Sep 15, 1998 - surface heating and offshore advection. This agrees with .... developed for application to the tropical Pacific and in- corporates ..... thermistor. Custom (WHOI) ..... water leads to a stable marine layer and development of internal ...
JOURNAL OF GEOPHYSICAL
RESEARCH, VOL. 103, NO. C10, PAGES 21,553-21,586, SEPTEMBER 15, 1998
Surface heat flux variability over the northern California
shelf
RobertC. Beardsley Departmentof PhysicalOceanography, WoodsHole Oceanographic Institution Woods Hole, Massachusetts
I•A•,,m,.AO Dever 4--•.•
VV
•1
•.•
•
ß
Centerfor CoastalStudies,ScrippsInstitutionof Oceanography Universityof California,SanDiego, La Jolla Steven J. Lentz and Jerome P. Dean
Departmentof PhysicalOceanography, WoodsHole Oceanographic Institution Woods Hole, Massachusetts
Abstract. Surfaceheat flux components are estimatedat a midshelfsite over the northern California shelf using moored measurementsfrom the 1981-1982
CoastalOceanDynamicsExperiment(CODE) and the 1988-1989Shelf Mixed Layer Experiment(SMILE). Time seriesof estimatedfluxesextendfrom early winter through summer upwellingconditions,allowingexaminationof seasonal variations as well as synoptic events. On a seasonaltimescale,the surfaceheat flux is stronglyinfluencednet surfaceheat flux are the annualvariation in incident
shortwavesolar radiation (insolation)and the atmosphericspringtransition. Between mid-November 1988 and late February 1989, insolation is weak and the
meandailyaveraged heatfluxis nearlyzero(absolute valuelessthan10W m-2), with a standarddeviationof -•50 W m-2. Beginning in March,insolationincreases
markedly, andtypicaldaily-averaged heatfluxes increase to greater than100W m-2 by the springtransitionin April or May. In June and July, the averageheat flux
is near200W m-2, with a standarddeviationof -•90 W m-2. In winter,the daily-averagedheat flux varieson periodsof severaldays. Net heat flux losses
canrangeup to 130W m-2. Theselosses arenot identified with anyonetype of event. For example, comparableheat flux lossescan occurfor very low relative
humidities(RHs), moderatewinds,and clearskies,and for highRHs, highwinds, and cloudyskies. In summer,surfaceheat flux variability is stronglyinfluencedby upwellingand relaxationevents.Upwellingis characterized by clearskiesand high equatorwardwinds,while relaxationis characterizedby the presenceof cloudsand low or northward winds. These conditions lead to opposingchangesin insolation
and in longwaveradiative coolingand latent heat flux. Variability in insolation dominates, and the daily-averagedheat flux into the ocean is greatest during
upwelling events(upto 350W m-2 or more)andleastduringrelaxation events (sometimes lessthan100W m-2). 1. Introduction
It has been well known since the pioneering coastal upwelling experiments conductedin the 1970s that at-
Halpern, 1976]. The Coastal Ocean DynamicsExperiment (CODE) was conductedover the northernCalifornia shelf during spring-summer 1981 and 1982 in part to understand the responseof the coastal ocean to
mosphericsurfaceforcing (i.e., wind stressand heat stronglocal surfaceforcing (i.e., persistentupwellingflux) can strongly influencevertical mixing, stratification, and currentsover the continentalshelf [e.g., favorablewindsand largepositiveheat flux) duringthe coastalupwellingseason[Beardsley and Lentz, 1987]. The ShelfMixed LayerExperiment(SMILE) wasconCopyright 1998 by the American GeophysicalUnion. Paper number 98JC01458.
0148-0227/98/983C-0145859.00
ducted in the same area during 1988-1989 to study the coastal ocean surface boundary layer responseto the more variable atmospheric forcing which characterizes
21,553
21,554
BEARDSLEYET AL.: SURFACEHEATFLUX OVERNORTHERNCALIFORNIASHELF CrescentCmty
39 ø 12'N
39 ø 00'N
LosAngele•
38 ø 48'N
to a doubling of the surface heat flux from the base case examined. However, the same study also found the presenceof surface heating was important in that surface heating greatly reduced the depth of the surface boundary layer compared to a no-surface-heating case. During SMILE, the surfacemixed layer deepens in responseto the winter surfacecoolingdiscussedhere. The spring transition to upwelling occurred in late
March beforeCode i (mid-April through July 1981) and in mid-April during Code 2 (late March through July 1982), so both observationperiodsincludemuch of the upwellingseason.SMILE (mid-November1988 through mid-May 1989) coveredwinter and the sub-
38ø36'N
38ø24'N Bay Marine Lab (BB) N
\ ßNDBC? 3
38ø12'N
38ø00'N ' 124ø00'W
'\' 123 ø30'W
'
Pt•.• Re 123 ø00'W
Figure 1. Map of CODE and SMILE regionbetweenPoint Reyesand Point Arena, showingthe mid-
sequent spring transition. Taken together, the CODE and SMILE measurementperiods coverall but fall, suggestingthat a careful analysisof the C3 heat flux time seriesshould provide a good comparisonwith existing climatology and new insight into heat flux variability on shorter timescales, especially those associatedwith synoptic weather events which characterize the northern California
shelf.
This paper is organized as follows. The basic formu-
shelf site C3 where the surface heat flux was estimated from moored measurementsand other sites providing
las used to estimate
44014 and 44013. The coastalmountain range is shown
SMILE, measurement uncertainty, and hourly time se-
heat
flux from moored
data
are
data used in this study. NDBC13 and NDBC14 de- presentednext (section2), followedby brief descripnote NOAA Data Buoy Center environmentalbuoys tions of the moored instrumentation used in CODE and
shaded,with lightlyand heavilyareasrepresenting ele- riesof the basicvariablesmeasuredat C3 (section3). vationsgreaterthan 305m and 610m, respectively. The C3 daily-averaged heat flux time series are presentednext (section4), ]•ollowed by a detailedlook at the influence of synoptic weather eventson the C3 heat
winteroffnorthernCalifornia[Alessiet al., 1991;Dever, flux (section5). The spatialvariabilityof someheat flux componentsis summarizednext (section6), fol1997]. In both CODE and SMILE, atmosphericand oceano-
lowedby a comparisonof the C3 monthly-averagedheat
graphicvariablesweremeasuredat a midshelfsite (de- flux time serieswith the monthly mean climatology of noted C3 in Figure 1) to allow estimationof surface Nelsonand Husby[1983](section7) and a summary (section8). AppendixA presentsa forcing. The resultant surfaceheat flux time serieshave and conclusions been used by different investigators to study a vari-
discussionof measurementproblemsand solutions,and
ety of dynamical processes overthe northernCalifornia Appendix B presentsa time seriesof surfaceheat flux shelf.For example,RudnickandDavis[1988]and Lentz estimated from moored measurements [1987a]haveshownthat the surfaceheat flux becomes Oregon shelf during summer 1972. important in summer to the shelf volume heat budget on timescalesof about onemonth or more, althoughit is small compared to advective fluxes on event timescales
2. Estimation
of Surface
From
Data
Moored
obtained
Heat
on the
Flux
of days to weeks. Similarly, Dever and Lentz [1994] showedthat the surface heat flux is important to the
meanheat balancein spring.The surfaceheat flux also influencesthe character of the oceanicsurfaceboundary
layer. Brink [1983]and Rosenreid [1987]foundevidence that the surface heat flux can modify diurnal variability in the surface mixed layer in coastal upwelling re-
gions.On subtidaltimescales, Lentz[1992]foundthat the magnitude of a positive surface heat flux did not strongly influencesurfacemixed layer depth during active upwelling, a fact he attributed to a balancebetween surfaceheating and offshoreadvection. This agreeswith
the Federiukand Allen [1995]modelstudyof summer upwelling circulation off Oregon which found the surface boundary layer structure was relatively insensitive
The net surface heat flux Qn into the ocean is the sum of four components
Qn = qi + Qb+ q s + Q•,
(1)
where Qi and Qb are the net incident shortwave and longwaveradiation fluxes, Qs the sensibleheat flux due to air-sea temperature difference, and Qt the latent heat flux due to water vapor transport. Here, positive
flux valuesindicate flux into the ocean. The following formulaswere usedto estimate thesecomponentsfrom hourly averaged moored data. Shortwaveheat flux was estimated by
qi = (1- Ab)I,•o,
(2)
BEARDSLEY ET AL.' SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF
where Ab is the ocean albedo indicating the fraction of
21,555
coverC wasestimatedusingthe Reed[1976]formula
the measured insolation Is•o which is reflected or scat-
teredupwardat the oceansurface.Payne[1972]showed that Ab depends primarily on the solar altitude and atmospheric transmittance Tr, and to a much lesser extent the surfaceroughness.Using softwaredevelopedby
R. Payne(personalcommunication, 1996),wefirst computed the instantaneoussolar altitude and the no-sky
insolation,Iswn, and then computedTr and interpolated hourly values of Ab using Table 1
of Payne [i972]. Typical aibedosover the northern California shelf ranged from 0.08-0.10 in winter to 0.05-0.07
in summer.
Note
that
use of an accurate
Ab reducesthe uncertainty in Qi to essentiallythat in the Is• measurement.
Longwave heat flux was estimated from measured downward longwave and shortwaveradiation and grey body upward radiation during SMILE and from a formula derived from daily-averaged cloud cover during CODE 1 and CODE 2. During SMILE, longwaveheat flux was computed using
C-
1 -(Is•o/I,•o•,)+.0019o•
/0.62,
(5)
whereI,• is the measureddaily meaninsolation, is the daily averageclear sky insolation, and c• is the local noon solar altitude. Both Iswc, and c• were com-
putedusingtheSmithsonian meteorological tables[List, 1984].FollowingReed[1976],cloudcoverslessthan 0.3 wo. ro..qo.t, t,c• 7,orc• in (d'l
In
.•MIT,I•,
tho
ctifForonr, c• in
daily-averaged bulkQbcomputed using(4) and(5) and themeasured Q• givenby (3) is -1.4-+-13.8W m-=. The differencede-correlationtimescaleis about two days, so that the impact of maximum differencesof order -+-40W m -2 diminish on timescalesof a few weeks. Sensible and latent heat fluxes were estimated from
Os -- pcpCh(Ta - Ts)U,
(6)
and
Q1- pL•C•(q,,- 0.98qs,,t)U,
(7)
Qt, -e((I•,• -O.O36I•,,) -o'T: ), (3)
where p is the air density,Cpis the heat capacityof
where I•o is the measured downward longwave radiation, the secondterm is a correction for solar heating of the longwavesensorusing I•,o and a constant coeffi-
air, Cn the sensibleheat flux transfer coefficient, U is the magnitude of the measuredwind velocity minus the
surfacecurrentvelocity(i.e., U = IU•- U,I), L• is the latent heat of evaporation of seawater,C• is the latent
cient of proportionality[Alados-Arboledas et al., 1988; heat flux transfer coefficient, qa is the specific humidDickey et al., 1994], e is the emissivityof the ocean ity (computedusingthe measuredrelativehumidityRH surface(taken as 0.98 [Dickeyet al., 1994]), • is the andTa), and 0.98q,•t is the specifichumidityjust above Stefan-Boltzmann constant, and T• is the surface temthe seasurface(whereRH is assumedto be 98ø7o above perature in degreesKelvin. salt water). Sincethe measuredwind velocitieswere During CODE, incident longwaveradiation was not measured,and we employeda bulk formula to estimate
generally much larger than the near-surface currents, we show wind velocity only to simplify the data presen-
Q•. Fung et al. [1984]comparedeightcommonlyused
tation.
bulk formulas for Qb against the results of a full radiative transfer model, and found that while some formu-
heat flux calculationsare shownby Rosenreid[1983], Limeburner[1985],and Alessiet al. [1991].
