Air-Sea Interaction Sea Surface Temp

To take an example, in calculating the ux mapsshown in Figure 4 many corrections were applied tothe VOS observations in an attempt to remove biasescaused by the observing methods. For example, airtemperature measurements were corrected for theheat island caused by the ship heating up in sunny,low-wind conditions. The wind speeds were adjusteddepending on the anemometer heights on differentships. Corrections were applied to sea temperaturescalculated from engine room intake data. Despitethese and other corrections, the global annual meanux showed about 30Wm2 excess heating of theocean. Previous climatologies calculated from shipdata had shown similar biases and the uxes had beenadjusted to remove the bias, or to make the uxescompatible with estimates of the meridional heattransport in the ocean. However, comparison of theunadjusted ux data with accurate data from airseainteraction buoys showed good agreement betweenthe two. This suggests that adjusting the uxesglobally is not correct and that regional ux adjust-ments are required; however, the exact form of thesecorrections is presently not known.In the future, computer models are expected to

provide a major advance in ux estimation, Recently,coupled numerical models of the ocean and of theatmosphere have been run for many simulated years,duringwhich themodeled climate has not drifted. Thissuggests that the airsea uxes calculated by themodels are in balance with the simulated oceanic andatmospheric heat transports. However, it does notimply that at present the ux values are realistic.Errors in the short-wave and latent heat uxes maycompensate one another; indeed, in a typical simula-tion the sea surface temperature stabilized to a valuethatwas, over large regions of the ocean, a few degreesdifferent from that which is observed. Nevertheless,the estimation of ux values using climate or NWPmodels is a rapidly developing eld and improvementswill doubtless have occurredby the time this article hasbeen published. There will be a continued need for in-situ and satellite data for assimilation into the modelsand for model development and verication. How-ever, it seems very likely that in future the most

accurate routine source of airsea ux estimates willbe from numerical models of the coupled oceanatmosphere system.

See also

Aerosols: Observations and Measurements; Role in Ra-diative Transfer. AirSea Interaction: Freshwater Flux;Sea Surface Temperature; Surface Waves. BoundaryLayers: Observational Techniques In Situ; ObservationalTechniquesremote; Surface Layer. Buoyancy andBuoyancy Waves: Optical Observations. Climate Var-iability: North Atlantic and Arctic Oscillation. CoupledOceanAtmosphere Models. El Nino and the South-ernOscillation:Observation.ReectanceandAlbedo,Surface. Weather Prediction: Regional PredictionModels.

Further Reading

Browning KA and Gurney RJ (eds) (1999) Global Energyand Water Cycles. Cambridge: Cambridge UniversityPress.

Dobson F, Hasse L and Davis R (eds) (1980) AirSeaInteraction, Instruments and Methods. New York: Ple-num Press.

Garratt JR (1992) The Atmospheric Boundary Layer.Cambridge: Cambridge University Press.

Geernaert GL and Plant WJ (1990) Surface Waves andFluxes, vol. 1, Current Theory. Dordrecht: KluwerAcademic.

Isemer H-J andHasse L (1987)The Bunker Climate Atlas ofthe North Atlantic Ocean, vol. 2, AirSea Interactions.Berlin: Springer-Verlag.

Josey SA, Kent EC and Taylor PK (1999) The SouthamptonOceanography Centre (SOC) OceanAtmosphere Heat,Momentum and Freshwater Flux Atlas, SOCReport No.6, 30 pp.1gs. (Available from The Library, Southamp-ton Oceanography Centre, European Way, Southamp-ton, SO14 3ZH, UK.)

Kraus EB and Businger JA (1994) AtmosphereOceanInteraction, 2nd edn. New York: Oxford UniversityPress.

Stull RB (1988) An Introduction to Boundary LayerMeteorology. Dordrecht: Kluwer Academic.

Wells N (1997). The Atmosphere and Ocean: A PhysicalIntroduction, 2nd edn. London: Taylor and Francis.

Sea Surface Temperature

W J Emery, University of Colorado, Boulder, CO, USA

Copyright 2003 Elsevier Science Ltd. All Rights Reserved.

