates across most important factorsin the seasonalvariationof the binedturbulentheat fluxes(latentand sensible) net surface heat flux. SMILE was one of the first coastal the composite front. They alsoobserved that the highoceanic field experiments to make direct observations est concentrationof cloudslies alongthe front, another
the surfaceheat flux, of downwardlongwaveradiation.In the SouthAtlantic importantfactorin determining Bight, Blantonet al. [1989]estimatedturbulentfluxes and that the wind directionchangesduring the passage of moisture,heat, and momentumas part of the Gen- of the front. Thepurpose oftheinterdisciplinary CoOPInnerShelf esisof Atlantic Lows Experiment(GALE). They atof the processes tributed variation in meteorologicalobservations,and Study was to increaseunderstanding hence in the estimated fluxes, to synopticweather sys- that affect larval distributions over the inner shelf, as tems. For instance,the passageof a cold-airoutbreak well as to increaseknowledgeof the physicaloceanog-
on January 27, 1986, was responsiblefor an estimated raphy of the inner shelf,a regionof the oceanwhere decrease in the surface heatfluxof nearly1400W m-2. there have been relatively few physicaloceanographic Mountain et al. [1996]useddata from mooredbuoys studies. The field programtook place betweenAuand coastal stations to estimate the annual cycle and
gustandDecember1994,whichbracketed the seasonal
interannualvariability of the net surfaceheat flux in the transitionfrom strongstratificationand surfaceheating Gulf of Maine between 1979 and 1987. In this study, to weak stratification and surfacecooling. The study annual variation in the shortwave insolation was pri-
site was located on the shallow shelf between Chesa-
marily responsiblefor the annual variability of the net peakeBay and CapeHatteras(Figure1) and included heat flux. On a larger scale,Bunker[1976]estimated two mooredbuoysinstrumentedwith the meteorologmonthlymean surfacefluxesoverthe entireNorth At- ical sensorsnecessaryto make bulk estimatesof the lantic Oceanusingdata from over 8 million shipboard surface heat flux and wind stress. weather observations and discussed heat flux variabilVariability in the meteorologyon timescalesof days to weeksduringthe CoOPInner ShelfStudycanbe atity on annual scalesfor the Mid-Atlantic Bight, among tributed to three basic scenarios,listed here in order of other specificregions. Enriquezand Friehe [1997]infrequency of occurrence: the passage of coldfronts,the vestigatedthe effect of coastalupwellingoff northern passage of tropical storms, and the passageof warm California on the stability of the air column, which in part determinesthe transfer of heat, momentum,and moisture between atmosphereand ocean. They found that the changein the transfer coefficientsdue to upwelling affected estimatesof surfacewind stress,but had a negligibleeffecton the turbulenttransferof heat.
fronts. The responseof the local meteorologicalvariablesto these weather systemswas distinctive. A key result of this study is that the predominanceof cold fronts as sourcesof variation and the particular response of the surface heat flux and wind stressto their passage
leadsto a strong relationshipbetweenwind direction consideredin the open ocean. The Frontal Air Sea In- and surface heat flux. teractionExperiment(FASINEX) wasan observational The rest of the paper is organizedas follows.Section program that studied the effectsof sea surfacetem- 2 describes the instrumentation and methods used to The effect of fronts on surface fluxes has also been
perature fronts and atmospheric fronts on open ocean surface heat flux and wind stress variability. During
estimate surface heat flux and wind stress. In section
3, meteorological time seriesand surfaceflux estimates FASINEX, Davidsonet al. [1991]observedsharpde- are used to examine the relationship between synoptic creases in surface heat fluxes and differences in wind meteorology and temporalvariationsin surfacefluxes. directionduring the passageof coldfrontsoverthe open Section4 is a discussionof cross-shelfgradientsin the ocean southeast of Bermuda, with decreasesof surface air and sea surfacetemperature fields and the implica-
heatfluxof up to 600W m-•' duringindividual frontal tions this has for the spatial distributionof surfaceheat passagesobservedduring January-May 1986. Also dur-
flux. Section 5 is a summary.
ing FASINEX, Friehe et al. [1991]studiedthe effectof sea surfacetemperature fronts on atmosphericbound- 2. Field Program, Methods ary layer structure, showingthat warm air blowingover 2.1. The Site cold water leads to a stable, shallow boundary layer, while cold air blowing over warm water leads to an unThe CoOP Inner Shelf Study took place offshoreof stable, growingboundarylayer. Mooerset al. [1976] the North Carolina Outer Banks, between Cape Henry studied the effects of cold fronts on surface heat flux and (at the mouth of the ChesapeakeBay) and Cape Hatwind stressusing a compositeof 34 low-pressuresys- teras, at the southernend of the Middle Atlantic Bight tems observedover the Middle Atlantic Bight between (Figure 1). The coastlineis relatively straightin this 1972 and 1975, for use as an idealizedforcingfield for region with an orientation of approximately 340øT. The ocean models. This compositelow-pressuresystemin- shelf is approximately 80 km wide and 60m deep at the cluded a trailing cold front with warm, moist air ahead shelf break, increasingin width to the north. On the of the front and cold, dry air behind the front. On the western side of the Outer Banks lie Currituck Sound, basis of the air temperature and moisture content on PamlicoSound,and AlbemarleSound,whichare large,
AUSTIN AND LENTZ' METEOROLOGICALFORCING ON THE NC SHELF
18,161
36.28
CoOP Inner Shelf Study Central line moorings
36.26
I
ß 36.24
340ø-
'.
!
5 km
--Fd3 (26m)
37
36.22
Cape Henry z
'o
36.2
•\c,..?,LO?•'. -I-d2 (21m)
36.5 04401
"+dl (13n:i).
36.18
e44006
Region of DItlill Atlantic
36.16
36.14
35.5
ß10m
'.20m Ca
36.12
35 76
-75.5
-75
-74.5
I
-75.75
-75.7
-75.65
-75.5
-75.55
-75.6 o
Longitude, E
Figure 1. A plan view of the CoastalOceanProcesses Inner Shelf Study (COOP ISS) observational site, showingthe study area, and the locationsof CoOP ISS buoys DP, dl, d2, and d3, and National Data Buoy Center buoys44006, 44019, and 44014. The inset is a regionalview.
-5
-10
E c.•
m -20
-25
-30
-35 0
5
10
15
20
25
30
35
40
45
Offshore distance, km
Figure 2. Cross-shelfconductivity-temperature-depth (CTD) sectiontaken from R/V Cape Hatteras on August 19, 1994, during upwelling-favorablewinds. The section extends 50km offshorefrom the Field ResearchFacility, perpendicular to the shore. The positions of the three CoOP ISS mooringsare indicated by vertical dashedlines.
50
18,162
AUSTIN AND LENTZ: METEOROLOGICAL
FORCING ON THE NC SHELF
28
'1•' ""/'"" L
(A)
O 26=24-
•2.tr,.]• ••,1• •
.I
•'22 -
,
t/.•.'..""l-' 'd.
i't ;;.•; '.
.j.:• ....
':r';
I
,
x
FRF
I ...... NDBC 44006
• i' r
........
d2
I ---
NDBC44019
:'.. •.•,, •z.• ½ {.•.l• - ,L•:•'" .: ,•:' :: '•.
f,•., / •:: :•i..7'
;";,'•• cl"'•: .,.:'...ß.x-.... •
::
:':•'•i::-: '.x ': x
',..... ;: ,. :............x ,'-;x''.x'' .. ß
• • ,•
.'..x , .',: ' .'ß:.x:..j'•'" x•x
ß 20-
,
o
,,.
"- "x
,
x
c/) 18-
CTD section (Figure 2)
z16I
I
i
Aug 8
Aug13
Aug18
14
Aug 3
I
i
I
Aug23 Aug28
Sep2
Se3
7
28 x
o
0 26
d2
(B)
NDBC 44006
NDBC 44019 =24
•'22 e 20 o
0318
Z16
tl
i
i
i
i
I
I
I
I
i
ct 2 Oct 7 Oct 12 Oct 17 Oct 22 Oct 27 Nov 1 Nov 6 Nov 11 Nov 16Nov 21
Figure 3. Near-surfacewater temperaturesfrom the FRF pier, d2, NDBC buoys 44006 and 44019,for (a) Augustand (b) October.FRF valuesare daily, othersare hourly. Arrow indicates timing of CTD sectionin figure 2.
shallow inland bodies of water, characterizedby high
to 6m thick, at a depth of approximately10m (Fig(northward)windsbrought surfacetemperatures duringsummer[Roelofs andBum- ure 2). Upwelling-favorable pus,1953].The mooredobservations focused on a cross- the thermoclineto the surface,resultingin cross-shelf shelf section located offshoreof the Army Corps of En- seasurfacetemperaturedifferences of up to 8øC (Figgineers'Field ResearchFacility (FRF) in Duck, North ure 3a), which play a significantrole in determining Carolina.
This section extended
16 km offshore and con-
the spatial distribution of the surface heat flux. Dur-
the watercolumnwasrelatively sistedof three main mooringsites,dl, d2, and d3, two ing October-November, of which (d2 and d3) wereinstrumentedwith meteoro- well mixed in temperature(as inferredfrom mooring logical equipment. data),andcross-shelf temperature differences weretypically lessthan 1øC (Figure3b). 2.2.
Oceanographic
Setting
The CoOP Inner Shelf Study took place during August and October-November1994, which represented two very distinct oceanicsettings. During August, the water columnwascharacterizedby a strongthermocline with a temperature differenceof 5øC to 9øC, usually 3
2.3.
Instrumentation
and
Other
Data
Sources
Many sourcesof data are utilized in this study to understandthe fluxesat the CoOP ISS site duringAugust and October-November 1994 and to put these observations into a wider regional and temporal context. The instrumentation
is summarized
in Table 1.