las estimatedQ• to within + 15 W m-2 underclear sky conditions, the differenceswere much greater during cloudy conditions. To determine the best formula to use during CODE, the eight formulas were used to estimate a bulk Q• during SMILE which was then com-
paredto (3). These comparisonsshowedthat a modified versionof
the Berliand and Berliand[1952]formulaprovidedthe best fit, with the original quadratic cloud correction
The near-surface
current
time series used in the
The sensible and latent heat flux transfer coefficients
Cn andC• usedin (6) and (7) werecomputed usingthe Tropical Ocean-GlobalAtmosphere/Coupled Ocean-
Atmosphere Response Experiment(TOGA/COARE) bulkformulation[Fairallet al., 1996]withoutcorrection for skin temperatureeffects. This code,basedon the
stability-dependent bulkapproach of Liu et al. [1979] andneutral10-mdragcoefficient of Smith[1988],was
developedfor applicationto the tropicalPacificand incorporatesincreasedheat fluxesdue to gustinessat low bC), whereb was chosento minimizethe least squares wind speeds. differencebetweenthe daily averagedbulk Qb and (3). It is an open questionhow well this (or any other The net longwave heat flux was thus estimated using open ocean) code estimatesturbulent air-sea fluxes
factor(1 -bC 2) replaced by the linearexpression (1-
over the continentalshelfwherelarge spatial and temporal changesin surfaceconditions(suchas aerodynamic roughnessand temperature)can occur. While this point will be discussed more in section4.3, re-
Qt, - (eaT•4 (O.39-O.O5eø'5)+4eaT•a (T,-T,• ))(1-bU), (4)
sultssummarized by Fairall andMarkson[1987]and wheree is the vaporpressure, Ta is the (absolute) air
Cartart [1990]suggestthat our atmospheric measuretemperature,and b - 0.75. The daily-averagedcloud ments were made sufficiently close to the ocean sur-
21,556
BEARDSLEYET AL.: SURFACEHEAT FLUX OVER NORTHERNCALIFORNIA SHELF Table
1. Moored
and Coastal Stations Water
Experiment
Site
Name CODE
38.61 38.52
123.47 123.67
94 402
Us,/•w, T•, T•, Us Us,/•w, T•, Ts, Us
38.20 39.20
123.30 124.00
125 306
P• P•
38.61
123.46
93
38.64 38.51 38.42 38.20 39.20
123.42 123.67 123.27 123.30 124.00
59 400 90 125 306
38.65
123.49
38.54 38.20 39.20
123.38 123.30 124.00
93 125 306
Ts T• T•
SP
38.66
123.40
28*
Ism, I•
BB
38.32
123.07
9*
C2 C5 R3 NDBC13 NDBC14
C3
SMILE
VariablesMeasured
m
C3
2
Depth
øW
NDBC13 NDBC14 CODE
Longitude
øN
C3 C5
1
Latitude
M3 NDBC13 NDBC14
93
Us, Ism, RH, Pa, Ta, Ts, Us Is• P• Ism, P• P•
U•, Is•o, Ii•o, RH, Pa, T•, Ts, Us
Is•
Moored and coastal stations which provided primary and secondarydata to estimate the surface heat flux in CODE
and SMILE
are listed.
The two coastal stations at Stewarts Point and the
BodegaBay Marine Laboratoryare denotedby SP and BB, respectively.The right-handcolumn lists the variablesmeasuredat eachstation: vectorwind (U•), insolation(Ism), incidentlongwave
flux(Itw), air pressure (Pa),relativehumidity(RH), air temperature (T•), watertemperature and oceancurrent *Sensorheights above sea level.
face that bulk formulations should provide good estimates of the turbulent air-sea fluxes, even during stable conditions when the atmospheric boundary layer tends to be thinner. For the northern California shelf, the heat and momentum fluxes computed using the
in this studywereobtained,and AppendixA presents the methods and assumptionsused to constructcomplete hourly time series of the basic variables listed above at C3 for each experiment. The mooredinstru-
mentationusedin CODE and SMILE and theirexperiTOGA/COARE code agree closelywith thoseca.lcu- mentaluncertainties are summarized next,followed by lated usingthe Large and Pond [1981,1982](LP)bulk a descriptionof the C3 time series. Additional informaformulation. In SMILE, aircraft-measured wind stresses tion aboutthe fieldprograms is presented by Rosenfeld agreeon averagewithin 0.014-0.02 N m-2 with buoy [1983], œimebur•er [19851, andAlessiel al. [1991]. wind stressesestimated using both TOGA/COARE andLP codes[Beardsley et al., 1997],andaircraft-based Moored Instrumentation estimates of the 10-m neutral transfer coefficients Cn
and Ce agree within uncertainty with the values used
The WoodsHoleOceanographic Institution(WHOI)
wind recorder(VAWR,)servedas the in both TOGA/COARE and LP [Enriquezand Friehe, vector-averaging sensingand recordingsystemin 1997]. (The MATLAB programsusedhereto compute basicmeteorological 1988;Trasket the surface fluxes may be found on the worldwide web CODE and SMILE [Deanand Beardsley, al., 1989]. Mountedon a 3-m toroidbuoy,the Code1 at http://crusty.er.usgs.gov/sea-mat.) VAWR measuredwind speedand direction,insolation, air temperature,and water temperature.In Code2, air 3. C3 Moored Measurements pressureand RH sensors wereadded(Figure2a). In Calculation of Qn using the above formulas requires SMILE, two integral VAWRs were mounted on a 3.5measurements of wind and ocean surface velocm discusbuoy at C3 to provide wind measurements at ities Ua and Us, the air and ocean surface tempera- several heights and provide redundant measurements tures Ta and Ts, insolation Isw, relative humidity RH, of other variables,including downwardlongwaveramoored
barometricpressurePa, and (for directestimatesof
diationand improvedRH sensors(Figure2b). Near-
downward longwave radiation Itw. Unfortunately, not surfaceoceancurrentsand temperature weremeasured all these variables were successfullymeasured at C3 in at C3 in all experimentswith either ScrippsInstituthe different field programs;Table 1 identifiesthe differ- tion of Oceanography(SIO) or vector-measuring curent moored and coastal sites where measurements
used
rent meters(VMCMs) [ Weller and Davis, 1980;Beard-
BEARDSLEY ET AL.' SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF
21,$$7
a' CODE-2
1
51PE VIEW
FRONT VIEW
b' SMILE _
2
1
51PE VIEW
FRONT VIEW
Figure 2. Schematics ofthemeteorological buoys deployed at C3in (a) CODE2 and(b) SMILE.
The CODE 1 C3 buoywasidenticalin designto CODE 2 but lackedair pressure andRH sensors. All C3 buoyshad steeringvanesto helporientthe meteorological sensors into the wind.
21,558
BEARDSLEY ET AL.' SURFACEHEAT FLUX OVER NORTHERN CALIFORNIA SHELF
Table 2. C3 SensorHeights Above Water During Each Experiment Variable
Wind speed,wind direction Current speed,direction Insolation
Longwave Air pressure Relative humidity Air temperature Water temperature Platform
CODE
1
CODE 2
3.5 -5.0
3.5 -5.0
3.5 -5.5
3.0
3.0
3.5
2.5
3.5 2.7
... 3.0 -1.0
2.7 3.0 -1.0
3.0 10.0 -6.0
toroid
toroid
discus
...... ...
SMILE
The C5 sensorheightsduringCODE-1 wereidenticalto the C3 buoy. The SMILE air temperatureheightis that at NDBC13. The SMILE C3 water temperature at 6 m is adjusted by the
near-surface (1-5 m) temperature difference at M3, 15km south of C3. (SeeAppendixA for details).
sley,1987].Table2 liststhe C3 sensor heightsfor each shelf winds drive coastal upwelling. This reducesT8 during summersothat trendsin T8 and Ta opposeeach Table 3 summarizesby instrumentand variablethe other. In summer,T• is generallylessthan Ta (leading to a stableatmosphericmarinelayer), while in winter, different sensors used in CODE and SMILE and estiwhen the along-shelfwinds are generally weaker and mates of their in situ measurement uncertainty. This more variable, Ta is generally less than the prevailing uncertaintyincludesthe inherentsensorerror, the sysmarinelayer). tem error introducedbetweenthe sensoroutput and the T• (leadingto an unstable
experiment.
recordeddigitalvalue,and any additionalerrorcaused The C3 time series also demonstrate strong variby unsteadiness in the marine layer, mooringmotion, ability on diurnal and synoptic(2-10-day) timescales. poor sensorplacement,imperfectprotectionof differ- Much of this variability is not independent. In all seaent sensorsfrom solar heating, etc. The contribution sons,there is a link betweenwind direction and cloud of these in situ measurement uncertainties plus other cover. Equatorward winds are often associatedwith measurement problems(AppendixA) to the uncertain- clear skies, and weak or poleward winds with cloudy ties in the computedheat flux components is discussed skies. This affects the downward longwave and shortin section 4. wave fluxesin oppositeways. In summer,weak or poleward along-shelfwinds also causerelaxation from upC3 Meteorological Time Series wellingwhichraisesT• [Sendet al., 1987;Lentz,1987a]. The C3 hourlymeteorological time seriesfromSMILE The cross-shelfwind velocity, although much weaker (Figure3), CODE2 (Figure4) andCODE1 (Figure5) than the along-shelfcomponent,can affect RH, espeprovideinsightinto the causesof temporalsurfaceheat cially in winter. The lowestRHs tend to occurfor offflux variability. These recordscoverthe followingpe- shelfwinds (negativeu) while higher RHs prevailfor on-shelf winds.
riods: SMILE (0800 UT, November14, 1988 to 0700 UT, May 14, 1989);CODE l (0800UT, April 13, 1981 to 0700 UT, July 3l, 1981); and CODE 2 (0800 UT, 4. C3 Heat Flux Time Series March 25, 1982to 0700 UT, July 28, 1982). Theseand In this section,we first estimatethe uncertainty in the subsequent time seriesare plotted usingUT, whereUT computed heat flux componentsbased on the in situ - local (PacificStandard)time + 8 hours.An on-shelf and along-shelfcoordinatesystemis usedto display the measurement uncertainties, and then present the C3 vector wind, with the positive on-shelfu componentori- daily-averaged heat flux times series for SMILE and ented toward 47øand the positive along-shelfv compo- CODE. A discussionof uncertainty in latent heat flux nent directed
toward
317 ø.
follows.
Taken together, the SMILE and CODE field proExperimental Uncertainties grams span most of the year, and seasonal trends in Here we combine the in situ measurement uncertainmost of the basic variables are evident. While Isw and Ta reach their maxima in summer months, the seasonal ties (Table 3) with the heat flux formulasand basic cycleof wind forcing[Strubet al., 1987]affectsTs. Fol- meteorologicaland oceanographictime seriespresented lowingthe atmospheric springtransition[Lentz,• 987b], aboveto constructestimatesof the resultinguncertain-
strong(v • - 10 m s- 1) persistent equatorward along- tiesin the differentheat flux components.The approach
BEARDSLEY ET AL.: SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF
21,559
taken is to compare the hourly heat flux component of the difference time series for individual measurement time seriescomputedwith the basicvariables(the base errors are summarized by componentand experiment in case)with that computedusinginput variablesbiased Table 4. Since It•owas not measuredin CODE, the unupward and downwardby the measurementuncertain- certainties in the CODE daily Qb serieswere estimated ties in Table 3. The means and standard deviations by comparing the daily Qb measured in SMILE with
Table 3. Meteorological andOceanographic Instrumentation Usedin CODEandSMILEOrganized by
Instrument
and Variables
Measured
In Situ
Variable
Sensor
Manufacturer
Accuracy Reference Experiment
VA WR
Wind speed/ direction
three-cup anemom-
magnetic compass
R. M. Young(6101)* EG&C (VACM)
three-cup anemom-
R. M. Young(6301)*
vane
Wind speed/ direction
q- 2%•
R. M. Young(6301)*
1
CODE
1
...
