active microwave workshop report - nasa · 192 active microwave workshop report part d large-scale...

192 ACTIVE MICROWAVE WORKSHOP REPORT

PART D

LARGE-SCALE PHENOMENA

This section deals with oceanic phenomenawith horizontal scales from approximately100 km up to the widths of the oceans them-selves. Their very size makes these phe-nomena difficult to monitor by surface-basedmethods, so that satellite-mounted active mic-rowave techniques are especially valuable andmay be the only way of gathering the infor-mation needed for scientific progress.

An important class of large-scale featuresconsists of those concerned either directly orindirectly with the vertical topography of theocean surface, estimated from the center ofthe Earth. This class includes the shape ofthe geoid, which is determined by the internalmass structure of Earth; the quasi-stationaryanomalies due to spatial variations in seadensity and steady current systems; and thetime-dependent variations due to tidal andmeteorological forces and to varying cur-rents. These features are discussed in sub-sequent sections. All these features make useof the techniques of radar altimetry, althoughcurrents can also be studied by satellite track-ing of floats.

Certain large-scale phenomena could, inprinciple, be usefully observed by satellite al-timeters ; however, it is unlikely that the nec-essary precision (approximately 1 cm) couldbe obtained in the near future. Steric varia-tions in sea level, due to time-dependentchanges in water density, notably at annualand semiannual periods, are known fromshore-based tide gages to have amplitudes ofless than 10 cm in most parts of the ocean.Tsunamis sometime reach meters of ampli-tude at a coastline, but in the open oceanwhere they would be most useful to measure,they merely consist of long waves of a fewcentimeters' amplitude and some tens of kilo-meters' wavelength. Prospects for detectingtsunamis by altimetry are discussed byGreenwood et al. (ref. 3-43).

MARINE GEODESY

Background

The primary function of geodesy is to de-termine the shape of the Earth by carefullymeasuring the geometric distance betweenpoints on its surface. For years, this ap-proach seemed to be satisfactory because thesurveyed areas were limited to portions ofcontinental crusts and the measurementswere not required to consider the micromove-ments of the solid Earth crust, which wasconsidered as a perfectly rigid body. Prob-lems related to the choice of a common refer-ence datum for all those local determinationswere approached through the use of a theore-tical surface called the geoid, which is anequipotential of the Earth gravity field.

The geoid on land is not a physical surface;it is determined through a numerical processthat involves the computation of an integralof gravity anomalies called the Stokes for-mula. The geoid on the oceans was long con-sidered to be in coincidence with the physicalsurface of the sea (refs. 3-44 and 3-45).

During the last 10 yr, the problems of geo-metric geodesy have been extended to the openoceans mainly through the needs of accuratenavigation and positioning. Marine geodesyhas appeared as a separate scientific endeavorbecause most of the hypotheses of practical"land" geodesy were found to poorly matchthe physical properties of the ocean environ-ment. At the same time, the accuracies of theavailable instruments for distance measure-ments were drastically improved so thatmovements of the crust of very small ampli-tude had to be considered.

Presently, serious difficulty is encounteredin coping with the shape of the Earth per se,and marine geodesy must be concerned withfour important surfaces (ref. 3-46) : thephysical surface of the Earth crust, the phys-

https://ntrs.nasa.gov/search.jsp?R=19760004732 2020-02-21T20:25:37+00:00Z

ACTIVE MICROWAVE REMOTE SENSING OF OCEANS 193

ical surface of seawater, the geoid, and themean sea level.

The physical surfaces are experimentallyaccessible but the others are concepts withdefinitions that unfortunately tend to varyslightly from one author to another. Follow-ing is a brief overview of the problems en-countered when dealing with those surfaces.

The physical surface of the Earth crust.—The Earth crust is the solid surface of Earth,including that portion underlying the seasand the oceans. Long regarded as perfectlyrigid, the Earth crust is, in fact, the outershell of a "pulsating" elastic medium in ever-lasting slow movement. Those displacementsare essentially as follows:

1. The solid Earth tides (20 to 40 cm, ver-tical).

2. The continental drift (a few centimetersa year, horizontal).

3. Movements caused by earthquakes andfault displacements (a few centimeters to afew meters in any direction).

4. Tilt of coastal areas caused by the risingand falling of ocean tidal motions (a few cen-timeters in any direction).

5. Movements caused by the modificationof loading conditions on the crust (erosion-sediments piling).

Some of those effects would have to be con-sidered very carefully when using a radaraltimeter with a 10-cm resolution capability.

The physical surface of the sea water.—Thewater surface changes position for many rea-sons, some of which are ocean tides; air pres-sure; winds; temperature, density, or pres-sure gradients; movements caused by suddenvariations in the physical surface of theocean floor (earthquakes, fault displace-ments) ; and geologic changes associated withthe melting of glaciers (ref. 3-47).