AUSTIN AND LENTZ: METEOROLOGICAL
FORCING ON THE NC SHELF
18,163
Table 1. Meteorological Instruments onthe Vector-Averaging WindRecorder andAssociated Moorings Parameter
Instrument
Height,m
Accuracy a
2.7
+0.4 ø
dœ VA WR instruments
Air temperature
Thermistor
YellowSprings
(WS > 5 ms-•)
5 K at 25 øC Barometric
pressure Longwave Radiation Relative
humidity
Paroscientific
2.7
model216-B-101 pyrgeometer
+0.6 mb
(WS < 20ms-1) 3.3
+5%
2.7
+ 5%
Eppley PIR Vais ala
Humicap 0062HMP
Insolation
Pyr anometer Eppley 8-48
3.3
4-5%
Wind direction
Integral vane
2.7
4- I bit
w/vane follower WHOI/EG2•G
Windspeed
R.M. Young
(5.6ø)
2.7
4-6%b
-1.1
•:0.5 øCb
3-cup anemometer
Water
thermistor
temperature(d2)
Water
Seacat
Other GoOP Buoy Instruments
-2.0
+0.5øCc
-4.0
4-1oCc
Air temperature Water temperature Wind speed
10 - 1.5 10
4-1øC 4-1øC
Wind direction
10
4-10 ø
Air temperature Water temperature Wind speed
5 -1 5
4-1øC 4-1øC
Wind direction
5
4-10 ø
temperature, dl, d3
Water
Seacat
temperature, DP NDBC
NDBC
NDBC
44006
NDBC Instrumentsd
4-1 ms -1 or 10%
44O19
4-1 ms -1 or 10%
44O14
2.3.1. Moored instrumentation. Meteorologi- midity, downwardshortwaveand longwaveradiation, cal instrumentationfor the experimentconsistedof two barometricpressure, seasurfacetemperature,andwind buoysequippedwith vectoraveragingwind recorders speed and direction to make bulk estimates of wind
(VAWRs,Table1), locatedat d2 andd3 [Alessiet al., stressand surfaceheat flux. The meteorological mea1996].Thesesiteswere5 and 16.4kmoffshore, on the surementsweremadeat heightsof 2.7 to 3.3m (Table 21 and 26-misobaths,respectively. Only data fromthe 1), and near-surface watertemperaturemeasurements
d2 VAWR were recovered. The d3 VAWR was lost during Hurricane Gordon on November 18. Each VAWR
recordedmeasurements of air temperature,relativehu-
were made at a depth of 1.1m. Data were recorded ev-
ery 7.5 min and weresubsequently block-averaged into hourly values. Each buoy was equippedwith an Ar-
18,164
AUSTIN
AND LENTZ:
METEOROLOGICAL
FORCING
ON THE NC SHELF
Table 1. (continued) Parameter
Instrument
Height,m
Accuracy a
Air temperature Water temperature Wind speed
5 -1 5
•-1øC •-1 øC 4-1ms -• or 10%
Wind direction
5
4-10 ø
20 19.5 19.5
n/a n/a n/a n/a
FRF
Instruments
Water temperature
bucketthermometer
Air temperature Wind Speed Wind direction
YSI thermistor F420 anemometer, NWS
Air temperature
R/V CapeHatterasInstruments R.M. Young41372C 15.25
n/a
Water temperature
YSI 701
2øCe
0
Abbreviationsare WS, wind speed;PIR, precisioninfraredradiometer;WHOI, WoodsHole Oceanographic Institution; COOP,CoastalOceanProcesses; NDBC, National Data Buoy Center;FRF, Field ResearchFacility; n/a, not available; NWS, National Weather Service. aValues denote estimated instrument accuracy;manufacturer'sspecificationsunlessotherwisenoted.
bValueis from Welleret al, [1990]. cAlthoughthe sensors themselves havegreateraccuracythan shownhere,the valuerepresents their estimatedaccuracy as a measure of surface water temperature.
dData are takenfromthe NationalData BuoyCenterwebsite,http://seaboard.ndbc.noaa.gov. eValue denotesresolution, not accuracy.
gostransmitter to transmit the meteorological data and
16.4km offshorein 25 m of water. All three moorings
buoy position to shore.
were instrumented
The d2 and d3 meteorologicalbuoysweredeployedon August 6, and recovery was planned for early December. However, failure of the surface moorings during severe storms and failure of the d3 Argos transmitter systemresulted in only 2.5 months of data from the d2
vertical structure of the water column. Hourly water temperature measurements were made near the end of the FRF pier at a depth of 4.0 m usinga SeabirdInstru-
site and almost no data from the d3 site. The d2 surface
mooring failed on September4 during a tropical storm and came ashore with
all of its instrumentation
intact.
The mooring was refurbishedand redeployedon October 4. It failed again on November 18 during Hurricane Gordon and came ashore. Although the tower with the VAWR was torn off the buoy as it came ashore, it was eventually recoveredwith all data intact. This sequence of events determined the two time periods considered here, which are designatedthe "August" time period, 0100 UTC on August 7, 1994, to 0000 UTC on September 4, 1994, and the "October" time period, from 1500 UTC on October 4, 1994, to 0800 UTC on November 18, 1994. The d3 surface buoy stayed in place until Hurricane Gordon on November 18, when it also came ashore.Unfortunately,the VAWR on the d3 buoy (and
with
thermistors
to determine
the
mentsSeacat(referredto asDP, for "DuckPier"). This is rather deep for estimating surfacetemperatures, as the mean temperature difference at dl between 4.6-m and 1.5-m depth is approximately0.6øC during August
(maximumdifferenceof 3.2øC), sothis measurement is most likely an underestimateof the sea surfacetemperature of order 1øC during August. During October, no data were available at the surface at dl, but the difference between the temperature at 4.6 and 1.1m at d2
was, on average,of the order or 0.02øC (with a maximuminstantaneous differenceof 1.1øC),suggesting that the temperature at 4-m depth is a reasonableproxy for the surface temperature. 2.3.2. Field Research Facility measurements. Wind speed and direction, air temperature, barometric pressure,and water temperature measurementshave been taken almost continuouslysince1982 at the Army
Corpsof Engineer'sFieldResearch FacilityFRF [Birkehenceall of the meteorological data from the d3 site) meier et al., 1985]. Wind speedand directionwere
was lost as the buoy came ashore. The d3 Argostransmitter failed on August 12, was repaired on September 1, and failed again on September8. The amount of data recoveredvia the d3 Argostransmitter was too small to be useful for this study. Other measurementsfrom the moored array consisted
of near-surface (2-m depth)watertemperatureat the dl site, 1.4 km offshorein 13 m of water, and at the d3 site,
measured using an anemometer located 19.4m above
sealevel at the end of the FRF pier (560m offshore). Air temperature was measuredin an instrument housing 40m onshoreand appearedto sufferfrom a diurnal instrument heating problem. Therefore only nighttime valueswere used. During onshorewinds, nighttime values typically differed from measurementsat d2 by less than 1øC, suggestingmeasurementerrors of the order
AUSTIN AND LENTZ: METEOROLOGICAL FORCING ON THE NC SHELF
18,165
of IøC, though no actual uncertainty valuesare avail- tum, heat, and moisturefluxes,the appropriateness of able. Sea surface temperature was measured using a the bulk formulasof Fairall et al. [1996]cannotbe bucket thermometer at the end of the pier, once per judged. The uncertaintyof the meanflux estimatesdue day, typically around 0700 local time. These data were to measurement uncertainty will be discussedin section highly correlated with buoy measurementsduring the 2.5. August and October 1994 periods, and were used to The surfacewind stressis estimatedusinga stabilitygain someperspectiveon the regional seasonaland in- dependent transfer coefficientthat relates the measured terannual variability, as well as qualitative information wind speed to the wind stress, on the cross-shelfvariation during the experiment. T ----pACd(Ua-- u)lu - u•l, (1) 2.3.3. NDBC buoy array. The National Data Buoy Center (NDBC) maintainedthree meteorological where PA is air density, Ca is a stability-dependent buoys,designations 44006,44019,and44014(Figure1), transfer coefficient, ua is wind velocity measured at a acrossthe shelfnear the FRF, which provedusefulto specifiedheight (in this case2.7m), and us is the surthis study. Thesebuoysrecordedhourly wind speed face velocity of the water. The direction of the wind anddirection,air temperature,andwatertemperature, stressis assumedto be the same as the velocity differwith quoted accuracieslisted in Table 1. While insufence vector Ua --Us. ficient for completeheat flux calculations,these data The net surfaceheat flux may be consideredthe sum were used to examine cross-shelf variations in the me-
teorology,estimate sensibleheat flux, and infer latent heat flux.
2.3.4. Shipboard data.
The R/V CapeHatteras
of four terms [Gill, 1982]:
QTOT-- QSWNE T + QLWNE T + QLAT + QSEN,(2)
performed conductivity-temperature-depth (CTD) tranQSWNET isnetshortwave radiation, QLWNET is sectson the North CarolinainnershelfduringAugust where net longwaveradiation (the differencebetweenupward andOctober1994aspart of the CoOPInnerShelfStudy radiation),QLAT is latentheat [Waldorfet al., 1995,1996].It made16transects of the anddownwardlongwave flux, and QSEN is sensibleheat flux. In this paper, shelfalongthe FRF to d3 centralmooringline in Au-
positive flux values always denote heat flux into the gust and 20 in October. In addition, cross-shelftran- ocean. sectsnorth and southof the centralmooringline were Thenetshortwave radiation fluxQSWNET duetosomadeto quantifyalong-shelf variationin the hydrogralar insolation between 0.28/zm and 2.8/zm, is estimated phy. During eachcruise,air temperature,surfacewater temperature, and wind velocity were measuredevery using 15 s. The air temperature measurementssufferedfrom QSWNET -- QSWDow N(1- Ab), (3) very low (2øC) resolution,sothe air temperaturemeawhere QSWDow N is measured using anEppley 8-48 surementsare usedonly for qualitative comparisons. pyranometer and Ab is the albedo of the sea surface, which is determinedempirically as a function of the so2.4. Estimation of the Surface Heat Flux and Wind
Stress
lar angleand atmospheric transmissivity[Payne,1972].
Thenetlongwave radiation fluxQLWNET isthedifference between the upward longwave radiation QLWu P heatfluxwereestimated usingbulkformulas developed The surface wind stress and the turbulent surface
by Fairallet al. [1996].Likemostotherbulkfluxalgo- due to blackbodyradiation from the oceansurface,cal-
rithms,theseweregenerated usingmeasurements made culated usingthe Stefan-Boltzmanlaw, and the downin the open ocean, away from boundariesand fronts. A comparisonof the heat flux and surface stressesti-
wardlongwave radiation QLWDow N duetoradiation
the Largeand Pond[1981,1982]formulationrevealed
is
from moisture in the atmosphere,and is measureddimatedwiththisformulation withthoseestimated using rectly. The formula for the net longwaveradiation flux no qualitative differences. The appropriateness of these bulk formulas to the
QLW•:T --e(QLWDow• --ate),
(4)
wheree - 0.98 [Dickeyet al., 1994]is the radiativeefof e varyfrom0.93to fewkilometers of the shoreline. Theirsuitabilityis es- ficiencyof the water(estimates 1.0 [Fung et al., 1984]), rr = 5.67 x 10-s W m-2K-4 is peciallyquestionable duringoffshorewinds,whenthe the Stefan-Boltzman constant, and Ts is the seasurface marineatmospheric boundarylayermustquicklyadjust to newsurfaceconditions. In addition,surfacewaves, temperature indegrees Kelvin. QLWDow Nisdue toinsteepenedin shallowwater, may affectthe transferco- fraredradiationin the range3.5 to 50/zmemittedby moistureandis measured directlyusingan efficients andhence estimates ofturbulent fluxes[Large atmospheric et al., 1995;Geernaertet al., 1986].It shouldbe noted Eppleyprecision infraredradiometer(PIR) pyrgeomecoastaloceanis not clear, especiallyto siteswithin a
that neither of these effectshas been taken into account in the estimatespresentedhere. In the absenceof di-
ter. The downwardlongwaveradiation was corrected
forinstrument heating bysubtracting 0.035 QSWDow N
rect (i.e., eddycorrelation)measurements of momen- [Alados-Arboledas et al., 1988].