1
SMILE
1,2
...
eter
8.5 ø
2%•
eter
magnetic compass
Custom(WHOI) EG&G (VACM)
pyranometer pyranometer pyranometer
EppleyLaboratory(8-48) EppleyLaboratory(PSP) Hy-Cal Engineering(8405)
5%•
3,4
5%• 5%•
3,4 1,4
pyrgeometer
EppleyLaboratory(PIR)
5%
5
SMILE
Air pressure
Digiquartz transducer gill pressure port
Paroscientific (215-AS) Paroscientific (215-AW)
0.6 mbar 0.6 mbar
3 3
CODE 2 SMILE
Relative
cellulose strip, strain gauge, gill shield capacitive-type sensor, gill shield
Hy-Cal Engineering
q- 6%
1
CODE
(HS-3552B) Viiisiilii (Humicap)
q- 5%
3
vane
Insolation
Longwave
8.5ø
CODE/SMILE SMILE CODE 2
radiation
humidity
Air temper-
thermistor
gill shield
YSI (Yellow Springs Instruments)
thermistor
Thermometrics
variable capacit-
< + 0.4 øC
SMILE
1,6
all
q- 0.1 ø
1
all
NDBC general service buoy payload
q- 1 mbar
7
CODE
thermistor
(seeair pressure)
q- 1øC
7
orthogonal propellors flux gate compass
SIO VMCM
-5%
8
Direction
SIO VMCM
q- 5 ø
Water temper-
thermistor
SIO VMCM
q- 0.10øC
-5
ature
Water temper-
_
2
(wind > 3 m/s) 1-3øC (lesswind plus strongsun)
ature
NDBC
13
Air pressure
ance transducer
Air temper-
1
SMILE
ature
$I0
VM CM
Current speed
CODE
1
ature
WHOI
VMCM
Current speed
orthogonal propellors
Direction
flux gate compass
EG&G (VMCM) EG&G (VMCM)
Water temper-
thermistor
YSt
ature
8
CODE 2/SMILE
9 2
CODE 2 SMILE
5ø 0.2 øC 0.1øC
21,560 Table
BEARDSLEY ET AL.: SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF 3.
(continued) In Situ
Variable
WHOI
Sensor
Manufacturer
System Accuracy
Reference
Experiment
0-20%overspeed-
8
CODE I (C5)
VA CM
Current speed
rotor
EG&G (VACM)
ing due to surface wave conditions
Direction
vane/magnetic EG&G (VACM)
q- 3øC
10
q- 0.1øC
9
compass
Water temper-
thermistor
ature
Thermometrics or YSI
BBML
Insolation
pyranometer
Li-Cor (LI-200SB)
q- 5
11
SMILE
Insolation
pyranometer
EppleyLaboratory(PSP)
q- 4%
5
SMILE
Longwave
pyrgeometer
EppleyLaboratory(PIR)
q- 5%
5
SMILE
Stewarts
Point
For each variable, the estimated in situ measurementuncertainty with referencesfor the system used in each ex-
perimentis listed. PSP denotesprecisionspectralpyranometer(Eppley);PIR denotesprecisioninfraredradiometer (Eppley). References are as follows:1, DeanandBeardsley [1988];2, Trasket al. [1989];3, Welleret al. [1990];4, MacWhorterand Weller[1991];5, R. E. Payne(personal communication, 1996);6, Payne[1987];7, Hamilton[1980]; 8, Beardsley [1987];9, Irish [1985];10, Bryden[1976];11, LiCor,Inc. technicalinformation.
* Modifiedat WoodsHole Oceanographic Institution(WHOI). The three-cupanemometors usedin CODE (model 06020Acupwheel)and SMILE (model12170Ccupwheel)weremodifiedat WHOI to interfacewith the VAWR counting circuitry.
t Doesnotincludeestimated effectof cupoverspeeding.
• Includes estimated effectof sensor tilt (theinsolation sensor wasungimballed). that computedusing(4) and (5) with differentinputs (e.g.,by replacingthe RH measuredin SMILE with its (constant)medianvalueto mimicCODE 1). In gen-
eral,the largestuncertainties arelessthan+15 W m2 and arise from the measurementuncertainty in Isw and RH.
Based on Table 4, estimates of the maximum uncertainties for each componenthave been made using the combination of modified input variables which lead to the largest differencefrom the base case. These estimates are listed in Table 5 together with estimates of the maximum uncertainty in Qn. In all experiments, the uncertainty in Qs was smaller than for other components, and in CODE 2 and SMILE when RH was measured,the uncertainty in Qt was largest. The decreasein uncertainty in Qi in SMILE in comparisonto CODE reflectsprimarily the reducedinsolationduring winter.
C3 Daily-Averaged
Heat Fluxes
C3 daily-averaged surface heat flux time series are
presentednext for SMILE (Figure 6), CODE 2 (Figure 7), and CODE 1 (Figure8), andtheir statisticsare summarized in Table 6. Daily-averaging eliminates di-
urnal variation in Qi, and the resulting time seriesbetter illustrate the importance of different componentson timescales of days to months. Between mid-November and the end of February, the averagenet heating is indistinguishable from zero. Net heat fluxes from March to May and in the summer months are large and positive. In each experiment, Qi is the largest single mean component, followed by Qb. Qt and Q, are weaker and about the same magnitude. This is generally true for
their fluctuations(as indicatedby the similar magnitude of their standarddeviations)as well as means. Both the mean Qt and Q• vary in sign. In winter and in spring prior to the spring transition, they almost always represent a transfer of heat from the ocean to the atmosphere. In summer, Q, is usually from atmosphere to ocean due to coastal upwelling, and the Qt varies in sign depending on RH.
The linkage between different heat flux components is demonstratedin Figure 9. The connectionis clearest
betweenthe net downward shortwave and net upward longwavefluxes, which are both greateston cloudless days.Cloudless daysareoftenassociated with equator-
ward winds,especiallyin the springand summerupwellingseason(as represented by the entire CODE 2
BEARDSLEY
ET AL'
SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA
(a)
SHELF
21,:561
U WindVelocity 10 .... ' ..... '.... ! ...' ..... ! ..... '..... ! ..... '..... ! ..... '..... ! ..... '..... I ..... :.........
•
! .........
'
o -10
(b)
V WindVelocity 10
E -10
(c)
Air Pressure
•1
e
""' ''
V ' ' ' ' ¾ ¾• ' W' •1.%/' '•'¾
(d)
DownwardShodwave
iillllllilllllllli I . ,
o .
(e)
.' ,.,
o
I
NDBC13 AirTemperature
, i. , , , , _.i , .... , , , ,
(f)
WaterTemperature i
i
i
; ..........
i
.:..
i
i
......
: ......
400--
,
,
!
,
!
,
,
200 I
I
I
I
I
I
I
I
15
........
.......
i
i
i
.:.........
i
i
: ........
i
i
i
:
..-: ....
,
g
!
i
I
i
I
-i
,,....
i
i
l
!
!
,
I
I
:.....
: ....
øø lO (g)
(h)
DownwardLongwave , g -,
I
I
I
II
RelativeHumidity 100
.
23
Nov 1988
3
Dec
13
23
2
.... I
I
I
I
I
I
I
I
I
I
I
I
I
12
22
I
11
21
3
13
23
2
12
22
2
12
Jan
Feb
Mar
Apr
May
1989
Figure 3. Hourly-averaged meteorological time seriesfromSMILE: (a) on-shelfwind; (b) alongshelfwind(with positivev poleward);(c) air pressure; (d) insolation; (e) air temperature; (f) water temperature;(g) downwardlongwaveradiation;and (h) relative humidity. SMILE spans winter and spring and includes the spring transition which starts May 1, 1989. To facilitate comparisonbetween Figures 3, 4, and 5, these time seriesare plotted with common horizontal and vertical axes lengths.
period), but also to someextent in winter (SMILE, NovemberthroughFebruary). BetweenNovemberand February, Qi and Qb meansand variability are similar in magnitude and tend to cancel, so that variation in Qi and Qb has relatively little effect on net heat flux variability during this time. Beginning in spring, Qi means and daily-averagedfluctuations become much larger, since the effect of clouds on reducing Isw is proportional to Isw, which increases significantly from winter to summer. This allows Qi
its variation. In summer, variability in the combined Qt and Q, remains similar in magnitudeto that in winter and continuesto be important to net heat flux variability. Uncertainties
in Positive
Latent
Heat
Flux
The C3 heat flux seriespresentedaboveshowlarge latent heat fluxes into the ocean of order 50 W m -2
or moreduringthe 1989springtransition(Figure6) and summer1982upwellingseason(Figure7). Asso-
to drive both the mean net heat flux and much of its
ciated with advection of very moist, warm marine air
variability in summer. In winter, the relative balance between Qi and Q• allows Qt and Q•, which are generally negative and correlatedto each other during this period, to becomeimportant to the mean Qn and
overcoolupwelledwateron the shelf,the resultingpositive air-seaspecifichumiditydifference leads(wethink) to direct condensationon the sea surface,thus causing positiveQt. The magnitudesof theselargepositive
21,562
BEARDSLEYET AL.: SURFACEHEAT FLUX OVER NORTHERNCALIFORNIASHELF U Wind Velocity
(a) 10
E -10
V Wind Velocity
(b)
........
10
,..... ! ..... ,..... .!..... ,....
• ..... ,..... ! ..... ,....
! ..... ,.................................
E -10
Air Pressure
I ' ' !' !' ' ' !' ' '
.• 1020 1000
Downward
(d)
ß
Shortwave
1000
E 500 0
Air Temperature
(e)
5
ß ß
ß
Water Temperature
iIIIIIIIIIIII :I ' ! ' ! ' ' '.-' ' '
ø 10 (g)
Cloud Factor
.o
0.5 0
(h)
............................
RelativeHumidity
lOO•, , , W•,¾ n,! •-
•. 5o
..................
I
Mar Apr
I
I
I
May
I
I
Jun
I
i
Jul
1982
Figure 4. Hourly-averaged meteorological time seriesfromCODE 2. Panelscorrespond to those in Figure3, exceptfor Figure4g whichshowsthe daily-averaged cloudfactor. CODE 2 includes the 1982 springtransitionwhich beganin April 14.
latent fluxes must be interpreted with caution for several reasons. Positive •t values depend strongly on the
oceanconditionswith negligiblespatial gradientsin air
accuracies of both RH measurements near 100% and
and ocean surface fields. As air flows over the northern
Theseformulaewere developedfor quasi-steady open
the surface skin temperature. The RH sensorsused Californiashelf,it mustadjustto the coastalorography in CODE and SMILE were routinely calibrated up to and changesin surfaceconditionssuchas aerodynamic valuesof about 94%, and valuesabove 94% may have roughness and temperature.During the upwellingsea-
uncertaintieslargerthan thosereportedin Table3 (Ap- son, the flow of warm marine air over cool upwelled pendix A). If the RH measurements are taken as cor- water leads to a stable marine layer and development rect, the magnitude of the positive latent flux is not of internal boundarylayersassociatedwith strongTs particular to the TOGA/COARE formulation(since changes.Scalingargumentssummarizedby Fairall and both TOGA/COARE and LP codesyield similarmag- Markson[1987]and (7arratt[1990]suggest that ouratnitudes). The differencebetweenthe true skin temper- mosphericmeasurements were made sufficientlyclose ature and our bulk measurement Ts at an effective to the oceansurface(_< 3.5 m) to be within a quasidepthat • 0.5 m (AppendixA) is difficultto estimate, equilibriumlayer where bulk formulationsshouldprobut we have assumedthat the differenceis insignificant vide good estimatesof the turbulent air-sea fluxesdursincethe wind was generally strong when insolationwas ing both unstable and stable conditions. While the stable caseof warm air flowing offshoreover large(i.e., duringthe upwellingseason).
Of more concernis the applicabilityof the bulk esti- coolwater hasbeenexaminedexperimentally[Cartart matesto the set of conditionsfoundin this shelfregion. andRyan, 1989]and numerically[Garratt,1987],much
BEARDSLEY ET AL.' SURFACEHEAT FLUX OVER NORTHERN CALIFORNIA SHELF (a)
21,563
U WindVelocity
-10 i
i
(b)
i
V WindVelocity
-1
.
NDBC 13 Air Pressure .......
,., 1020
:m,'• •.'. A ,' •
. •,/•_ ._' J•.• .'.:., •, .•,.,,.,• ,•.., ß
ß
ß
.
.
,
Downward Shortwave
(d) 1000
E 500 0
(e)
Air Temperature
• .
i
i
•
'
I
'•
i
'
',
'
',
'•
....
........ i ..... i..... ;..... ,..... ;..... i..... ;..... ,..... ;..... ,..... ; ..... i...........................
