Mar 15, 1997 - tion of Oceanography, the National Marine Fisheries Service, and the ..... area to have an average value of 5 x 10 -s m s -⢠over the upwelling ..... Latent Heat. The bulk latent heat flux in watts per square meter is taken.
JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO. C3, PAGES 5641-5653, MARCH 15, 1997
Seasonal patterns of surface wind stress and heat flux over the Southern California Bight Clinton
D. Winant
and Clive E. Dorman
Center for CoastalStudies,ScrippsInstitutionof Oceanography, La Jolla, California
Abstract. Patternsof wind stressand heat flux betweenthe atmosphereand the ocean over the SouthernCaliforniaBight are describedbasedon observations from buoysand ships.During the winter, the wind stressis spatiallyhomogeneous and temporallyvariable, with strongeventscorresponding to low-pressure systems sweepingthroughthe area. During the summer,spatialpatternsare more persistent,with largegradients.Inshoreof a line runningapproximately betweenPoint Conceptionand Ensenada,Mexico,windsare weak. Offshorewind speedsare comparablein magnitudeto thosefound over the continentalshelfnorth of Point Conception.The boundaryis the locationof maximum wind stresscurl, and the spatialresolutionaffordedby CaliforniaCooperativeFisheries Investigation(CalCOFI) observations suggests maximumwind stresscurlsover 3 times larger than the valuesproposedby Nelson[1977].Net heat flux estimatesderivedfrom the CalCOFI measurements are somewhatlarger than the valuesproposedby Nelsonand Husby [1983], due to differencesin latent heat flux estimates.Possiblemechanisms responsible for the spring-summer spatialstructurein the wind and the relationship betweenthesegradientsand the propertiesof the underlyingoceanare discussed. [1983],andBakunandNelson[1991],amongothers,havepublished quantitative estimatesof the fluxesin this area. Those The SouthernCaliforniaBight (SCB) is often describedas descriptionsare basedon shipreportsaveragedon gridswhich having moderate weather. The area is sheltered from the cover 1ø of latitude and longitude.Descriptionsof the atmostrongand persistentwindswhich characterizethe rest of the sphericvariabilityover the California coastalarea north of San California coast. The extent to which this shelteringtakes Francisco[Winant et al., 1988] and near Point Conception place,the seasonalvariationin the atmosphericcharacteristics [Brinket al., 1984]demonstratethat largegradientscan existin in the SCB, and the resultingfluxesof momentum and heat the atmosphereover scalesconsiderablyless than 1ø. Morebetweenthe atmosphereand the oceanare describedbasedon over,Hickey[1978],Lynn and Simpson[1987],and othershave a combinationof buoy and ship-bornemeasurements. shownthat oceanpropertiesin this area exhibitfeatureswhich Since 1981, the National Data Buoy Center (NDBC) has are smallerthan the gridsusedto analyzethe shipreports.This maintainedan arrayof meteorologicalbuoyson the westcoast paper attemptsto improvethe spatialand temporal resolution of the continentalUnited States,as illustratedin Figure 1. As of the available descriptionsby combiningthe CalCOFI and part of this effort, two buoys(47 and 48) were deployedin a NDBC measurementsand to showthat significantgradientsin transectoff San Diego. Several other buoyshave been main- the ocean-atmosphere fluxesoccuron spatialscaleswhich are tained at differentlocationsin the SCB, notablybuoys25 and similar to those which characterizethe variability of ocean 45, while buoy 23, locatedjust off Point Conception,provides properties. a measure of the atmosphericforcing at the northwestern extremity of the bight. The locationsof each buoy and the 2. Background periodsfor which measurementsare availableare summarized Nelson[1977] providedthe first detaileddescriptionof surin Table 1. In a separate effort the California Cooperative Fisheries face wind stressover the coastalwaters adjacentto the west 1.
Introduction
Investigation(CalCOFI), sponsored jointly by ScrippsInstitution of Oceanography,the National Marine FisheriesService, and the California Department of Fish and Game, has conductedshipsurveysof water massand atmosphericproperties on a regulargrid since1949. The portion of that grid which is relevantto this studyis illustratedin Figure 1. Nominal station spacingis 66 km, and this spacingis reduced near the coast. Details of the samplingschemeand station location are describedby Lynn and Simpson[1987]. The subjectaddressedhere has alreadyreceivedconsiderable attention;Nelson[1977],Hickey[1978],Nelsonand Husby Copyright1997by the American GeophysicalUnion. Paper number 96JC02801. 0148-0227/97/96J C-02801 $09.00 5641
coast of the continental United States,based on surface marine
observationsprovided by ships.Most observationswere acquired after 1950, although some date back to the midnineteenthcentury.Nelsonand Husby [1983] (hereafter referred to as NH) extendedthis analysisto computeheat flux betweenthe atmosphereand the ocean.In both casesthe ship observationswere grouped in 1ø squares,and climatological average propertieswere computed in each square for each month of the year. The wind stresswas determined from average wind speedand direction,while the heat flux was computed asthe sumof shortwaveradiation,longwave radiation, andlatent (dueto evaporation)andsensiblefluxes.During the monthsof June and July, maximumwind stressesin excessof 0.15 Pa were reported along the northern California coast, extendingseveralhundredkilometersoffshore.In this area the
5642
WINANT
AND
DORMAN:
SEASONAL
WIND
STRESS
AND
HEAT
FLUX
winds is generally good, the computedwinds are found to poorlyrepresentobservationsover the SCB during the spring
36 ø
and summer 35'
,
*
32•
ß
ß
season.
An analysisof 10-year-longtime seriesof buoy observations [Dorman and Winant, 1995] showsthe SCB shelteredfrom strong wind forcing throughout the year. This sheltering is most pronouncedduring the spring and summer,when the centraland northernCalifornia coastis subjectedto persistent and strong equatorward winds. Sea surface temperature reachesmaximumvaluesin the SCB during the late summer, and monthly mean air-sea temperature differencesalso reach maximumvalues(of about 1.5øC)then. Hickey [1978] usesthe Nelson [1977] stressand stresscurl estimatesto interpret the spatial and seasonalcharacteristics of the ocean circulation,as describedby the CalCOFI measurements.The curl of the wind stressmay be responsiblefor severalcirculation features in the region of interest here, including the Davidson Current and the Southern California
ß ß
Countercurrent.