Introduction

As the controlling variable of heat, momentum, salt,and gas uxes between the ocean and the atmosphere,

100 AIRSEA INTERACTION / Sea Surface Temperature

the sea surface temperature (SST) has always been atopic of interest to scientists. It is also the easiestoceanographic parameter to observe and observationsof some form of SST extend back to the time of theearly Greek scientists. In addition to its relationship tooceanatmosphere uxes, the SSTalso relates directlyto a number of human concerns. For example, thesuccess of sheries and shermen can be enhanced by aknowledge of the SSTpattern. Formany years SSTwasmeasured by taking a bucket sample of the surfacewaters and then measuring its temperature. As shipsevolved to powered vessels, this practice of SST bucketsampling had to be abandoned and the practice ofusing ameasurement of the temperature of the coolingwater coming to the ships engines was used. This wasknown as ship injection temperatures (the SSTsensorwas injected into the water stream), and these datasuffer from many basic problems (engine room heat-ing, depth of the water intake, etc.). The advent ofsatellite-tracked drifting buoys introduced a platformthat could measure the SST and then report it in nearreal time. Thought to be less noisy and more accuratethan ship SSTs, the drifting buoy SST data became thestandard for both the calibration and validation ofsatellite infrared estimates of SST.Over the years thesebuoy SSTs have been used to calculate the infraredalgorithm coefcients for the computation of the SST.The problem is that a buoy cannot measure thetemperature of the 10mm thin skin of the ocean,which is the layer that radiates out into space. Thus,satellite infrared measurements of SSTare of this skintemperature and not of the deeper bulk SST measuredby the ships and buoys. It is the difference between theskin andbulk temperature that is directly related to thewind speed and the net airsea heat ux. In the past thebulk SST has been used in the computation of airseaheat ux terms using empirical bulk formulas. Thesecontinue to be used, but research efforts are underwayto transform these computations into a satellite-onlycalculation. These efforts will explicitly involve skinand bulk SST, giving a more physical basis to theconnections between heat ux and the skin and bulkSSTs. These improvements are critical for the mode-ling of climate change, since the SST is the boundarythat connects the model atmosphere to the modelocean. Any incorrect specication of the global SSTpatterns will lead directly to errors in the modelsimulations.

History of SST Measurements andApplications

One of the earliest uses of sea surface temperature(SST) was when Benjamin Franklin mapped the

position of the Gulf Stream from SST measurementsmade from the mail packet ships that he rode to andfrom Europe. Fortunately for Franklin, the GulfStream has a very sharp thermal contrast on itswestern edge where the cold shelf waters comingdown from the north meet the warm Gulf Streamwaters advecting to the north east. Together withFolger, he published a map (Figure 1) that describedthe position of the Gulf Stream, advising ship captainsnot to sail against this current, which was strongenough to hold a sailing vessel still. This practicalapplication of SST made it possible to reduce thecrossing times for ships sailing from the Americas tothe continent.As one of the easiest measurements in oceano-

graphy, SST became widely observed whether as asample taken from a pier or beach or as a bucketsample from a ship on the open ocean. The collectionof a bucket of water whose temperature wasmeasuredas an estimate of SST became common practice amongthe sailing ships carrying theworlds commerce.Whenthe ships logs became the source of global informationon winds and currents, the SST information was alsocompiled from the ships logs. This information wasroutinely published, alongwith the currents andwindsfrom the ships logs, as part of sailing directions putout rst by the USNavy and later by the Coast Guard.The same practice was introduced later in Europe andbecame an operational reality for most ocean-goingvessels.The collection of a bucket sample from sailing ships

that travel at speeds between 5 and 15 knots was easy.For powered ships traveling alongat 2030knots itwasno longer possible to collect a bucket sample for SSTmeasurement. Instead, the temperature of the coolingwater used to cool the ships engines was measured asthe SST. This temperature was called the injectiontemperature because the sensor was injected into thepipe carrying this cooling water to the heat exchangers.One of the fundamental problems with these injectiontemperatures was the location of the inlet pipe for thiscooling water, which was positioned far down on theships hull, generally collecting water from about 5mdepth far away from the sea surface. In addition, theheat in the engine roomwas known to heat the coolingwater and the associated thermometer, resulting in awarm bias for the SSTs.Other sources of error were thereading of the analog gauge, the hand recording of thedata in the ships log and the radio broadcast of theSST to be included in the SST database.