18,166
AUSTIN AND LENTZ: METEOROLOGICALFORCING ON THE NC SHELF
The latent heat flux QLAT representsthe heat releasedor gained when water evaporatesfrom or condenseson the oceansurface.Althoughoften interpreted as representingonly evaporation,there is a significant portionof the Augusttime seriesduringoffshorewinds when it appearsthat heat was being gained owing to
as possibleusing manufacturer'sspecifications(Table 1) and literature valuesof uncertaintyto providea context for interpreting the surfaceflux estimates.This error analysisrevealswhat terms of the surfaceflux (and what instruments)contributemost significantlyto uncertainty in the estimate of the total surfaceheat flux
condensation at the seasurface[Beardsley et al., 1998]. and wind stress, given what is known about the unThe latent heat flux is related to the moisture gradient at the ocean surface and the air-water velocity differenceusing the bulk formula
=
certainties of the measurements. For a more thorough treatment of uncertainty in the VAWR measurements and how they apply to flux estimates, see Weller et al.
[1990]and Beardsley et al. [1998].Resultsof the error
- q0)lu - ul,
analysisare summarized in Table 2. The quoted accuracy of the wind speed measurewhere L is the latent heat of evaporation; qz is the spe- ment is 2%, but there have been indications that cup cific humidity measuredat height z abovesealevel (in anemometersare prone to overspeedingby as much as
this case,z - 2.7 m); q0 is the estimatedhumidity at
the sea surface, calculated by assumingthe air at the water surfaceis the same temperature as the water and
6% [Weller et al., 1990]. Usinga nominalvalueof 6% for the uncertainty of the wind speedestimate and assuming that error in the estimate of wind stressis due
that the air is saturated(assumingthat the saturation primarilyto this measurement (as opposedto the meahumidity for air over salt water is 0.98 of the saturation humidity of air over fresh water of the same temper-
surement of surface ocean currents, which are relatively
small),overspeeding couldbe responsible for uncertainature); and Ce is a stability-dependent transfercoeffi- ties in the stressestimatesof up to 12%. cient,estimatedusingthe Fairall et al. [1996]formulaAn analysis of error propagation in the latent heat
tion.
flux bulk equation suggeststhat the main sourceof in-
The sensible flux QSENis relatedto the temperature strument error is the measurementof relative humidity, difference between the air and sea surface and the airassumedto be •- 5% [Weller et al., 1990]. This error
water velocity differenceusing the bulk formula
=
causesan uncertainty in the estimate of the mean latent heat flux of order 12 W m -2. The largest sourceof uncertainty in the sensibleheat
- Ts)lu - ul,
whereCpis the heat capacityof water,T$ is the seasur-
flux is the determination
of the difference
between
air
face temperature, TA is the air temperature, and Ch is and surfacewater temperatures. The uncertainty in the a stability-dependent transfer coefficient,alsoestimated differenceis greatest during strong insolationand weak winds, when the air temperature sensor suffers from usingthe Fairall et al. [1996]formulation. overheatingand the water temperature sensorunderestimates the surfacetemperature owing to formation of 2.5. Estimation of Instrument-Induced Flux stratification in the upper meter of the water column. Uncertainty During strong winds, both of these effects are small. Althoughit is impossible to takeinto accountall po- Sincethe sensibleflux is proportional to the wind speed, tential sources of error inherentin makingsurfaceflux uncertainty in the temperature differenceduring wcak
estimates,it is essentialto make as good an estimate
winds leads to small uncertainties
in the sensible flux.
Table 2. Statisticsof Augustand OctoberTime Seriesof Heat Flux ComponentsMeasuredat d2. Mean
Aug.
Standard
Oct.
Deviation
Subtidal
Standard
Aug.
Oct.
Aug.
Oct.
QSWNE T QLWD OWN QLWu P QLWNE T
2204-11 377 4-19 -4224-7 -454-20
1444-7 325 4-16 -3994.5 -744-17
301 27 8 28
222 37 8 37
83 38 31 29
59 39 23 35
QSEN QLAT QTOT
54-4 -334-13 1474-26
-154-4 -944-12 -394-22
19 72 317
31 98 258
22 68 135
30 93 138
Deviation. a
Uncertaintiesreflect instrument-baseduncertaintiesonly.
Allvalues areinWm-2. Subscripts denote SWNET, netshortwave flux;LWDOWN , downward longwave flux;LWup,
upwardlongwaveflux; LWNET, net longwave flux;SEN, sensible heatflux;LAT, latent heatflux; TOT, total heatflux. aThe subtidal standard deviationrepresentsthe standarddeviationof the low-passedtime series.
AUSTIN AND LENTZ: METEOROLOGICAL FORCING ON THE NC SHELF
18,167
1000 (A)Net ................. •-..t............................. •.. •... •........ 500........... •........... •............ I
I
I
I
I
II
I
I
I
.
.,
.
•
• •
I
o 50
N
.I
I
I
I
I
II
I
I
...... •... o .(B) et.longwave -50
..
-100
....
I
-150
200
,
,
(C)Latent • -200-'
ß,
100
.
.
,
•• I
.,
,
•
• , ..........,,........... •.....il .-•........ I
I
....
(D)Sensible ......
- 100 1000 500 0 -5OO
0.2 (F)Wind Stress [
•
I
I
[
I[
[
I
I[
Aug 8
I
ß ................. , .... ,,........ ......... I
-0.2 •< •h•re Aug 3
]
[
Aug13
[ i ,
Aug18
[•
,
Aug23
Aug28
•
I
i,
Sep 2
i
Sep 7
Figure 4. Termsof theheatfluxandthewindstress fortheAugusttimeperiod,measured at
d2' (a) netshortwave, (b) netlongwave, (c) latent,(d) sensible, and(e) netsurface heatflux (hourly (solid) andlowpass filtered (dashed)) andwindstress (lowpassed andsubsampled every 6 hours).Thewindstress is defined withalong-shelf upandoffshore to theright.Thevertical
dashedlinesmark the meteorological eventslistedin Table4.
With an estimatedtemperaturedifference uncertainty ationis about7Wm-2 in August. In October,the of order 1.0øC,muchlargerthan the air temperature uncertainty in the estimation of e is dominant, and the andwatertemperatureinstrumenterrorsalone,the cor- uncertainty is around5 W m-2 . respondinguncertainty in the sensibleflux is of order The downward longwaveradiation was measureddi12 W m -2. rectly, and the instrumentuncertaintyis 5%. Beardsley Upward longwaveradiation is a function of the es- et aL. [1998]comparedrecordsfrom two PIR records
timated seasurfacetemperatureand the emissivityof and founda differenceof 4.2%, but recentstudiessugthe seasurface.Assumingthe uncertaintyin seasur- gest that even this may overestimatethe uncertainty facetemperaturein Augustis IøC (due primarilyto [Fairall et al., 1999]. Using5% as an estimatein the near-surfacevertical temperaturegradients;the value uncertaintyof the measurement,the uncertaintyin the for Octoberis mostlikelysmaller)andthe uncertainty downwardlongwaveflux estimateis of order19W m-• in the emissivity to be +0.01 [Dickeyet al., 1994],the in Augustand 16W m-• in October. total uncertaintyin the meanupwardlongwaveradiIt will be assumed that most of the error in the es-
18,168
AUSTIN AND LENTZ' METEOROLOGICAL FORCING ON THE NC SHELF
-,ooo "'(A)'sh0rave, 500........
I
,
,
I
o
,
,
I
'•
l',
lOO
'
I
'
I
'
'
I '
(B) Net Lengwave ß ':
o
'
'1
' I
] I
I'
•[
'I
' I
'
I
-100 -200
0
( [
'
-200
' '
I ]
[
I'
'1
'I
I
I
'
I •
' I
' I
,
•
I[
•
I[
'
ß
-400
100
'
'
J
(D)Sensible ,I.... ..... -100
•
/•,
,I
. [.
[
..,,
-ß
I
I
(E) N
500
0
lUx
I
[
I
. .
I
[
I
.
I
[
I
I
ß
'
-500
L
[
i
i
i
I [
I [
o.2j(F)Wind •tress , -0.2J--....... J
Oct 2
. •//•/.////.
I '/,,/•/I
Oct 7
Oct12
[
:.1.................... I
I
Oct17
I
Oct 22
I[
,
/./VJ[.... I I
Oct 27
[I
, I.
I I
Nov 1
I I
i
I[
'
.• /.! ./../..I..
. ' ..
, ,
I I"
Nov 6
II
Nov11
i
Nov16
Nov 21
Figure 5. Sameas figure4 but for the Octobertime period.
timation
of shortwave radiation
is in the measurement
3.
Observations
and Results
itself,asopposed to the altitude-dependent albedo.The During the Augustand Octobertime periods,the uncertaintyin the measurement for an ungimballed sen- surfaceheat flux (Table2 and Figure4 (August)and soris 5% [Wetteret at., 1990],corresponding to mean Figure5 (October))variedon two distincttimescales:
uncertainties of 11Wm-2 in Augustand7Wm-2 in diurnal(daily)andsynoptic (timescales of 2-7 days). October.
As the instrumentsthat are primarily responsiblefor the uncertaintyare differentfor eachterm on the net surfaceheat flux, the uncertaintiescan be considered independent. Making this assumption,the estimated uncertainty in the mean net surface heat flux due to
This study focuseson the synopticvariability,which is evidentin all of the meteorological time series(Ta-
ble 3 andFigure6 (August)andFigure7 (October)), and hencethe surfaceheat flux components.Diurnal variabilitywas due almostentirelyto the daily shortwaveradiationcycle.This variabilitywasremovedfrom
measurement uncertainty isapproximately 26W m-2 in the data usingthe p164lowpass-filter[Beardstey et at., Augustand22W m-2 in October.Thelargestsources1985].However,thereis alsoa largedifference between of error, givenwhat is known about the error character- the mean surfaceheat fluxesduring August and Octoistics of the measurements,are the downward longwave ber, presumablyassociatedwith seasonalvariationin and latent
heat flux estimates.
the meteorology, whichis discussed first.
AUSTIN AND LENTZ: METEOROLOGICAL FORCING ON THE NC SHELF
18,169
Table 3. Statisticsof Augustand October-November Meteorological Time Series Mean
Deviation
Oct.
Aug.
23.1 1018 393 86.2 236 4.9
17.1 1019 338 78.6 161 6.2
1.7 3.8 28 8.0 310 2.4
2.16 5.7 36 11.0 234 3.3
21.2 21.7 22.2 23.3
18.0 18.6
1.8 1.7 1.3 0.9
1.4 1.1
Air temperature, øC Barometric pressure,mbar
Downwardlongwave,W m- •' Relative humidity, %
Shortwaveradiation, W m- •' Wind speed,m s-1 WT (DP, 4m depth), øC WT (dl, 2m depth), øC WT (d2, 1.1m), øC WT (d3, 2m), øC
Standard
Aug.
Subtidal
Oct.
Standard
Aug.
Deviation
Oct.
1.5 3.7 25 6.9 89 2.1
2.0 5.6 33 10.6 64 3.1
1.7
-
1.6
-
1.2 0.9
1.4 1.1
WT is water temperature.
30
I
I
I
.
