JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 96, NO. C6, PAGES 10,467-10,486, JUNE 15, 1991

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 96, NO. C6, PAGES 10,46710,486, JUNE 15, 1991 Smultaneous Observatons of Ocean Surface Wnds and Waves by Geosat Radar Altmeter and Arborne Synthetc Aperture Radar Durng the 1988 Norwegan Contnental Shelf Experment NELLY M. MOGNARD, JOHNNY A. JOHANNESSEN, CHARLES E. LIVINGSTONE,SDAVID LYZENGA, 4 ROBERT SHUCItMAN, 4AND CATHY RUSSEL 4 Quassmultaneous measurements of the ocean surface by the Canada Centre for Remote Sensng (CCRS) Convar580 (CV580) synthetc aperture radar (SAR) were acqured on March 20, 1988, along an ascendng pass of the Geosat radar altmeter as part of the prelaunch ERS1 Norwegan Contnental Shelf Experment (NORCSEX '88). Over a regon where the SAR look drecton s parallel to the wnd vector, a relatonshp between the ocean scatterng crossectons measured by the Geosat altmeter and the SAR s obtaned. In the regons where the wnd drecton s changng, estmate of the wnd drecton s derved from the dfferences measured between the altmeter and the SAR scatterng crossectons. Wavelength and wave drectons derved from the SAR wave spectrare n good agreement wth the sea truth data obtaned wth the wave drectonal buoys and wth the altmeterderved swell estmates. Sgnfcant wave heght measured by the Geosat altmeter s compared wth the buoy measurements and s used to assess the valdty of the sgnfcant wave heght deduced from the SAR. 1. INTRODUCTION other one. For nstance, the ocean scatter cross sectons from the altmeter are used to derve the wnd drecton from the On March 20, 1988, as part of the Norwegan Contnental Shelf Experment (NORCSEX'88) prelaunch ERS1 wndwavecurrent nvestgaton [NORCSEX'88 Group, 1989], the Canada Centre for Remote Sensng (CCRS) Convar580 (CV 580) synthetc aperture radar (SAR) flew a data acquston pattern over the Haltenbanken regon of the Norwegan contnental shelf. The purpose of ths flght was to provde nearsmultaneous supportng data for ocean surface measurements made durng an ascendng pass of the Geosat altmeter. To our knowledge, ths s the frst tme that concdent observatons of the ocean surface from satellte radar altmeter SAR along part of the track. The altmeter sgnfcant wave heght measurements are also used to assess the valdty of the SAR sgnfcant wave heght determnaton. Fgure! dentfes the major features of the arcraft data set, the locaton of the four drectonal wave measurement buoys that montored sea state and atmosphercondtons throughout the NORCSEX '88 experment, and the locaton of two drllng platforms used to provde ancllary surface and meteorologcal data. The SAR swaths n ths dagram are 22 km wde and span the ncdence angle range from 0 ø (arcraft nadr) to 74 ø. W polarzaton was used for mage formaton. The frst lne (1/1) was synchronzed wth the Geosat and arborne SAR have been acqured quassmultaneously overflght tme so that the arcraft and spacecraft data sets along a track several hundred klometers long (Fgure 1). Sea state parameters such as wnd speed, sgnfcant wave heght (SWH), and wavelength are nferred from the radar would be concdent tme and space n the vcnty of buoy 4. To provde data for modelng expected correlatons between future satellte altmeters and spaceborne SARs, lne 1/1 was observatons and are compared wth correspondng parameters postoned so that the center of the Geosat altmeter track from surface observatons. The measurements from one radar corresponded to the SAR mage at 30 ø ncdence angle. The are used to mprove and complement the capabltes of the second lne (2/2) retraced 1/1 from the seaward end to measure shortterm, temporal varatons of the sea surface roughness and to provde data needed to resolve the wave propagaton drecton ambguty nherent n SAR 1 Centre Natonal d'etudes Spatales, Toulouse, France. 2 Nansen Envronmental Remote Sensng Center, Bergen, Norway. measurements of waves. Durng lne 2/2 the radar look drecton (wth respecto the sea surface) of lne 1/1 was mantaned. Flght lnes 3/3 and 4/4 were flown to provde 3 Canada Centre for Remote Sensng, Ottawa. multple aspect angle data n the vcnty of buoy 4 and to lnk the arborne measurements to the buoy 1 data set. 4 Envronmental Research Insttute of Mchgan, Ann The smultaneous arborne SAR and Geosat altmeter data Arbor. Copyrght 1991 by the Amercan Geophyscal Unon. Paper number 91JC /91/91JC ,467 were acqured durng a local storm that started on March 19 wth strong wnds blowng from the southwest along the coast. Durng the Geosat pass at 0800 UTC on March 20, an occluded front was located over the ocean, parallel to the Norwegan coast.

2 10,468 ]VIOONARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 2/2 400' Fg. 1. Map of the March 20, 1988, experment area showng the SAR swaths, surface measurement platforms, and major features of the SAR data set overlad on the bathymetry contours of the test area. Large sold arrows dentfy the flght lne numbers and the flght drectons. (Lne 2/2 ends at "A".) Small crcled numbers 1 to 4 dentfy the wnd front locatons (dashed lne) observedurng flght lnes 1/1 to 4/4. Numbered crcleshow the locatons of wave scan buoys 1 to 4. The open square represents the drllng platform West Delta; the sold square s the drllng platform Polar Poneer. B dentfes regons of wave refracton; H dentfes the "hgh wnd speed" area. Small sold arrows ndcate the wnd drecton observed from wnd streaks n the SAR data, openheaded arrow show the calculated wnd drecton on the "low speed" sde of the front, and dashed arrows show the wnd drecton calculated from combned SARGeosat data. The dashed lne along the edge of the lne 1/1 swath s the center of the Geosatrack.