This fast-moving surface is obviously diffi-cult to match with the present concept of thegeoid over the open seas; a fresh look at theproblem is badly needed. Other important in-teractions that may occur between the twosurfaces are the cross-coupling of ocean tidesand solid Earth tides and the movements ofthe sea surface due to movements of the

ocean floor. Both these interactions tend tobe very difficult to decipher in the coastalzone areas. -

The geoid.—The only satisfactory defini-tion of the geoid from the mathematical view-point is that of an equipotential of the Earthgravity field with a certain absolute poten-tial. However, an empirical approach hasbeen imposed by practical considerations sothat local geoids have been defined by eachnational geodetic agency as equipotential sur-faces passing through a defined point (datumpoint), usually chosen in the countries havingoceanic borders near the mean sea level. Un-fortunately, those local geoids do not coincideover the total surface of the Earth becausethe datum points have not been chosen on thesame equipotential.

One of the greatest contributions of satel-lite altimetry (using coast-to-coast trackingacross the Atlantic Ocean) will be to provideexperimental data ultimately useful in thechoice of a common datum for North Amer-ica and Europe. This datum should be chosenas a reference for the North Atlantic marinegeoid.

Three ways of mapping the shape of thegeoid are available today.

1. Measurement at sea of gravity and in-tegration through the Stokes formula.

2. Measurement at sea of the vertical de-flection of gravity and integration throughthe Veining-Meinez formula.

3. Measurement of the altitude of a satel-lite over the physical surface of the sea lead-ing to the shape of the geoid with the use ofappropriate correction and filtering of thedata.

Gravity at sea cannot be measured, in theforeseeable future, to an accuracy better than0.01 m/sec2, thus leading to uncertainties inthe geoidal heights computed from the sea ofat least a few meters. Deflections of the ver-tical could be measured to an accuracy of 0.5sec of arc with the GEON system (ref. 3-48)and to a few seconds of arc using an inertialnavigation system (ref. 3-49).

The wealth of geoidal information avail-able from radar altimetry has been ade-


quately demonstrated by data from the Sky-lab S193 altimeter experiment (ref. 3-50).Figures 3-25 and 3-26 are two examples.Figure 3-25 shows the profile of the PuertoRican trench area, which represents a massdeficiency, and figure 3-26 shows the effectof a sea mount, which corresponds to a massconcentration. Because the geophysical valueof such information is well known, this dis-cussion will be directed toward geoidal fea-ture resolution available from altimetry.

To discuss the large-scale resolution capa-bilities of radar altimetry, it is first necessaryto examine geoid spectrum and altimeter-related characteristics. Figure 3-27 shows ageoidal power spectral density of the PuertoRican trench area, computed by using SkylabS193 data from McGoogan et al. (ref. 3-50)and Miller and Brown (ref. 3-51). This areawas selected because it should contain short-wavelength components of substantial mag-nitude; therefore, the analysis should yieldan approximate upper boundary on geoidaldata processing requirements for GEOS-C.Figure 3-28 shows the spatial filter transferfunction of the altimeter, which was com-puted based on the specific tracking systemused in the GEOS-C altimeter. The mini-

120

100

| 80

> 60

I

; 40

20 Trench area(large localized ,.gravity anomaly)'

; Pass 6''

Puerto Ricolandmassi i I

0 20 40 60 80 100 120 140Elapsed time, sec

FIGURE 3-25.—Geoidal profile of the Puerto Ricantrench area from Skylab passes 4 and 6 (ref. 3-48).

mum-mean-square linear filter for processingaltimeter data may be derived based on thepower spectrum and spatial filter models dis-cussed and on known altimeter random errorcharacteristics. Optimum impulse responsefunctions are shown in figure 3-29. Refer-ring again to figure 3-27, the half-powerbandwidth of the optimum filter is seen tobe 40 km for the GEOS-C design.

These findings indicate that future altime-ter systems will be able to provide substantialimprovement in resolution of features towavelengths as short as 5 km.

The mean sea level.—The geodesist, lookingfor a reference surface from which landheights could be measured, adopted the aver-age height of the ocean surface in coastalareas as defining segments of its referencesurface, the hypothesis being that the geoidand the mean sea level coincide all along thecoastal lines.

The basic idea was that proper averagingin time of tide gage logs taken in ports alongthe coast would provide reliable estimates ofthe true position of the geoid. Coastal ge-odesists have developed complex schemes forfiltering out all tidal influences. Unfortu-nately, other systematic effects are still pres-ent in their computations. It is well knownthat a 10-m discrepancy exists between thetwo ends of the eastern coast of the UnitedStates. High correlations of the residualswith atmospheric pressure variations (ref. 3-52) have been noted by oceanographers allalong the Mediterranean coast.