I
25 -(A)
Water Temperature • Air Temperature
15
I
I
,,
20
•
•'
I
-I
ii
I
i
..
,
,
,
,
,
, •
,
,
,
,
0.015
"'0.01 0.005 1030
I (C)Barometric ,øressure
I
il
I
'1
,_ 1020 E
1010 I
I
I
I
I
I
I
1000
180
90 0
(D) WindDirecti
,'
'
I ..........
-90 -180
..
5
10
i
i
(E)Wind Speed ''
i
[[ I
i
I
-ii.•I -•I
ß
i
[
I
4 I
[
[ ./
I
I
[
I
5
i
I
I
[ ,,•.[I I
'
ß
_
I
I I
o
Aug 3
Aug 8
Aug 13
Aug 18
Aug 23
Aug 28
Sep 2
Se)
7
Figure 6. Meteorological time seriesfor August'(a) air andwatertemperature, (b) specific humidity,(c) barometricpressure, (d) wind direction,and (e) windspeed.Wind directionis defined suchthat 0ø represents windsblowing directlyoffshore and90ørepresents windsblowing alongshelfpoleward.
18,170
AUSTIN AND LENTZ' METEOROLOGICAL FORCING ON THE NC SHELF 30
I
I(A)
I
'
I
I
I
'
I
'
- Water Temperallure 25 L-'. / .... - :' '• ..... i.... • .... •..................
i..............
I---- Air/emDeraturE•
I
'
'
,
,
-
'1
, i
i ..........
i
I
'
I
i ...... i................
20
15
/
0.02 l
i
I
i
I
i
i
I •f •
i I•1 •
i I
i•
I
i
I
i
I
I
i I
I i
I
i
ii
,
,
,
,
I
I
I
I
I
I
I
' .
'1
(B) Specific,Humidity •
I
0.015 E
"'
0.01
/•i
I
...
0.005 1030
I
,... 1020 E
1010
I
.....
-
I
I'
I
I
i
I
I
I
I
I I
180(D)Wind•l•'rectiol•l •, 90
,
I
I
I.......
- v..-•. .......
/(C) Barometric Pressure i
I
' I................
/ .......................... •
1 ooo I
I
' ',........
I .......
I '
'• '•/'......
/'l
I!
i ..... •'
I
•
[ I
Ii
I k I ,....
I-
I
•
, ....
•
I
-90 -180
.............
i..........................
......
15
(E WindS
10 I
5
0
Oct 2
,
I ,
I , I
¾' •,l,
I,
Oct 7 Oct 12 Oct 17 Oct 22 Oct 27 Nov 1 Nov 8 Nov 11 Nov 18 Nov 21
Figure 7. Sameasfigure6 but for the Octobertime period.
3.1.
Seasonal
Variation
The mean net surface heat flux was 147 W m -2 in
ence. A decreasein relative humidity also contributed to the increased
latent
heat loss from the ocean.
To put the August and October 1994 observationsin the context of the seasonalcycle, 13 year recordsof air 39Wm-2 in October(Table2). About40%of this temperature and wind data collectedbetween 1982 and decrease was due to reduction in shortwave radiation. Comparison of measured shortwave radiation with cal- 1994 and an 11 year record of near-surfacewater tem-
August (positive indicatesflux into the ocean) and-
culatedaverage clear-sky radiation(321W m-2 in Au- perature collected between 1984 and 1994 at the FRF gustand 184Wm-2 in October[U.S.NavalObserva-wereaveragedinto monthlyvalues(Figure8). In gentory, 1978]) indicatesthat the reductionin shortwave eral, August is a period of weak winds and is one of the warmest months in terms of both air and water, of incidenceof sunlight,as opposedto increasedcloud with the air temperature 1.1øC warmer, on average,
radiation was due to the seasonalreduction in the angle
cover. About 33% of the decrease in the net surface heat flux was due to an increase in latent heat flux loss,
than the water temperature(3.1øC warmerin 1994).
October, on the other hand, occursduring the period associatedprimarily with a decreasein air temperature of most rapid cooling of both the water and air, and the air temperature is 1.8øC colder than the water temrelative to the decrease in the surface water temperature (Table3), whichincreases the specific humiditydiffer- peratureon average(3.5øC colderin 1994). At the d2
AUSTIN
30
AND LENTZ:
METEOROLOGICAL
.(A) I.Ai.r. Temperature I I I I
I
FORCING
I
•
ON THE NC SHELF
xI
I
x
I
18,171
I
x
20
10 x
•
I
I
I
I
I
Jan Feb Mar Apr May Jun I
I
I
I
I
I
I
I
I
I
I
I
Jul Aug Sep Oct Nov Dec I
I
I
I
I
I
(B) Water Temperature 2O
10
Jan Feb Mar Apr May Jun
Jul Aug Sep Oct Nov Dec
10
x
-5 x
x x
I
-10
I
I
I
I
I
Jan Feb Mar Apr May Jun 10
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
(D) WindSpeexd '7 •
"
x
•
• •
6......x..-• x
•
x
I
x x
.•.ß-..• x
I
Jul Aug Sep Oct Nov Dec
•
x
................ • x X
Jan Feb Mar Apr May Jun
x
x
•x
•
.....,........... •
Jul Aug Sep Oct Nov Dec
•i•ure 8. Monthly •ver•es of •R• •rchived d•t• (198•-1994). Solid line represents1994 d•t•. VMuesof (•) •ir temperature(outliersbetweenAugust•nd November•re from firs• year of op•tio• •d m• •p•s•t i•st•m•t •o•), (b) •t• t•mp•t•, (c) •i• minusw•ter temperature,•nd (d) wind speed.
site, the air temperature- water temperaturedifference was of the same sign but smaller in magnitude during both months(Table 3). Winds in Octoberare typically strongerthan in August, as was observedin 1994. The seasonalcyclesin Figure 8 and the mean heat fluxes for August and October are consistentwith the analysisof Mid-Atlantic Bight climatologyby •t•r•/•er[1976]. These results suggestthat the differencesin the meteorologicalvariablesbetweenthe August and October time periods,and hencein the heat flux terms, are due
to seasonalvariations and that 1994 was a typical year in this sense.
3.2.
Synoptic Variation
The standarddeviationof the low-passednet surface
heatflux wasapproximately 135Wm-2 (Table2) in boththe Augustand Octobertime series.The largest contributionsto this variability came from the latent heat flux and shortwave radiation. Shortwave radiation
variability was largestin August, and the latent heat
18,172
AUSTIN AND LENTZ' METEOROLOGICAL
FORCING ON THE NC SHELF
Table 4. Characteristicsof MeteorologicalEvents
Date
EventType
AT,
ASH x 10a,
øC Aug. 15
Clouds
WindDirection,
kgm-a
øT
MaximumWinds,
A SHF
ms-x
Wm-2
Aug.16
warmfront
coldfront
-4.5
-4.5
4
yes
no
45 -•-135
180-->0
8
10
-160
Aug. 22 Aug. 29 Sept. 1 Sept. 4
coldfront coldfront coldfront tropicalstorm
-3.5 -4 -4.5 ?
-7 -5 -7 -
yes yes yes yes
0 -->-135 30 -->-150 0 -+ -135 -170 -+ -135
9 7 8 16
-300 -250 -370 -320
Oct. 10 Oct. 15 Oct. 26
cold front tropicalstorm cold front
-6 -4 -6
-6.5 -5 -6
yes yes yes
45 -+ -135 -170 -+-135 0 --> -170
13 15 11
-610 -360 -490
Nov. 1 Nov. 7
cold front cold front
-4.5 -6.5
-8 -6.5
no no
45 --> -45 45 -->-135
14 14
-530 -470
Nov. 10
cold front
-6.5
-6.5
yes
30 -->-135
14
-530
4
(+100)
The changes in air temperature AT, specific humidityASH, andsurface heatfluxASHF, wereestimated byjudging the maximumchange in theseparameters in a 24 hourwindowaroundthe frontalpassage. Lowpassfilteredsurface heat flux wasusedfor this purpose.Wind directionis definedsuchthat 0ø is directlyoffshore, 90ø is alongshore poleward.
flux variability was largest in October. The most common source of variability on synoptic timescales was the passageof atmosphericfronts associatedwith lowpressuresystems. The net surfaceheat flux and the pattern of variability depend on the track of the lowpressurecenter relative to the study site. During the CoOP Inner Shelf Study, variability associatedwith three basic storm tracks was observed.
The most com-
mon pattern of variation was due to low-pressuresystems passingfrom west to east-northeast,to the north of the study site. In this case,a trailing cold front associated with the low-pressuresystem passedover the
and 5 times during the Octobertime period (October 10 and 26 and November1, 7 and 10). The secondpattern was associatedwith tropical storms, low-pressure systemsmoving north to the east of the site, which occurred on September 4 and October 15. The third pattern consistedof the passageof a warm front associated with a low-pressurecenter developingover the southeastern United States and moving north, to the west
of the site, which occurredonce (August 18). Each of these cases had a distinct pattern of variation in the meteorology,surfaceheat flux, and wind stress. Table 4 containsbasiccharacteristicsof eachsynopticweather
study site. This occurred 4 times during the August event. Historical data from the FRF suggestthat the
time period (August15, 22, and 29, and September1), number of low-pressuresystemsobservedduring 1994
••
!
•
/
•
Direction of
propaganon
I -:e:.•
The relationship between synoptic weather systems and meteorological forcing on the North Carolina inner shelf JayA. Austin1 Massachusetts Institute of Technology/WoodsHole OceanographicInstitution Joint Program in Physical Oceanography,Woods Hole, Massachusetts Steven J. Lentz Department of Physical Oceanography,Woods Hole OceanographicInstitution, Woods Hole, Massachusetts
Abstract. A strong relationshipis observedbetween synopticweather systemsand atmosphericforcing of the oceanas estimated from buoy measurementsmade on the North Carolina inner shelfduring August and October-November1994 as part of the
CoastalOceanProcesses (COOP) Inner ShelfStudy. Synopticvariation (timescales of daysto weeks)in the meteorological time serieswas primarily associatedwith the passageof atmosphericfrontal systems. The most common synoptic weather pattern observedwas the passageof a low-pressurecenter to the north of the study site, which causedthe associatedcold front to pass over the study region. Before passageof the cold front, warm, moist northeastwardwinds increasedthe heat flux
intothe ocean,whereas afterthe coldfrontpassed, cold,dry southwestward winds decreasedthe heat flux into the ocean. In addition, in the presenceof oceanic stratification, northeastward winds drove coastal upwelling, bringing colder water to the surface, further increasingthe air-sea temperature contrast and hence the heat flux into the ocean inshore of the surface front between cool upwelled water and warmer water offshore. The decreasein surface heat flux during the passage
of a coldfront wasof order400W m-2, due primarilyto a decrease in latent heat flux. Although other synoptic patterns were observed,including one warm front passageand two tropical storm systems,the dominanceof cold fronts as a sourceof variability resultedin a strongpositivecorrelationbetweenthe along-shelf component of wind stressand the surfaceheat flux. To addressthe issueof spatial variation in the surface heat fluxes, data from several different sourceslocated along a cross-shelftransect were analyzed. This analysissuggeststhat the temperature of the atmospheric boundary layer undergoesadjustment when warm air blows over cold water but not when cold air blows over warm water. This producescross-shelf gradients in the bulk estimatesof turbulent heat fluxes during offshorewinds but not during onshorewinds. 1.