ß
WaterTemperature
' .......
lO
i.................... ""1
01oud Factor
(g)
•o.•i '
OI- ......................... 12
Apr
22
2
May
12
22
I
11
Jun
'
21
I
11
..., ..... 21
31
Jul
1981
Figure 5. Hourly-averaged meteorological time seriesfrom CODE 1. CODE 1 beginsafter the 1981 spring transition.
less is known about the caseof warm, moist airflow overcoldwaterwell awayfrom land. We canfind only
onepublished report[Anderson andSmith,1981]of direct eddy flux measurements of downwardvapor flux over water corresponding to (•t • 20 W m-2 made duringstableatmospheric boundarylayerconditions on the outerScotianshelf.For comparison, for highdewfall over land, (•l can reach 50 W m-2 for 10-m wind
Winter
Events
'Windsover the northern California shelfduring winter are characterizedby strong poleward and equator-
ward fluctuationswith timescales of severaldays(Figure 3) [Deverand Lentz,1994;Dormanet al., 1995]. These fluctuationstend to be strongerin the along-shelf
directiondue to the coastalmountainrange(Figure1) and are causedby the passageof both low (cyclonic) speeds lessthan6 rns-1 [Garrattand$egal,1988].For pressure systems overthe coast lack of additionalinformation,we will presentQl as andhigh(anti-cyclonic) from the northwest. These wind events are typically computedusing(7) but note that the magnitudesof separatedby periods of weaker and more variable winds. the large positivelatent fluxesmay be overestimated during extreme condensationconditions. Figure 10 illustratesa periodof initially weakand variablewindsfollowedby polewardand equatorward
5. Surface Heat Flux Events in Different Seasons
winds. The synopticweatherpatternfor this period is shownin Figure11. By February4, a large-scale lowhaddeveloped overtheU.S.southwest whilea high formedoverthe U.S. centralnorthwest.This cyclone-
Atmosphericforcingand the surfaceheat flux also anticyclonepair carriedvery cold, dry continentalair
exhibitsignificant variabilityontimescales ofhours,es- westward and southward over northern California. This peciallyduringsynoptic weatherevents.Togaininsight synopticpattern persistedfor severaldaysuntil a new into this variability,we presentnext hourlytime series lowcrossed thesouthern California coastonFebruary 9, duringseveralcharacteristic eventsin winter,spring, bringingwarm, moist marine air over the northern Caland summer.
ifornia
shelf.
21,564
BEARDSLEY ET AL.' SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF Table 4. Sensitivity of Heat Flux Componentsto Individual MeasurementUncertainty in the Different Field Experiments CODE 1 Sensitivity to Individual Measurement Errors Ta q- 0.4 ø
Qi
T8 q- 0.1ø
RH q- 5%
mean
s.d.
Ql
mean
s.d.
Q8
mean
s.d.
I•
q- E
U - 6%
+13
mean
s.d.
Qbest
Is• q- 5% +17
4-2 +1
0 0
......... .........
4.6 4.3
+2 4.1
...... ......
0 4.1
4.5 4.2
4.1 4.1
...... .........
-1 :t:1
+7 +8
CODE 2 Sensitivity to Individual Measurement Errors
Ta q- 0.4ø Qi Qbest
Q8
RH q- 5%
I•0 q- 5%
h•, q- E
mean
.........
4.12
......
s.d.
.........
4.16
......
+ 1
4.6
......
4.1
4.8
......
mean
s.d.
Ql
T• q- 0.1ø
4.2
4.1
U - 6%
0
0
me::u
4.7
:t:3
4.13
......
2
s.d.
4.5
4.4
+ 7
......
4.3
Mean
4.4
4.1
0
......
0
Std
4.2
4.1
0
......
4.1
SMILE Sensitivity to Individual Measurement Errors Ta q- 0.4 ø
Qi
.........
mean
ß.. ß..
4.1 0
... ...
4.2 0
4.1 0
4.1 4.1
4.5 4.3
4.2 -t-2
4.9 4.5
4.3 4.2
4.1 0
mean mean s.d.
Qs
I•0 q- 5%
h•. q- E
U - 6%
mean
s.d.
Ql
RH q- 5%
s.d. s.d.
Qbest
T• q- 0.1 ø
mean s.d.
4.12
0 0
0 0 4.2 4.4
...... 4.15 4.2
... ...
...... ......
...... ......
2 4.2
...... ......
0 4.1
The mean and standard deviation are between hourly heat flux componentscalculated with the C3 basic data and those biased by the uncertainty in each variable. Units are
W m-2 and are roundedto nearestinteger.The uncertainty in downward longwave measurements E are q- 10 W m-2 in dayand q- 5 W m-2 at night.
Despite initially weak winds (often lessthan 5 m Weakwindscontinue onFebruary8 accompanied by s-x) andclearskies,the daily-averaged Qn losses on increasingRH, Ta, and cloudinessas the offshorelow
February6 and7 areamongthe largestobserved during SMILE. This occursbecausethe Qb lossnearlycancels out the Qi gain and the very cold, dry continentalair flowingoverthe shelfboostsboth Q• and Qs losses.On these days, diurnal cyclingis most evident in Ta and
RH (duein part to nighttimecooling overland)rather
approaches fromthesouthwest. In response to this,the magnitudesof all heat flux componentsdecrease.On
February9, the windbecomes poleward andstronger, causingonlya smallincrease in Q• andQ• a.sthe air-sea temperaturedifference decreases andspecifichumidity rises. Diurnal variabilityin Qt and Q, fluxesis not evidenton thesedays. As polewardwindsweakenon February10 and 11, Qn becomes positive. The weak Q• fluxesbetweenFebruary9 and 11 are closeto the
than in the wind. It affectshourly Qs and Qt fluxes with maximumlosses occurringin (local)earlymorning hours.Althoughthe daily-averaged fluxesarelarge in magnitudein comparison to otherSMILE days,they average for SMILE between November and the end of remain dwarfedby the diurnal variationin Qi. February.
BEARDSLEY
ET AL.: SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA
Table 5. Sensitivity of Heat Flux Components to the Combined
Maximum
Measurement
Un-
certaintiesin the Different Field Experiments CODE
Qi Qb
Qt Q8 Q,•
Listed
1
CODE
2
SHELF
21,565
ing this 2-day period,moderatelylow T• (about 8øC) and RH (70-76%) combineto producelargeQt and Q8
losses (about140W m-2) whichapproximately equal the net surfaceheat losson these two days. In winter, the correlation of Q, and Qt together with the compensationbetween Qi and Q• causesQn to be
SMILE
mean
+ 13
+ 12
+ 7
s.d.
+ 17
+ 16
+ 12
controlledlargelyby the air-sea(Qt plusQ,) heatflux.
mean
+ 9
+ 10
+ 16
s.d.
-1-8
•: 8
-1-2
mean
+ 8*
+ 20
+ 15
During this time, low T• is associatedwith low RH resulting from the offshoreadvection of cool, dry continental air by high- and low-pressuresystemsas they
s.d.
+4
+11
+
mean
+ 6
+ 5
+ 4
s.d.
-1-3
-1-3
-1-3
mean
:t:35
+ 47
•: 42
s.d.
+ 20
+ 22
+ 15
are the means and standard
move eastward
9
deviations
of
across the coast.
Although atmosphericforcing continuesto be driven by storms until the spring transition, the relative balancebetweendifferent heat flux componentschangesin March when strengthening insolation causesa persistently positive daily-averagedQ• and becomesa dominant part of its variability.
the difference time series computed using C3 basic hourly data and the C3 data biased by the combination of measurement uncertainties which gives
the maximum positive negative) heat flux biases. The uncertainties in Qn were computed assuming the uncertainties in the four heat flux components
areindependent. UnitsareW m-2. *The uncertaintiesgivenfor Qt in CODE-1 (where RH was not measured)reflect the uncertaintyin other input variables. Actual Q• uncertainty in CODE I is unknown, but probably comparable to or larger than that observedin CODE 2.
The atmospheric spring transition marks the end of
the winterstormseason andonsetof the spring/summer coastal upwelling season. During this transition, the Aleutian low weakensand splits into weaker lows over Asia and the Aleutian Islands, the North Pacific subtropical high strengthensquickly and spreadsnorthward and eastwardtoward California, and a persistent
low pressure cell develops over•he southernUnited
States and Mexico [Lentz, 1987b; Strub and James, 1988]. Over the U.S. west coastshelfnorth of Point Conception,strong and persistentequatorwardwinds On February12, the wind becomesequatorwardand developin severaldayswhichinitiatecoastalupwelling is accompaniedby clear skiesand a slightlylower RH. [Strubet al., 1987].This basiclarge-scale patternthen Again, changesin Qi and Qb net compensate.The Qt persists well into summer. loss increases to near 100 W m -2 due to the reduced The timing of the springtransitionvariesfrom year RH, whilethe Q8 lossremainsfairly low (around20 W to year as does its effect on the surface heat flux. Here m-2). The daily-averaged Qn is negative andfairly we examinethe surfaceheat flux during the May 1989 weakduringthistime(between 20and40W m-•). Di- spring transition and contrastit with that during the urnal variabilityin Qt is againpresent,but duringthis April 1982springtransition. Both springtransitionsare period it is forced by diurnal wind variability rather markedby a periodof strongequatorwardwindslasting than Ta or RH (whichremainfairly constant).Maxi- approximately5 daysand subsequent coastalupwelling mum Qt losseson February12-13 (and to a lesserex- which causesa decreasein T• of 3ø-4øC. Despitethese tent on February14) occurduringlocal afternoonand similarities,other meteorologicalconditionslead to net eveninghours when wind speedpeaks. heat fluxes which are quite different. Figure 10 demonstrates several characteristicsof the The 1989 springtransition beginson May 1, slightly surface heat flux variability in winter. As mentioned later than the climatological average[Dormanand Winabove,daily-averagednet Qi and Qb fluxestend to can- ant, 1995]andfollowsthe presence of a warmlensof off-
cel eachother (Figure 9). This allowsvariabilityin other meteorologicalvariablesto have an appreciable affecton the net flux, sothat heatfluxesof similarmagnitudecanoccurfor quitedifferentconditions(seeFigures3, 6, and 10). For example,the largestheat lossin Figure 10 occurson February 6 during clear skieswith
shorewateron the shelf[Largieret al., 1993].Although equatorwardwindsare generallyaccompanied by clear skies,cloudsare often presentduring this springtransition (Figure 12). Despitethis, insolationdominates Q•, and its fluctuation drives much of the net flux vari-
ability.Net Q• loss,initiallynear100W m-2, declines weakwind(about3 m s- •) accompanied byverylowTa to lessthan 50 W m-•. This is dueprimarilyto cloud (aboutaøC) and RH (about35%). Similarheat losses cover and high RH and only secondarilyto the decline occur on December25 during clear skieswith strong in T• with the onset of upwelling. Upwelling does afwinds(about8 m s-•), and on December 24 during fect Q8 and Qt in an important way. Initially, sea and cloudyskieswith weakerwinds(about4 m s-•). Dur- air temperaturesare similar so that Q• is weak despite
21,566
BEARDSLEY ET AL.' SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF Wind u (-) and v (-) components (a)
15 10 5
•
0 -5
-10
-15
Net Heat Flux
(b) 400
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
200
_
E
0
-200
Shortwave(--) and Longwave(--) Heat Fluxes (c) 400
200 E
0
............
-200
Latent(--) and Sensible(--) Heat Fluxes (d) 100 50
•, E
•
0
i
i
i
-
-100
Nov 1988
i
i
i
i
i
i
i
i
i
i
i
i
i
........
-5o
-150
i
I
Dec
I
I
Jan 1989
_
I
I
I
I
Feb
Mar
Apr
May
Figure 6. Daily-averaged wind velocityand heat flux time seriesfrom SMILE: (a) on-shelf(u) and along-shelf(v) wind components; (b) net heat flux; (c) shortwaveand longwavefluxes;and (d) latent and sensibleheat fluxes. To aid comparison betweenFigures6, 7, and 8, thesetime seriesare plotted with commonhorizontal and vertical axeslengths and line types.
of C3 high wind speeds.Q• lossis significantdue to the spe- m-2. The useof NDBC13 T• in the absence cific humidity differenceand high wind speed. Later, as measurementsalso complicatesmatters. A comparison Ts dropsbelow Ta and the wind remainsstrong, Q s be- of the NDBC13 Ta with that at NDBC14 during this pecomes positiveandstronger (over50W m-•' at times). riod showsdifferencesranging from -1 ø to •-2øC, with The decline in T• has a more striking effect on Qt as sign changeson timescalesof 0.5 day or less. Overall, estimatedby (7). By May 4 throughMay 9, T• is 3- the NDBC14 Ta is slightly lower. Use of the NDBC14 4øC lessthan Ta. This is accompaniedby RHs of near rather than NDBC13 Ta doesreducelatent heat fluxes, 100% and leads to a large positiveQt associatedwith especiallyon May 6, but they remain positiveand range condensationon the sea surface. Hourly Qt fluxes ap- up to 70 W m-2. Sensorheightdifferences between proach100W m-•' on May 5 andMay 6. The addition NDBC13 and C3 are expected to causelessof a problem of Qi to these large positive Qt and Q• fluxes leads to as the TO(]A/COARE codeallowsfor differentsensor the largest heat flux gains observedduring SMILE. Qt heights(Table 2). and Q• fluxes alone nearly cancel out the Qb loss, so The 1982 spring transition beginson April 14 after a that during the night, Qn remains near zero or is even short period of weak winds and clouds,similar air and slightly positive. water temperatures, and high RH as marine air flows As discussedin section4.3, the large positive Qt es- over the shelf (Figure 13). As the equatorwardwind timates between May 4 and May 9 must be interpreted increases,the sky clears and the Ta and RH drop sigwith caution. Heat flux estimates under these condinificantly as cooler, drier continental air flows offshore tions are sensitive to errors in T, and RH. If RH mea- into the marine layer. On April 15, daily-averaged surementswerebiasedhigh by 5% duringthis time, the is nearzero(about9 W m-2), dueto the largeQ• and actual positive Qt fluxes would be reduced by 15 W Q s losseswhich offsetthe increasednet radiation gain.