Satellite SST

In spite of all these limitations, the ship SSTs wereinitially used to match and adjust infrared satellite

AIRSEA INTERACTION / Sea Surface Temperature 101

measurements of the SST. Later, satellite-trackeddrifting buoys were equipped with temperature portssticking out from the hull to measure the SST as thebuoy traveled around the ocean. Most of these buoySST sensors were initially calibrated to 70.11C, butsince the buoys are considered expendable there is nopost-deployment calibration and it is not known howwell the buoys retain their calibration. There are aboutthree different hull types used today in these buoyswith slightly different congurations of the hull SSTsensors. All of these buoys oat in the active near-surface layer and move up and down with the waveeld. As a consequence, the buoy SST represents thetemperature between the surface and 12m depth.The best estimate of the accuracy of these data can bemade by considering the mean difference and varia-bility of the difference between contemporary SSTmeasurements. This value is 0.41C,which can be takenas an overall error limit for buoy-measured bulk SST.It is important to realize that past practice has been

to compute the satellite infrared SST algorithm coef-

cients from regression with nearly coincident buoySST data. In this approach one not only assumes thatthe satellite and buoys measure the same SST, but alsothat the buoy SSTs have no errors themselves. That isnot to say that people assume the buoy SSTs have noerror, but the practice of using them to nd thealgorithm coefcients through linear regression im-plies that the buoy SSTs are error-free. It appears thatthe source of this conict is the fact the satelliteinfrared systems and their calibration systems do notretain their calibrations over long periods of time andthat regression to in situ SST is required to overcomethese drifts. Since buoy SSTs are the best of the presentin situ SSTs, they are used to supply the in situ SSTobservations in spite of the fact that they do not andcannot measure the skin SST, as any direct physicalcontact with the skin layer will temporarily disrupt it.Observations of the effects of breaking waves on thepresence of the skin layer have shown that while abreaking wave does indeed destroy the skin layer, theskin forms again after just a few seconds. So, in spite of

Figure 1 The Poupard version of the FranklinFolger Gulf Stream chart published in Philadelphia in 1786 with Franklins MaritimeObservations.


strongwinds and breakingwaves, the skin SST layer ispresent most of the time andmust be considered in theremote sensing of SST. Before various groups andagencies will transform their computations to skinSST, there must be a source of in situ skin SST toreplace the buoy SSTs that are used at present. We willdiscuss this point further later in this article.

SST Climatology

Oneof the benets of being relatively easy to observe isthe abundance of data that exists both geographicallyand over time. This excellent data coverage makes itpossible to compute a climatology of SST conditions,which is a long-term average map of SST conditionsusually over the 12 month period of an annual cycle.Many different climatologies have been computed forSST, starting with strictly ship measurements beforeabout 1970 and a mix of satellite, ship, and buoy SSTssince that time. One must always be careful todetermine exactly which periods are covered by sucha climatology and what data went in to the compu-tation of the climatology.For this article we will use a climatology that

includes satellite infrared, buoy, and ship SST data. Aspart of this analysis, satellite SSTs were ltered to

eliminate obvious outliers. The input buoy and shipSST data sets were also ltered to remove large errors.All three of these input data setswere to save space.Weonly present four images to represent the 12 monthannual cycle. The rst is a global map of SST for themonth of January (Figure 2), which represents SSTconditions for the Northern Hemisphere winter.In this map a number of basic features are readily

obvious. The warmest temperatures are in the tropics,particularly in the Pacific and Indian Oceans. Thewarmest temperatures are found in the warm pool ofthe western tropical Pacific. There are cold featuresalong the west coasts of South America and SouthAfrica, which correspond to upwelling events mostactive in the austral summer. Cold temperaturesextend a bit farther north in the Southern Oceanthan they do in the Northern Hemisphere, which isprimarily a consequence of the open character of theSouthern Ocean compared with the geographicallyrestricted waters of the Arctic.The representative Northern Hemisphere spring