ß
31-122' -121'
ß
-120'
-119'
-118'
-117'
Lynn and Simpson[1987] describethe seasonalvariation of
-116' physicalvariables and circulation in the California Current
Figure 1. Map of the SouthernCalifornia Bight. The square system,including the SCB. Generally, the variability in the symbolsare the positionof the NDBC buoysusedin thisstudy; systemcan be attributed to changesin amplitude and location the circles are the locations of the CalCOFI stations. of severalcurrentsystems. The CaliforniaCurrent, the eastern limb of the North Pacific gyre, flows toward the equator throughoutthe year. During the fall andwinter, near the coast, stressis directedequatorwardand thus is favorableto coastal the Inshore Countercurrent carries surface water poleward. upwelling.Along the coastof southernand Baja California,the Finally, the California Undercurrent flows poleward at depth wind stressis describedas favorablefor upwellingthroughout along the North American coast. Hickey [1991, 1992] conductedan extensivesurveyof the the year but with weaker amplitudesthan along the northern California coast. The reduced wind stressin the SCB during circulation in the waters off Santa Monica and San Pedro in the SCB. The wind stressfield usedto representatmosphericforcthe summergivesrise to a strongcyclonicwind stresscurl. Bakun and Nelson [1991] extend the Nelson[1977] analysis ing includes local winds over the Santa Monica-San Pedro to four subtropicaleasternboundarycurrent regions.Each of basin as observedat buoy 25, as well as a representationof thesecoastalupwellingsystemsis shownto be characterizedby large-scalewinds derived from models.The latter winds have cyclonicwind stresscurl near the continent,and anticyclonic persistentmean monthlyvaluesdirectedtoward the southeast, further offshore,while the wind stressis predominantlyori- asalongthe remainder of the U.S. westcoastin this season,but ented toward the equator. The maximum stressoccurssome the local winds differ significantlyin the same season,with a distanceoffshore,the decaytoward the coastgivesrise to the meandirectiontowardthe northeastandmuchloweramplitudes. Brink et al. [1984] summarizeatmosphericand oceanicobcycloniccurl, and the offshoredecayleadsto the anticyclonic curl. Largest values of cycloniccurl are noted to occur near servationsin the area around Point Conceptionduring spring 1981. Rapid surveysof the wind field from an aircraft flying at capesduring the summer. The 1ø squaresusedby Nelson[1977] result in a sampling a nominal altitude of 152 m are characterizedby strong grawhich is coarsecomparedto the dimensionof a local region dientsin the wind field to the southand eastof Point Concepsuch as the SCB. Dotman [1982] averagedship observations tion when the overallwind pattern was toward the south.The over 0.2ø of latitude and longitude to show the structure of map which representsaverage southwardairflow showsthe observationsin a transect off San Diego. Wind speedswere windspeeds decreasing from 10 m s-• (or about0.15Pa) to of order10km,nearthecoasteastof found to decreaseby a factor of 2 or more from 140 km nearzeroovera distance offshoreto the coast.This finer-scaleanalysisrevealsthat the Point Conception.While Nelson [1977] reports comparable summer wind stress near the coast is to the north or in the valuesof stress,the grid schemeimposedby the shipobservaoppositedirectionfrom the directionsuggested by the larger- tion samplingspreadsthe differencesover larger distances.It scale analyses. may then be expectedthat the curl can reachlargervaluesthan Halliwell and Allen [1987] examinedthe large-scalecoastal wind field along the west coast of North America for the summersof 1981 and 1982 and the interveningwinter, using Table 1. Locationsand Deployment Periodsfor the observedand model-derivedwinds. In this analysis,summer NDBC Buoys wind fluctuationswere found to be driven by the combined Station Latitude Longitude DeploymentPeriod North Pacific high and the southwestthermal low, with some NDBC 47 32042 ' 119036 ' influenceexercisedby propagatingatmosphericsystemsto the Dec. 4, 1991 to Aug. 31, 1993 north of the area. In the winter, wind fluctuations are driven
primarily by the propagatingcyclonesand anticyclonesand have larger variance and space scalesthan in the summer. While the agreement between observedand model-derived
NDBC NDBC NDBC NDBC
48 23 25 45
32054 34018 33042 33048
' ' ' '
117054 120042 119006 118024
' ' ' '
Dec. 5, 1991 to Aug. 31, 1993 April 7, 1982 to Aug. 31, 1993 April 21, 1982 to Aug. 31, 1993 Jan. 30, 1991 to Aug. 31, 1993
WINANT
AND
DORMAN:
SEASONAL
given by Nelson'sanalysis,but these large values occur over fairly small areas.
3.
Meteorological Observations
In December 1991,the NDBC, under the sponsorshipof the Minerals ManagementService,deployedtwo buoys(47 and 48) on a transectextendingoff San Diego (Figure 1). These platforms,in conjunctionwith buoys25 and 45, located near Los Angeles,and buoy 23 locatedoff Point Conception,provide an opportunityto describeatmosphericsurfacefieldsover the SCB free from questionsassociatedwith effectsof land obstacles.The locationsof these five stationsand the periods of deployment are summarizedin Table 1. The baroclinic Rossbyradius of deformationcan be estimatedfor the atmosphereusingthe scaleanalysisof Overland[1984] and varies between 100 and 250 km for the range of parametersin this area. The buoysare locatedwell within a Rossbyradiusfrom the coast and thus experiencethe steering effect of coastal topography,but they are sufficientlyfar from the coastthat the observationsare not affectedby microscaleland features(of the order of a few kilometers). All buoyswere instrumentedto measure wind speed and direction, atmosphericpressure,and air and sea temperature 1 m beneath the surface.The accuracyof the wind speed
WIND
STRESS
AND
HEAT
FLUX
5643
been converted to stressvalues using the Large and Pond [1981] procedure.In later years,more atmosphericvariables were sampled, including surface pressure,dry and wet bulb temperature, and cloud cover. Seasonal averages of atmospheric observationswere obtained by computing means and standard deviationsof all observationsavailableat each station over a 3-month period: winter (December-February),spring (March-May), summer (June-August),and fall (September-November). The number of observations
available
in each season for each station
is
presentedin Appendix A, alongwith an analysisof the errors associatedwith the seasonalaverages. While the usualmeteorologicalconventionis to indicatethe wind direction as that from which it is blowing, the oceanographicconventionis usedthroughoutthispaper.The statistics of the mean wind vector are computedin the referencecoordinate system.The magnitudeof the fluctuationsis described with respectto the principal axes,that is, along the direction for whichthe covariancebetweencomponentsis null. Analyses [e.g.,Beardsleyet al., 1987] showthat the principal axesare usually closelyaligned with topographyin a coastal environment and thusrepresenta natural coordinatesystemin which to considerfluctuatingwinds.The ratio of major to minor axis amplitudesdeterminesthe degreeof polarizationof wind fluctuations.Along the coastof central California, where winds are sensor, asreportedbyHamilton[1980]is _ 1 m s- •, the direc- persistentlyequatorwardduringspringand summer,the mean tion sensoris _+10ø,the barometricpressuresensoris _+1 hPa, wind is greaterthan the standarddeviationalongthe principal and the air and sea temperature