Introduction
estimates from observations taken off the coast of North
Carolina, north of Cape Hatteras during August and October/November1994 as part of the CoastalOcean Processes (COOP)InnerShelfStudy(ISS)fieldprogram [Burman,1994]. The primary purposeof this paper is
The surfaceheat flux and wind stressplay a crucial role in determining the behavior of the upper ocean, especially in the shallow coastal zone, where the entire water column can be directly influencedby atmospheric
forcing[WinantandBeardsley, 1979;Leeet al., 1989]. to describethe effectof synopticweathersystemson the
This paper presents surface heat flux and wind stress temporal and spatial variation in meteorologicalforcing. There have been few previousobservationalstudiesof the surface heat flux over U.S. continental shelves and even fewer which use direct in situ measurements of the
1Nowat theCollege ofOceanic andAtmospheric Sciences, OregonState University, Corvallis Copyright1999bytheAmericanGeophysical Union. Papernumber1999JC900016. 014g-O227/99/1999JC900016509.00
radiative fluxes. On the U.S. west coast, Beardsleyet
al. [1998]estimatedsurfacefluxesduringthe Coastal OceanDynamicsExperiment(CODE) and the Surface Mixed Layer Experiment (SMILE). They studiedseasonal and synoptic variation during both experiments. 18,159
18,160
AUSTINANDLENTZ:METEOROLOGICAL FORCINGONTHENCSHELF
They foundthat variationin shortwaveradiationand either sideof the front, they estimateda sharpdecline, ofthecomsea surfacetemperature(due to upwelling)were the oftheorderof400W m-2, in bulkestimates across most important factorsin the seasonalvariationof the binedturbulentheat fluxes(latentand sensible) net surface heat flux. SMILE was one of the first coastal the composite front. They alsoobserved that the highoceanic field experiments to make direct observations est concentrationof cloudslies alongthe front, another
the surfaceheat flux, of downwardlongwaveradiation.In the SouthAtlantic importantfactorin determining Bight, Blantonet al. [1989]estimatedturbulentfluxes and that the wind directionchangesduring the passage of moisture,heat, and momentumas part of the Gen- of the front. Thepurpose oftheinterdisciplinary CoOPInnerShelf esisof Atlantic Lows Experiment(GALE). They atof the processes tributed variation in meteorologicalobservations,and Study was to increaseunderstanding hence in the estimated fluxes, to synopticweather sys- that affect larval distributions over the inner shelf, as tems. For instance,the passageof a cold-airoutbreak well as to increaseknowledgeof the physicaloceanog-
on January 27, 1986, was responsiblefor an estimated raphy of the inner shelf,a regionof the oceanwhere decrease in the surface heatfluxof nearly1400W m-2. there have been relatively few physicaloceanographic Mountain et al. [1996]useddata from mooredbuoys studies. The field programtook place betweenAuand coastal stations to estimate the annual cycle and
gustandDecember1994,whichbracketed the seasonal
interannualvariability of the net surfaceheat flux in the transitionfrom strongstratificationand surfaceheating Gulf of Maine between 1979 and 1987. In this study, to weak stratification and surfacecooling. The study annual variation in the shortwave insolation was pri-
site was located on the shallow shelf between Chesa-
marily responsiblefor the annual variability of the net peakeBay and CapeHatteras(Figure1) and included heat flux. On a larger scale,Bunker[1976]estimated two mooredbuoysinstrumentedwith the meteorologmonthlymean surfacefluxesoverthe entireNorth At- ical sensorsnecessaryto make bulk estimatesof the lantic Oceanusingdata from over 8 million shipboard surface heat flux and wind stress. weather observations and discussed heat flux variabilVariability in the meteorologyon timescalesof days to weeksduringthe CoOPInner ShelfStudycanbe atity on annual scalesfor the Mid-Atlantic Bight, among tributed to three basic scenarios,listed here in order of other specificregions. Enriquezand Friehe [1997]infrequency of occurrence: the passage of coldfronts,the vestigatedthe effect of coastalupwellingoff northern passage of tropical storms, and the passageof warm California on the stability of the air column, which in part determinesthe transfer of heat, momentum,and moisture between atmosphereand ocean. They found that the changein the transfer coefficientsdue to upwelling affected estimatesof surfacewind stress,but had a negligibleeffecton the turbulenttransferof heat.
fronts. The responseof the local meteorologicalvariablesto these weather systemswas distinctive. A key result of this study is that the predominanceof cold fronts as sourcesof variation and the particular response of the surface heat flux and wind stressto their passage
leadsto a strong relationshipbetweenwind direction consideredin the open ocean. The Frontal Air Sea In- and surface heat flux. teractionExperiment(FASINEX) wasan observational The rest of the paper is organizedas follows.Section program that studied the effectsof sea surfacetem- 2 describes the instrumentation and methods used to The effect of fronts on surface fluxes has also been
perature fronts and atmospheric fronts on open ocean surface heat flux and wind stress variability. During
estimate surface heat flux and wind stress. In section
3, meteorological time seriesand surfaceflux estimates FASINEX, Davidsonet al. [1991]observedsharpde- are used to examine the relationship between synoptic creases in surface heat fluxes and differences in wind meteorology and temporalvariationsin surfacefluxes. directionduring the passageof coldfrontsoverthe open Section4 is a discussionof cross-shelfgradientsin the ocean southeast of Bermuda, with decreasesof surface air and sea surfacetemperature fields and the implica-
heatfluxof up to 600W m-•' duringindividual frontal tions this has for the spatial distributionof surfaceheat passagesobservedduring January-May 1986. Also dur-
flux. Section 5 is a summary.
ing FASINEX, Friehe et al. [1991]studiedthe effectof sea surfacetemperature fronts on atmosphericbound- 2. Field Program, Methods ary layer structure, showingthat warm air blowingover 2.1. The Site cold water leads to a stable, shallow boundary layer, while cold air blowing over warm water leads to an unThe CoOP Inner Shelf Study took place offshoreof stable, growingboundarylayer. Mooerset al. [1976] the North Carolina Outer Banks, between Cape Henry studied the effects of cold fronts on surface heat flux and (at the mouth of the ChesapeakeBay) and Cape Hatwind stressusing a compositeof 34 low-pressuresys- teras, at the southernend of the Middle Atlantic Bight tems observedover the Middle Atlantic Bight between (Figure 1). The coastlineis relatively straightin this 1972 and 1975, for use as an idealizedforcingfield for region with an orientation of approximately 340øT. The ocean models. This compositelow-pressuresystemin- shelf is approximately 80 km wide and 60m deep at the cluded a trailing cold front with warm, moist air ahead shelf break, increasingin width to the north. On the of the front and cold, dry air behind the front. On the western side of the Outer Banks lie Currituck Sound, basis of the air temperature and moisture content on PamlicoSound,and AlbemarleSound,whichare large,
AUSTIN AND LENTZ' METEOROLOGICALFORCING ON THE NC SHELF
18,161
36.28
CoOP Inner Shelf Study Central line moorings
36.26
I
ß 36.24
340ø-
'.
!
5 km
--Fd3 (26m)
37
36.22
Cape Henry z
'o
36.2
•\c,..?,LO?•'. -I-d2 (21m)
36.5 04401
"+dl (13n:i).
36.18
e44006
Region of DItlill Atlantic
36.16
36.14
35.5
ß10m
'.20m Ca
36.12
35 76
-75.5
-75
-74.5
I
-75.75
-75.7
-75.65
-75.5
-75.55
-75.6 o
Longitude, E
Figure 1. A plan view of the CoastalOceanProcesses Inner Shelf Study (COOP ISS) observational site, showingthe study area, and the locationsof CoOP ISS buoys DP, dl, d2, and d3, and National Data Buoy Center buoys44006, 44019, and 44014. The inset is a regionalview.
-5
-10
E c.•
m -20
-25
-30
-35 0
5
10
15
20
25
30
35
40
45
Offshore distance, km
Figure 2. Cross-shelfconductivity-temperature-depth (CTD) sectiontaken from R/V Cape Hatteras on August 19, 1994, during upwelling-favorablewinds. The section extends 50km offshorefrom the Field ResearchFacility, perpendicular to the shore. The positions of the three CoOP ISS mooringsare indicated by vertical dashedlines.
50
18,162
AUSTIN AND LENTZ: METEOROLOGICAL
FORCING ON THE NC SHELF
28
'1•' ""/'"" L
(A)
O 26=24-
•2.tr,.]• ••,1• •
.I
•'22 -
,
t/.•.'..""l-' 'd.
i't ;;.•; '.
.j.:• ....
':r';
I
,
x
FRF
I ...... NDBC 44006
• i' r
........
d2
I ---
NDBC44019
:'.. •.•,, •z.• ½ {.•.l• - ,L•:•'" .: ,•:' :: '•.
f,•., / •:: :•i..7'
;";,'•• cl"'•: .,.:'...ß.x-.... •
::
:':•'•i::-: '.x ': x
',..... ;: ,. :............x ,'-;x''.x'' .. ß
• • ,•
.'..x , .',: ' .'ß:.x:..j'•'" x•x
ß 20-
,
o
,,.
"- "x
,
x
c/) 18-
CTD section (Figure 2)
z16I
I
i
Aug 8
Aug13
Aug18
14
Aug 3
I
i
I
Aug23 Aug28
Sep2
Se3
7
28 x
o
0 26
d2
(B)
NDBC 44006
NDBC 44019 =24
•'22 e 20 o
0318
Z16
tl
i
i
i
i
I
I
I
I
i
ct 2 Oct 7 Oct 12 Oct 17 Oct 22 Oct 27 Nov 1 Nov 6 Nov 11 Nov 16Nov 21
Figure 3. Near-surfacewater temperaturesfrom the FRF pier, d2, NDBC buoys 44006 and 44019,for (a) Augustand (b) October.FRF valuesare daily, othersare hourly. Arrow indicates timing of CTD sectionin figure 2.
shallow inland bodies of water, characterizedby high
to 6m thick, at a depth of approximately10m (Fig(northward)windsbrought surfacetemperatures duringsummer[Roelofs andBum- ure 2). Upwelling-favorable pus,1953].The mooredobservations focused on a cross- the thermoclineto the surface,resultingin cross-shelf shelf section located offshoreof the Army Corps of En- seasurfacetemperaturedifferences of up to 8øC (Figgineers'Field ResearchFacility (FRF) in Duck, North ure 3a), which play a significantrole in determining Carolina.
This section extended
16 km offshore and con-
the spatial distribution of the surface heat flux. Dur-
the watercolumnwasrelatively sistedof three main mooringsites,dl, d2, and d3, two ing October-November, of which (d2 and d3) wereinstrumentedwith meteoro- well mixed in temperature(as inferredfrom mooring logical equipment. data),andcross-shelf temperature differences weretypically lessthan 1øC (Figure3b). 2.2.