3 MOONARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 10,469 The SAR magery obtaned March 20 s characterzed by welldefned wave felds throughout all passes and systematc slow varatons n radar return wth sudden dscontnutes at the crossng of the frontal feature n the vcnty of buoy 4 (marked by 1, 2, 3, and 4 n Fgure 1). Ths front s very sharply defned and s seen by the SAR at all ncdence angles. Ths feature s not assocated wth any clear surfac evdence of a current boundary n the ocean surface layers. Smlar features observedurng other NORCSEX flghts were probed by the research vessel H kon Mosby and were found to be sea surface manfestatons of rapd wnd shfts [Johannessen et al., ths ssue]. The domnant swell feld observed n the SAP, magery came from the southwest and was augmented by a wnd sea especally to the east of buoy 4. The complex meteorologcal and oceanographc condtons that were present durng the SAP, flght are descrbed n secton 2. The next secton ntroduces the SAP, system and the altmeter sea state measurements. The SAP, and the altmeter ocean scatterng cros secton are analyzed n secton 4 and, based upon the wnd speed determnaton wth the altmeter, wnd vectors are derved from the SAR crosssecton measurements over part of the flght when the wnd drecton was changng. The next secton presents the wave analyss from the radar altmeter and the arborne SAR and s followed by a dscusson and summary of the results. 2. THE METEOROLOGICAL AND OCEANOGRAPHIC CONDmONS The altmeter and SAR tracks crossed two drectonal wave measurng buoys (Wavescan) provdng tme seres of wave heght, wave propogaton drecton and wnd speed at regular ntervals of 3 hours (Fgure 1). At these buoy postons, two current meter moorngs were also deployed recordng the subsurface current and temperature feld from 25 m below the surface to wthn 25 m of the bottom. Meteorologcal data (at 3hour ntervals) from all surface platforms n the measurement area except buoy I show the passage of a storm system through the area over a 9hour perod endng durng the SAR data acquston. Near the common ntersecton of the SAR passes, the drllng platform West Delta and buoy 4 report a large drop n wnd speed over the reportng epoch endng at 0900 UTC. Wnd drectons reported by buoy 4 [Barstow and Berken, 1988] are anomalous n the context of the rest of the data set, probably owng to the applcaton of the magnetc declnaton correcton n the wrong sense. When the data are adjusted for ths error, the surface data set s nternally consstent. Durng the evenng of March 19, all four buoys measured maxmum sgnfcant wave heghts, whch slowly decreased untl approxmately the tme of the flght, durng whch condtons changed rapdly due to the passage of a storm. Pror to and durng the flght, the man wave propagaton drecton measured by the buoys came from the southwest wth perods varyng between 11.8 and 12.8 s correspondng to wavelength between 220 and 250 m. These waves resulted from a storm whch occurred on March 19 off the east coast of Scotland. The storm was characterzed by strong wnds blowng from the southwest along the Norwegan coast. The reported arsea temperature dfferences ndcate that neutral stablty condtons at the arsea nterface are satsfed over the experment area. The water temperature the drllng platform West Delta s 2 ø warmer than buoy 4 (30 km westnorthwest), ndcatng the presence of a frontal structure. The mean oceanographc condtons n the SARaltmeter crossover regon are schematcally summarzed n Fgure 2. The bathymetry along the track shows consderable varatons; on the broad contnental shelf wth a wdth of about 250 km the mean depth s 300 m. West of the shelf break, whch runs northward, the bathymetry deepens rapdly nto the Norwegan Sea wth a mean depth of 3000 m. The presence on the shelf of a seamount rsng to approxmately 100 m below surface delmts the western boundary of a narrow, nearshore channel, 50 km wde, runnng northeastward parallel to the coast. The general ocean crculaton s substantally steered by these sgnfcant bathymetrc features [Haugan et al., ths ssue]. The relatve cold and fresh Norwegan Coastal Current (NCC) forms a strong baroclnc jet n the channel wth a maxmum speed of 0.70 m/s drected northeastward. The man flow of the North Atlantc Norwegan Current (NANC) s steered northward along the outer shelf break at a mean speed of 0.25 m/s. Between these two currents, a weak barotropc eddy flows dockwse around the seamount. Ths predomnant mesoscale crculaton system can generate dfferent wavecurrent and wavebottom refracton patterns, whch are observed at buoys 1 and 2 and are further dscussed by Haugan et al. [ths ssue]. 3. THE GEOSAT ALTIMETER AND THE CCRS CV580 SAP, 3.1. The Geosat 41tmeter Measurements The Geosat satellte was n a nearly crcular orbt wth an nclnaton of 108 ø at an alttude of 800 km. The satellte carred a shortpulse (3.125 x 109s) nadrvewng radar altmeter operatng at 13.5 GHz. The geophyscal measurements obtaned from the radar altmeter were as follows: (1) the alttude of the altmeter above the ocean surface derved from the tme delay between transmsson and recepton of the radar altmeter sgnal, (2) the sgnfcant wave heght computed on board the satellte from the slope of the mean return waveform leadng edge, and (3) the ocean surface wnd speed deduced from the ocean back scatter coeffcent determned from the automatc gan control (AGC) loop used to normalze the ampltude of the ocean return sgnal. The Geosat radar altmeter system s descrbed by MacA hur et al. [1987]. For the measurements of sea surface elevaton and SWH, the effectve footprnt sze on the sea surface vares from 2.4 to 11.6 km dependng on the sea state. The ocean backscatter coeffcent s derved from the AGC, whch drves the sum of 63 range gates of the return sgnal to a constant value. Sxty of the sample gates are separated by ns wth three gates halfway between gates 29, 30, 31, and

4 10,470 MOONARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 20 cm/s 300m 66 ø I Current ncrease along shelf break. Waves propagate along the shelf break. Clockwse rotatng eddy. SAR observes rapd wnd shft. 65 ø 200m Weak flow around sea mount. CMT 2 64 ø U Strong and jet lke current. SAR observes wavecurrent refracton, 63 ø R6 Fg. 2. Measured current drectons and magntudes at 25m depth n the experment area. Statons CMT1 to 3 reported current vector data throughout the experment perod. The current vectorshow the condtons present on March 20, For reference, the swath maged by the frst flght lne of the arborne SAR s shown as a double lne. Some features of nterest (crcled numbers) are as follows: (1) The SAR mageshow a wave refracton event n the vcnty of the shelf break. A strong current s known to exst here. (2) A clockwse rotatng eddy has been observed n the vcnty of the edge of the bank. Supportng evdence s found n the CM3 current vector. (3) A weak flow has been observed around the peak of the Haltenbanken sea mount. (4) A strong and jetlke (spatally confned) current n the deep, nearshore channel. Ths feature has been observed by both the currentmeter at staton CM2 and the SAR. In the SAR magery the surface manfestaton of the current jet s a localzed regon of wave refracton. The Geosat altmeter measuresgnfcant rms heght anomales. 32. The alttude tracker n the altmeter electroncs package centers the mean sea level at gate The footprnt of the sphercally expandng altmeter pulse projected on the mean sea level s determned by the area covered over the tme nterval from gate 30.5 to gate 60 (a total of ns). Ths gatelmted footprnt corresponds to a crcle wth a dameter of 9.5 km. The SWH and ocean crosssecton measurements are acqured along the satellte track every second, (over a dstance of 6.7 km) and are used for sea state analyss The CCRS CV580 SAR System Durng NOR CSEX '88 The CCRS CV580 C (5.30 GHz) and X band (9.25 GHz) SAR system was used as a data acquston tool durng the NORCSEX '88 flght program from March 11 to March 21, Durng ths perod, the C band SAR was nearng the end of ts commssonng phase, but the X band was newly nstalled and provded ts furst extensve ocean magery. The radar systems are descrbed n detal by Lvngstone et al. [1987, 988]. Both C and X band radars employ selectable transmtter polarzaton and smultaneous phase coherent, dualpolarzed recevers to allow the acquston of smultaneous and crosspolarzed mages. Both antennas are carred on the same twoaxs (azmuth/elevaton) controlledrve, they vew the same terran and cannot be ponted ndependently. For the NORCSEX experment, both radars were operated exclusvely n a Wpolarzed sngle channel mode for ocean measurements. The wde swath magng mode (18 m range resoluton, 61km swath at 15m range pxel separaton) was

5 ..... MOONARD ET AL.' SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 10,471 used for nvestgaton of largescale sea surface structure, and the nadr mode (5.7m range resoluton, 16.4kn swath at 4m pxel separaton) was used to measure ocean waves. The nadr mode geometry allows magng from ncdence 0 ø to 74 ø, and the realtme processed mage s normally recorded n the slant range plane to mnmze feld errors arsng from errors n arcraft atttude estmaton ? lo.o l I 1 # JDED FRONT ALTIM R AND SAR OCEAN SCATTERING CROSS SECTION AND WIND PARAMETER MEASUREMEN7 z The Altmeter Measurements 9.0 The altmeter antenna measurespecular returns from the nadr sea surface. On board Geosat, the ampltude of the oceanreturn sgnal s normalzed va the AGC loop. Properly calbrated, the AGC settng measures the backscatter coeffcent at the ocean surface that n turn s dependent on I I I I I Ill Illl wnd speed at the ocean surface. The scatterng cros secton LATITUDE of the ocean surface at the nadr can be modeled by specular (degrees N) ponts assumng that the sea surface slopes are nearly Gaussan Fg. 3a. Geosat altmeter measurements of normalzed radar and sotropc n ther dstrbuton [Barrck, 1974]. Several wnd speed algorthms have been developed for satellte altmeters. A comparson of these algorthms by scatterng crosssecton o ø, at nadr, along the SAR swath of lne 1/1 (supermposed s the locaton of the atmospherc front). Dobson et al. [1987], shows that the algorthm frst developed wth Geos3 [Mognard and Lago, 1979; and Brown, 1979] and later fnetuned wth the NOAA buoys for the Seasat altmeter 25 "LI! I I I I I I! I 'rr'lt! ' '"l '1 'd'l'rltirtr [Brown et al., 1981] yelds the smallest rms dscrepancy (1.7 c' = : GEOS,T 0oooo BuOY S m/s) and overall bas (0.5 m/s) for the 7month comparson of 06:00 (4) Geosat wnd speed data to the Natonal Data Buoy Center 0 network. Ths comparson used 5s averages of Geosat data. A study wth Geosat of the algorthms performances over varous meteorologcal stuatons (cold, warm, and occluded fronts, sharp sobarc gradents, antcyclones) showed that the dfferent algorthms measure qualtatvely smlar wnd speed gradents over the ocean but somewhat dfferent quanttatve values [Mognard et al., 1987]. The Geosat data used n ths OCCLUDED FRO'4T paper were processed wth the Brown algorthm [Brown et al., 1981] whch yelds valdated wnd speeds from 0 to 21 m/s. The slghtly modfed Brown algorthm [Goldhrsh and "uo:uu (2) Dobson, 1985] has not been valdated for wnd speeds hgher O than 14 m/s and s thus not used here. The varatons of the altmeter ocean crosssecton along wth the derved wnd speed are presented n Fgures 3a and 3b. The altmeter senses the varatons n speculareflecton along the track and thus responds to changes n wnd speed that are opposte to the SAR responses. For nstance, the Fg. 3b. Varatons of the Geosatderved wnd speed along poston of the front s sharply defned by an ncrease n the lne 1/1, front locaton, and buoy wnd speed measurements altmeter ocean cros secton of 0.6 db (Fgure 3a) that over the 3hour perod close to the Geosat overpass, at 0600 corresponds to a decrease n wnd speed from 12 to 10.5 m/s and 0900 UTC, or 0800 and 1100 UTC. (Fgure 3b). The Geosat pass starts north of buoy 2 whch has a sheltered poston near the Norwegan coast. On March 20, buoy 2 measured a wnd speed of 8.5 m/s at 0500 UTC and 8.9 m/s at 0800 UTC, wth a stable wnd drecton from the southsouthwest. A local fetchlmted sea s present at buoy 2, where the wnd speed measurements are n good agreement wth the Geosat wnd speed of 9.3 m/s at the start of the satellte pass, slghtly west of buoy 2. In the area between buoys 2 and 4, the altmeter wnd speed frst ncreases steadly to 13 m/s to reach a plateau (between the lattudes 65.0øN and 65.3øN) where the Geosat wnd speed oscllates between 12 : 0 5:00 (2) 06:00 09:00 I 09:00 (4) I I I I I I I o, I I I o I I I I I I I I I I I LATITUDE (degrees N)