SST map in Figure 3 shows a modest shift in thistemperature distribution. The warmest temperatureshave increased, particularly in the equatorial IndianOcean, which has increased from about 281C to about301C. The Western Pacific Warm Pool has expandedslightly and increased in temperature. There is now a

90 N

60 N

30 N

0

30 S

60 S

90 S

Latit

ude

180 W 150 W 120 W 90 W 60 W 30 W 0 30 E 60 E 90 E 120 E 150 E 180Longitude

0C 4C 8C 12C 16C 20C 24C 28C 32CSea surface temperature

Figure 2 SST climatology for January. (Reproduced with permission from Reynolds and Smith, 1995.)


distinct band of warm temperatures in the equatorialPacific. The upwelling zones off southwestern SouthAmerica and South Africa have decreased in size. Thecolder SSTs have remained largely the same.Turning to the Northern Hemisphere summer, we

look at the July SST climatology in Figure 4. In thiscase the maximum temperatures in the tropics haveactually reduced slightly, reecting the decrease insolar insolation on a global level. The warm band inthe tropical Atlantic has weakened, as has the equa-torial warm band in the tropical Indian Ocean. Eventhe warm pool in the western tropical Pacific hasweakened in both magnitude and areal coverage. Theupwelling zones off western South America andwestern South Africa have again expanded, and bothshow warm tongues that extend westward out fromthe northernmost extent of the colder upwellingwater.There is a corresponding cold upwelling region offNorth America consistent with the seasonal shift tonortherly upwelling winds off that coast. The same istrue off northwest Africa.To complete the cycle we look at the Northern

Hemisphere fall (October)map inFigure 5. In thismapall of the west coast upwelling regions have weakenedslightly. In particular, the northwest Africa upwellingregion has disappeared. In the Southern Hemispherethe upwelling regions appear to have stretched farther

to the west. The equatorial warm regions haveweakened in magnitude, while they are about thesame in geographic coverage as they were in July. Thecolder waters have not changed much since the lastseason (summer).All of these maps have used a mixture of satellite

infraredmeasurements, drifting andmoored buoy SSTmeasurements, and ship SST measurements. In thisapplication the satellite skin SST is adjusted to matchcoincident drifting buoy SSTs, as introduced earlier.Thus, these maps really represent a pseudo bulk SSTdue to the overwhelming number of infrared SSTobservations as compared to the in situ buoy and shipobservations. Still, the general seasonal pattern of theSST is clearly apparent, and it is not likely that anadjustment to skin SSTwould show any substantiallydifferent SST patterns. SST patterns computed fromvarious satellite SSTalgorithms all look very similar; itis the absolute temperature value that is different andthe skin SST must be considered when addressingquestions such as airsea heat and gas exchange.

Skin SST

Due to its very high emissivity, the ocean is consi-dered to very nearly approximate a blackbody. This

90 N

60 N

30 N

0

30 S

60 S

90 S

Latit

ude

180 W 150 W 120 W 90 W 60 W 30 W 0 30 E 60 E 90 E 120 E 150 E 180 Longitude


Figure 3 April SST climatology. (Reproduced with permission from Reynolds and Smith, 1995.)


90 N

60 N

30 N

0

30 S

60 S

90 S

Latit

ude

180 W 150 W 120 W 90 W 60 W 30 W 0 30 E 60 E 90 E 120 E 150 E 180 Longitude


Figure 4 Mean July SST. (Reproduced with permission from Reynolds and Smith, 1995.)

Sea surface temperature32C28C24C20C16C12C8C4C0C

Longitude180 150 E120 E90 E60 E30 E030 W60 W90 W120 W150 W180 W

Latit

ude

90 S

60 S

30 S

0

30 N

60 N

90 N

Figure 5 Mean October SST. (Reproduced with permission from Reynolds and Smith, 1995.)