BEARDSLEY
ET AL.' SURFACE HEAT FLUX OVER NORTHERN
CALIFORNIA
SHELF
21,567
Wind u (-) and v (-) components (a)
15
............
lO
5 o
-5 -lO
-15
Net Heat Flux
(b)
400
..........
o
-200
i
i
i
i
i
i
i
i
I
i
Shortwave(-) and Longwave(-) Heat Fluxes (c)
400
............
200
0
-200
'
'
'
'
'
'
'
'
'
'
'
'
Latent (-) and Sensible(-) Heat Fluxes (d)
100 50
0
-5o -100 -150
3
13
23
Mar Apr
3
13
May
23
2
Jun
12
22
2
12
22
Jul
1982
Figure 7. Daily-averagedwind velocityand heat flux time seriesfrom CODE 2.
As the spring transition progresses,Qn becomesposi- able) winds which last up to 3 weeks,separatedby tive and increasesover the next 5 days as RH increases shorterperiodsof weak or polewardwinds [Beardsley and Ts drops to match Ta, both causingthe Ql and Qs et al., 1987]. The marine layer in summeris capped loss to decrease. by a strong temperature inversiondue to subsidencein The biggest difference between the 1982 and 1989 the North Pacific high. Over the shelf, this inversion spring transitions occursin Qt. It is persistently nega- is typically located at a 30 to 200-m height, well below tive and much larger in 1982 while positive and weaker the top of the coastalmountain range. This situation, in 1989. This is due to the lower RHs, higher wind plus the larger-scalesynoptic pattern, coastlinecurvaspeeds, and negative air-sea temperature differences ture, and the coastalorography,leadsto a low-leveljet in 1982. The larger-scale synoptic conditions caused and supercriticalflow in the marine layer overthe shelf, the 1982 spring transition to advect initially cool, dry causing windsto typicallyexceed 10m s-1 duringequacontinental air along the shelf, while relatively warm, torwardwind periods[ZembaandFriehe,1987; Winant moist marine air was carried over the shelf during the et al., 1988; $amelson,1992]. Theseextendedperiods 1989 spring transition. Local wind directionsduring of equatorward winds are generally cloud-free. the 1982 spring transition were nearly parallel to the
coast(within 5ø) with a slightoffshorecomponent on most daysbetweenApril 14 and 19, while the local wind direction during May 1 - 6, 1989 had an onshorecom-
ponent(9ø-16ø onshore of alongshore) (Figure12). Summer Upwelling and Relaxation Summer
winds over the northern
Events
California
shelf ex-
Shorterperiodsof weakor poleward windsaretypi-
callyassociated with cloudyskiesand an increaseof T• on the shelf due to a northward flow of warm surface
water from southof Point Reyes,and are termed "relax-
ation" events[Lentz,1987a;Sendet al., 1987;Rudnick and Davis, 1988]. Severalmechanisms lead to •hese windrelaxationsandreversals, mostnotablythe northward propagationof Kelvinwavesand gravitycurrents
hibit periodsof strongequatorward(upwelling-favor- in the marine layer along the central California coast
21,568
BEARDSLEY ET AL.: SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA
SHELF
Windu (-) and v (-) components (a)
15
i
i
i
i
i
i
i
i
10
5 0 E
-5 -10
-15
Net Heat Flux
(b)
400
i
20O E
0
-200
(c)
Heat Fluxes , , ,Shortwave , , (-)and, Longwave , , (,_), ,
400
i
i
i
i
i
i
i
i
i
i
I
i
200 E
0
-200
Latent(-) and Sensible(-) Heat Fluxes i
i
i
i
i
i
i
i
i
(d) 100 t 50
•,E
0
-lOO
-15o
f
1•2212• 1'22'2 '1 1'12'1 I1 1'12'13•1
Apr
May
Jun
Jul
1981
Figure 8. Daily-averagedwind velocityand heat flux time seriesfrom CODE 1.
[Beardsley et al., 1987;Dotman,1985,1987].Bothpro- in the midafternoon.Daily-averaged Q,• duringthis periodis strongand positive(between280and380W
cesses cause the inversion to lift over the shelf from
south to north, resulting in the formation of a cloud band which progressesnorthward.
m-2). Asinthe1989spring transition, thehighair-sea
ure 14). Low-levelaircraft flightsmadeon July 14 and
noon).
temperature difference and RH causebothQt andQ• We examineherethe surfaceheatfluxbetweenJuly 12 gainsasestimated by (6) and(7). Thesetermsroughly and 21, 1982. This includesan initial periodof super- balancethe net Q• lossso that Q,• flux is near zero critical equatorwardflow over the shelfand a portion andsometimes positive at nightandquitelargeduring of the relaxationeventwhichimmediately follows(Fig- the day (hourly valuesapproach1000W m-2 at local 18 contrast the coherent spatial structure of the surThe largeshortwaveand longwavefluxesfounddurface wind field during upwelling-favorable wind condi- ing this upwellingperiodare fairly typicalof summer tionswith the lack of structureduringrelaxationevents (which isdominated byactiveupwelling), asispositive
Winant et al., 1988,seeFigures5 and 9]. The relaxation event which starts on July 13-14 in the Southern
downwardQ•. The positivedownwardWt shownin Figure14 betweenJuly 12 and July 14 is lesscommon.
CaliforniaBightreachesthe northernCaliforniashelfby It ispartof a periodbeginning July10whichrepresents July 18, andis described by Dotman[1987]asa gravity the lengthiest andstrongest occurrence of positiveQt current in the marine layer. in summer1982(Figure7). Earlierupwelling events Strongupwelling-favorable windsnear 10 m s-1 occur are accompaniedby generally lower RHs and smaller from July 12 to July 16. Thesewindsare accompanied air-seatemperaturedifferences, leadingto a negative by generallyclearskiesand high specifichumiditydif- or near-zerorather than positiveQt. ference as warm moist marine air flows over cold shelf The relaxationfrom upwelling(as markedby the water. T• is about 1ø-3øCaboveT• and typicallypeaks rapiddecrease in equatorward winds)begins onJuly18
BEARDSLEY ET AL.' SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF
Table 6. Statistics of the Daily-Averaged Heat Flux Time Series for Each Experiment
CODE 1, Summer Mean
241
258
-46
Maximum Minimum s.d.
411 61 80
336 76 68
-17 -79 20
5
23
62 -22 15
78 -5 18
due to high RHs and positive air-sea temperature differences.
During relaxation events, Qn decreasesdue primarily to low-level cloudsand the resultingdecreasein Qi. This is to some extent mitigated by an associatedincreasein downward longwaveradiation. This increase in downwardlongwaveradiation is greater than the increasedgrey body radiative lossassociatedwith higher
T, (Figure 14). In response to low wind speedsand reduced air-sea temperature differences,Qt and Q• also becomequite small during relaxation events. Low wind
CODE 2, Summer
I
Mean Maximum Minimum s.d.
180 382 -60 88
237 337 26 77
-49 - 17 -101 24
-15 62 -151 31
7 55 -33 17
SMILE, Winter and Spring Mean Maximum Minimum s.d.
48 343 - 132 96
135 319 10 77
-59 9 - 136 30
-27 75 - 111 31
-1 45 -43 13
SMILE, November Through February Mean Maximum Minimum s.d.
-9 102 -132 49
98 195 10 37
-64 9 -136 29
-38 23 -111 28
-5 16 -43 10
SMILE, March Through May Mean Maximum Minimum s.d.
130 343 - 55 87
189 319 28 87
-52 4 - 109 30
21,:569
-13 75 -84 29
6 45 - 17 14
Units are W m -9'.
1 .....
J
J-l__
_.E'_
:
•1
1 .....
,.-1•4-.1,.•
4- .....
zero[Lentz,1992],effectivelytrappingthe surfaceheat flux near the surface. This causes a strong diurnal cycling in Ts relative to upwelling events, despite decreased Q n.
6. Spatial Variability Components
in Heat Flux
We investigate next to what extent the C3 heat flux time seriesdiscussedabove are representativeof the entire shelf near C3, sinceshort spatial differencesin wind velocity,cloud coverand seasurfacetemperaturein particular can affect the surfaceheat flux pattern over the shelf. Our approach is to compare daily-averagedheat flux variables measured simultaneouslyat two or more locations during the different field experiments. For the radiative fluxes Qi and Qb, we considerIs• and Iz•. For Qt and Qs, we examine the effects of variability in U, T, and T,. Unfortunately, reliable measurementsof RH exist only at C3, so that we can only speculateas to its effect on Qt. Insolation
was measured
at two or more
locations
duringCODE and SMILE (Table 1). In Table 7a, we and continuesbeyond the end of Figure 14 through July 23 (Figure4). Duringthis period,Ts warmsabout 3øC to roughly match T, and the daily mean RH remains high (> 80%), which in combinationwith the drop in wind speed causesQ1 and Qs to decreasein
magnitudeto nearzerovalues(lessthan 5 W m-2). The increased cloudinessduring this relaxation event reducesboth the shortwavegain and longwaveloss,so
that Qn is reducedto roughly150 W m-2 from the July 12-16 dailymeanof 300W m-2.
considerI,w measured at C3 and C5 during CODE 1, at C3 and R3 during CODE 2, and at C3, Stewarts
Point (SP), and the BodegaBay Marine Laboratory (BB) during SMILE (Figure 1). Both the SP and BB recordscontainednumerousgapsof varying length. Gaps of 2 hours or lesswere interpolated using a spline fit; days with longer gaps were not consideredin the comparison. CODE 1 and CODE 2 comparisonsencompassspring and summer months when I,w is highest. SMILE I,• comparisonsare divided into winter and spring periods based on common coverageat the
In summer,heat flux variability is determinedby the presenceor absenceof upwelling-favorablewinds. Clear skies accompany periods of strong equatorward winds and lead to maximum net shortwave gain. In summer, Qi gain associatedwith clear skiesis greater than
SP and BB locations.
Correlation
coefficients
(Figure 7). However,in July, it can representa gain
extends 50 km offshore of the northern
between
all record pairs are highly significant,and linear regressions show a one-to-one relationship between locations to within 95% confidencelimits. While it is possible coastally trapped cloud cover could lead to differences the longwaveloss(Figure 14) alsoassociated with clear between the CODE 1 C3 and C5 average insolation, skies. Q, fluxes are positive during equatorward winds their comparison showslittle evidenceof this, suggestas the air temperature is greater than the cold, upwelled ing that coastal cloud cover usually extendsoffshoreof water along the coast. The sign and magnitude of Qt C5 (28 km from the coast)when present. This is in vary during equatorward winds. Over the summer of agreementwith Dotman [1985],whopresented satellite 1982, it is generally a loss term during May and June images suggestingcoastally-trapped cloud cover often California
coast.
21,570
BEARDSLEY ET AL.' SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF (a)
Winter (x)DailyValues :,ß .•,
1
:
x)/(: x
:
x
.......:_ :...
0.8
(b)
:
•: •xx ,•
.....
xx
-- - ß:•.x-.:..X •..•'X
..i...x. '
•o.e
..... i" ' i..... ?'X'X}< ''i•( ........ i...... ! ..... .•!. •.ix ....