Oceanographic
Setting
The CoOP Inner Shelf Study took place during August and October-November1994, which represented two very distinct oceanicsettings. During August, the water columnwascharacterizedby a strongthermocline with a temperature differenceof 5øC to 9øC, usually 3
2.3.
Instrumentation
and
Other
Data
Sources
Many sourcesof data are utilized in this study to understandthe fluxesat the CoOP ISS site duringAugust and October-November 1994 and to put these observations into a wider regional and temporal context. The instrumentation
is summarized
in Table 1.
AUSTIN AND LENTZ: METEOROLOGICAL
FORCING ON THE NC SHELF
18,163
Table 1. Meteorological Instruments onthe Vector-Averaging WindRecorder andAssociated Moorings Parameter
Instrument
Height,m
Accuracy a
2.7
+0.4 ø
dœ VA WR instruments
Air temperature
Thermistor
YellowSprings
(WS > 5 ms-•)
5 K at 25 øC Barometric
pressure Longwave Radiation Relative
humidity
Paroscientific
2.7
model216-B-101 pyrgeometer
+0.6 mb
(WS < 20ms-1) 3.3
+5%
2.7
+ 5%
Eppley PIR Vais ala
Humicap 0062HMP
Insolation
Pyr anometer Eppley 8-48
3.3
4-5%
Wind direction
Integral vane
2.7
4- I bit
w/vane follower WHOI/EG2•G
Windspeed
R.M. Young
(5.6ø)
2.7
4-6%b
-1.1
•:0.5 øCb
3-cup anemometer
Water
thermistor
temperature(d2)
Water
Seacat
Other GoOP Buoy Instruments
-2.0
+0.5øCc
-4.0
4-1oCc
Air temperature Water temperature Wind speed
10 - 1.5 10
4-1øC 4-1øC
Wind direction
10
4-10 ø
Air temperature Water temperature Wind speed
5 -1 5
4-1øC 4-1øC
Wind direction
5
4-10 ø
temperature, dl, d3
Water
Seacat
temperature, DP NDBC
NDBC
NDBC
44006
NDBC Instrumentsd
4-1 ms -1 or 10%
44O19
4-1 ms -1 or 10%
44O14
2.3.1. Moored instrumentation. Meteorologi- midity, downwardshortwaveand longwaveradiation, cal instrumentationfor the experimentconsistedof two barometricpressure, seasurfacetemperature,andwind buoysequippedwith vectoraveragingwind recorders speed and direction to make bulk estimates of wind
(VAWRs,Table1), locatedat d2 andd3 [Alessiet al., stressand surfaceheat flux. The meteorological mea1996].Thesesiteswere5 and 16.4kmoffshore, on the surementsweremadeat heightsof 2.7 to 3.3m (Table 21 and 26-misobaths,respectively. Only data fromthe 1), and near-surface watertemperaturemeasurements
d2 VAWR were recovered. The d3 VAWR was lost during Hurricane Gordon on November 18. Each VAWR
recordedmeasurements of air temperature,relativehu-
were made at a depth of 1.1m. Data were recorded ev-
ery 7.5 min and weresubsequently block-averaged into hourly values. Each buoy was equippedwith an Ar-
18,164
AUSTIN
AND LENTZ:
METEOROLOGICAL
FORCING
ON THE NC SHELF
Table 1. (continued) Parameter
Instrument
Height,m
Accuracy a
Air temperature Water temperature Wind speed
5 -1 5
•-1øC •-1 øC 4-1ms -• or 10%
Wind direction
5
4-10 ø
20 19.5 19.5
n/a n/a n/a n/a
FRF
Instruments
Water temperature
bucketthermometer
Air temperature Wind Speed Wind direction
YSI thermistor F420 anemometer, NWS
Air temperature
R/V CapeHatterasInstruments R.M. Young41372C 15.25
n/a
Water temperature
YSI 701
2øCe
0
Abbreviationsare WS, wind speed;PIR, precisioninfraredradiometer;WHOI, WoodsHole Oceanographic Institution; COOP,CoastalOceanProcesses; NDBC, National Data Buoy Center;FRF, Field ResearchFacility; n/a, not available; NWS, National Weather Service. aValues denote estimated instrument accuracy;manufacturer'sspecificationsunlessotherwisenoted.
bValueis from Welleret al, [1990]. cAlthoughthe sensors themselves havegreateraccuracythan shownhere,the valuerepresents their estimatedaccuracy as a measure of surface water temperature.
dData are takenfromthe NationalData BuoyCenterwebsite,http://seaboard.ndbc.noaa.gov. eValue denotesresolution, not accuracy.
gostransmitter to transmit the meteorological data and
16.4km offshorein 25 m of water. All three moorings
buoy position to shore.
were instrumented
The d2 and d3 meteorologicalbuoysweredeployedon August 6, and recovery was planned for early December. However, failure of the surface moorings during severe storms and failure of the d3 Argos transmitter systemresulted in only 2.5 months of data from the d2
vertical structure of the water column. Hourly water temperature measurements were made near the end of the FRF pier at a depth of 4.0 m usinga SeabirdInstru-
site and almost no data from the d3 site. The d2 surface
mooring failed on September4 during a tropical storm and came ashore with
all of its instrumentation
intact.
The mooring was refurbishedand redeployedon October 4. It failed again on November 18 during Hurricane Gordon and came ashore. Although the tower with the VAWR was torn off the buoy as it came ashore, it was eventually recoveredwith all data intact. This sequence of events determined the two time periods considered here, which are designatedthe "August" time period, 0100 UTC on August 7, 1994, to 0000 UTC on September 4, 1994, and the "October" time period, from 1500 UTC on October 4, 1994, to 0800 UTC on November 18, 1994. The d3 surface buoy stayed in place until Hurricane Gordon on November 18, when it also came ashore.Unfortunately,the VAWR on the d3 buoy (and
with
thermistors
to determine
the
mentsSeacat(referredto asDP, for "DuckPier"). This is rather deep for estimating surfacetemperatures, as the mean temperature difference at dl between 4.6-m and 1.5-m depth is approximately0.6øC during August
(maximumdifferenceof 3.2øC), sothis measurement is most likely an underestimateof the sea surfacetemperature of order 1øC during August. During October, no data were available at the surface at dl, but the difference between the temperature at 4.6 and 1.1m at d2
was, on average,of the order or 0.02øC (with a maximuminstantaneous differenceof 1.1øC),suggesting that the temperature at 4-m depth is a reasonableproxy for the surface temperature. 2.3.2. Field Research Facility measurements. Wind speed and direction, air temperature, barometric pressure,and water temperature measurementshave been taken almost continuouslysince1982 at the Army
Corpsof Engineer'sFieldResearch FacilityFRF [Birkehenceall of the meteorological data from the d3 site) meier et al., 1985]. Wind speedand directionwere
was lost as the buoy came ashore. The d3 Argostransmitter failed on August 12, was repaired on September 1, and failed again on September8. The amount of data recoveredvia the d3 Argostransmitter was too small to be useful for this study. Other measurementsfrom the moored array consisted
of near-surface (2-m depth)watertemperatureat the dl site, 1.4 km offshorein 13 m of water, and at the d3 site,
measured using an anemometer located 19.4m above
sealevel at the end of the FRF pier (560m offshore). Air temperature was measuredin an instrument housing 40m onshoreand appearedto sufferfrom a diurnal instrument heating problem. Therefore only nighttime valueswere used. During onshorewinds, nighttime values typically differed from measurementsat d2 by less than 1øC, suggestingmeasurementerrors of the order
AUSTIN AND LENTZ: METEOROLOGICAL FORCING ON THE NC SHELF
18,165
of IøC, though no actual uncertainty valuesare avail- tum, heat, and moisturefluxes,the appropriateness of able. Sea surface temperature was measured using a the bulk formulasof Fairall et al. [1996]cannotbe bucket thermometer at the end of the pier, once per judged. The uncertaintyof the meanflux estimatesdue day, typically around 0700 local time. These data were to measurement uncertainty will be discussedin section highly correlated with buoy measurementsduring the 2.5. August and October 1994 periods, and were used to The surfacewind stressis estimatedusinga stabilitygain someperspectiveon the regional seasonaland in- dependent transfer coefficientthat relates the measured terannual variability, as well as qualitative information wind speed to the wind stress, on the cross-shelfvariation during the experiment. T ----pACd(Ua-- u)lu - u•l, (1) 2.3.3. NDBC buoy array. The National Data Buoy Center (NDBC) maintainedthree meteorological where PA is air density, Ca is a stability-dependent buoys,designations 44006,44019,and44014(Figure1), transfer coefficient, ua is wind velocity measured at a acrossthe shelfnear the FRF, which provedusefulto specifiedheight (in this case2.7m), and us is the surthis study. Thesebuoysrecordedhourly wind speed face velocity of the water. The direction of the wind anddirection,air temperature,andwatertemperature, stressis assumedto be the same as the velocity differwith quoted accuracieslisted in Table 1. While insufence vector Ua --Us. ficient for completeheat flux calculations,these data The net surfaceheat flux may be consideredthe sum were used to examine cross-shelf variations in the me-
teorology,estimate sensibleheat flux, and infer latent heat flux.
2.3.4. Shipboard data.
The R/V CapeHatteras
of four terms [Gill, 1982]:
QTOT-- QSWNE T + QLWNE T + QLAT + QSEN,(2)
performed conductivity-temperature-depth (CTD) tranQSWNET isnetshortwave radiation, QLWNET is sectson the North CarolinainnershelfduringAugust where net longwaveradiation (the differencebetweenupward andOctober1994aspart of the CoOPInnerShelfStudy radiation),QLAT is latentheat [Waldorfet al., 1995,1996].It made16transects of the anddownwardlongwave flux, and QSEN is sensibleheat flux. In this paper, shelfalongthe FRF to d3 centralmooringline in Au-
positive flux values always denote heat flux into the gust and 20 in October. In addition, cross-shelftran- ocean. sectsnorth and southof the centralmooringline were Thenetshortwave radiation fluxQSWNET duetosomadeto quantifyalong-shelf variationin the hydrogralar insolation between 0.28/zm and 2.8/zm, is estimated phy. During eachcruise,air temperature,surfacewater temperature, and wind velocity were measuredevery using 15 s. The air temperature measurementssufferedfrom QSWNET -- QSWDow N(1- Ab), (3) very low (2øC) resolution,sothe air temperaturemeawhere QSWDow N is measured using anEppley 8-48 surementsare usedonly for qualitative comparisons. pyranometer and Ab is the albedo of the sea surface, which is determinedempirically as a function of the so2.4. Estimation of the Surface Heat Flux and Wind
Stress
lar angleand atmospheric transmissivity[Payne,1972].