long-wave heat emission is directly proportional to theskin SST, which is the only SST that interacts with theoverlying atmosphere. Having a thickness of between5 and 10 mm, the skin of the ocean can easily bedestroyed by breaking ocean waves. When this hap-pens the skin reforms within 36 s, which means thatthe skin of the ocean is most generally present. Thisskin layer is the molecular boundary between theturbulent atmosphere and the turbulent ocean. It isthis skin layer that transfers heat, gases, and momen-tum between the atmosphere and the ocean. Thetemperature of this ultrathin layer can only bemeasured radiometrically, since any contact with theskin layer will disturb it. Thus, this layer cannot bemeasured directly by drifting or moored buoys or byany ship. Equipped with infrared radiometers, shipscould measure this skin layer temperature radiomet-rically without the attenuating atmosphere in betweenthe ship radiometer and the sea surface. Such meas-urements could provide validation and possibly cali-bration information for satellite-based infraredradiometric measurements of skin SST.An important question is: What is the precise

definition of bulk SST? Drifting buoys measure atemperature somewhere between 0.5 and 1.5m belowthe sea surface. This is a result of the buoys interactionwith the waves at the sea surface. As described earlier,ship cooling water intake ports may be locatedanywhere from 2 to 5m beneath the waterline. So isthe bulk SST located at 25m beneath the surface? Ifwe consider the temperature prole in the shallowupper layer of the ocean (Figure 6), we can see that theskin SST is always slightly colder than the temperaturejust below it.Note the logarithmic abscissa in this plot.At night this temperature slightly below the coolingskin layer is isothermal down to a few meters. In thiscase the bulk temperature could bemeasured at any ofthe depths down to5m.During the day, however, solarinsolation can heat up a shallow layer sufciently sothat the temperature of the cool skin is a little higherthan the isothermal layer below the warm diurnallayer. Often referred to as warm skin SST, thiscondition can only exist during daytime under rela-tively clear sky conditions when the shallow surfacelayer is heating.The relationship between the skin and the bulk

temperatures depends on two forcing factors: thewindand the net airsea heat ux. A good example of this isshown in Figure 7, which is taken from two differentoceanographic research cruises. Here we have plottedthe difference between the skin SST and the bulktemperature (taken as a temperature between 2 and5m) as a function of wind speed and net airsea heatux. The wind speed was observed from the researchvessel while the net airsea heat ux was calculated

from the traditional bulk formulas using routinemeteorological measurements also made from theship. At the top is a series from a cruise on the FSMeteor in the North Atlantic. Here, interestingly

0

1 mm

1 cm

1 dm

1 m

10 m

z

T T

1b1a

DAY

T < 0 T > 0

Figure 6 Temperature prole in the shallow upper layer of theocean.

0.600.500.400.300.200.100.00

400

300

200

100

0

400

300

200

100

0

2 46 8

10

2 46 8

10

T (K

)

0.600.500.400.300.200.100.00

T (K

)

Observed night time T from CEPEX

Observed night time T from F/S meteor

Wind (m s_1)

Wind (m s_1)

Heat flux (W m _1)

Heat flux (W m _1)

Figure7 Skin and bulk temperatures as a function of wind speedand net airsea heat ux.


enough, the DT between the skin SST and the bulktemperature both rises and falls with increasing windspeed. At low heat ux levels DT decreases as windspeed increases. This is consistent with traditionalwisdom that increasedmixing due towind stirringwillhomogenize the upper layer, eventually making skinand bulk temperatures the same. At higher heat uxvalues, however, DT actually increases as the windspeed increases. In the lower gure a data set wascollected from the tropical Pacific. A much morelimited change in DT is found for this tropical sample,but there is still a decrease inDT with increasingwind.These effects can be modeled with a semiempirical

formula that includes the effects of wind and net heatux on the DT difference between skin SST and bulktemperature:

DT QNrwcpk1=2

CshearnZ0un3

1=2(

Cconv nrwcpagQN

1=2"

CshearnZ0un3

1=2#e Rfcr=Rf0

)1=21

In eqn [1],QN is the net heat ux, rw is the density ofwater, cp is the specific heat capacity of sea water at aconstant pressure, k is the thermal diffusivity of water,Cshear is the proportionality constant for shear-driventime scale, Cconv is the proportionality constantfor convective driven time scale, n is the kinematicviscosity, z0 is the momentum roughness length, u

n isthe friction velocity of water, g is the acceleration dueto gravity, a is the thermal expansion coefcient,Rf0 isthe surface Richardson number, and Rfcr is the criticalRichardson number.This equation connects the skin SST with the