sensorsare _+iøC. The height axes,whereasthe oppositeis true alongthe coastof Washingat which the wind observationswere made depends on the ton and Oregon, where the vector means are small but only buoy and payloadconfigurationusedand variesfrom location becauseof cancellationbetweenstrongwind eventsin opposite to location and in time at any location, as different buoyswere •;rections [Dorman and Winant, 1995]. The net heat flux, Q not,from the atmosphereto the ocean is movedto a givenlocationin the courseof the deployment.The different platforms deployedoff the west coast include 3-m usually divided into four parts: discusbuoys,6-m Navy Oceanographicaland Meteorological AcquisitionDevice (NOMAD) buoys,and 10- and 12-m discussbuoys [Hamilton, 1980]. Individual observationsof wind is the heatdueto shortwaveradiation,Q•....g is speedand direction providedby NDBC include the height at whereQshort the heat due to long wave radiation, Q latentis the latent heat which the measurement was made, ranging between 5 and lossby the ocean due to evaporation,and Q ... ibleis the sen13.8 m. When 10-m observationswere not available,they were sible heat transfer due to conduction. The four heat flux comcomputed from the measurementsassuminga logarithmic ponentswere estimatedusingthe proceduredescribedby NH wind profile using the method described by Dotman and and are summarizedin Appendix B. The procedure requires !4/inant[ 1995]. measurementsof wind speed,atmosphericpressure,wet and Individual atmosphericsensorson the buoys are sampled dry bulb temperature, sea surface temperature, and cloud once each secondfor an 8.5-min period each hour. Hourly cover. Insufficient information is available from the NDBC valuesof the wind speedand direction are convertedto north buoysto computeheat fluxes.Only a subset(between a half and eastcomponents.Hourly valuesof the wind stressare then and a third dependingon season)of the CalCOFI stationsused estimatedusingthe proceduresuggestedby Large and Pond for estimating the seasonallyaveraged wind stressprovides [1981] for a neutral atmosphere.Fluctuationsat periodsof 1 sufiScient information to computeheat fluxes. day or lessare removedby filtering, usingthe PL64 low-pass filter [Limeburner,1985] which removesalmost all the energy at periodsshorter than 38 hours.Over the period of deploy4. Wind Stress and Heat Flux Patterns ment, severalgapsof varyinglengthsoccurredat each station. Monthly means and standard deviations (along principal In the casewhen gapswere shorterthan 2 days,observations were estimatedby linear interpolation. For longer gaps, no axes) of the wind stressobservationsat each buoy location effort was made to replace the missingobservations. were computedto determinethe seasonalpattern in the Wind The CalCOFI programhasconductedsurveysin the SCB at stress.In this analysis,all the measurementsavailable at each irregular intervalssince1949. The CalCOFI stationpattern is station were used, so that the averagesfor different buoys illustrated in Figure 1. Stationsare located from the southern include measurements in different years (Table 1). The tip of Baja California to north of San Franciscoand extendto monthlyaveragesand standarddeviationellipsesillustratedin about 1000 km offshore.The surveyswere primarily designed Figure 2 suggestthat the wind stressfield can be divided into to measurethe verticaldistributionof physical,biological,and two distinct regimes.The high wind zone is the area offshore chemicalpropertiesof the water at each station. Wind speed from a line betweenPoint Conceptionand Ensenada,Mexico, and direction were measured at nearly all stations and have and including buoys 23 and 47, where the monthly mean
5644
WINANT
AND
DORMAN:
SEASONAL
WIND
STRESS AND
HEAT
FLUX
cantly different from zero, suggestingthat the scale of the atmosphericdisturbancesresponsiblefor the weather is at least as large as the SCB, even thoughthe amplitudeof the fluctuationsvaries with distancefrom the coast.During the summer,wind stressesare significantlycorrelatedin two distinct regions,between the offshorebuoys (23 and 47) and betweenthe inshorebuoys(25, 45, and 48), but correlations betweenbuoyslocatedoffshoreand closeto the coast(e.g.,23 and 45) are not significantly differentfrom zero. This resultis consistentwith differentregimesexistingon either sideof the boundaryseparating the high-speed windfrom the shelteredarea. The distancebetween the buoysis too great to locate this boundaryprecisely.The grid of CalCOFI station,with a nominal stationseparationof 66 km and a smallerseparationbe-
36'
35'
34'
33'
tween stations located near the coast, can be used to resolve 32'
31'
-122'
-121 ø
-120 ø
-119'
-118'
-117 ø
-116ø
Figure 2. Annual cycle of wind stressas observedat the NDBC stations,computedwith all observationsavailableat each station (Table 1). Solid vectorsrepresentthe monthly averagedwind stressvector. Each time seriescorrespondsto the closeststation as illustratedby the square symbol.The standarddeviationellipsesare drawnsuchthat the semimajor axis correspondsto the standarddeviation along the major axis,usingthe samescaleas the mean:when the meanvector extendsbeyondthe ellipse,the mean is largerthan 1 standard deviation.
The wind stress and standard
deviation
units are
pascals.
stressesare large toward the southeastand frequentlylarger than the major axisfluctuations.Inshoreof this line, monthly averagedwind stressesare smaller, the direction is usually onshore,and the meansare frequentlylessthan the standard deviationalong the major axis. Descriptionsof the atmosphericvariabilityalongthe entire west coast [Halliwelland Allen, 1987;Dorman and Winant, 1995]describethe annualcyclein termsof two seasons. During the fall and winter, when the atmosphereis dominated by travelingcyclones andanticyclones with time scaleslessthan 1 month,the stressvectorsillustratedin Figure2 are everywhere lessthan fluctuations.During the springand summerseasons, winds offshoreor near Point Conceptionare persistentand energetic,drivenby the North Pacifichighcombinedwith the southwestthermal low. At the same time, winds near the coast
spatialvariationsin statisticalpropertiesof the wind stressfield with better resolutionthan providedeither in Nelson's[1977] summaryor with the NDBC mooredarray. The averagewind stressvectorsat eachstationand in each seasonare illustratedin Figure 3. An analysisof the sampling errorsassociated with theseaveragesis includedin Appendix A. The standarderror ellipsesare alsoshownin Figure 3. The spatialpattern and seasonalvariation derived from the CalCOFI
observations
and exceed 0.2 Pa. In this area and offshore, the wind stress
vectorsare oriented toward the southeast.Average stressvalues first increasewith distancesouthfrom Point Conception and then decreasesouthof the latitude of San Diego. Closeto the coast,wind stressesare typically 10 times lessthan the maximum
meanwind stresses are nearly10 timeslargerat buoy23 than The
structure
of correlations
between
wind stress fluctua-
tions along the major axis at the differentbuoy locationsdescribesthe spatial structurequantitatively.Correlationswere computedfor the winter (December-February)and summer (June-August),avoidingthe periodswhentransitions between regimesoccur.The unlaggedcorrelationsare summarizedin Table 2. Laggedcorrelationswere alsocomputed,but they do not differ appreciablyfrom these results.During the winter, correlationsbetween the various stations are always signifi-
and directed
toward
the coast.