Thenetlongwave radiation fluxQLWNET isthedifference between the upward longwave radiation QLWu P heatfluxwereestimated usingbulkformulas developed The surface wind stress and the turbulent surface
by Fairallet al. [1996].Likemostotherbulkfluxalgo- due to blackbodyradiation from the oceansurface,cal-
rithms,theseweregenerated usingmeasurements made culated usingthe Stefan-Boltzmanlaw, and the downin the open ocean, away from boundariesand fronts. A comparisonof the heat flux and surface stressesti-
wardlongwave radiation QLWDow N duetoradiation
the Largeand Pond[1981,1982]formulationrevealed
is
from moisture in the atmosphere,and is measureddimatedwiththisformulation withthoseestimated using rectly. The formula for the net longwaveradiation flux no qualitative differences. The appropriateness of these bulk formulas to the
QLW•:T --e(QLWDow• --ate),
(4)
wheree - 0.98 [Dickeyet al., 1994]is the radiativeefof e varyfrom0.93to fewkilometers of the shoreline. Theirsuitabilityis es- ficiencyof the water(estimates 1.0 [Fung et al., 1984]), rr = 5.67 x 10-s W m-2K-4 is peciallyquestionable duringoffshorewinds,whenthe the Stefan-Boltzman constant, and Ts is the seasurface marineatmospheric boundarylayermustquicklyadjust to newsurfaceconditions. In addition,surfacewaves, temperature indegrees Kelvin. QLWDow Nisdue toinsteepenedin shallowwater, may affectthe transferco- fraredradiationin the range3.5 to 50/zmemittedby moistureandis measured directlyusingan efficients andhence estimates ofturbulent fluxes[Large atmospheric et al., 1995;Geernaertet al., 1986].It shouldbe noted Eppleyprecision infraredradiometer(PIR) pyrgeomecoastaloceanis not clear, especiallyto siteswithin a
that neither of these effectshas been taken into account in the estimatespresentedhere. In the absenceof di-
ter. The downwardlongwaveradiation was corrected
forinstrument heating bysubtracting 0.035 QSWDow N
rect (i.e., eddycorrelation)measurements of momen- [Alados-Arboledas et al., 1988].
18,166
AUSTIN AND LENTZ: METEOROLOGICALFORCING ON THE NC SHELF
The latent heat flux QLAT representsthe heat releasedor gained when water evaporatesfrom or condenseson the oceansurface.Althoughoften interpreted as representingonly evaporation,there is a significant portionof the Augusttime seriesduringoffshorewinds when it appearsthat heat was being gained owing to
as possibleusing manufacturer'sspecifications(Table 1) and literature valuesof uncertaintyto providea context for interpreting the surfaceflux estimates.This error analysisrevealswhat terms of the surfaceflux (and what instruments)contributemost significantlyto uncertainty in the estimate of the total surfaceheat flux
condensation at the seasurface[Beardsley et al., 1998]. and wind stress, given what is known about the unThe latent heat flux is related to the moisture gradient at the ocean surface and the air-water velocity differenceusing the bulk formula
=
certainties of the measurements. For a more thorough treatment of uncertainty in the VAWR measurements and how they apply to flux estimates, see Weller et al.
[1990]and Beardsley et al. [1998].Resultsof the error
- q0)lu - ul,
analysisare summarized in Table 2. The quoted accuracy of the wind speed measurewhere L is the latent heat of evaporation; qz is the spe- ment is 2%, but there have been indications that cup cific humidity measuredat height z abovesealevel (in anemometersare prone to overspeedingby as much as
this case,z - 2.7 m); q0 is the estimatedhumidity at
the sea surface, calculated by assumingthe air at the water surfaceis the same temperature as the water and
6% [Weller et al., 1990]. Usinga nominalvalueof 6% for the uncertainty of the wind speedestimate and assuming that error in the estimate of wind stressis due
that the air is saturated(assumingthat the saturation primarilyto this measurement (as opposedto the meahumidity for air over salt water is 0.98 of the saturation humidity of air over fresh water of the same temper-
surement of surface ocean currents, which are relatively
small),overspeeding couldbe responsible for uncertainature); and Ce is a stability-dependent transfercoeffi- ties in the stressestimatesof up to 12%. cient,estimatedusingthe Fairall et al. [1996]formulaAn analysis of error propagation in the latent heat
tion.
flux bulk equation suggeststhat the main sourceof in-
The sensible flux QSENis relatedto the temperature strument error is the measurementof relative humidity, difference between the air and sea surface and the airassumedto be •- 5% [Weller et al., 1990]. This error
water velocity differenceusing the bulk formula
=
causesan uncertainty in the estimate of the mean latent heat flux of order 12 W m -2. The largest sourceof uncertainty in the sensibleheat
- Ts)lu - ul,
whereCpis the heat capacityof water,T$ is the seasur-
flux is the determination
of the difference
between
air
face temperature, TA is the air temperature, and Ch is and surfacewater temperatures. The uncertainty in the a stability-dependent transfer coefficient,alsoestimated differenceis greatest during strong insolationand weak winds, when the air temperature sensor suffers from usingthe Fairall et al. [1996]formulation. overheatingand the water temperature sensorunderestimates the surfacetemperature owing to formation of 2.5. Estimation of Instrument-Induced Flux stratification in the upper meter of the water column. Uncertainty During strong winds, both of these effects are small. Althoughit is impossible to takeinto accountall po- Sincethe sensibleflux is proportional to the wind speed, tential sources of error inherentin makingsurfaceflux uncertainty in the temperature differenceduring wcak
estimates,it is essentialto make as good an estimate
winds leads to small uncertainties
in the sensible flux.
Table 2. Statisticsof Augustand OctoberTime Seriesof Heat Flux ComponentsMeasuredat d2. Mean
Aug.
Standard
Oct.
Deviation
Subtidal
Standard
Aug.
Oct.
Aug.
Oct.
QSWNE T QLWD OWN QLWu P QLWNE T
2204-11 377 4-19 -4224-7 -454-20
1444-7 325 4-16 -3994.5 -744-17
301 27 8 28
222 37 8 37
83 38 31 29
59 39 23 35
QSEN QLAT QTOT
54-4 -334-13 1474-26
-154-4 -944-12 -394-22
19 72 317
31 98 258
22 68 135
30 93 138
Deviation. a
Uncertaintiesreflect instrument-baseduncertaintiesonly.
Allvalues areinWm-2. Subscripts denote SWNET, netshortwave flux;LWDOWN , downward longwave flux;LWup,
upwardlongwaveflux; LWNET, net longwave flux;SEN, sensible heatflux;LAT, latent heatflux; TOT, total heatflux. aThe subtidal standard deviationrepresentsthe standarddeviationof the low-passedtime series.
AUSTIN AND LENTZ: METEOROLOGICAL FORCING ON THE NC SHELF
18,167
1000 (A)Net ................. •-..t............................. •.. •... •........ 500........... •........... •............ I
I
I
I
I
II
I
I
I
.
.,
.
•
• •
I
o 50
N
.I
I
I
I
I
II
I
I
...... •... o .(B) et.longwave -50
..
-100
....
I
-150
200
,
,
(C)Latent • -200-'
ß,
100
.
.
,
•• I
.,
,
•
• , ..........,,........... •.....il .-•........ I
I
....
(D)Sensible ......
- 100 1000 500 0 -5OO
0.2 (F)Wind Stress [
•
I
I
[
I[
[
I
I[
Aug 8
I
ß ................. , .... ,,........ ......... I
-0.2 •< •h•re Aug 3
]
[
Aug13
[ i ,
Aug18
[•
,
Aug23
Aug28
•
I
i,
Sep 2
i
Sep 7
Figure 4. Termsof theheatfluxandthewindstress fortheAugusttimeperiod,measured at
d2' (a) netshortwave, (b) netlongwave, (c) latent,(d) sensible, and(e) netsurface heatflux (hourly (solid) andlowpass filtered (dashed)) andwindstress (lowpassed andsubsampled every 6 hours).Thewindstress is defined withalong-shelf upandoffshore to theright.Thevertical
dashedlinesmark the meteorological eventslistedin Table4.
With an estimatedtemperaturedifference uncertainty ationis about7Wm-2 in August. In October,the of order 1.0øC,muchlargerthan the air temperature uncertainty in the estimation of e is dominant, and the andwatertemperatureinstrumenterrorsalone,the cor- uncertainty is around5 W m-2 . respondinguncertainty in the sensibleflux is of order The downward longwaveradiation was measureddi12 W m -2. rectly, and the instrumentuncertaintyis 5%. Beardsley Upward longwaveradiation is a function of the es- et aL. [1998]comparedrecordsfrom two PIR records
timated seasurfacetemperatureand the emissivityof and founda differenceof 4.2%, but recentstudiessugthe seasurface.Assumingthe uncertaintyin seasur- gest that even this may overestimatethe uncertainty facetemperaturein Augustis IøC (due primarilyto [Fairall et al., 1999]. Using5% as an estimatein the near-surfacevertical temperaturegradients;the value uncertaintyof the measurement,the uncertaintyin the for Octoberis mostlikelysmaller)andthe uncertainty downwardlongwaveflux estimateis of order19W m-• in the emissivity to be +0.01 [Dickeyet al., 1994],the in Augustand 16W m-• in October. total uncertaintyin the meanupwardlongwaveradiIt will be assumed that most of the error in the es-
18,168
AUSTIN AND LENTZ' METEOROLOGICAL FORCING ON THE NC SHELF
-,ooo "'(A)'sh0rave, 500........
I
,
,
I
o
,
,
I
'•
l',
lOO
'
I
'
I
'
'
I '
(B) Net Lengwave ß ':
o
'
'1
' I
] I
I'
•[
'I
' I
'
I
-100 -200
0
( [
'
-200
' '
I ]
[
I'
'1
'I
I
I
'
I •
' I
' I
,
•
I[
•
I[
'
ß
-400
100
'
'
J
(D)Sensible ,I.... ..... -100
•
/•,
,I
. [.
[
..,,
-ß
I
I
(E) N
500
0
lUx
I
[
I
. .
I
[
I
.
I
[
I
I
ß
'
-500
L
[
i
i
i
I [
I [
o.2j(F)Wind •tress , -0.2J--....... J
Oct 2
. •//•/.////.
I '/,,/•/I
Oct 7
Oct12
[
:.1.................... I
I
Oct17
I
Oct 22
I[
,
/./VJ[.... I I
Oct 27
[I
, I.
I I
Nov 1
I I
i
I[
'
.• /.! ./../..I..
. ' ..
, ,
I I"
Nov 6
II
Nov11
i
Nov16
Nov 21
Figure 5. Sameas figure4 but for the Octobertime period.
timation
of shortwave radiation
is in the measurement
3.
Observations
and Results
itself,asopposed to the altitude-dependent albedo.The During the Augustand Octobertime periods,the uncertaintyin the measurement for an ungimballed sen- surfaceheat flux (Table2 and Figure4 (August)and soris 5% [Wetteret at., 1990],corresponding to mean Figure5 (October))variedon two distincttimescales:
uncertainties of 11Wm-2 in Augustand7Wm-2 in diurnal(daily)andsynoptic (timescales of 2-7 days). October.