temperature just below it, but does not account forthe further transfer of heat downward or upwardwithin the water column. To do this a mixed-layermodel must be added to the skinbulk parametriza-tion. Such a model has been employed to show thedevelopment of the full upper layer temperatureprole. Using this model combination it was possibleto trace the temperature history of the upper layer for acouple of consecutive days of eld measurements.There is excellent agreement between the observedbulkskin DT temperature difference observed andthatmodeled using thewind speed andnet airsea heatux. Thus, it appears that this combination of modelscan go a long way to simulating the temperaturebehavior of the upper layer of the ocean. It should benoted that these models do not account for surfacewave effects, or the effects of turbidity and foamon the

sea surface. It is likely that these effects are smaller inmagnitude than those driven by the wind and net heatux.Muchmore sophisticatedmodels will be requiredto include these phenomena in modeling DT.

In situ Measurement of Skin SST

Since infrared satellite sensors are only able tomeasureradiation from the skin of the ocean, the challenge is toprovide in situmeasurements of the skin SST that canbe used to calibrate/validate the satellite radiances interms of temperature. As mentioned earlier, ships canbe equipped with radiometers to measure directly theskin SST without the atmosphere attenuating theinfrared signal. In principle these same radiometerscould be installed on moored buoys to continuouslymeasure skin SST. The problem with both of theseinstallations is that the radiometer optics must beprotected from sea spray, which is difcult to do in anautonomous installation. At least on the ship theradiometer can be examined each day and cleaned offto maintain a clear optical path.Another requirement of these radiometers is that

they are verywell calibrated. The best approach to this

Figure 8 Radiometer requirements.


requirement is to equip the radiometers with twoblackbodies with one at ambient temperature and theother heated a few degrees above ambient. To correctfor reected infrared sky radiation, these radiometersmust look up and down to view the sky, the ocean, andboth blackbodies each scan. The easiest way toimplement these requirements is to use a rotatingmirror to channel the radiation to the detector, as seenin Figure 8. In this setup the sensor is a lowcostthermal infrared thermometer with a supplementalrotating mirror to collect radiation from the skythrough the top hole, the sea surface through thebottom hole, and both of the blackbodies. In this caseno special efforts have been taken to protect themirrorfrom sea spray. The addition of a cover that wouldmove into place in cases of high seas and/or rainfall hasbeen considered. Other protection schemes use amoving plastic wrap window that would carry thesea spray or rainfall away from the radiometer view.One suggestion was to use windshield wipers fromautomobile headlights. It will be a challenge to seewhich of these methods is successful in protecting theradiometer from sea spray and rainfall.It is difcult to predict just how many ships need to

be operating to provide the global coverage needed toregularly calibrate and validate the satellite radio-

meters. It is clear that the need for continuing samplingmeans that the ships must be merchant ships travelinglong routes on a regular basis. A plot of the presentship coverage for a year (Figure 9) reveals that thereare plenty of ships to choose from in the NorthernHemisphere, but in the Southern Hemisphere theavailable selection is much more restrictive. Shiptracks that make the long transits to Australia andNew Zealand are good candidates, as are those routesfrom Asia to southern Chile. Only experience willreveal just how many ships over which routes will berequired to supply the in situ skin SSTs needed tocalibrate/validate the satellite infrared radiances.

Summary and Conclusions

SST is andhas been one of themostmeasured variablesin the ocean, and as such it has receivedmuch scientificattention. Using a combination of ship SSTs, mooredand drifting buoy SSTs, and satellite infrared SSTs, wehave characterized the global and seasonal patterns ofSST. Realizing that the satellite infrared SSTs can onlyrepresent the SST of the 10 mm thick skin layer of theocean, future efforts should be aimed at separating thesatellite skin SSTs from the 15m deep bulk SSTmeasured by the buoys and ships. The skin layer is

Stddev = 8.45275Mean = 18.7291 Num points = 1132447

Figure 9 Global merchant ship routes for 1996.


simply the molecular layer that interfaces between aturbulent ocean and a turbulent atmosphere. Thetemperature difference between the skin and the bulktemperatures is directly related to the wind speed andnet airsea heat ux. Thus, an improved understand-ing of this relationship can help us to better resolve thenet heat andmomentum uxes between the ocean andthe atmosphere. In part, this understanding and theshift to the computation of skin SST from the satelliteinfrared data depends on the creation of a network ofship-of-opportunity based skin SST radiometers col-lecting global and continuous samples of skin SSTground truth data (i.e., without an interveningatmosphere).