To emphasizethe largegradients,the amplitudeof the componentof wind stressperpendicularto CalCOFI lines (along 150øT)is shownas a functionof positionalongthe CalCOFI lines in Figure 4, for the springseason.Maximum valuesare reachedabout 100 km offshore,exceptat Point Conception where the maximumis at the coast.The rate of changeof this componentof wind stressalongthe line is the majorcontribution to the wind stress curl and reaches values in excess of 1
/xPam- • alongtheboundary line described above,evenwhen
Table 2. UnlaggedCorrelationBetweenObservationsof the Major Axis Componentof Wind Stressat Different Stations
in the SCB are generallyweak and directed onshore.The largestgradientsin the wind stressfield occurduringthe spring and particularly during the month of April, when monthly within the bight.
are consistent with the station observations
illustratedin Figure 2. The seasonallyaveragedwind stresses are weakerduringfall andwinter than duringspringand summer, althoughin all seasonsthe wind stressincreaseswith distancefrom the coastin the SCB. Maximum averagedwind stressesoccurin the springabout 200 km west of San Diego
Station Station Station Station Station
23 47 25 45 48
Station 23
Station 47
Station 25
Station 45
Station 48
1.00 0.84 0.!9 -0.10 0.15
0.92 1.00 0.30 -0.04 0.30
0.70 0.78 1.00 0.66 0.80
0.53 0.61 0.72 1.00 0.65
0.69 0.77 0.89 0.79 1.00
Correlationslisted in the upper right-handside are for the winter season(December 1991 to February 1992 and December 1992 to February1993). Summerseasoncorrelationsare listedin the lower left-handside (June-August1992 and June-July1993). The period over which correlationswere computedwas determinedby the availabilityof observations from buoys47 and 48. Correlationslessthan or equalto 0.30 are not significantlydifferentfrom zero.
WINANT AND DORMAN: SEASONALWIND STRESSAND HEAT FLUX 36'
36'
34'
34'
32'
E .
-122'
-120'
32'
-118'
-116'
-122'
36'
36'
34'
4'
32 ø
-120'
-118'
-116'
-120'
-118ø
-116 ø
32'
-122'
-120'
-118'
-116ø
-122'
Figure 3. Seasonalpatternsof wind stressbasedon CalCOFI observations. The root of eachvectoris
located atthelocation ofthecorresponding CalCOFIstation. Theellipses represent thestandard errorinthe
mean as describedin AppendixA.
32'
i10.1 Pa 30 •
•
-124'
1
•
-122'
,1
-120ø
i
I
-118'
Figure4. Trackplotsof theamplitude of thewindstress component along150øT (perpendicular to the CalCOFIlines)inthespring. Thecircles oneachtracklinerepresent thelocation of theCalCOFIstation. The
dashedlinesrepresentthe standarderror aboutthe mean.
5645
5646
WINANT
AND
DORMAN:
SEASONAL
WIND
36'
36'
35'
35'
34'
34'
33'
33'
32'
32'
31-122' -121'-120'-119'-118'-117'-116' 36'
35'
35'
4
34'
33'
33'
32'
32'
-122' -121' -120' -119' -118' -117' -116'
AND
HEAT
FLUX
-121' -120 ø -119' -118 ø -117' -116'
36'
31'
STRESS
-121' -120'
-119' -118'
-117' -116'
Figure 5. Seasonalpatternsof wind stresscurl basedon the CalCOFI observations. The unitsare micropascalsper meter, which is numericallyequal to the unitsusedby Nelson[1977]. Contoursintervalsare 0.5/xPa
m-•. Dark shadedareascorrespond to negative (anticyclonic) valuesof thecurl. account
is taken
of the standard
error
in the individual
wind
stress estimates.
differencebetween estimatesis due to the different grid sizes used to estimate
The wind stress curl leads to vertical
velocities
the wind stress curl.
in the under-
Thesevaluesof wind stresscurl are large. If the divergence lyingocean,asthe divergencein the Ekman fluxescorrespond- in the horizontalmasstransportdue to sucha curl is balanced ing to a wind stresscurl is balancedby verticalmasstransport, by verticaltransportwith velocitiesw, thesemaybe computedas the mechanismresponsiblefor open oceanupwellingor downV welling.In a rectangularcoordinatesystem,the wind stresscurl V is
V•
O,y ox
oy
where x and y representthe east and west directions,respectively, and ß representsthe seasonallyaveragedwind stress. The north and east componentsof the averagedwind stress vectorshave been placed onto a grid extendingover the area illustratedin Figure 2, with grid pointsspacedevery0.2ø, using the gridding algorithm, based on splinesunder tension, describedby Smithand Wessel[1990].The partial derivativesare estimated as centered differencesthroughout the interior of the grid, and forward or backwarddifferenceswere usedon the boundariesof the grid. Contoursof V computedin eachseason are illustrated in Figure 5. The resultingspatial patternsvary markedly between sea-
Thecorresponding maximum verticalvelocities are4 x 10-5 m s-•. For comparison, RudnickandDavis[1988]estimated vertical velocitiesin the northern California coastal upwelling
areato havean averagevalueof 5 x 10-s m s-• overthe
upwellingfavorableseason. Monthly averagesof air and seatemperature(measured1 m beneaththe surface)computedfrom the NDBC measurements are illustratedin Figure 6. The monthly averagedwater temperature alwaysequals or exceedsthe air temperature by a greater amount duringthe summerthan duringthe winter and fall. In the central and southernportion of the SCB, minimum water temperaturesoccurin January.Near Point Conception, where coastalupwelling reachesmaximum intensityin early spring,minimumwater temperaturesoccurin April. Maps of seasonallyaveragedair temperaturesderivedfrom sons,and a maximum wind stresscurl, located 150 km south of the CalCOFI observationsare illustrated in Figure 7. MiniSantaBarbaraand exceeding 3 /xPam-•, occursin spring. mum air temperaturesare generally observedin the winter. Errors associatedwith this estimationprocedureare estimated Exceptin fall the minimum temperatureis found in the vicinity in AppendixA to be of the orderof 1 /xPam-•, sothat the of Point Conception.During the fall and winter the air temmaximumvaluesare significantlydifferentfrom zero. For com- perature differenceacrossthe map is about 3øC.During the parison,the maximum curl error estimatedby Nelson [1977] spring, an area of warmer air begins to develop near San for this area is 0.93 /xPam-• for the monthof May. The Diego. In the summer,the air continuesto warm in this area,