0.4
0.2 0 -15
-10
-5
0
5
::: ....
00'4 0.2 0 -15
10
AIRn..g.-Sh.e,lf•Wi.n,d ' (,rn s-1) w•mer(x)uauyvmues
(c)
Summer(0) DailyValues
:
-10'
-5
0
5
10
ong-Shelf Wind(m s-1)
(d)
ummer(o) DailyValues
0 x
-so-.••
-50
....:-.......... '.........
-100
•. -1 O0 -150
-150
-200
0
1 O0
' 300
200
-200
400
(e)
(f)
Winter (x)DailyValues
400 [ . 300[- ß ,.
2oo r.............. -, 1001-
X 35 m s-1 [Deanand Beardsley,causedpartial shadingon clear dayswith equatorward 1988]. While comparisons betweenbuoy VAWR and winds (about 48% of the C3 record). This shading NCAR aircraft wind data collectedin CODE [Frieheet causeda maximum error in the daily total insolationon al., 1984]and SMILE [Beardsley et al., 1997]exhibited a clear day of -2.4% at C3 and -1.5% at C5. Sincethe no evidenceof significant(>5-10%) systematicover- pattern of shadingwas quite consistentin eachrecord, speeding,severallater comparisonsbetweenan integral the insolationtime serieswere correctedby removing VAWR and an R. M. Young model 5103 wind moni- the obviousshadows.Someconsistentshadingwasalso observedin CODE 2, and the obviousshadowswerealso
tor integral propeller and vane mounted at the same height on the same discussuggestthat systematicdifferencesin wind speed between these two sensorsmay vary from 0% to about 6% in differentexperiments,with
removed.The CODE 2 C3 recordhad one177-hourgap which was filled using C2 insolationdata sincedailyaveragedC2 and C3 data generallyagreedwithin q- 10%
the VAWR readinghigh [Trasket al., 1989].
(although the C3 insolation wasgenerally higher).
The questionsconcerningflow distortion are difficult to address.
A numerical
model was used to examine the
In SMILE, two pyranometerswere mountedslightly abovethe other sensorsand shadingwas not observed. However, the C3 Eppley precisionspectral pyranome-
influenceof the VAWR sensorsupportsand housingon steady potential flow past the cup area; the supports ter (PSP) and the 8-48 data includedmanytime gaps and vanecausedabout 0.3% increasein the model "cup due in part to the partial floodingof both VAWR housspeed,"while the cylindrical housingcontributed about ings. The PSP recordstopped85 daysafter deployment 1.1% [Norment,1992].While flowdistortiondueto the and the 8-48 returned data until the C3 mooringwas rest of the buoy superstructure may be more impor- recovered.While the PSP and 8-48 serieswere highly tant, the CODE and SMILE wind sensorswere placed correlated,the 8-48 valueswere consistentlysmallerby to minimize the influenceof other sensorsand the sup- about 15% in comparisonwith the C3 PSP and a second port structure, and a large steering vane was mounted PSP deployedat StewartsPoint (Figure 1). Sincethe on the aft end of the buoy superstructureto keep the C3 and Stewarts Point PSPs daily-averagedserieswere wind velocity,air temperature, RH and air pressuresen- also highly correlatedand differed in the mean by only
sorsorientedinto the wind (Figure 2). Largeet al. [1995]recentlypresented evidencethat large surfacegravity wavesincreasethe vertical shearin the near-surfacemean wind profile. This meansthat use of the log profile formula to computethe wind speedat l0 m from a wind speed measurementmade at a lower height will underestimatethe true 10-m wind speedunless corrected
for wave distortion.
For conditions
found
2.3q-8.2 W m-2, the C3 PSP recordwasjudgedto be correct and the lower values of the 8-48 were attributed
to scalingerrors within the VAWR circuitry. Linear regressionwas then used to correct the amplitude of the 8-48 series, and any remaining gaps were filled using
StewartsPoint PSP data. A Lycor (modelLI-2005B) pyranometer mounted on the roof of the Bodega Bay
(BB) Marine Laboratory(Figure1) returneddata durin CODE 1 (wind sensorheight, 3.5 m; mean wind ing SMILE with some small gaps. The B B insolation speed,8 m s-1; mean significantwaveheight, 2 m time serieshad a nightly minimum of approximately45 [Beardsley, 1987]), the Largeet al. [1995]modelsug- W m-2, whichwassubtractedfromthe entirerecordto
BEARDSLEYET AL.' SURFACEHEAT FLUX OVER NORTHERNCALIFORNIASHELF
21,$81
producethe edited seriesusedhere. The mooredinsola- NDBC14 hourly time seriesbeing 0.2 + 1.4 mbar. Pa tion recordsmay have somesmall error due to sensortilt was measured at both C3 and C5 in CODE 2. The (noneweregimballed)[MacWhorterand Weller,1991], C3 record had gaps of 21 days at the beginning of the but no clear evidence was found for this in the SMILE deployment and 26 days at the end. Since the mean data. and standard deviation of the C3 minus C5 hourly time serieswere -0.5 q-0.6 mbar, the C3 gaps were mostly A.3. Incident Longwave Radiation filled with C5 data, with the remaining gap filled with l•w was not measured in CODE. In SMILE, l•w was NDBC13 data. measured using two Eppley precisioninfrared pyrogeA.6. Air Temperature
ometers(PIR1 and PIR2) mountedon the C3 buoyand •
1-)TO
,..1,.,.,,.,1 ....
,..1 ,-,•- C•- ......
+•
D,.-,;•+
A 11 DT'Do
•-,•+ .....
,4
usefuldata for muchof SMILE, with somegaps,primarily in the two buoy records during the last 2 months of the experiment. The three PIR recordswere compared over a 124-day common period, indicating that the C3 PIR1 and PIR2 recordsdisagreedby 4.2% on average, with PIRI>PIR2, while the C3 PIR2 record was larger than the Stewarts Point PIR by 1.0% on average. Basedon the better agreementbetweenStewarts Point and PIR2 data, the Stewarts Point data were used to fill gaps in the PIR2 record to produce a best C3 l• series.
A.4. Relative
CODE I and at C3 in CODE 2. Unfortunately, only one SMILE VAWR was instrumented for T•, and it returned no data. T• was measured for the entire experiment at NDBC13 and for January 12 to MW 2, 1989, at NDBC14. For the overlapping period, the mean and standard deviation of the hourly NDBC14 minus NDBC13 T• serieswere -0.2 ø •0.8øC, with the maximum hourly differencesbeing about •4øC. The decorrelation
timescale
of the difference
series is about
1 day, so the maximum differencesdid not last long. The NDBC13 T• record was used at C3, with an un-
certaintyin the daily-averaged T• (as indicatedby the standarddeviationof its differencewith NDBC14) of
Humidity
•0.5øC.
RH was not measured in CODE 1, so the median
value observedat C3 in CODE 2 was used (89%) at both C3 and C5 at 2.7 m height. This is slightly higher
A.7. Water Temperature
The CODE I VAWRs returned completeT• records.
than the average(85%) of the shipboardmeasurements In CODE 2, the C3 T• recordhad one 177-hourgap, made near C3 during the CODE I hydrographicsurveys whichwasfilled by temperaturedata obtainedat 5 m by (J. Huyer,personalcommunication, 1985). In CODE 2, a vector-measuring currentmeter (VMCM)(corrected measured values of RH between 100 and 105% occurred for bias [seeLentzand Trowbridge, 1991]). No VAWR frequently during periods of weak winds and reduced T• was obtained at C3 in SMILE. Instead, a C3 T• seinsolation(especiallyduring relaxationevents). The ries was estimated using the 6-m C3 WHOI VMCM RH sensors used in CODE and SMILE were calibrated record plus the difference between the M3 1-m VAWR up to valuesof about 94%, and valuesabove94% may and 5-m SIO VMCM temperature records. The mean have uncertaintieslarger than thosereportedin Table 3 adjustmentwas +0.10øC, while the medianadjustment
(R. Payne,personalcommunication, 1994). Assuming
was under +0.03øC.
the true maximum RH during thoseperiodswas 100%, Ideally, for heat flux computations,the ocean skin RH was corrected by clipping humidities above 100% temperature is desired. An experiment in the Arabian to 100%. An alternative scheme was also tried in which
Sea(R. •ask, personalcommunication, 1996)recently
humiditiesbetween94% and 105%werelinearlytransformed to between 94% and 100%. Resulting latent heat flux estimateswere nearly identicalto the clipped RHs (includingpositivelatent heat flux estimates).In
confirmedearlier NDBC and WHOI tests showingthat a 1-m temperature measurementunder a 0.5-m-deep discusbuoy correspondsto the temperature at about 0.5 m depth. Sincethe wind was generallystrongwhen
the absenceof any calibration information above 94%, insolationwas large (i.e., during summerupwelling), we elected to present results for the clipped RHs. One no additional corrections were made to the CODE and 177-hour gap was filled with the record median value. SMILE 1-m T• recordsfor skin temperature effects. In SMILE, one VAWR returned a complete good RH record. Isolated valuesabove 100% were again clipped A.8. Surface Current to 100%. A.5.
Air
The surface currents used in the heat flux computations
Pressure
Pa was not measured in CODE 1, so the NDBC13 time series was used at both C3 and C5. The alongshelf scale of air pressurewas large in CODE 1, with the mean
and standard
deviation
of NDBC13
minus
were taken
from the shallowest VMCM
at C3
andvector-averaging currentmeter(VACM) at C5 (see Table3 for sensordepth). Measurements of velocitydifferencesover the top 5 m during SMILE usinga vertical array of acousticcurrent sensorsindicate that they were
generally small,lessthan0.03m s-1 [Santala, 1991].
21,582
BEARDSLEY ET AL.' SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF (a)
Appendix B' Description of Surface Heat Flux Over the Oregon Shelf During July 1973
Wind Velocity
We presenthere surfaceheat flux estimatesobtained
the CoastalUpwellingExperiment(CUE II) [Halpernet al., 1974;ReedandHalpern,1974].In doingso,wehave two primary motivations. The first is to enablea comparisonand contrastwith the summerCODE heat flux obtained
over the northern
California
II data.
,
,
i
i
i
i,
-10
(b)
Wind Velocity 10 0
shelf.
The secondis to presenton its own merits the net surface heat flux and its componentsin this region. As the Oregon shelf remains an active area of modelingand observationalstudy, we believe this is an appropriate time to revisit the CUE
,
10
from measurementsmade over the Oregon shelf during
estimates
,
-10
H5 Air Pressure
(c)
1020
The surface heat flux
obtained from these data, though mentioned in several
papers[e.g.,Brydenet al., 1980;Lentz,1992;Federiuk and Allen, 1995],hasyet to be presentedin detail in a
1000
Downward
(d)
Shortwave
1000
widely available format.
CUE II took placeover the Oregonshelfin July and August 1973. For the purposesof determiningthe net surfaceheat flux, the limiting measurements are July
500 0
insolationmeasurements [Reedand Halpern,1974]at (e) buoy B3 (location45ø15.8•N124ø7.8•W).Accordingly,
Air Temperature
we considerthe heat flux there from 0800 UT, July 5, 1973 to 0700 UT, July 30, 1973. Precisesystemaccuolo racy for these measurementsis unknown,but many of 5 the systemsused in 1973 including the basic VAWR, Eppley model 8-48 pyranometer, and VACM were of (f) similar design to those used in the CODE 1 experi15 ment, and it is believed that in situ systemaccuracy shouldbe similar(Table 2). The CUE II measurements o
also resemble the CODE
1 measurements
in that
Water Temperature
no
continuous measurements of RH or Itw exist. For this
reason, we adopt a constant RH of 85% at a height
Cloud Factor
(g)
of 4 m (R. Smith,personalcommunication, 1996),and
1
Qt estimates should be viewed with caution. Also, air pressuremeasurementsusedcamefrom the H5 mooring
(107 km offshoreof B3), but this is of minor concern given the small pressure effect on air-sea heat fluxes and the larger uncertainty due to RH. Other measurements usedwere acquired at the B3 buoy. Measurement heightsfor the particular instrumentsusedare given in Table
B1.
Table B1. B3 Sensor Heights Above Water During CUE II Variable
Height
Wind speed, wind direction
2.4 m
Insolation
4.0 m
Air pressure Air temperature Water temperature Platform
1.8 m 1.6 m -0.7 m toroid
The air pressureheight is that at H5.