6 10,472 MOGNARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS and 13 m/s (Fgure 3b). At the tme of the Geosat overflght cross secton of the sea surface wll be some functon of the an atmospherc front was passng over buoy 4. The wnd speed wnd velocty V, the radar wavelength X, the ncdence angle 0, record from the buoy shows the wnd decreasng from 20.5 and the wnd azmuth angle p(measured from the surface m/s at 0600 UTC to 5.7 m/s at 0900 UTC; by 1200 UTC, the projecton of R). wnd speed has agan ncreased (to 10.7 m/s) and the wnd drecton has shfted to 150 ø. From arborne scatterometer measurements of the PROMESS experment n 1984, and from Nordsee tower measurements, Attema et al., [1986] have derved a scatterng 4.2 SAR Measurements cros secton model for C band, Wpolarzed radars. The model uses wnd measurements at 10 m above the sea surface Background. At ncdence angles near nadr, the and s vald for a neutrally stable boundary layer, all azmuth radar reflectvty of the sea surface s domnated by specular angles, and ncdence angles n the 20* to 50* range. reflectons. Ths s the Geosat altmeter measurement regme, From Attema et al. [1986], the emprcal normalzed and t also governs SAR returns from the sea surface at scatterng cros secton of the sea surface (n meters per ncdence angles less than 20*. square meter, not decbels) s At larger ncdence angles the radar returns are governed by Bragg scatterng from surface structures whose slant range projected wave lengths are of the order of half the radar wave length. Smallscale waves can, n turn, be related to the wnd oo(v,o,) = c(o)v(o) 1 +/ (0, + V)cos, v) + 2(0, frcton velocty V* at the sea surface and to the stablty of the ar sea nterface. (Donelan and Person [1986] show that the The varables V, 0, o have been defned prevously and the wnd speed gradent at the sea surface s a physcally more satsfyng parameter to use. It s not, however, accessble from coeffcents C, 'Y, bl, b 2 are lsted n Table 1. The angle of qo s the March 20, 1988, NORCSEX data set and would also need equal to 0 when the radar beam s drected nto the wnd. The to be estmated from models.) model o ø of equaton (2) s ambguous wth respect to the sgn Followng PanofsM [1963], the frcton velocty can be related of,p. When the wnd speed s suffcently hgh, wnd streaks to the measured wnd speed V at heght z above the surface by appearng n the SAR magery can be used to resolve the,p ambguty. The ambguty can also be resolved by magng a regon of the ocean at multple aspect angles f the wnd V(z) = V*'xJO.O41Inz/zo, ] velocty s temporally stable over the measurement perod. where z 0 s a roughness length and p s the arsea nterface In equaton (2) the term o ø = C(O)V (O) s the normally used power law expresson for the radar cross secton of the sea stablty functon. surface. Keller et al. [1989] note that there s no present Investgatons by Keller et al. [1989] based on 1984 data from the research tower "Nordsee" show that when the ar theoretcal justfcaton for ths functonal form Expermental results. SAR mage transects for the temperature dfference AT = rarrse a s less than2øc, the March 20, 1988, VVpolarze data were constructed by arsea nterface s unstable and both the C band VVpolarzed extractng radar returns correspondng to the 25 ø to 35 ø scatterng crosssecton at 45 ø ncdence and V* are hgh. 2øC ncdence angle range, averagng these n range (angle), and < AT < 2øC defnes the neutral stablty regme. For AT > 2øC the C band radar cross secton and V* decrease wth then averagng the results n 96m blocks along the flght track. The lne 1/1 transect, shown n Fgure 4, les near the center ncreasng AT wth the rate of ncrease smallest at hgh wnd lne of the Geosat altmeter measurement profle. Snce each speeds. For frcton veloctes less than 0.2 m/s, Keller et al. pont along the transect s computed from a large number note a dependence of radar cros secton on rms wave slope (3800) of ndependent samples, SAR speckl effects can be whch vanshes at hgher frcton veloctes. No relatonshp gnored. The hgh (spatal) frequency components n Fgure 4 between radar cros secton and water temperatures (predcted vary n ampltude by up to 1.2 db over tme less than Geosat by Donelan and Person [1986]) was found. altmeter reportng nterval and measure the local varatons of The dependence of V* on surface currents was not o ø due to wnd and wave features n the data. Smlar transects nvestgated by Keller et al. The relatonshps between the wnd stress vector V*, the surface current vector U, and the were constructed for lnes 2/2 and 4/4. surface current shear are expected to affect the capllary wave A typcal plot of the SAR cros secton as a functon of spectrum and thus the radar scatterng crosssecton of the ncdence angle oø(0), and the correspondng Geosat o 0 ocean. Whle nadrlookng radars provde nondrectonal measurements of sea surface roughness, SAR measurements are dependant on the vectorelatonshps between the wnd estmate s shown n Fgure 5 for a hghreturn porton of the SAR mage where the wnd vector was upwnd (maxmum return) compared wth the SAR look drecton. Ths curve spans both the specular and quasspecular scatterng regme stress V* (and thus V) and the radar range R to the measured (0 < 10 ø) correspondng to the Geosat observatons and the element. Settng asde the surface current effects on the generaton and dsspaton of capllary waves, the normalzed scatterng dffuse scatterng regme (20 ø <0 <70ø). The scatterng cross sectonshown n Fgure 5 are comparable to measurements by Daley dscussed by Valenzuela [1978]. +/ 2(0,V)co v) j (2)