WINANT
AND
DORMAN:
SEASONAL
WIND
STRESS
AND
HEAT
FLUX
5647
latter exceedingthe wet bulb temperatureby betweenIøC and Maps of seasonallyaveraged sea temperaturesmeasured 1 m beneath the surfaceduring the CalCOFI cruisesare illustrated in Figure 8. The sea temperature patterns generally correspondto the air temperature patterns,with the sea temperature exceedingthe air temperature by about IøC, in all seasons. The seasurfacetemperaturegradientalongthe coast, betweenSan Diego and Point Conception,is near 6øCduring the summer,and the patternis similarto that derivedfrom the CalCOFI observationsat 10-m depth,asreportedbyLynn et al. [1982]. Seasonaldifferencesbetween temperaturesillustrated in Figure 8 are consistentwith observationsfrom moored buoys,illustrated in Figure 6. Seasonalmapsof the heat flux from the atmosphereto the ocean are illustratedin Figure 9. These maps differ from the resultsof NH in two significantrespects.First, net heat exchangevalues computed from the CalCOFI observationsare
35'N
33'N 47
32'N
generallylargerby 50 W m-2. Second,while the patterns
31'N
122'W
i2 'W
120'W
119'W
I18'W
117'W
116'W
illustratedin Figure 9 are fairly uniform in fall and winter, the summer and spring maps are characterizedby large spatial Figure 6. Annual cycle of temperatures observed at the gradients, whereasthe NH maps do not showmuch structure NDBC stations.The circles represent the sea temperature in any season, over areas comparable to that illustrated in (measured1 m beneaththe surface),whichequalsor exceeds Figure 9. the air temperature. Quarterly averagesof the NH estimatesfor two different 1ø squaresare compared,in Tables3a and 3b, to heat fluxesbased reachingmaximum seasonallyaveragedvaluesof about 20øC. on the CalCOFI observationsaveragedin the same1øsquares. The maximum air temperature difference over the area The net heat flux values indicate that CalCOFI observations mappedin Figure 7 exceeds6øCin that season.Patternsof wet predicta net fluxabout50 W m-2 greaterthanNH. Of the bulb temperaturesare similarto the air temperature,with the %ur constituentswhich make up the net heat, only the latent
36 ø
36 •
35'
35 •
34 ø
34'
33 ø
33 ø
32 ø
32 ø
31' -122 ø -121 • -120 • -1!9'
31 ø
-!18 • -117 • -116 •
.!22 • .121 ø -120 ø .119 ø -118 ø -117 ø -116 ø
36 •
36'
35 •
35 ø
34'
34'
'17
33ø ( / 32'/• (/
33•
32'
3[122ø .121 • .120 ø.119 • -118 • .117 • -116 ø 3•122• .121 ø.!20 • .119 ø.1!8ø.117 ø.116 ø Figure 7. Seasonalpatternsof air temperaturebasedon the CalCOFI observations.The units are degrees Celsius.
5648
WINANT AND DORMAN: SEASONAL WIND STRESSAND HEAT FLUX 36'
36'
35'
35'
34'
34'
33'
33'
32'
32 ø
31'
31'
36'
36'
35ø
35'
34'
34ø
33'
33ø
32' 31' -122' -121' -120' -119' -118' -117' -116'
32' 31' -122ø -121' -120ø -119' -1!80 -117' -116'
-122' 421' -120' -119' -118' -117' -116ø
•
•
•
-122' -121• -120' -119' -118' -117• -116ø
Figure8. Seasonal patterns ofseatemperature based ontheCalCOFIobse•ations. Theunitsaredegrees Celsius.
the windspeedinheat differssubstantially. NH estimatelatent heat transfer (MABL) to turn.In springandsummer rapidlyto the east(Figure fromtheocean totheatmosphere ofabout50W m-2,whereas creasesto the southand decreases 3). A similarphenomenon hasbeenobserved to occurin the offPointArena[Winant etal., 1988],where icallya fewwattsper squaremeter.The methodusedto com- springandsummer puteheatfluxes(AppendixB) is similarto thatusedby NH. the southward flow accelerates as it turns south due to the off Differencesin averagedobserved valuesusedto estimatethe changein coastalorientation.Atmosphericsoundings the CalCOFI observationsresult in much smaller values, typ-
latent heat must therefore be responsiblefor the different northern California reveal a vertical structurein which a strong estimates. The latent heat is a function of the difference be-
inversion separates the MABL fromthefree atmosphere. The tweensaturationvaporpressureat the seasurfacetemperature depthof the MABL is of the orderof 100m, or lessthanthe
and the watervaporpressure,and NH reportsubstantiallyheightof the coastalmountainrange.If similarconditions largervaluesfor thisdifference. An attemptwasmadeto com- prevail,theflowisforcedto expandaroundPointConception pute dewpointtemperature depressions from the variables to conformto therapidchangein orientationof thecoast.The listedin NH. Thesewere found to be 2 or 3 timesgreater than speedof gravitywavestravelingat the interfacebetweenthe the depressions obtainedfrom CalCOFI.The CalCOFI pro- MABL andthe free atmosphere aboveis of the orderof 10 m ceduresfor measuring dewpointhavebeenreviewed,andthe s-•. Averageflow velocitiesin the MABL can exceedthis dewpointdepressions reportedby CalCOFI,of the orderof value,and the MABL can then developas a supercriticalreiøC,appearto be reliable.It hasnotbeenpossible to resolve ducedgravityflow.In a supercritical statean expanding flow thesedifferencesany further, and it is concludedthat there accelerates asit turns,and the layerheightdecreases dramatexistdiscrepancies oftheorderof 50W m-2 in theestimationically,as requiredby continuity.The latter conditionsetsa of air-sea heat fluxes. limiton themaximumanglebywhichtheflowcanexpand,and The largestcontributorto the net heat exchange is Qshort, separation occursif the boundary changes by a largerangle. the shortwave radiation term. As noted in AppendixB, Q short The Bernouilli conditionsetsa maximumvalue for the velocity
is dependent on the cloudcover.The CalCOFIobservationsof the flow downstreamof the expansion. suggest persistent patterns of variationin cloudcover,andthe Samelson [1992]constructed a numericalmodelof the maspatialstructure of Qnetillustrated in Figure9 is dueto vari- rinelayeralongthenorthernCaliforniacoastandshowed that ationsin Qsho•t whichin turn are due to variationin cloud manyof theobserved features, notablytheacceleration which cover.Thesetakeplaceonsmallerscales thanresolved byNH. accompanies theexpansion, arewellpredicted asfeaturesof a reducedgravitysupercritical flow.The effectsof frictionwere 5.