6
8
10
12
14
16
18
20
22
24
26
28
Jul 1973
Figure B1. Hourly-averaged meteorological time series from CUE II. Note that the daily-averaged cloud factoris alwaysgreaterthan0, evenonapparently clear days.Seetext for a discussion of the uncertainty in the
cloud factor.
The hourly meteorologicaldata and the daily-averaged cloud factor are presentedin Figure B1. The CUE II data displaysomesimilaritieswith the summer CODE
data taken some 750 km to the south. Winds
are primarily equatorward, and, while statistical relia-
bility for the relatively short 25-day recordis suspect, the insolation data indicate equatorwardwinds are often accompaniedby clear skies. One differencefrom the
BEARDSLEY ET AL.' SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF northern California shelf is in the relative air and sea
21,583
Wind u (-) and v (-) components
temperatures.The CUE II data indicatethat evenin {a) summer,Ta is often lessthan Ts overthe Oregonshelf. This has implicationsfor both Qt and Qs. One puzzlingobservation is that the daily-averaged cloudfactor, determinedfollowingReed[1976] is neverzero, even -r on severalapparentlyclear days. This is becausethe [=
15
observeddaily-averagedinsolationremainswell below the clear sky insolationpredictedby the Smithsonian
[$eckelandBeaudry,1973]formulausedin the estimation of cloudcover•Reed[!975]first,nnt,o.delo.•,rskyin-
Net Heat Flux
solation inCUE II was only about 93% ofthat predicted (b) 400 by the Smithsonianformula but found other midlati-
,.
,.
,
,.
tude insolation measurementsagreedmore closelywith the Smithsonian
formula.
Measurement
error on the
200
part of the Eppley 8-48 pyranometerat B3 is unlikely •,
to account forthisdiscrepancy asit hadbeen compared •. to calibrated instruments and found to agree to within ß
1% [Reedand Halpern,1974]. Uncertaintyas to the
.
.
actual cloud cover is therefore a large sourceof uncertainty in the estimated longwaveheat flux and will be
.
.
ß
-200
,
,
;
....
;
.
;
;
....
Shortwave(-) and Longwave(--) Heat Fluxes
discussed.
Thedaily-averaged heatfluxcomponents andheat (½)
''1
flux statistics are shown in Figure B2 and Table B2. As in the CODE observations,Qi is the largest term
in the positivenet heat flux. However,the mean Q•
m.
offOregon issubstantially smaller thantheJulyQ• off
E
northern California. Thisisdueto thelargeQtloss • at 45øN and to Q• which is weakly negative off Oregon rather than positiveas observedoff northernCalifornia. Although the lack of RH measurementscausessignificant uncertainty in the CUE II and CODE 1 Qt esti-
0
-200
Latent(-) and Sensible(-) Heat Fluxes
mates,there is reasonto believethe true Q• lossover (d)
]00 5O
Table B2. Statistics of the Daily-Averaged July Heat Fluxes off Oregon and Northern California
Q•
Q•
Q•
Q8
Q•
148 256 23 68
216 313 97 73
California: CODE I C3 Mean Maximum Minimum s.d.
190 319 99 76
218 332 123 73
California: CODE 2 C3 Mean Maximum Minimum s.d.
245 362 99 62
Units are W m-2.
249 334 126 69
-40 -20 -65 15
-25 -3 -63 12
-3 20 -16 9
July 1-30, 1981 -35 -14 -71 19
-5 17 -22 11
12 38 -5 15
July 1-27, 1982 -38 - 14 -68 17
13 62 -10 19
0
-5o -lOO -15o
Oregon: CUE II B3 July 5-30, 1973 Mean Maximum Minimum s.d.
?
20 55 -2 18
6
8
10 12
14
16
18 20
22
24
26
28
Jul 1973
Figure B2. Daily-averagedwind velocityand heat flux time series from CUE
II.
Because CUE
II did not in-
clude relative humidity measurements,the latent heat flux is only an estimate based on a constant relative humidity.
the Oregon shelf in CUE II is greater than that over the northern California shelf during CODE. Over the Oregon shelf, Ta is often less than T,. Hence, even if the RH approached100% severalmeters abovethe sea surface,it would drop below 100% at the sea surface. In contrast, over the northern California shelf, summer Ta is often above T,, so an RH near 100% observed severalmeters abovethe sea surfacecan imply an even
21,$84
BEARDSLEY
ET AL.' SURFACE HEAT FLUX OVER NORTHERN
CALIFORNIA
SHELF
greater RH at the seasurface.Of course,the difference OceanographicInstitution and a Mellon Foundation postdoctoral scholarshipawarded to Dever at the ScrippsInsti-
in relative Ta and T8 also affects Q s.
Comparedto Q• and Q• differences,mean insolation tution of Oceanography.In addition to the many peoplewho
differences between CUE II and CODE
are small dur-
ing CODE I but larger during CODE 2. Althoughthe computedmean July Qb lossis approximatelythe same over the Oregonand northern California shelves,there is considerableuncertainty as to its magnitude during CUE II. The mean CUE II Qb lossin Table B2 is based on the cloudcoverestimatedafter Reed[1976]without any modificationsto the Smithsonianformula. If the CUE II insolation on apparently clear days is a better measure than the Smithsonian
formula of the true clear
helped make the CODE and SMILE field programssuccessful, we also want to acknowledgethe following individuals who provided useful input during the preparation of this report: S. Anderson, J. Austin, V. Chow, T. Dickey, C. Dorman, J. Edson, C. Fairall, C. Friehe, W. Large, R. Payne, R. Weller and two anonymousreviewers. C. Alessi, J. Cook, and A.-M. Michael helped with the final preparation of the manuscript, figures, and tables. The Oregon shelf data was generouslyprovided by D. Halpern. The NDBC13 time se-
riesdata wereobtainedfromthe coastaldata archive("Data Zoo") constructedand maintainedby the Centerfor Coastal Studies, ScrippsInstitution of Oceanography.This is WHOI contribution
9406 and U.S. GLOBEC
contribution
103.
sky insolation over the Oregon shelf in July, then we have overestimated cloud cover and underestimated Q• References loss.To gaugethis effect,we recalculatedCUE II cloud coverbasedon a clearskyinsolationequalto 93% of the Alados-Arboledas, L., J. Vida, and J. I. Jimenez, Effects of solar radiation on the performance of pyrogeometers Smithsonian formula, and used it in the estimation of with silicon domes, J. Atmos. OceanicTechnol., 5, 666the Q•. With the lowercloudcover,the meanJuly 1973 670, 1988.
Q• lossincreased by 8 W m-2. The maximumincrease Alessi, C.A., S. J. Lentz, and R. C. Beardsley, Shelf Mixed in the daily-averaged Q• losswas27 W m-2, andthe LayerExperiment(SMILE) programdescriptionandcoasstandard deviation of the difference time series was 8 W m
-2
.
A comparisonof the 1973 CUE II mean heat flux components with the NelsonandHusby[1983](NH)climatologyis problematic,sincethe 1ø x 1ø squarecon-
tainingthe B3 mooringsite (centeredat 45øN,124øW) contained only 64 ship observationsduring July. The resulting July monthly mean values and standard er-
rorsare (in W m-•) Qn - 209 q- 8; Qi - 243 q- 9; Qb - -35 q- 3; Q• - -10 q- 3; and Q• - 11 q- 2. With the exceptionof Q•, the other heat flux compo-
tal and moored array data report, Tech.Rep. WHOI 9139, 211 pp., Woods Hole Oceanogr. Inst., Woods Hole, Mass., 1991. Anderson, R. J., and S. D. Smith, Evaporation coefficientfor the sea surfacefrom eddy flux measurements,J. Geophys. Res., 86, 449-456, 1981. Austin,J. A., andS. J. Lentz, 1998,The relationship betweensyn-
opticweather systems andmeteorological forcingontheNorth Carolinainnershelf,J. Geophys. Res.,in press,1998. Beardsley,R. C., A comparisonof the vector-averaging current meter and new Edgerton, Germeshausen,and Crier, Inc., vector-measuring current meter on a surface mooring in Coastal Ocean DynamicsExperiment 1, J. Geophys.Res., 92, 1845-1859, 1987.
nentsweresmallerthan climatology by 14-27 W m-•, and Qn duringCUE II was about 70 W m-• smaller Beardsley,R. C., and S. J. Lentz, The CoastalOceanDythan climatology. The primary conclusionfrom this is that while the NH climatologyprovidesa useful qualitative description of the seasonalvariation in the surface heat flux over the Oregonshelf,future field experiments designedto look at upper oceanprocessesshould include surfacemeteorologicaland oceanographicmeasurements
sufficient
to allow
direct
estimation
of the
surfacewindstressandheatflux (f•ollowing equations 2, 3, 6, and 7). Only through direct measurementof in-
namicsExperiment(CODE) collection:An introduction,
J. Geophys.Res., 92, 1455-1463, 1987. Beardsley,R. C., C. E. Dorman, C. A. Friehe, L. K. Rosenfeld, and C. D. Winant, Local atmosphericforcing during the Coastal Ocean Dynamics Experiment, 1, A description of the marine boundary layer and atmospheric conditions over a northern California upwelling region, J. Geophys.Res., 92, 1467-1488, 1987. Beardsley,R. C., S. J. Lentz, R. A. Weller, J. D. Irish, K. H. Brink, R. Limeburner,and J. D. Manning, A preliminary descriptionof the heat budget, southernflank of Georges Bank, Spring-Summer, 1995: Part I. Surface heat flux,
cident shortwave and longwaveradiation, wind speed, (abstract), Eos, Trans.AGU, 77(46), Fall Meet. Supp., RH, Ta, and T• can accurateestimatesof surfaceforcing F372, 1996. over the shelf be achieved. Fortunately, the technology Beardsley,R. C., A. G. Enriquez, C. A. Friehe, and C. A. to make these measurementson a mooredplatform with Alessi, Intercomparisonof aircraft and buoy measurements of wind and wind stressduring SMILE, J. Atmos. high data return has been improvingand is now becomOceanic Technol., 1•, 969-977, 1997. ing routine. Given the importanceof surfaceforcingon Berliand, M. E., and T. G. Berliand, Measurementof the efmany shelf processes,we encouragemaking air-sea flux fective radiation of the Earth with varying cloud amounts measurementsa key componentof future shelf physical (in Russian), Izv. Akad. Nauk SSSR, Ser. Geofiz., 1, 1952. oceanographicexperiments. Brink, K. H., The near-surfacedynamics of coastal upAcknowledgments. The meteorologicaland oceanowelling, Prog. Oceanogr., 12, 223-257, 1983. graphic measurements made in CODE and SMILE were Bryden, H. L., Horizontal advectionof temperaturefor lowfundedby the National ScienceFoundation.Supportfor the frequencymotions, Deep Sea Res., 25, 1165-1174, 1976. final analysisand synthesisof the measurementspresented Bryden, H. L., D. Halpern, and R. D. Pillsbury,Importance here was provided by NSF grants OCE 91-15713, OCE 93of eddy heat flux in a heat budget for Oregon coastal 13671, and OCE 96-32357 to the authors at the Woods Hole
waters, J. Geophys. Res., 85, 6649-6653, 1980.
BEARDSLEY ET AL.: SURFACE HEAT FLUX OVER NORTHERN CALIFORNIA SHELF
21,585
Choppin, P. A., An examination of cup anemometer overspeeding, Meteorol. Rundsch., 35, 1-11, 1982. Dean, J.P., and R. C. Beardsley,A vector-averagingwind
Hamilton, G. D., NOAA Data Buoy Office programs, Bull. Amer. Meteerol. $oc., 61, 1012-1017, 1980. Huyer, A., and P.M. Kosro, Mesoscalesurveysover the
recorder(VAWR) systemfor surfacemeteorological measurementsin CODE (Coastal Ocean DynamicsExperi-
shelfand slopein the upwellingregionnear Point Arena, California, J. Geophys.Res., 92, 1655-1681, 1987.
ment),Tech.Rep.WH01-88-20, 74pp.,WoodsHoleOcean- Irish,J. D., Mooredtemperature andconductivity observaogr.Inst.,WoodsHole,Mass.,198._•. tions,in CODE 2: MooredArrayandLarge-Scale Data Dever, E. P., Wind-forced cross-shelf circulation on the northern California shelf, J. Phys. Oceanogr., 27, 15661580, 1997. Dever, E. P., and S. J. Lentz, Heat and salt balances over the northern California shelf in winter and spring, J. Geophys.Res., 99, 16,001-16,017, 1994.