7 , MOONARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 10,473 TABLE 1. ESA Wnd Model Parameters From Attema et al. [1986] Parameter Value 00(6,,, v) c(o) 3`(0) t,l(o, v) C(O)V7(O)(1 +bl(o,lo.cos +b2(o,lo.cos2,p)/(1 +, V) + b2(o, V)) 10( '302) (U c 4 Ud)/(UcU d + 2u d + Uc) b2(o, V) 1 4Ud/(UcU d + 2u d + Uc) At 30* Incdence Angle C '3 (29.53 db).0538 bl(v) b2(v) 2 4( V)/( V '4V 2) 14( V)/( V '4V 2) Here, u c s the upwnd/crosswnd rato (u c = ( ) V), u d s the upwnd/downwnd rato (u d = ( ' ) I/. V s the 10m wnd n meters per second, 0 s the ncdence angle n degrees, and,p s the azmuth angle n degrees. 16 March 20, 1988 Lne 1/1 SAR Cross Secton at 30 Degrees '2.2 ' 6'.4 '6' '5 " ' 65' Lattude n Degrees Fg. 4. SAR measurements of the normalzed radar scatterng crosssecton at 30 ø ncdence angle, 00(30ø), for flght lne 1/1. Each data pont n ths graph s the average of 3800 SAR mage pxels, and thus speckle nose s not sgnfcant here. The nterval AA defnes the regon of overlap of the SAR and Geosat data sets. In the common data regon, the nterval from 64.6øN to 65.3øN shows the effect of varyng wnd speed when the SAR beam looks drectly nto the wnd. The sharp decrease near 65.3øN s an atmospherc front at whch both the wnd speed and drecton change over less than 100 m. The fne structure mmedately followng the front can be nterpreted ether as wnd speed and drecton varatons n the center of the lowpressure regon bordered by the front or as evdence of a current shear known to be present n the area. From 65.5øN to 66.1øN wnd speed and drecton both change. Pror to the start of the nterval AA, the 1.5dB step n o ø near 64.5øN s assocated wth the SAR sgnature of the nearshore current jet referred to n Fgure 2.

8 10,474 MOONARD ET AL.: SATELLITE AND MRBORNE SEA STATE OBSERVATIONS 1o 5 lo 15 o E o o lo o Go 70 8o Incdence Angte In Degrees Fg. 5. SAR measurements of o 0 (0)(n db) for ncdence angle range of the March 20 SAR mages. The sold curve s the best ft thrdorder polynomal. The sold nverted trangle s the Geosat altmeter measurement of o ø for the same segment of the ocean surface. When the 200 to 50 ø ncdence angle nterval of Fgure 5 s compared wth European Space Agency (ESA) wnd model predctons of equaton (2) (converted to decbels) at 12.5m/s wnd speed (measured by Geosat for the Fgure 5 data block), the constant term 101og10 C(O) must be reduced from 18.8 db to 5.0 db to match the two o ø estmates to wthn db. The adjusted wnd scatterng model underestmates nadr measurements of o ø by 3.8 db (Geosat) and 8 db (SAR). When the model s left unscaled, t overestmates the nadr crossectons by 15 db (Geosat) and 10.8 db (SAR). For the remander of ths paper the scaled verson of the model wll be used. When Fgure 4 s examned n the context of the correspondng SAR mage, three dstnct regons are evdent over the pass segment common to both SAR mage and the Geosat altmeter Regon 1: Wnd streaks n the SAR magery for the frst regon (lattude 64.55øN to 65.35øN along the pass) show that the wnd drecton s constant wth respecto the SAR look drecton and s orented parallel to the ground projecton of radar range vector. From observatons n lne 4/4 the radar s lookng unambguously upwnd. The scatterng cross secton measured by the SAR and the altmeter are lnearly related n the power doman by O0Geosat =218 soar(30 ø) (3) The data cluster n logarthmc form (db) and equaton (3) are shown n Fgure 6. The pont scatter n Fgure 6 llustrates the dfferences between the SAR and Geosat altmeter responses to smallscale varatons n the surface roughness of the ocean. Contrbutng factors are dfferences between the scatterng processes responsble for the Geosat altmeter (specular scatterng) and SAR (dffuse scatterng) measurements and dfferences between the szes of the regons averaged to yeld each measurement pont (approxmately 55 km 2 for Geosat and 0.2 km 2 for the SAR) Regon 2: The second regon conssts of a sharply defned atmospherc front (O0SAR(30 ø) decreases 7 db n less than 30 m; Fgure 7), a regon of reduced scatterng cross secton and a further decrease of 2 db 5 km further down the SAR vs GEl]SAT SIGMA O, MAR 20 L1 CONSTANT /IND DIRECTION 10.5 œ 9.5.p o, ;20 I I I I I SAR Sgma 0 at 30' Incdence, db Fg. 6. Scatter plot of the SAR and Geosat o ø measurements pror to the wnd front. The sold curve s the best ft straght lne to the data n the power (as opposed to logarthmc) doman.

9 MOGNARD ET AL.' SATELLITE AND AIRBORNE SEA STATE OBSF. RVATIONS 10,475 Norcsex March 20 Pass 1 na, o I Azmuth I I I km Fg. 7. SAR mage secton for lne 1/1 showng the "dscontnuty" n the ocean reflectvty across the atmospherc front. pass. The sudden change n radar cros secton can be attrbuted to a sudden change n both wnd speed and drecton at the front. From equaton (2), the change n wnd speed can be expressed n terms of radar crosssecton change as ' vhcre V 1 and 1 correspond to, r 2 and 2 correspond to a2 0, and, ar expressed n decbels. A smlar approach s used n Johannessen et al. [1990]. An teratve soluton to equaton (4), takng nto accounthe wnd data from the drllng platform West Delta at 0900 UTC, [ 0%ø/ ø ø/ ø) l +bl(v1)cos Pl+b2(V1)cos2 + O/:)cø' z+ zo/z)cø' )O+ O/O (4)

10 10,476 MOGNARD ET AL.' SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS yelds a change n the wnd feld from 12.5 m/s at 220 ø + 5 ø on the hghreturn sde of the front to m/s at 340 ø + 5 ø on the lowreturn sde of the front. The second radar crosssecton decrease n regon 2 can be suppresson) wth scale szes rangng from 1 km to 5 km show the wnd to be varable n both speed and drecton. The wnd front contrast for ths pass s reduced to less than 0.5 db by the drectonal terms n equaton (4) and frontal structure, n the nterpreted as a wnd speed reducton wth or wthout a change form of a lne of m cells, s vsble along the front n wnd drecton. For a wnd drecton of 350 ø the boundary. The wnd drecton on the hghspeed sde of the correspondng wnd speed s m/s. The data reported by West Delta and buoy 4 at 0900 UTC show the wnd speed and drecton to be varable n the lowwnd regme. Ths s supported by SAR mage features observed n lne 3/3. For the Geosat altmeter wth ;.ts relatvely large measurement cell (approxmately 9.5 km), the wnd speed and radar crosssecton "dscontnuty" at the front approxmates an deal step change. Combnng the Geosat and SAR data for ths regon, the half power step response dstance of Geosat o ø measurement (plus algorthm) s 9.8 km along track Regon 3: The thrd regon (lattude 65.45øN to 66.15øN) contans an ncrease of the SAR cros secton from the low return (low wnd) condton and an nterval of relatvely stable return. Snce the spatal gradent of the sea front (225 ø 5 ø) s vsble n the radar magery as wnd streaks. From equaton (4) the best estmate for the mean wnd speed on the lowwnd sde of the front boundary s 5.5 m/s from 335 ø + 5ø; on the hgh wnd sde the wnd velocty s m/s from 227 ø + 5 ø. In the pass 4 SAR data set the wnd drecton s unambguously defned n the magery by the wnd shadow of the drllng platform Polar Poneer. The wnd speed and drecton derved from SAR data near the front, m/s from 230 ø 5 ø, agree reasonably wth the 0900 UTC meteorology log from Polar Poneer (at 0800 UTC 12.8 m/s from 220 ø + 5ø). The 30 ø nddence angle transect for lne 4/4, Fgure 9, shows a reduced wnd velocty at Polar Poneer poston, 10 + m/s, wth an unchanged wnd drecton. At surface roughness (o ø) s suffcently small over ths regon, the the tme of the pass 4 overflght, 1017 UTC, a second wnd Geosat cross secton can be accepted as a wnd speed measure, front s close to the Polar Poneer poston as shown n Fgure whle the SAR data provdes a measurement of wnd drecton. 1. The front south of Polar Poneer s not as sharply defned as Usng the regon 1 o ø plateau as a reference, equaton (4) was the more northerly front and (from equaton (4)) s assocated resolved for wnd angle n regon 3 as s shown n Fgure 8. wth a lowreturn sde wnd of m/s from 180 ø as The ambguty n equaton (4) was resolved by the wnd streak observed by buoy 1 at 0900 UTC. The decrease n SAR return south of the front les outsde of orentaton n the SAR magery to create the dashed wnd drecton ndcators n Fgure 1. The sold wnd drecton lnes the observed wnd speed range of buoy 1 (2.3 m/s at 180 ø + 5 ø are observed from wnd streak data. would be requred) and may be, n part, due to the wnd Wnd nformaton extracted from passes 2 to 4 confrmed nteracton wth a current system that s known to flow through pass 1 results n the regons of overlap as well as provdng ths area as shown n Fgure 2. further nformaton on the spatal dstrbuton of wnd speed Behavor Of the 'horthern" wnd front: Over the and drecton (Fgure 1). course of the SAR data acquston perod the wnd front near 65.2øN, 7.0øE was observed four tmes over an nterval of 2.3 The lowwndspeed regon s best observed by the SAR look drecton n pass 3, snce the radar beam s orthogonal to the hours. Durng that perod the front moved a total of 11.5 km wnd vector n the hghwndspeed regon. In the lowwnd 0.4 km at 132 ø wth a mean speed of 5 km/h whle regme, patches of radar return enhancement (and mantanng ts approxmate orentaton as shown n Fgure 1. The moton n the SAR record s n agreement wth the Wnd Azmuth Angle rom :SAR/Geosat Data observed eastwar drft of the assocated weather system descrbed n the 0900 UTC weather maps for March 20, (Fgure 10) ALTIMETER AND SAR WAVE MEASUREMENTS 5.1. The Geosat Sgnfcant Wave Heght Measurements I I I I Lattude n Degrees Fg. 8. The wnd azmuth angle so (from north) s calculated by combnng SAR and Geosat altmeter data n equaton (2). The data used starts on the low wnd sde of the lne 1/1 front and contnues to the seaward end of the lne. The effect of waves n the altmeter's footprnt s to stretch the leadng edge of the return radar pulse because of early returns from wave crests and later returns from wave troughs. The slope of the leadng edge s nversely related to the wave heght and s reduced to SWH estmates on board the satellte. The altmeter crosstrack sea surface footprnt vares from 2.4 to 11.6 km for SWH varyng from 0 to 20 m. For 3m SWH, the typcal altmeter footprnt s 5 x 6.7 km 2 over s. From the 7month comparson between SWH measurements from the Geosat altmeter and the NOAA buoy network, Dobson et al.