Discussion
alsoshownto be importantin understanding the development of the expansion. Samelson andLentz Southof PointConception the changein coastalorientation of the flow downstream
forces the flow in the marine atmosphericboundarylayer
[1994]haveestimated termsin thehorizontal momentum bal-
WINANT AND DORMAN: SEASONALWIND STRESSAND HEAT FLUX 36'
36 ø
35'
35'
34'
34'
33 •
33 ø
32 ø
32'
31 ø
31 ø
-122 ø -121' -120 • -119 ø -118 • -117 ø -116 ø
-122' -121 ø -120 ø -119 •' -118 ø -117 ø -116 ø
36•
36ø
35'i
3so
34'
34•
33ø
33*
32•
32*
31 • -122 * -121 ø .120 ø .119 • .118 ø .117 ø -116 •
5649
x 31 ø
Figure9. Seasonal patternsof net heatfluxfrom the atmosphere to the oceanbasedon the CalCOFI observations. The unitsare wattsper squaremeter.Contourintervalis 20 W m-2 ance for the MABL off northern California and have shown
commonlyavailable.Dortnanand Winant[1995]compared
that duringperiodsof strongsouthwardflow, the cross-shore soundinginformation from severalcoastalstationsto aircraft momentum balance is between the acceleration and the cross-
soundingover the oceanand concludedthat the coastalsound-
shorepressure gradient,with the latterbeinglargerthanthe geostrophic gradient,and the alongshoremomentumbalance isbetween verticalstress divergence andthealongshore pressuregradient,consistent withSatnelson's [1992]modelresults. Couldthe flow overthe SCB,particularlynear PointCon-
ingsoverestimate the MABL depthoverwaterby a factor between2 and3. Withoutreliableestimates of the heightof
ception, be similar to that observed off northern California?
The conditionfor supercriticalflow in the MABL is that the Froudenumber,Fr, definedasthe squaredratio of the verti-
the MABL it is not possibleto concludewhether or not the flow is supercritical.
Threeobservations are, however,consistent with a supercriticalexpansion in theleeof PointConception. Thespeedof thelayerandthe corresponding windstressincreaseasthe flow
expands downstream. A supercritical flowcanonlyturnby a callyaveraged velocity in thelayerto thespeedof gravity waves finiteangle(lessthan90ø),beyond whichtheflowmustsepaat the interface,be greaterthan1. Windvelocityobservationsrate.The sheltered regioneastof thepointcouldcorrespond are commonly available.The wavespeeddependson the rel- to thisseparated flowarea.Finally,seasonal patternsof cloud ativedensityof the MABL andthe free atmosphere aboveas cover deduced from the CalCOFI observation show relative wellastheheightof theMABL. Thesetwovariables canonly minimasouthof Point Conception,the areawheremaximum be deducedfrom soundings of the atmosphere, whichare not speedsand minimumMABL heightsare expected.The low cloudcovercouldbeexplained bytheheightdecreasing below the minimum lifting condensation level, suppressing cloud Table 3a. Seasonal Estimates of Heat Fluxes Estimated cover in that area.
From CalCOFI ComparedWith NH Values
Season N Winter Spring Summer Fall
778 796 858 875
Qnet 60/14 190/169 243/215 112/87
The major findingreportedhere is that gradientsin wind
Qshort Qlong Qlatent Q.....ble 100/120 225/242 282/271 144/172
40/56 36/49 37/40 31/50
2/46 3/28 1/19 2/33
- 1/4 -4/-3 0/-3 -1/2
Table 3b. SameasTable3a, but for a 1ø SquareCentered on 121øW, 33øN
Season N
The CalCOFIestimates, beforethe slash,are averaged frommea- Winter surements in a 1ø squarecenteredon 119øW,34øN.The NH values, Spring after the slash,are for the samesquare.The numberof observations Summer usedto make up the NH estimatesis listed in the columnheadedN.
Standard errorsin netheatestimates byNH aretypically 5 W m 2.
Fall
644 674 639 617
Qnot 56/- 1 194/115 179/146 107/55
Qshort Qlong Qlatcnt Q.....ble 96/114 226/223 177/224 151/154
34/45 36/40 7/25 36/38
7/62 3/64 -3/49 8/56
0/8 - 8/4 7/3 1/5
5650
WINANT
AND
DORMAN:
SEASONAL
WIND
STRESS
AND
HEAT
FLUX
0.14
O. lO
0.08
0.04 I
I
i
I
I
I
oo
lOO
26.4
200
- 26.4
300 400
---"-26.8
27.0
•27.0
27.0
• 500
0.10
0.08 0.04
Summer I
I
I
I
0
'0
lOO
26.0
2o0•' ...•.. 26.4 300 •
26.8
400
500
120
100
90
80
70
60 53 45
37
28
CCOFIStation Number 32
'
•
600
•
'
400
•
'
200
Distance (km)
0
Figure 10. Seasonalpatternsof wind stressand corresponding densityof underlyingoceanalongCalCOFI line 90. The density(%) contoursare redrawnfrom Lynn et al. [1982].Line 90 is locatedon Figure 1.
stresslarger than thosereported by Nelson [1977] or Bakun and Nelson [1991] occur over fairly short distances.Is this important?Hickey [1978] describesthe seasonalvariation of the California Current systemand relates dynamicalfeatures to the wind stressand the wind stresscurl. South of Cape Mendocino,a northwardmean flow is reported,with maximum amplitudewhere the wind stresscurl is locallystrongest.The CalCOFI observationswere used by L•vnnet al. [1982] and Lynn and Simpson[1987] to describethe seasonalvariation of the structureof isopycnalsurfacesin the California Current system.In the SBC, vertical sectionsof propertiesperpendicular to the coastand surfacemapssuggestthat an area of high
density,correspondingto low temperatures,is often found about200 km off SanDiego. Seasonalsectionsof density(%) alongCalCOFI line 90 are illustratedin Figure 10.The doming of isopycnalsnear station 53 occursin all seasonsbut is particularlynoticeableduring springand summer.Featuressuch asthesemaybe interpretedin termsof positivewind stresscurl resultingin open oceanupwelling.The wind stressacting on the ocean along line 90 does exhibit stronggradients,correspondingto cycloniccurl abovethe areawhere domingoccurs. The distributionof density in the SCB waters can only be explainedin terms of a model which includeshorizontal and vertical advectionas well as fluxesat the boundary.The colt-
WINANT
AND DORMAN:
SEASONAL
WIND
36'
36'
34'
34'
32'
32'
-122'
-120'
-118'
-116ø
-122ø
36'
36'
34 ø
34 ß
STRESS AND HEAT
FLUX
5651
-120'
-118'
-116'
.120 ø
-1!8 •
.116*
32 •
-122•
.120ø
.118 ø
.116ø
.122 ø
Figure A1. Numberof CalCOFI stationswherewind speedand directionobservations are availablefor each seasonin the period 1949-1994.The diameterof eachcircleis proportionalto the numberof observations.
cation of isopycnaldoming and wind stresscurl illustrated in These standarderrors can be comparedto standarderrors for Figure 10 doessuggest,however,that knowledgeof the wind the sameparameterestimatedby Nelson[1977]which range stresspatternson scalesof order of tensof kilometersmight be between0.01 and 0.001 Pa, with a typicalvalue near 0.003 Pa, requiredfor sucha model to successfully predict featuressuch or about 5 times lessthan the value computedwith the samas are illustrated in Figure 10, in addition to the evaluation of pling approachused here. This differenceis due to larger horizontaland vertical advectiveprocesses. samplesizein the shipdata setusedby Nelson.Along anyline within the SCB the standard error increases with distance from
Appendix A: Error Estimates Seasonallyaveragedpropertiessuchasthosepresentedhere basedon CalCOFI observationsare subjectto both systematic and random errors.Bakun and Nelson [1991] present a comprehensivereviewof both types.Here attentionis restrictedto random samplingerrors, which for a scalar quantity such as temperatureare taken to be the squareroot of the varianceof all the observations in a season to the number of observations.
