Jan 15, 1997 - should be 180 ø for a perfect inverse barometer response. This result is in .... Another objective is to provide a more precise correction of this effect in altimetric ... terranean Sea to atmospheric pressure forcing is below 1 day,.
JOURNAL
OF GEOPHYSICAL
RESEARCH,
VOL. 102, NO. C1, PAGES 973-984, JANUARY
15, 1997
Responseof the Mediterranean mean sea level to atmospheric pressure forcing Pierre-YvesLe Traon and Philippe Gauzelin Direction Oc6anographieSpatiale,CollecteLocalisationSatellites,Toulouse,France
Abstract. The responseof the Mediterraneanmean sea level to atmosphericpressure forcingis analyzedusing3 yearsof TOPEX/POSEIDON data. Coherenceanalysis betweenmean sea level and atmosphericpressureshowsa significantdeparturefrom a
standard inverse barometer effectat frequencies higherthan30 days -•. At high frequenciesthe phasedifferencebetweensea level and pressureis about 100ø,while it shouldbe 180ø for a perfect inversebarometer response.This result is in agreementwith previousfindingsand confirmsthe role of the Straitsof Gibraltar and Sicilyin limiting the water exchange(and thus the responseto atmosphericpressureforcing) at high frequencies.The responseof the Mediterranean mean sea level is then investigatedusing the Candela[1991] analyticalmodel which takes accountof friction in the Straitsof Gibraltar and Sicily.The model explainsa large part of the variancein TOPEX/POSEIDON mean sealevel variations(50% for the westernbasinand 38% for the easternbasin). Comparedto an inversebarometercorrection,it givesa smoother responsewith a phasedelay at high frequencies.It also explainsmore variancein
TOPEX/POSEIDON meansealevelvariations (5 cm2 and7 cm2 for thewestern and easternbasins,respectively).This demonstratesthat this simplemodel providesan improvedcorrectionof atmosphericpressureeffectsin TOPEX/POSEIDON data. As the two correctionshave an rms differenceof 2-3 cm with maximumdifferencesof up to 10 cm, the impact on the mappingof oceaniccirculationis not negligible.This is exemplified throughthe comparisonof sealevel anomalyderivedfrom the two corrections. 1.
[1989] and Candela[1991] have proposeda simple analytical
Introduction
model
In the open oceanthe responseof sealevel to atmospheric pressureis closeto the well-knowninversebarometer (IS) effect [e.g., Wunsch,1972; Ponte, 1993; Fu and Pihos, 1994; Gasparand Ponte,1996].The IS correction(ISC) (to be subtractedto the seasurfaceheight)is simplygivenby
IBC = - 1/p#(P - Pref)
(1)
where Pref is the mean pressureover the ocean which, in practice,is often assumedto be constant(p is density,# is gravity).A 1-mbar increasein atmosphericpressurethus inducesa sea level depressionof about 1 cm. In a closedbasin,becausemassmustbe conserved,Pref must be made equal to the mean pressureover the basin;the mean sealeveldoesnot thusrespondto the meanpressure,andonly adjustmentsof sea level to pressureanomalies(e.g., due to spatialvariationsof pressure)are possible.In a semi-enclosed sea suchas the Mediterraneanthe responseis more complex. The sealevel canrespondto the meanpressureover the basin if there is an adjustmentthroughthe straits.From the work of Lacombe[1961] and Crepon[1965]it hasbeenrecognizedthat this adjustmentoccurs.Crepon[1965] observeda correlation betweenthe flux throughthe Strait of Gibraltar and the mean atmosphericpressure.Actually,the continuityequationwould imply a correlationwith the time derivativeof pressure,not pressureitself. Severalpapershave dealt with the problem [e.g.,Garrett,1983;Garrettand Majaess,1984]. Candelaet al. Copyright1997by the AmericanGeophysicalUnion. Paper number 96JC02777. 0148-0227/97/96JC-02777509.00 973
which
takes account
of the friction
in the Straits
of
Gibraltar and Sicily. The model providesa responseof sea level to a given spectrumof atmosphericpressure.At low frequenciesthe straitsdo not constrainthe flow. At high frequenciesthe Mediterranean cannot fully adjust, and the responsediffers significantlyfrom a standard IB effect. As a consequence, perfectcorrelationwith pressurechangescannot be expectedwhichpartly solvedthe contradiction.This model was testedwith tide gaugesand explaineda large part of the sea level variancesignal[Candela,1991]. The nonisostaticresponseof sea level to atmosphericpressureforcingin the MediterraneanSeahasbeen observedfrom the analysisof tide gaugedata. The responseis mainlyunderisostaticat high frequenciesand overisostaticat low frequencies[Lascaratosand Gacic, 1990;Tsimplisand ViahaMs,1994]. Tide gaugesgive the sea level responseat a givenplace.This local responsewill be different,however,from the basinaverage response.It will include,in particular,the effect of wind whichis unlikelyfor the mean sealevel response[e.g.,Garrett, 1983]. Thanksto the highaccuracyof TOPEX/POSEIDON (hereafter T/P) data, satellitealtimetrycan now be usedto quantitativelymeasurethe mean sea level (and oceaniccirculation variations)in the MediterraneanSea[Larnicolet al., 1995].T/P providesglobal coverageof the Mediterranean which should improveanalysisof mean sealevelvariationsand in particular of the influenceof atmosphericpressure.In addition, to analyze other signalssuch as oceaniccirculationvariations,it is important to correct altimetric data for atmosphericpressure effects.In their analysisof T/P data in the Mediterranean Sea,
974
LE TRAON
AND
GAUZELIN:
MEDITERRANEAN
MEAN
SEA LEVEL
46
3O
o5
15
36
Longitude(degrees)
Figure 1. Location of TOPEX/POSEIDON tracksin the MediterraneanSea. Tracks in bold are sampled during the first 3 daysof the cycle.
Larnicol et al. [1995] found a mean responseof sea level to usingthe BM4 parameterizationfor both instruments[Gaspar atmosphericpressureclose to an inverse barometer effect. et al., 1994], Cartwrightand Tayler[1971] solid tides,Ray and However, the sea level variations corrected for an IB effect still Sanchez[1989]tidal loading,and P. Canceillet al. (Barotropic showedhigh-frequency (