11 MOGNARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 10,477 MARCH 20 L4 SCATTERING CR[]SS SECTI[]N (db) AL[]NG TRACK œ ß 63, , ,2 65, La!:!;ude n Beõr'ees Fg. 9. SAR measurements of o ø at 30 ø ncdence angle along a transect through the lne 4/4 SAR mage. Ths data shows the exstence of two wnd fronts. The northern front (near 65.2øN) s observedurng all passes. The southern front (near 64.4øN) defnes the southern lmt of the storm system. The decrease n o ø south of 64.1øN s beleved to be a result of surface current modulaton of the wnd stress. [1987] found a mean dfference of 40 cm, ndcatng a slght underestmaton of the Geosat SWH compared to the buoys, and arms dscrepancy of 36 cm. Smlar results were obtaned from the statstcal analyss performed durng NORCSEX '88 comparng the Geosat and buoy SWH measurements (O. M. Johannessen and T. A. Johansen, personal communcaton, 990). E Cotnparson Of the Geosat and Buoy Sgnfcant Wave Heght Measurements The Geosat SWH varatons and the buoy measurements along the satellte track are presented n Fgure 11. The SWH varatonshow a steady ncrease from the start of the pass to almost the end of the contnental shelf from 3 to 4.5 m. El0 N66 N65 N64 Fg. 10. Weather map from 0900 UTC, March 20, 1988, clearly showng the occluded lowpressure regon that generates the wnd front observed by the SAR and the Geosat altmeter.

12 , ,478 MOGNARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS z I I I '1 I I I I o GEOS,T (X:x:x:x3 BUOY 06:00 (4) 08:00 (2) 11 :O0 (2) 09:00 (4) 08:00 o,,, 9.OC (1/,, I I I I I I I I I I I I I I I I I I I I I I I I LATITUDE (degrees N) Fg. 11. Geosat sgnfcant wave heght (SWH) varatons and correspondng buoy SWH measurements at 0600 and 0900 UTC, or 0800 and 1100 UTC. Beyond the contnental shelf, the mean SWH remans stable at 4.5 m. An analyss of the buoy data on March 20 [Barstow and Berken, 1988] shows that buoy 2 was measurng a local fetchlmted sea wth westerly wave drecton. Long waves were refracted by the Haltenbanken plateau toward buoy 2, whch occupes a sheltered locaton behnd the bank. At 0800 UTC, the SWH measured by buoy 2 was 2.1 m, whch s low compared wth the altmeter measurement of 3.2 m at the start of the Geosat pass (Fgure 11). Ths 1m dfference between Geosat and buoy 2 mght be due to the sheltered locaton of buoy 2 (at the same tme, buoy 3 was measurng 4.7 m and buoy 13.6 m at 0900 UTC, n good agreement wth Geosat). Ths shelterng effect mght explan the almost constant overestmaton of Geosat SWH n the vcnty of buoy 2 durng NORCSEX '88 [Barstow and Bjerken, 1988]. At 0900 UTC, a SWH of 5.3 m was measured at buoy 4, n good agreement wth Geosat 4.8 m measurements. The crossng of the atmospherc front near buoy 4 s observed n the Geosat and SAR scatterng crosssecton varatons (Fgures 3 and 4). Across the front, the altmeter wnd speed decreases from 12 m/s to 10.5 m/s (Fgure 3b), whch would correspond to a dmnuton of the fully developed wnd wave from 3.6 m to 2.6 m. However, the sea has not had tme to reach equlbrum because of the front moton. A decrease s not observed on the altmeter SWH whch n ths area oscllates between 4.6 m and 4.3 m. The crossng of the atmospherc front s not defned by a s fcant varaton of the SWH measurement. s swelldomnated. 0 Ths s an ndcaton that the sea state 5.3. The Geosat Mnmum Sgnfcant Swell Heght From the smultaneous measurements of SWH and wnd speed (V) by the Geosat altmeter, a mnmum sgnfcant swell heght can often be nferred. Ths s the case when the SWH measured by the altmeter s greater than the fully developed wave heght derved from V. The fully developed wave heght s computed usng the PersonMoskowcz expresson of the spectrum [Person and Moskowcz, 1964]. When the fully developed wave heght assodated wth Vs smaller than the altmetermeasured SWH, the dfference between the energy n the altmeter SWH feld and the fully developed wave feld s due to the presence of swell, and a mnmum sgnfcant swell heght can be computed [Mognard, 1984]. The Person Moskowcz spectrum used to derve the mnmum sgnfcant swell heght from the wnd speed measurement assumes that the wnd sea s fully developed. In the mddle of a storm, when the condtons are not fully developed and the energy n the altmeter SWH measurement s lower than the energy n the fully developed sea (as derved from 1O, the altmeternferred mnmum sgnfcant swell heght cannot be estmated. The Geosat mnmum sgnfcant swell heghts are presented n Fgure 12 along wth the 0800 UTC hndcast values from the Norwega numercal wave model WINCH. At the start of the Geosat pass, near the Norwegan coast, the mean altmeter derved swell s 2.5 m, n good agreement wth the hndcast values; north of the atmospherc front, the altmeter mean swell heght reaches 3.5 m, slghtly lower than the hndcast values from WINCH. South of the atmospherc front where the hghest wnds were measured, the swell heght cannot be derved from the altmeter SWH and U measurements. Z I I eeeee GEOS,T 0 : :CX:)CX:)O WINCI 4 ' 0 0 O ß 'ßß ßß% ß oc)ß ß ß () ß ß ß ß. 0 I I I I I I J I I I I I I I I I I I I LATITUDE (degrees N) Fg. 12. Geosatderved mnmum sgnfcant swell heghts and sgnfcant swell heght hndcasts from the WINCH model.