For the wind stressvector, the variance between observations
the coast(Figure 4), so that eventhe smallstressvectorsnear the coastare larger than their standarderror. In each season, averagestressestimatesare ignored for any of the three following reasons: if the maximum standard error exceeds the mean, if the maximum standarderror exceeds0.1 Pa, or if the number of observations is less than 18. The latter criterion was
selectedas an effectiveway of screeningout stationswhich were only sampledduring a limited portion of the 45-year period.
The consistentpattern of wind stressthroughoutthe area considered lendscredenceto the result,althoughthis is more covarianceis null), and the standarderrorsare taken to be the the spatialaveraging squareroot of the ratio of the variancealong each axisto the difficultto quantify,exceptby increasing (therebyincreasing the numberof samplescontributingtoward number of observations. the spatialresolution),but the value The number of observationsthus stronglyaffectsthe stan- the meanand decreasing of the present analysis lies in the enhancedspatialresolution dard error of the mean. In this work, CalCOFI observations over the 45-year period 1949-1994 have been used. For the afforded by the CalCOFI measurements.The values of wind wind stressaverages,the number of observationsat each site stresswhich result from both CalCOFI and ship observations whichsupportsthe generalvalidity and in each seasonis illustrated in Figure AI, in a similar are remarkablyconsistent, displayto that adopted by Lynn and Simpson[1987]. The of the two data sets. The wind stresscurl is estimatedas a centereddifference,or largestnumber of observationsare along lines 80, 90, and 93, while lines 83 and 87 have about half the densityof measure- overtwo grid points,sothe characteristic lengthoverwhichthe ments.Typicalvaluesof the major axisstandarderror are 0.015 curl is estimatedis 40 km. Nelson[1977] estimatesthe error in Pa for the spring and summer and 0.025 Pa for the winter. curl astwicethe standarderror in stressdividedby the distance
in eachseasonis first rotatedinto principalaxes(where the
5652
WINANT
AND
DORMAN:
SEASONAL
Table A1. Standard Error in SeasonallyAveraged Air Temperature, Wet Bulb Temperature, SurfaceTemperature, Salinity, Cloud Cover, and Net Heat Flux
WIND
STRESS
AND
HEAT
FLUX
shortwave is adjustedupwardby addingin the indirect short wave radiation in a formula given by Qo = Ia + (0.91 - Ia)0.5
S(rair) , S(rwet), S(SSr), S(Salt), S(cloud), S(qtot),
øC Winter
0.44
øC
øC
0.51
0.23
ppt 0.042
octants W m-2 0.11
15.5
Spring
0.40
0.45
0.21
0.046
0.10
24.8
Summer Fall
0.42 0.38
0.43 0.44
0.31 0.24
0.047 0.042
0.08 0.10
18.7 18.7
I a is the direct incomingradiation at the Earth's surface.C is the total cloudcover(rangingbetween0 and 1 but is set to 0 if less than 0.25 of cloud cover). Here h is the noon solar altitude in degrees. B3.
Back
Radiation
To compute the net upward back radiation in watts per square meter, a modified formula from Brunt [1932], with of 10-7 Pa m-•. An analogous procedure yieldsan erroresti- empirical constantsof Budyko [1956], and the linear cloud mate of 10 -6 Pa m -• for the calculation included here. The correctionof Reed [1976] are employed,resultingin differenceis partly due to the larger standarderrors for mean over which the curl is estimated,resultingin typicalcurl errors
stress and the smaller distance over which the curl is estimated.
Q,ong = 5.50x 10-8(rs+ 273.16)4(0.39-0.05vø'S)(1 - 0.9C)
The curl estimatesovermostof the field (Figure5) are of this Ts is the seatemperaturein degreesCelsiusand v is the water order, exceptfor the area southof the ChannelIslands,during vapor pressurein hectopascals.C is the total cloud cover. the springand summer,where curl estimatesare 2 and even3 times larger than the expectederror. Standard errors in temperaturesand heat fluxes are estimated as the squareroot of the ratio of the variancein each seasonto the number of availablesamples.Since someof the variablesrequiredfor estimatingheat fluxwere not sampledin the earlier part of the program, the number of samplesavailable dependsnot only on location and seasonbut also on the individual variable. For instance, cloud cover, which is an im-
portant parameter in estimatingboth long and short wave radiation,was only reported after 1960. Accordingly,with the exceptionof sea surfacetemperatures,the number of observations availableto determine seasonallyaveragedproperties is generallylessthan the number of observationsillustratedin Figure A1, by as much as a factor of 2. Estimates of the standard error in the various parameters are presentedin Table A1. The standarderror in sea surface temperatureis abouthalf the air and dew point errors,because the varianceis lower and the number of observationsis larger. The standard error in total heat flux varies seasonally,as the long wave radiation, and is due for the most part to the relatively large variability between cloud cover measurements.
Appendix B: Heat Flux Formulations B1.
Total Heat Exchange
The net heat exchangeacrossthe sea surfaceQ net,in watts per squaremeter, is given by
B4.
Latent
Heat
The bulk latent heat flux in watts per squaremeter is taken from Kraus [1972] and has the form
Qlatent = 9LCe(qo- q•o)U•o.
Assuming the densityof air 9 - 1.22kg m-3, latentheatof vaporization L - 2.45 x 106 J kg-•, the dragcoefficient Ce = 0.0013, and the pressureP = 1013.25, the bulk latent heat flux becomes
Qlatent: 2.38(0.98Vsca- V)U
where Vsea is the saturationvapor pressureat seasurfacewater temperaturein hectopascals, multiplied by 0.98 to accountfor salinityeffects[Miyake,1952].V is the atmospheric watervapor pressurein hectopascals.U is the 10-m height wind speedin meters per second. B5.
Sensible
Heat
The bulk sensibleheat flux in wattsper squaremeter is taken from Kraus [1972] and has the form
Q....ible= pcpC14(Ts - Ta)S10 Assumingthe samevaluesas for the latent heat, specificheat
of airCp= 1.00 x 103J kg-• perøC-• andCH = 0.0013, the bulk sensible heat flux becomes
Qnet: Q short q- Qlong q- Qlatent q- Q ....ible where the factorson the right-handsideof the equationare the short wave radiation, the back radiation, the latent heat, and the sensible heat fluxes. B2.