Dickey, T. D., D. V. Manov, D. A. Siegel,and R. A. Weller, Determination of longwave heat flux at the air-sea interface using measurementsfrom buoy platforms. J. Atmos. Oceanic Technol., 11, 1057-1078, 1994. Dorman, C. E., Evidence of Kelvin waves in California's marine layer and related eddy generation, Mon. Weather Rev., 113, 827-839, 1985. Dorman, C. E., Possible role of gravity currents in northern California's coastal summer wind reversals, J. Geophys. Res., 92, 1497-1506, 1987. Dorman, C. E., and C. D. Winant, Buoy observations of the atmosphere along the west coast of the United States, 1981-1990, J. Geophys.Res., 100, 16,029-16,044, 1995. Dorman, C. E., A. G. Enriquez, and C. A. Friehe, Structure of the lower atmosphere over the northern California coast during winter, Mon. Weather Rev., 123, 2384-2404, 1995. Enriquez, A. G., and C. A. Friehe, Bulk parameterization of momentum, heat, and moisture fluxes over a coastal upwelling area, J. Geophys. Res., 102, 5781-5798, 1997. Fairall, C. W., and R. Markson, Mesoscale variations in surface stress, heat fluxes, and drag coefficient in the marginal ice zone during the 1983 Marginal Ice Zone Experiment, J. Geophys.Res., 92, 6921-6932, 1987. Fairall, C., E. F. Bradley, D. P. Rogers, J. B. Edson, and G. S. Young, Bulk parameterization of air-sea fluxes for Tropical Ocean-Global Atmospheric Coupled OceanAtmospheric ResponseExperiment, J. Geophys.Res., 101, 3747-3764, 1996. Federiuk, J., and J. S. Allen, Upwelling circulation on the Oregon continental shelf, II, Simulations and comparisons with observations, J. Phys. Oceanogr., 25, 1867-1889, 1995. Friehe C. A., R. C. Beardsley,C. D. Winant, and J.P. Dean, Intercomparison of aircraft and surface buoy meteorolog-
Report, edited by R. Limeburner, Tech. Rep. WH01-850ø5,pp. 133-164, Woods Hole Oceanogr. Inst., Woods Hole, Mass., 1985.
Kelly, K. A., The influenceof winds and topography on the sea surface temperature patterns over the northern California slope, J. Geophys.Res., 90, 11,783-11,798, 1985.
Large, W. G., and S. Pond, Open ocean momentum flux measurementsin moderate to strongwinds, J. Phys. Oceanogr., 11, 324-336, 1981. Large, W. G., and S. Pond, Sensible and latent heat fluxes over the ocean, J. Phys. Oceanogr., 12, 464-482, 1982.
Large, W. G., J. Morzel, and G. B. Crawford,Accountingfor surface wave distortion of the marine wind profile in lowlevel Ocean Storms wind measurements,J. Phys. Oceanogr., 25, 2959-2971, 1995. Largier, J. L., B. A. Magnell, and C. D. Winant, Subtidal circulation over the northern California shelf, J. Geophys. Res., 98, 18,147-18,179, 1993. Lentz, S. J., A descriptionof the 1981 and 1982 springtransitionsover the northern California shelf, J. Geophys.Res., 92, 1545-1567, 1987a. Lentz, S. J., A heat budget for the northern California shelfduring CODE 2, J. Geophys.Res., 92, 14,491-14,509, 1987b.
Lentz, S. J., The surfaceboundarylayer in coastalupwelling regions, J. Phys. Oceanogr., 22, 1514-1539, 1992. Lentz, S. J., and J. H. Trowbridge, The bottom boundary layer over the northern California shelf, J. Phys. Oceanogr., 21, 1186-1201, 1991.
Limeburner,R. (Ed.), CODE 2: Moored Array and LargeScale Data Report, Tech. Rep. WH01-85-$5, 234 pp., Woods Hole Oceanogr. Inst., Woods Hole, Mass., 1985.
Limeburner, R., and R. C. Beardsley,CTD observationsoff northern California during the Shelf Mixed Layer Experi-
ment (SMILE), November1988, Tech.Rep. WH01-89-25, 272 pp., Woods Hole Oceanogr.Inst., Woods Hole, Mass., 1989a.
Limeburner, R., and R. C. Beardsley,CTD observationsoff northern California during the Shelf Mixed Layer Experiical data during CODE 1, J./,Atmos. OceanicTechnol., ment (SMILE), February-March1989, Tech.Rep. WH011, 79-86, 1984. 89-•1, 218 pp., Woods Hole Oceanogr.Inst., Woods Hole, Fung, I. Y., D. E. Harrison, and A. A. Lacis, On the variability of the net longwave radiation at the ocean surface, Rev. Geophys., 22, 177-193, 1984. Garratt, J. R., The stably stratified internal boundary layer for steady and diurnally varying offshore flow, Boundary Layer Meteorol., 38, 369-394, 1987. Garratt, J. R., The internal boundary layer- A review, Boundary Layer Meteorol., 50, 171-203, 1990. Garratt, J. R., and B. F. Ryan, The structure of the stably stratified internal boundary layer in offshoreflow over the sea, Boundary Layer Meteorol., J7, 17-40, 1989. Garratt, J. R., and M. Segal, On the contribution of atmospheric moisture to dew formation, Boundary Layer Meteorol., •5, 209-236, 1988. Halpern, D., Structure of a coastal upwelling event observed off Oregon during July 1973, Deep Sea Res., 23, 495-508,
Mass., 1989b. Limeburner, R., and R. C. Beardsley, CTD observations off northern California during the Shelf Mixed Layer Ex-
periment (SMILE), May 1989, ,Tech. Rep. WHOI-89-J2, 239 pp., Woods Hole Oceanogr.Inst., Woods Hole, Mass., 1989c.
List, R. J., Smithsonian Meteorological Tables, 527 pp., Smithsonian Inst. Press, Washington, D.C., 1984.
Liu, W. T., K. B. Katsaros,and J. A. Businger,Bulk parameterizations of the air-sea exchangeof heat and water vapor including the molecular constraints on the sea sur-
face, J. Atmos.$ci., 36, 1722-1735, 1979. MacWhorter, M. A., and R. A. Weller, Error in measure-
mentsof incomingshortwaveradiation made from ships and buoys,J. Atmos.Oceanic Technol.,8, 108-117, 1991. 1976. Mooers,C.N.K., J. Fernandez-Partagas, and J. Price, MeteHalpern, D., J. R. Holbrook, and R. M. Reynolds, A comorologicalforcingfieldsof the NewYork Bight (firstyear's pilation of wind, current, and temperature measurements: progressreport). Tech.Rep. TR76-8, 151 pp., Univ. of Oregon, July, and August 1973, Tech.Rep. M7•-75, 190 pp., Miami, Miami, Fla, 1976. Univ. of Wash., Seattle, 1974. Mountain, D. G., G. A. Strout, and R. C. Beardsley,Sur-
21,586
BEARDSLEYET AL.: SURFACEHEAT FLUX OVERNORTHERNCALIFORNIASHELF
face heat flux in the Gulf of Maine, Deep Sea Res., •3, GLOBEC Spec. Issue, 1533-1546, 1996.
Nelson, C. S., and D. M. Husby, Climatology of surface heat fluxes over the California Current region, NOAA Tech.Rep.NMFS SSRF-763, 155 pp., Natl. Oceanicand Atmos. Admin., Woods Hole, Mass., 1983.
Norment, H. G., Airflow about the WHOI VAWR cup anemometer, 178 pp., final report, Atmos. Sci.Assoc., Concord, Mass., 1992.
Payne,R. E., Albedoof the seasurface,J. Atrnos.Sci., 29, 959-970, 1972.
Payne,R. E., Air temperatureshieldtests,Tech.Rep. WHOI87-•i0, 22 pp., WoodsHole Oceanogr.Inst.,WoodsHole, Mass., 1987.
Reed,R. K., An evaluationof formulasfor estimatingclearskyinsolationoverthe ocean,Tech.Rep.ERL 352 - PMEL 26, 25 pp., Natl. Oceanicand Atmos. Admin., Seattle, Wash., 1975.
Reed, R. K., On the estimationof net long-waveradiation from the oceans,J. Geophys.Res., 81, 5793-5794, 1976.
Reed,R. K., and D. Halpern, Radiationmeasurements off the Oregoncoast:July/August 1973, Tech.Rep.MZi-18, 51 pp., Univ. of Wash. Seattle, 1974.
Rogers,D. P., D. W. Johnson,and C. A. Friehe,The stable internal boundary layer over a coastal sea, I, Airborne measurements of the mean and turbulence struc-
ture, J. Atrnos.$ci., 52, 667-683, 1995.
Rosenfeld, L. K. (Ed.), CODE 1: MooredArray andLargeScale Data Report, Tech. Rep. WH01-83-œ3, 185 pp., Woods Hole Oceanogr.Inst., Woods Hole, Mass., 1983.
Rosenfeld,L. K., Tidal band current variability over the northern California shelf, Ph.D. thesis,Mass.Inst. of Tech-
Send,U., R. C. Beardsley,and C. D. Winant, Relaxation from upwellingin the CoastalOceanDynamicsExperiment, J. Geophys. Res., 92, 1683-1698, 1987. Smith, S. D., Coefficientsfor sea surfacewind stress,heat flux, and wind profilesas a functionof wind stressand temperature,J. Geophys. Res., 93, 15,467-15,472,1988. Strub, P. T., and C. James, Atmospheric conditionsduring the spring and fall transitions in the coastal ocean off western United States, J. Geophys.Res., 93, 15,56115,584, 1988.
Strub, P. T., J. S. Allen, A. Huyer, and R. L. Smith, Largescalestructure of the springtransition in the coastalocean off North America, J. Geophys.Res., 92, 1527-1544, 1987. Trask, R. P., J.P. Dean, J. R. Valdes, and C. D. Marquette, FASINEX (Frontal Air-Sea Interaction Experi-
ment) mooredinstrumentation,Tech.Rep. WH01-89-3, 60 pp., WoodsHole Oceanogr.Inst., WoodsHole, Mass., 1989.
Weller, R. A., and R. E. Davis, A vector measuringcurrent meter, Deep Sea Res., œ7A,565-582, 1980. Weller, R. A., D. L. Rudnick,R. E. Payne,J.P. Dean, N.J. Pennington,and R. P. Trask, Measuringnear-surfacemeteorologyoverthe oceanfrom an array of surfacemoorings in the SubtropicalConvergence Zone, J. Atmos.Oceanic Technol., 7, 86-103, 1990.
Winant, C. D., C. E. Dorman, C. A. Friehe, and R. C. Beardsley,The marine layer off northern California: An exampleof supercriticalchannelflow, J. Atmos.$ci., •5, 3588-3605, 1988.
Zemba,J., and C. A. Friehe,The marineatmospheric boundary layerjet in the CoastalOceanDynamicsExperiment, J. Geophys.Res., 92, 1489-1496, 1987.
nol./WoodsHoleOceanogr. Inst.,WoodsHole,Mass.,1987. Rudnick, D. L., and R. E. Davis, Mass and heat budgetson the northernCaliforniashelf,J. Geophys. Res.,93, 14,013-14,024, 1988.
Samelson,R. M., Supercritical marine layer flow along a smoothly-varyingcoastline,J. Atmos.$ci., •9, 1571-1584, 1992.
Santala, M. J., Surface-referenced current meter measure-
ments,Ph.D. thesis,Mass. Inst. of Technol./Woods Hole Oceanogr.Inst., WoodsHole, Mass., 1991. Seckel,G. R., and F. H. Beaudry,The radiationfrom sun
R. C. Beardsley, J.P. Dean, and S. J. Lentz, Physical OceanographyDepartment, Mail Stop 21, WoodsHole Oceanographic Institution, WoodsHole, MA 02543-1541. E. P. Dever, Center for Coastal Studies, ScrippsInstitution of Oceanography,Universityof California, San Diego, La Jolla, CA 92093-0209.
and sky over the north PacificOcean, (abstract),Eos (ReceivedMay 15, 1997; revisedApril 7, 1998; Trans. AGU, 5•i, 1114 1973. acceptedApril 14, 1998.)