13 MOONARD LeT AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 10,479 Durng the SAR flght, the domnant swell and wnd wave felds observed n the SAR magery come from the southwest. The four buoys n the NORCSEX area are measurng, on March 20, a swell comng from the southwest that s stll present and domnant durng the SAR flght. Ths swell s comng from the regon between Scotland and Norway and had been generated the precedng day by the same storm system. The only Geosat passes n the regon on March 19 are a descendng pass at 0200 UTC and an ascendng pass at 0900 UTC that both crossed the NORCSEX area. These two Geosat passes do not probe the most ntense part of the storm that s located southwest of the NORCSEX regon. The hgh sea state measured by the Geosat altmeter on March 19 are obtaned at 0900 UTC near the Norwegan coast at 64øN and 2øE wth SWH of 6 m and U of 18 m/s. If we assume that these measurements are representatve of sea state condtons not too dfferent from the hghest sea state generated by the storm, a resultng swell perod of 11 s can be nferred from the Geosat altmeter measurements usng the Jont North Sea Wave Project (JONSWAP) form for the spectrum vald for underdeveloped sea state [Hasselrnann et al., 1973; Mognard et al., 1986]. The storm that occurred on March 19 was responsble for the generaton of the southwest swell that reached the four buoys n the evenng of March 19 wth measured perods varyng between 11.8 and 12.8 s. The Geosat underestmaton of the swell perod (11 s) confrms that on March 19 the Geosat altmeter dd not acqure data over the mddle of the storm. On March 20 at 0900 UTC, buoy 4 measured a bmodal wave system wth a resdual southwesterly swell of 12.5s perod. were lowpass fltered, and the orgnal mages were dvded by the correspondng fltered mages to remove the lowfrequency nose n the data, ncludng that caused by the resdual antenna pattern. Each of the 5!2 x!024 mage segments was then dvded nto two 5!2 x 5!2 subsets whch were Fourer transformed, and the squared magntudes of each par of Fourer transforms were averaged together to form an estmate of the mage spectrum for each data segment. These mage spectra were dsplayed n two dfferent ways and were also used to estmate the SWH as descrbed n secton Fgure!3 s a map of the regon showng the approxmate locaton of the SAR spectra. Twodmensonal gray level dsplays of the mage spectr are shown n Fgure 14. The spectra were also reduced to onedmensonal wave number and wave drectonal spectra shown n Fgures 15 and 5.4. SAR Processng Estmaton of sgnfcant wave heght from SAR data. There s a consderable amount of uncertanty wth regard to SAR mage formaton was performed n realtme on board the mechansms by whch ocean waves modulate the SAR the arcraft. Ths processng used the onboard navgaton mage ntensty, partcularly for waves travelng at large angles system to mantan good spatal control of the mage locatons wth respecto the radar look drecton. Nevertheless, there s and to compensate for arcraft moton effects on the mage a body of nformaton about these mechansms whch can be focus. The onboard realtme processor performed a tme exploted to yeld at least a frstorder estmate of the SWH for doman correlaton between seven defned subbeams and a comparson wth the Geosat estmates. stored azmuth reference functon (computed for the radar In the general case, the magng of ocean waves by SAR s magng geometry and ntal ground speed at the start of each nfluenced by surface moton (velocty bunchng and azmuth flght lne) then recombned the tmeshfted, detected snglefalloff) effects, as well as by "real" radar crosssecton look mages to form an output magntude mage. For the modulaton (tlt and hydrodynamc) mechansms. In the data March 20, 1988, data, the realtme processed SAR magery set consdered here, however, snce the peak of the spectrum s was recorded n a slant range format. n the radar range drecton, surface moton effects are Estmaton of wave spectra from SAR mage. SAR expected to be mnmal and have been neglected n ths spectra were extracted from the standard sevenlook processed analyss. We assume therefore that the mage spectrum S(k ) mages collected n nadr mode as descrbed n sectons! and 3. Data segments consstng of 512 pxels n range by!024 can be related to the wave heght spectrum Sw(k ) by the pxels n azmuth were selected at alongtrack ntervals of 6.7 equaton km, correspondng to the!s Geosat sample locatons. The crosstrack locatons of these segments were chosen to be centered at an ncdence angle of approxmately 35 ø. =Im 12 + (s) These segments were frst resampled onto a regular 4 x 4 m ocean surface grd to remove the dstorton caused by recordng the data n the slant plane. The resampled mages where rn s the ocean waveradar modulaton transfer functon (mtf) and Ssp(k ) represents the contrbuton of coherent speckleffects to the mage spectrum. The latter term was 16. Plots of the wave numberntensty spectra provde nformaton about wavelength and wave propagaton drecton (Fgure!4); wave numbers n azmuth and range are ndcated along the horzontal and vertcal axs, respectvely, whle the crcles dentfy wavelengths of 100, 200, and 400 m. The twodmensonal and onedmensonal SAR spectra (Fgures 14 and 15) show only one consstent peak at a wavelength of approxmately 250 m across the length of the pass. However, on the bass of the local wnds estmated from Geosat to be of the order of!012 m/s at the tme of the overpass, there was expected to be a second peak at a wavelength of about 100 m. Although there s som evdence of such a peak n a few of the twodmensonal spectra (Fgure!4), ths feature does not appear consstently n the SAR data as expected. In addton to observng the general shape of the SAR spectra, we have also attempted to estmate the SWH from the SAR data.

14 10,480 MOGNARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 4 o 6 o 8 o 10 ø 12 ø 66* 65* 4 Fg. 13. Locaton, along lne 1/1, of the SAR wave spectra presented n Fgure 14. 9O2OO78 R1 estmated by computng the mage spectrum over a regon near shore where no vsble features were apparent n the mage. Ths term was then subtracted from the mage spectr at the other locatons n the mage, and the remander was dvded by Jm 12 k 2 to yeld an estmate of the wave heght spectrum, whch was then ntegrated over k to obtan the sgnfcant wave heght,.e., E H s = 4 I m I '2k'2[s (k)ssp(k)] where the mage spectrum s der'reed such that the ntegral over k s equal to the mage varance. The value of the modulaton transfer functon used n ths analyss was calculated from theoretcal expressons for the tlt and hydrodynamc effects,.e., rn = m t + m h. For a perfectly conductng surface and a surface wave heght spectrum of the form Sw(k ) = A k'p near the Bragg wave number, the tlt modulaton transfer functon can be wrtten as (6) 4 sn0 cos0 m t = p cot0 1 + sn20 (7) where 0 s the ncdence angle. The hydrodynamc modulaton transfer functon, for the smplest case where relaxaton effects are neglected, s gven by m n = p + 1/2 (8) The total modulaton transfer functon s equal to Iml = [ Imtl 2 + Ir%1211/2 (9) whch has the value 6.2 for œ = 4 and 0 = 35 ø. The sgnfcant wave heght varatons obtaned from ths analyss are plotted versus lattude n Fgure 17. It can be seen from Fgure 17 that for the whole regon the SAR and the Geosat SWH are n far agreement. However,

15 MOGNARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 10,481 A F H Fg. 14. Twodmensonal SAR wave spectra obtaned every 20 km along the Geosat pass for lne 1/1 (see Fgure 13 for locatons). Horzontal axs s alongtrack wave number and vertcal axs s crosstrack wave number. Crcles ndcate wavelength of 100, 200, and 400 m. The domnant feature s a long wavelength between 200 and 300 m. Indcatons of the presence of shorter wavelength can be found on the SAR spectra D, E, and F, and also at the end of the pass n the open ocean on the spectra K and L.

16 10,482 MOONARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 0.01,3 1D Wavenumber P!of the SAR appears to underestmate the wave heght n regons where the wnd speed s hgh ( 64.7øN to 65.3øN, and north of 65.5øN) and to overestmate the wave heght where the wnd speed s low (65.3 ø to 65.5øN). Ths dscrepancy due to the wnd speed dependence of the ocean waveradar modulaton D Angle Plof O.Oll 0.22 ' ' ' I ' ' ' ' I ' " ' ' I ' ' ' ' I O.OLO G Wovenumber L 0.02 I Fg. 15. Onedmensonal SAR wavenumber spectr along the Geosat pass for lne 1/1 at locatons concdent wth the 1s 0.00 Geosat measurements. Letters on the rght sde ndcate spectra correspondng to the twodmensonal plots shown n Fgure 14. The domnant wavelength that was present n all the two dmensonal spectra s located here n the wave theta (degrees N) Fg. 16. Onedmensonal SAR wave drectonal spectralong numbe range between 0.02 and 0.03 rad/m whch corresponds flght lne 1/1 (see Fgure 13 for locatons). The domnant to wavelengths between 200 and 300 m. wave drecton comes from the southwest.