Short
Wave
Radiation
The downward short wave radiation in watts per square meter is computedfrom Reed [1977] by
Qshort = (1 -- aOQo(1 - 0.62C + 0.019h) Here a • is the fraction of incomingradiation reflectedfrom sea
surfacewhich is extractedfrom tablesby Payne[1972].Qo is direct and diffuseradiation from cloudlesssky with an atmospherictransmissioncoefficientof 0.7 [List, 1949]. The direct
Q .... ible= 1.59(Ts-
Ta)U
Here Ts is seasurfacetemperaturein degreesCelsius,Ta is the air temperature in degreesCelsius,and U is the 10-m height wind speedin meters per second.
Acknowledgments. The work reported here was sponsoredby the Minerals ManagementServiceof the U.S. Department of the Interior under cooperativeagreement 14-35-0001-30571and by the Office of Naval Research under grant N00014-94-1-0232.The original manuscriptwasconsiderablyimproveddue to the helpful suggestions of A. Huyer and two other reviewers.We would like to expressour appreciation for the remarkable servations.
effort which resulted
in the CalCOFI
ob-
WINANT
AND
DORMAN:
SEASONAL
References Bakun,A., and C. S. Nelson,The seasonalcycleof wind-stresscurl in subtropicaleasternboundarycurrentregions,J. Phys.Oceanogr., 20, 1815-1834, 1991.
WIND
STRESS
AND
HEAT
FLUX
5653
seasonalvariability of its physicalcharacteristics, J. Geophys.Res., 92(C12), 12,947-12,966,1987. Lynn, R. J., K. A. Bliss, and L. E. Eber, Verticaland Horizontal Distributionsof SeasonalMean Temperature, Salinity,Sigma-t,Stability, Dynamic Height, Oxygenand OxygenSaturationin the California Current,1950-1978,CalCOFIAtlas30, 513 pp., Stateof Calif. Mar.
Beardsley,R. C., C. E. Dorman, C. A. Friehe, L. K. Rosenfeld,and C. D. Winant, Local atmosphericforcingduringthe CoastalOcean Res. Comm., La Jolla, 1982. DynamicsExperiment,I, A descriptionof the marineboundarylayer Miyake, Y., A table of the saturatedvapor pressureof sea water, and atmosphericconditionsover a northern California upwelling Oceanogr.Mag., 4, 95-118, 1952. region,J. Geophys.Res.,92(C2), 1467-1488,1987. Nelson, C. S., Wind stress and wind stress curl over the California Brink, K. H., D. W. Stuart, and J. C. Van Leer, Observations of the current,NOAA Tech.Rep. NMFS SSRF-714,87 pp., Natl. Oceanic coastalupwellingregionnear 34 x 30'N off California:Spring1981, and Atmos.Admin., SilverSpring,Md., 1977. J. Phys.Oceanogr.,14, 378-391, 1984. Nelson, C. S., and D. M. Husby, Climatologyof surfaceheat fluxes Brunt, D., Notes on radiationin the atmosphere,Q. J. R. Meteorol. over the California current region,NOAA Tech.Rep. NMFS SSRFSoc., 58, 389-420, 1932. 763, 94 pp., Natl. Oceanicand Atmos.Admin., SilverSpring,Md., Budyko,M. I., The Heat Balanceof the Earth's Surface(in Russian), 1983. 255 pp., Gidrometeorol.Izd., St. Petersburg,Russia,1956. (English Overland, J. E., Scale analysisof marine winds in straitsand along translation,Off. of Tech. Serv.,U.S. Dep. of Commer.,Washington, mountainouscoasts,Mon. WeatherRev., 112, 2530-2534, 1984. D.C., 1958.) Payne, R. E., Albedo of the sea surface,J. Atmos. Sci., 29, 959-970, Dorman, C. E., Winds between San Diego and San Clemente Island, 1972. J. Geophys. Res.,87(C12), 9636-9646, 1982. Reed, R. K., On estimation of net long-wave radiation from the Dorman, C. E., and C. D. Winant, Buoy observationsof the atmooceans,J. Geophys. Res.,81(33), 5793-5794,1976. sphere along the west coast of the United States, 1981-1990, J. Reed, R. K., On estimatinginsolationover the ocean,J. Phys.OceanGeophys. Res.,100(C8), 16,029-16,044,1995. ogr., 7, 482-485, 1977. Halliwell, G. R., Jr., and J. S. Allen, The large scalecoastalwind field Rudnick, D. L., and R. E. Davis, Mass and heat budgets on the alongthe west coastof North America, 1981-82, J. Geophys.Res., northern California continentalshelf, J. Geophys.Res., 93(Cll), 92(C2), 1861-1884, 1987. 14,013-14,024, 1988. Hamilton, G. D., NOAA Data Buoy Officeprograms,Bull. Am. Me- Samelson,R. M., Supercriticalmarine-layer flow along a smoothly teorol.Soc.,6•(9), 1012-1017,1980. varyingcoastline,J. Atmos. Sci., 49, 1571-1584, 1992. Hickey, B. M., The California Current system--Hypothesesand facts, Samelson, R. M., and S. J. Lentz, The horizontal momentum balance Prog. Oceanogr.,8, 191-279, 1978. in the marine atmosphericboundarylayer during CODE-2, J. AtHickey, B. M., Variability in two deep coastalbasins(Santa Monica mos. Sci., 49, 1571-1584, 1994. and San Pedro) off SouthernCalifornia,J. Geophys.Res.,96(C9), Smith, W. H. F., and P. Wessel, Gridding with continuouscurvature 16,689-16,708, 1991. splinesin tension,Geophysics, 55, 293-305, 1990. Hickey, B. M., Circulationover the SantaMonica-SanPedro basinand Winant, C., C. Dorman, C. Friehe, and R. Beardsley,The marine layer shelf, Prog.Oceanogr.,30, 37-115, 1992. off northern California: An exampleof supercriticalchannelflow,J. Kraus, E. B., Atmospheric-OceanInteraction, 275 pp., Oxford Univ. Atmos. Sci., 45, 3588-3606, 1988. Press, New York, 1972.
Large,W. G., and S. Pond,Open oceanmomentumflux measurements C. E. Dorman and C. D. Winant, Center for CoastalStudies,Scripps in moderate to strongwinds,J. Phys.Oceanogr.,11,324-336, 1981. Institution of Oceanography,9500 Gilman Drive, La Jolla, CA 92093Limeburner, R. (Ed.), CODE-2: Moored array and large-scaledata 0209. rcport, WHOI Tech. Rep. 85-35, CODE Tech. Rep. 38, 220 pp., Woods Hole Oceanogr. Inst., Woods Hole, Mass., 1985. List, R. J., SmithsonianMeteorologicalTables,6th ed., Smithson.Misc. Collect.,114, 527 pp., 1949. (ReceivedSeptember22, 1995; revisedAugust 27, 1996; Lynn, R. J., and J. J. Simpson,The California current system:The acceptedSeptember5, 1996.)