See also

AirSea Interaction: Freshwater Flux; Gas Exchange;Momentum, Heat and Vapor Fluxes; Storm Surges; Sur-face Waves. Global Change: Surface TemperatureTrends. Observation Platforms: Buoys. Observationsfor Chemistry (Remote Sensing): IR/FIR. SatelliteRemote Sensing: Temperature Soundings.

Further Reading

EmeryWJ and Yu Y (1997) Satellite sea surface temperaturepatterns and absolute values. International Journal ofRemote Sensing 18: 323334.

Emery WJ, Baldwin DJ, Schluessel P and Reynolds RE(2001) Accuracy of in situ sea surface temperatures usedto calibrate infrared satellite measurements. Journal ofGeophysical Research 105: 23872405.

Emery WJ and Baldwin D, Schluessel P and Reynolds RE(2001) Accuracy of in situ sea surface temperatures usedto calibrate infrared satellite measurements. Journal ofGeophysical Research 23872405.

McClain EP, Pichel WG,Walton CC, Ahmad Z and Sutton J(1983) Multi-channel improvements to satellite-derivedglobal sea surface temperatures. Advances in SpaceResearch 2: 4347.

McClainEP, PichelWGandWaltonCC (1985)Comparativeperformance of AVHRR-based multichannel sea surfacetemperatures. Journal of Geophysical Research 90:1158711601.

Reynolds RW and Smith TM (1994) Improved global seasurface temperature analysis using optimum interpola-tion. Journal of Climate 7: 929948.

Reynolds RWand Smith TM(1995)A high resolution globalsea surface temperature climatology. Journal of Climate8: 15711583.

Richardson PL (1979) The Benjamin Franklin andTimothy Folger charts of the Gulf Stream. In: Sears Mand Merriman D (eds) Oceanography the Past,pp. 703717. New York, Heidelberg, Berlin: Springer-Verlag.

Walton CC, Pichel WG, Sapper JF andMay DA (1998) Thedevelopment and operational application of nonlinearalgorithms for the measurement of sea surface tempera-tures with the NOAA polar-orbiting environmentalsatellites. Journal of Geophysical Research 103: 2799928012.

Wick GA (1995) Evaluation of the Variability and Predict-ability of the BulkSkin Sea Surface TemperatureDifference with Application to Satellite-MeasuredSea Surface Temperature. PhD thesis, University ofColorado, Boulder, CO.

Storm Surges

R A Flather, Proudman Oceanographic, Prenton, UK

Copyright 2003 Elsevier Science Ltd. All Rights Reserved.

Introduction and Definitions

Storm surges are changes in water level generated byatmospheric forcing, specifically by the drag of thewind on the sea surface and by variations in the surfaceatmospheric pressure associated with storms. Stormsurges last for periods ranging from a few hours to2 or 3 days and have spatial scales that are large com-pared with the water depth. They can raise or lowerthe water level in extreme cases by several meters;a raising of level is a positive surge, a lowering anegative one. Storm surges are superimposed onthe normal astronomical tides generated by variationsin the gravitational attraction of the Moon andSun. The storm surge component can be derived

from a time series of sea levels recorded by a tide gaugeusing

surge residual observed sea level predicted tide level 1

producing a time series of surge elevations. Figure 1shows an example.Sometimes, the term storm surge is used for the sea

level (including the tidal component) during a stormevent. It is important to be clear about the usage of theterm and its significance to avoid confusion. Stormsalso generate surface wind waves that have periods oforder seconds and wavelengths, away from the coast,comparable to or less than the water depth.Positive storm surges combined with high tides and

wind waves can cause coastal oods which in terms ofthe loss of life and damage are probably the mostdestructive natural hazards of geophysical origin.

AIRSEA INTERACTION / Storm Surges 109

Documents

Air-Sea Interaction Sea Surface Temp