17 MOONARD ET AL.' SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 10,483 SAR vs. GEOSAT Sgnfcant Wave Heght 20FEB Lffude (deg) LEGEND SAR GEOSAT Fg. 17. Comparson of sgnfcant wave heght along the Geosat pass as measured by the altmeter wth that estmated from the SAR data. transfer functon, as observed n tower measurements by Plant Although ths wnd speedependence of the radar mtf s a et at. [1983] at X band and by Schr6ter et at. [1986] at C band. complcatng factor n attemptng to make quanttatve use of Accordng to Schrbter et al., the mtf can be plotted as a SAR ocean wave data, t s qute concevable that correctons functon of the nondmensonal varable]u/g, where f s the can be made for ths dependence based on wnd speed wave frequency, U s the wnd speed, and g s the acceleraton estmates from the SAR data tself. These correctons are of gravty. Furthermore, for frequences near the peak of the beyond the scope of ths paper, however. wave heght spectrum, the magntude of the mtf s approxmately equal to the nverse of ths nondmensonal 6. DISCUSSION AND SUMMARY OF RESULTS varable. Thus a decrease of the wnd speed by a factor of 2 s capable of explanng the varaton of the estmated SAR SWH Concdent observatons were acqured on March 20, 1988, n ths regon. In terms of the model dscussed above, ths may by the Geosat altmeter and the arborne SAR over a also be nterpreted as a change n the slope of the wave heght partcularly complex oceanc stuaton where wnd, swell, and spectrum at hgh wave numbers, whch s consstent wth current were nteractng wthn an occluded lowpressure current (lmted) knowledge of ths part of the spectrum. system that was located about 200 km off the coast of Norway.

18 10,484 MOGNARD ET AL.' SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS In ths low, the wnds computed from SAR scatterng cross secton measurements are lght (2 to 5 m/s) and, as observed from SAR magery, are varyng randomly over dstances of the order of 5 km. Because of the footprnt sze of the altmeter, ths geographcally very narrow regon of low wnd speed s not sensed by the altmeter. From the SAR ocean backscatter crosssecton measurements acqured from lne 1/1 (Fgure 4), the approxmate wdth of the low s 20 km. Along the eastern edge of the low s a welldefned wnd front whch s observed by both the SAR and the altmeter. On the southeastern sde of the front, the wnd feld s relatvely unform (Fgure 1), wth wnd vector m/s from 227 ø + 5 ø, n the zone parallel to the front. The northwestern (low wnd speed) sde of the front s delneated on the SAR mages by 500mscale cellular dsturbances along the front boundary and by wnd vectors m/s from 345 ø + 5 ø. Across ths atmospherc front, the wnd speed computed from the altmeter scatterng crosssecton decreases from 12 m/s to 10.5 m/s over a dstance of approxmately 23 km. Because of ts footprnt sze (about 9.5 km n dameter) and of the tme response of the AGC, the altmeter behaves as a spatalowpass flter [Glazman and Plorz, 1990]. The altmeter wnd speed estmates are domnated by the step response of the nstrument n a regon of hgh spatal varablty probed very accurately wth the SAR. Across the lowpressure center the wnds turn through approxmately 160 ø and ncrease n magntude to produce SARGeosat measured wnd vectors of m/s from 14 ø + 5 ø. In the regon 130 km south of buoy 4, whch s not sampled by Geosat, a second front s observed by the SAR wth wnd vectors of 10 + m/s from 220 ø + 5 ø along ts northern boundary and m/s from 180 ø + 5 ø to the south. From meteorologcal measurements at buoy ths front defnes the southern lmt of the storm system. In both cases the second decrease n crosssecton s accompaned by a local, cusplke, crosssecton ncrease (Fgures 4 and 8) assocated wth the presence of a current shear. The surface current measurements n these areas (Fgure 2) supporths current nteracton hypothess. In the area near the begnnng of lne 1/1, the wnds blows off the coast from 170 ø (Fgure 1). The local wnd sea s fetchlmted near the coastal slands; thus equatons 2 and 3 are not vald for wnd speed estmates. About 10 km from the coast at 64.3øN (Fgure 4) the wnd vector calculated from equaton 3 s m/s from 175 ø + 5 ø. The SAR magery shows a smooth shft n wnd drecton from ths pont to buoy 2, located 30 km further down the SAR swath where the wnd vector s m/s from 196 ø + 5 ø. A contnung smooth wnd drecton transton s observed at the start of the Geosat altmeter data set at 64.4TN where the wnd vector s m/s from 225 ø + 5 ø. In the vcnty of the nshore current jet descrbed n secton 2, the SAR magery shows largerscale ntensty modulaton whch s not related to nstrument parameters or to the wnd drecton. These perturbatons appear n Fgure 4 as a 1.5dB steplke devaton n radar cross secton supermposed on the otherwse smooth ncrease o ø wth alongtrack poston. Ths feature s attrbuted to the surface current and current shear n ths area. In the SAR data analyss, the ESA wnd model ofattema et al. [1986] has been used to relate SAR o ø measurements to wnd velocty. Wnd parameters computed from ths model are n excellent agreement wth other data sources when the model s scaled to SAR cros sectons n the 200 to 50 ø ncdence angle range. Attempts to scale the SAR o ø to the model predctons n ths angle range resulted n SAR o ø errors at nadr that when compared to Geosat measurements are n excess by 10 db. Snce the condtons present durng ths experment are very smlar to those on whch the model s based, no satsfactory explanaton for the dscrepancy has been found at ths tme. The sea state condtons observed wth the altmeter and SAR are swelldomnated. The storm located east of Scotland on March 19 s responsble for the generaton of the swell famly that s measured n the NORCSEX area durng the Geosat overpass on March 20. Ths southwestern swell s detected by the four Wavescan buoys, the altmeter, and the SAR. The crossng of the atmospherc front near the locaton of buoy 4 does not produce sgnfcant changes n the wave feld as observed by buoy 4, the Geosat altmeter, and the SAR. Along the Geosatrack, the SWH measured by the altmeter ncreasesteadly from 3 m near the Norwegan coasto 4.5 m near the end of the contnental shelf. Over the Norwegan Sea, west of the contnental shelf break, the altmeter SWH reaches a plateau at about 4.5 m. The Geosat SWH measurements are n good agreement wth the buoy measurements (Fgure 11). When compared wth the Geosat SWH measurements, the SWH values nferred from the SAR mage spectrum are of the same order of magntude (Fgure 17): there s a far qualtatve agreement between both radars. The smple expresson used for the SAR modulaton transfer functon (equaton(9)) does not take nto account a wnd dependence or a sea state parameter. In the regon of low wnd speed, on the west sde of the atmospherc front, the SAR SWH values exhbt an artfcal, welldelmted ncrease that yelds SWH values that are overestmated compared wth the Geosat values. Outsde the lowwndspeed regon, the SAR SWH are underestmated compared wth the Geosat measurements. Because of the varable wnd condtons along the Geosat track, a constant S. R mtf cannot be expected to gve accurate SWH values. The dfferences between the Geosat and the SAR SWH emphasze the mportance of wnd contrbuton on the SAR mtf. Mnmum sgnfcant swell heghts varyng from 2.5 m near the Norwegan coasto 3.5 m beyond the contnental shelf break are derved from the altmeter measurements of SWH and wnd speed. An approxmate wavelength of 180 m s determned from the altmeter measurements of the hghest sea state along the Geosat pass that sampled the northeast corner of the March 19 swellgeneratng storm. The Geosatderved swell wavelength s underestmated compared wth the buoys and the SAR because the satellte dd not acqure data n the center of the storm.