As the instrumentsthat are primarily responsiblefor the uncertaintyare differentfor eachterm on the net surfaceheat flux, the uncertaintiescan be considered independent. Making this assumption,the estimated uncertainty in the mean net surface heat flux due to
This study focuseson the synopticvariability,which is evidentin all of the meteorological time series(Ta-
ble 3 andFigure6 (August)andFigure7 (October)), and hencethe surfaceheat flux components.Diurnal variabilitywas due almostentirelyto the daily shortwaveradiationcycle.This variabilitywasremovedfrom
measurement uncertainty isapproximately 26W m-2 in the data usingthe p164lowpass-filter[Beardstey et at., Augustand22W m-2 in October.Thelargestsources1985].However,thereis alsoa largedifference between of error, givenwhat is known about the error character- the mean surfaceheat fluxesduring August and Octoistics of the measurements,are the downward longwave ber, presumablyassociatedwith seasonalvariationin and latent
heat flux estimates.
the meteorology, whichis discussed first.
AUSTIN AND LENTZ: METEOROLOGICAL FORCING ON THE NC SHELF
18,169
Table 3. Statisticsof Augustand October-November Meteorological Time Series Mean
Deviation
Oct.
Aug.
23.1 1018 393 86.2 236 4.9
17.1 1019 338 78.6 161 6.2
1.7 3.8 28 8.0 310 2.4
2.16 5.7 36 11.0 234 3.3
21.2 21.7 22.2 23.3
18.0 18.6
1.8 1.7 1.3 0.9
1.4 1.1
Air temperature, øC Barometric pressure,mbar
Downwardlongwave,W m- •' Relative humidity, %
Shortwaveradiation, W m- •' Wind speed,m s-1 WT (DP, 4m depth), øC WT (dl, 2m depth), øC WT (d2, 1.1m), øC WT (d3, 2m), øC
Standard
Aug.
Subtidal
Oct.
Standard
Aug.
Deviation
Oct.
1.5 3.7 25 6.9 89 2.1
2.0 5.6 33 10.6 64 3.1
1.7
-
1.6
-
1.2 0.9
1.4 1.1
WT is water temperature.
30
I
I
I
.
I
25 -(A)
Water Temperature • Air Temperature
15
I
I
,,
20
•
•'
I
-I
ii
I
i
..
,
,
,
,
,
, •
,
,
,
,
0.015
"'0.01 0.005 1030
I (C)Barometric ,øressure
I
il
I
'1
,_ 1020 E
1010 I
I
I
I
I
I
I
1000
180
90 0
(D) WindDirecti
,'
'
I ..........
-90 -180
..
5
10
i
i
(E)Wind Speed ''
i
[[ I
i
I
-ii.•I -•I
ß
i
[
I
4 I
[
[ ./
I
I
[
I
5
i
I
I
[ ,,•.[I I
'
ß
_
I
I I
o
Aug 3
Aug 8
Aug 13
Aug 18
Aug 23
Aug 28
Sep 2
Se)
7
Figure 6. Meteorological time seriesfor August'(a) air andwatertemperature, (b) specific humidity,(c) barometricpressure, (d) wind direction,and (e) windspeed.Wind directionis defined suchthat 0ø represents windsblowing directlyoffshore and90ørepresents windsblowing alongshelfpoleward.
18,170
AUSTIN AND LENTZ' METEOROLOGICAL FORCING ON THE NC SHELF 30
I
I(A)
I
'
I
I
I
'
I
'
- Water Temperallure 25 L-'. / .... - :' '• ..... i.... • .... •..................
i..............
I---- Air/emDeraturE•
I
'
'
,
,
-
'1
, i
i ..........
i
I
'
I
i ...... i................
20
15
/
0.02 l
i
I
i
I
i
i
I •f •
i I•1 •
i I
i•
I
i
I
i
I
I
i I
I i
I
i
ii
,
,
,
,
I
I
I
I
I
I
I
' .
'1
(B) Specific,Humidity •
I
0.015 E
"'
0.01
/•i
I
...
0.005 1030
I
,... 1020 E
1010
I
.....
-
I
I'
I
I
i
I
I
I
I
I I
180(D)Wind•l•'rectiol•l •, 90
,
I
I
I.......
- v..-•. .......
/(C) Barometric Pressure i
I
' I................
/ .......................... •
1 ooo I
I
' ',........
I .......
I '
'• '•/'......
/'l
I!
i ..... •'
I
•
[ I
Ii
I k I ,....
I-
I
•
, ....
•
I
-90 -180
.............
i..........................
......
15
(E WindS
10 I
5
0
Oct 2
,
I ,
I , I
¾' •,l,
I,
Oct 7 Oct 12 Oct 17 Oct 22 Oct 27 Nov 1 Nov 8 Nov 11 Nov 18 Nov 21
Figure 7. Sameasfigure6 but for the Octobertime period.
3.1.
Seasonal
Variation
The mean net surface heat flux was 147 W m -2 in
ence. A decreasein relative humidity also contributed to the increased
latent
heat loss from the ocean.
To put the August and October 1994 observationsin the context of the seasonalcycle, 13 year recordsof air 39Wm-2 in October(Table2). About40%of this temperature and wind data collectedbetween 1982 and decrease was due to reduction in shortwave radiation. Comparison of measured shortwave radiation with cal- 1994 and an 11 year record of near-surfacewater tem-
August (positive indicatesflux into the ocean) and-
culatedaverage clear-sky radiation(321W m-2 in Au- perature collected between 1984 and 1994 at the FRF gustand 184Wm-2 in October[U.S.NavalObserva-wereaveragedinto monthlyvalues(Figure8). In gentory, 1978]) indicatesthat the reductionin shortwave eral, August is a period of weak winds and is one of the warmest months in terms of both air and water, of incidenceof sunlight,as opposedto increasedcloud with the air temperature 1.1øC warmer, on average,
radiation was due to the seasonalreduction in the angle
cover. About 33% of the decrease in the net surface heat flux was due to an increase in latent heat flux loss,
than the water temperature(3.1øC warmerin 1994).
October, on the other hand, occursduring the period associatedprimarily with a decreasein air temperature of most rapid cooling of both the water and air, and the air temperature is 1.8øC colder than the water temrelative to the decrease in the surface water temperature (Table3), whichincreases the specific humiditydiffer- peratureon average(3.5øC colderin 1994). At the d2
AUSTIN
30
AND LENTZ:
METEOROLOGICAL
.(A) I.Ai.r. Temperature I I I I
I
FORCING
I
•
ON THE NC SHELF
xI
I
x
I
18,171
I
x
20
10 x
•
I
I
I
I
I
Jan Feb Mar Apr May Jun I
I
I
I
I
I
I
I
I
I
I
I
Jul Aug Sep Oct Nov Dec I
I
I
I
I
I
(B) Water Temperature 2O
10
Jan Feb Mar Apr May Jun
Jul Aug Sep Oct Nov Dec
10
x
-5 x
x x
I
-10
I
I
I
I
I
Jan Feb Mar Apr May Jun 10
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
(D) WindSpeexd '7 •
"
x
•
• •
6......x..-• x
•
x
I
x x
.•.ß-..• x
I
Jul Aug Sep Oct Nov Dec
•
x
................ • x X
Jan Feb Mar Apr May Jun
x
x
•x
•
.....,........... •
Jul Aug Sep Oct Nov Dec
•i•ure 8. Monthly •ver•es of •R• •rchived d•t• (198•-1994). Solid line represents1994 d•t•. VMuesof (•) •ir temperature(outliersbetweenAugust•nd November•re from firs• year of op•tio• •d m• •p•s•t i•st•m•t •o•), (b) •t• t•mp•t•, (c) •i• minusw•ter temperature,•nd (d) wind speed.
site, the air temperature- water temperaturedifference was of the same sign but smaller in magnitude during both months(Table 3). Winds in Octoberare typically strongerthan in August, as was observedin 1994. The seasonalcyclesin Figure 8 and the mean heat fluxes for August and October are consistentwith the analysisof Mid-Atlantic Bight climatologyby •t•r•/•er[1976]. These results suggestthat the differencesin the meteorologicalvariablesbetweenthe August and October time periods,and hencein the heat flux terms, are due
to seasonalvariations and that 1994 was a typical year in this sense.
3.2.
Synoptic Variation
The standarddeviationof the low-passednet surface
heatflux wasapproximately 135Wm-2 (Table2) in boththe Augustand Octobertime series.The largest contributionsto this variability came from the latent heat flux and shortwave radiation. Shortwave radiation
variability was largestin August, and the latent heat
18,172
AUSTIN AND LENTZ' METEOROLOGICAL
FORCING ON THE NC SHELF
Table 4. Characteristicsof MeteorologicalEvents
Date
EventType
AT,
ASH x 10a,
øC Aug. 15
Clouds
WindDirection,
kgm-a
øT
MaximumWinds,
A SHF
ms-x
Wm-2
Aug.16
warmfront
coldfront
-4.5
-4.5
4
yes
no
45 -•-135
180-->0
8
10
-160
Aug. 22 Aug. 29 Sept. 1 Sept. 4
coldfront coldfront coldfront tropicalstorm
-3.5 -4 -4.5 ?
-7 -5 -7 -
yes yes yes yes
0 -->-135 30 -->-150 0 -+ -135 -170 -+ -135
9 7 8 16
-300 -250 -370 -320
Oct. 10 Oct. 15 Oct. 26
cold front tropicalstorm cold front
-6 -4 -6
-6.5 -5 -6
yes yes yes
45 -+ -135 -170 -+-135 0 --> -170
13 15 11
-610 -360 -490
Nov. 1 Nov. 7
cold front cold front
-4.5 -6.5
-8 -6.5
no no
45 --> -45 45 -->-135
14 14
-530 -470
Nov. 10
cold front
-6.5
-6.5
yes
30 -->-135
14
-530
4
(+100)
The changes in air temperature AT, specific humidityASH, andsurface heatfluxASHF, wereestimated byjudging the maximumchange in theseparameters in a 24 hourwindowaroundthe frontalpassage. Lowpassfilteredsurface heat flux wasusedfor this purpose.Wind directionis definedsuchthat 0ø is directlyoffshore, 90ø is alongshore poleward.
flux variability was largest in October. The most common source of variability on synoptic timescales was the passageof atmosphericfronts associatedwith lowpressuresystems. The net surfaceheat flux and the pattern of variability depend on the track of the lowpressurecenter relative to the study site. During the CoOP Inner Shelf Study, variability associatedwith three basic storm tracks was observed.
The most com-
mon pattern of variation was due to low-pressuresystems passingfrom west to east-northeast,to the north of the study site. In this case,a trailing cold front associated with the low-pressuresystem passedover the
and 5 times during the Octobertime period (October 10 and 26 and November1, 7 and 10). The secondpattern was associatedwith tropical storms, low-pressure systemsmoving north to the east of the site, which occurred on September 4 and October 15. The third pattern consistedof the passageof a warm front associated with a low-pressurecenter developingover the southeastern United States and moving north, to the west
of the site, which occurredonce (August 18). Each of these cases had a distinct pattern of variation in the meteorology,surfaceheat flux, and wind stress. Table 4 containsbasiccharacteristicsof eachsynopticweather
study site. This occurred 4 times during the August event. Historical data from the FRF suggestthat the
time period (August15, 22, and 29, and September1), number of low-pressuresystemsobservedduring 1994
••
!
•
/
•
Direction of
propaganon
I -:e:.•