19 MOONARD ET AL.' SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS 10,485 The SARdcrvcd wavelengths vary between approxmately sgnfcant wave heghts usng buoy data,.j. Geophys. Res., 200 and 300 m along the Gcosat pass. In the hghwnd regon 92(C10), 10,71910,731, cast of the atmospherc front, the mcan wavelength generally Donclan, M. A., The effect of swell on the growth of wnd oscllates between 200 and 250 m. West of the front, the SAR waves, Johns Hopkns APL Tech. Dg., 8, 1823, derved wavelengths ncrease to a mcan value that oscllates Donclan, M. A., and W. J. Person, A twoscale Bragg between 250 and 300 m. Sporadc ndcatons of energy at scatterng model for mcrowave backscatter from wnd lower wavelength between 100 and 150 m arc found n some of generated waves, Proceedngs of IGARSS'86, Eur. Space the SAR wave spectra (Fgure 14). However, a consstent peak Agency Spec. Publ., ESA SP254, , n energy between 100 and 150 m correspondng to a wnd Glazman, R. E., and S. H. Plorz, Effects of sea maturty on wave component n agreement wth the measured altmeter satellte altmeter measurements, J. Geophys. Res., 95(C3), wnd speed s not obtaned n the SAR wave spectra , There arc several possblexplanatons for ths apparent Goldhrsh, J., and E. B. Dobson, A recommended algorthm dscrepancy. However, wc beleve that the lack of a second for the determnaton of ocean surface wnd speed usng a wnd wave peak n the SAR spectra may bc duc to the satellteborne radar altmeter, Tech. Rep. S1R85U005, mechansm dscussed by Donelan [1986] wheren the presence Appl. Phys. Lab., Johns Hopkns Unv., Laurel, Md., of a swell nhbts the growth of a hgherfrequency wnd wave Hasselmann, K., et al., Measurements of wndwave growth and nstead causes energy nput from the wnd to bc deposted and swell durng the Jont North Sea Wave Project n the swell tself. The reason for ths effect s not well (JONSWAP), Dtsch. Hydrogr. Z., 8 (12), suppl. A, 95 pp., understood but has bccn hypotheszed by Donclan to bc duc to a "dctunng" of the resonant nonlnear nteractons whch Haugan, P.M., G. Evensen, J. A. Johannessen, O. M. transfer energy to the long waves. Johannessen, and L. H. Pettersson, Modeled and observed Swell drectons determned from SAR wave spectra arc mesoscale crculaton durng the 1988 Norwegan mostly comng from the southwest (Fgure 16). However, at Contnental Shelf Experment, J. Geophys. Res., ths ssue. the start of the pass, near the Norwegan coast, and at the end Johannessen, J. A., R. A. Shuchman, O. M. Johannessen, K. L. of the pass, beyond the contnental shelf break, a larger scatter Davdson, and D. R. Lyzenga, Synthetc aperture radar n the energy dstrbuton from the SAR wave spectra ndcates magng of upper ocean crculaton features and wnd fronts, the possblty of nteractons wth the local current features. J. Geophys. Res., ths ssue. Keller, W. C., V. Wsman, and W. Alpers, Towerbased Acknowledgments. The authors of ths paper would lke to measurements of the ocean Cband radar backscatterng thank Lassc Pcttcrsson, the coordnator of the NORCSEX '88 crosssecton, J. Geophys. Res., 94(C1), , campagn for hs role and help durng the experment. Wc also Lvngstone, C. E., A. L. Gray, R. K. Hawkns, R. B. Olsen, J. wsh to thank the varous groups nvolved n data acquston G. Halbertstoa, and R. A. Deane, CCRS Cband arborne and reducton: the Canadan ar crew on board the CV580, radar system: system descrpton and test results, Proc. the HSkonMosby crew from the Unversty of Bergen for the Canadan Symp. on Remote Sens., 11th, Waterloo, Ontaro, acquston of the oceanographc data, and the OCEANOR , group n charge of the buoy deployment and data reducton. Lvngstone, C. E., A. L. Gray, R. K. Hawkns, and R. B. Olsen, CCRS C/X arborne synthetc aperture radar: An REFERENCES R&D tool for the ERS1 tmeframe, IEEE Trans. Aerosp. Electron. Syst., 3(10), 1120, Attema, E. P. W., A. E. Long, and A. L. Gray, Results of the MacArthur, J. L., P. C. Marth, and J. G. Wall, The Geosat ESA arborne Cband scatterometer campagns, Proceedngs of IGARSS'86,.Eur. Space Agency Spec. Publ., ESA SP254, altmeter, Johns Hopkns APL Tech. Dg., 8(2), , , Barrck, D. E., Wnd dependance of quasspecular mcrowave Mognard, N.M., Ocean wave parameters extracton usng sea scatter, IEEE Trans. Antennas Propag., AP22, , satellte shortpulse radar altmeter, Proceedngs of a Workshop on ERS1 Radar Altmeter Data Products, Eur. Barstow, S. F., and S. Bjerken, Wave measurements from Space Agency Spec. Publ., ESA SP221, 3741, moored drectonal wave buoys durng NORCSEX'88, Mognard, N.M., and B. Lago, The computaton of wnd speed ODAP Rep.. 108, p. 146, Oceanographc Data Acquston and wave heght from Geos3 data, J. Geophys. Res., 84(B8), Project, Trondhem, Norway, , Brown, G. S., Estmaton of surface wnd speeds usng satellte Mognard, N.M., W. J. Campbell, R. E. Cheney, J. G. Marsh, borne radar measurements at normal ncdence, J.. Geophys. and D. B. Ross, Southern ocean waves and wnds derved Res. 84, , from Seasat altmeter measurements, n Wave Dynamcs and Brown, G. S., H. R. Stanley, and N. A. Roy, The wnd speed Rado Probng of the Ocean Surface, edted by O. M.Phllps measurement capablty of spaceborne radar altmeters, and K. Hasselmann, pp , Plenum, New York, IEEE J. Oceanc Eng., 6(2), 5963, Mognard, N.M., W. J. Campbell, and E.G. Josberger, Geosat Dobson, E., F. Monaldo, J. Goldhrsh, and J. Wlkerson, surface wnd speed estmates and comparsons n the North Valdaton of Geosat altmeterderved wnd speeds and Atlantc for March 1987, paper presented at the

20 10,486 MOONARD ET AL.: SATELLITE AND AIRBORNE SEA STATE OBSERVATIONS WOCE/NASA Altmeter Algorthm Workshop, sponsored by the U.S. Scence Steerng Commttee for WOCE, Corvalls, Oreg., NORCSEX '88 Group, NORCSEX '88, A prelaunch ERS1 experment, EOS Trans. AGU, 70(49), , , Panofsk, H. A., Determnaton of stress from wnd and temperature measurements, O. J. R. Meteorol. Soc., 98, 85 94, Person, W. J., and L. Moskowcz, A proposed spectral form for fully developed wnd seas based on the smlarty theory of S. A. Ktagorodsk, J. Geophys. Res., 69, , Plant W. J., W. C. Keller, and A. Cross, Parametrc dependance of the ocean waveradar modulaton transfer functon, J. G½ophys. Res,., 88, , Schr6ter, J., F. Fendt, W. Alpers, and W. C. Keller, Measurements of the ocean waveradar modulaton transfer functon at 4.3 GHz, J. Geophys. Res,, 91, , Valenzuela, G. R., Theores for the nteracton of electromagnetc and oceanc waves: A revew, Boundary Layer Mete0rol., 13, 6185, J. A. Johannessen, Nansen Envronmental Remote Sensng Center, Edvard Gregsve 3A, 5037 Solhemsvk/Bergen, Norway. C. E. Lvngstone, Canada Centre for Remote Sensng, 2464 Sheffeld Rd., Ottawa, Ontaro, Canada K1A 0Y7. D. Lyzenga, C. Russel, and R. Shuckman, Envronmental Research Insttute of Mchgan, P.O. Box 8618, Ann Arbor, MI N.M. Mognard, Centre Natonal d'etudes Spatales, 18 Avenue E. Beln, ToulouseCedex, France. (Receved May 30, 1990; revsed September 15, 1990; accepted August 6, 1990.)