The mean-sea-level computations could becorrected with radar altimeter measure-ments, but this is a very difficult task indeed.Radar altimeters take instantaneous profilesof the physical surface of the sea, whereasmean sea level is computed from time logstaken at fixed points. The only method ofcomparison would be by the averaging ofmany profiles measured by the altimeter.Such a time-consuming task would only befulfilled many years from now after manysuccessful launches of radar altimetersaboard satellites.


f^y 28 aCaboFrip'' 11691538 \1700177Q

V)V 62 60 2° 2°/ 900 | , /A7 AO 7/ nn" i-ain 11004 1 X

7000.' 1310 |i«U- « X 1734

13 "71 70 88 / 153() 7170/i857 7.05llha de Sao Sebastiao -----551 "••••- t^ 2139 1315

45.....---""1?l 108° 1684 /fy^O..-•'"'400 I • \

1629 21002209 1235

(a)

Mount A-'

^Marsh-Vincent geoid

Mount B'

Altimeter residualsfrom S-band orbit

14:37:20 14:3&00

(b)

Time, hr:min:sec14:39:00 14:39:30

FIGURE 3-26.—Effects of subsurface features on altimeter data; Skylab 3, pass 11.(a) Geographic area and ocean bottom topography (numbers indicate waterdepth). (6) Altimeter relative profile.


1000 r.

--0.05 rrT/Hz

(GEOS-C based on

.1 1.0Frequency, Hz

1000 100 10Distance, km 31.4

FIGURE 3-27.—Geoid undulation spectrum of PuertoRican trench and Wiener filter transfer function.

Figure 3-30 presents ideas on how an inte-grated system for marine geodesy could bebuilt around the short-arc coast-to-coasttracking of a radar altimeter. Signals Tlt T2,and T3 are time references transmitted by thesatellite to synchronize all the ocean andEarth tides logs.

Applications

The need for high-accuracy geoidal infor-mation to support studies such as crust de-formation and solid-Earth models is wellknown (ref. 3-53). Marine geodesy is con-cerned with two main applications: the po-sitions of objects on the Earth solid surfacein a marine environment, and the averageposition of an object on the water surface.

Requirements of users for each categoryare summarized in tables 3-V and 3-VI (ref.3-54) and are expressed in terms of neededprecisions in geocentric (#, A.) position andheight (h) above the geoid. Many of the spe-cifications can only be fulfilled if a marinegeoid has been determined with a relativeprecision of better than 1 m (==50 cm) andan absolute precision of a few meters (*»5m).

A secondary requirement for high-accuracy

7.85 3.14xs, km

FIGURE 3-28.—Spatial filter transfer function, wherealtimeter altitude is 440 km and antenna beam-width is 1.5° (rectangular gate). (From ref. 3-50.)

(<10 cm) geoidal information is that of sup-porting ocean topography activities. If ocean-ographic applications of remote-sensing ac-tivities are to achieve their full potential, itwill be necessary to define the geoidal, orpermanent, contours to a resolution compara-ble to the magnitude of oceanographic effetcs.

OCEANIC TIDES

Historically, tidal research has concen-trated on the immediately practical problemsof computing tide tables for the major ship-

— Undulation niter

FIGURE 3-29.—Derived weighting function for geoidaldata processing.


T.. T,

TS 'Tracking station (laser)ETO • Earth tides observatoryCTG - Coastal tide gage

DSTG • Deep sea tide gageRB«Time relay buoyT3= Time references

FIGURE 3-30.—Concepts of an integral coast-to-coastradar altimeter system for marine geodesy.

ping ports. Unfortunately, this can be donequite accurately by a purely empirical processthat teaches one almost nothing about tidaldynamics. Since the work of the 18th- and19th-century French mathematician, La-place, it has been clear that the proper under-standing of tidal dynamics demands a knowl-edge of the tides in the open ocean, wherethey are generated by the Newtonian gravityfield of the Moon and the Sun. Such knowl-edge would not only be a fundamental con-tribution to oceanography but would also, inprinciple, enable the computation of thepropagation of tidal energy into the shallowseas and estuaries and, hence, give predic-tions of both tidal elevations and streams inany location that may be required for futurepractical purposes.

No one has yet succeeded in attaining thisknowledge. Computing the disturbing gravi-

TABLE 3-V.—Positions of Motionless Objectson the Solid Surface of the Earth

Precision, m A

Activity

Coastal landmarks . .Island positioning . .Deap sea mining . . .Drilling platforms . .Shallow water

topography . .Geologic mapping . .

0,\

1201020

2050

h I

0.5203

bsolute (A)or

Relative (R)

RAAR

RR

tational field that drives the system is fairlyeasy, and Laplace's field equations have beenwritten in countless learned scientific papers,but their solution for the complex geometryand topography of the actual ocean presentsgreat difficulties. The difficulties arise partlyfrom the large number of possible normalmodes of oscillation, partly from the fact thatall the ocean basins interact (so that partialsolutions for a single ocean are impossiblewithout independent tidal measurementsaround its entire boundary), and partly fromthe necessity to model the dissipation that isconcentrated in the intricate network of shal-low seas. Analytical solutions in geometri-cally defined basins are only of qualitativevalue. Numerical solutions for actual oceantopography tax even the largest modern com-puters and ultimately only show the extremesensitivity to fine topographical detail and tothe yielding of the Earth crust, factors pre-viously considered negligible.

The lack of definitive global maps for theoceanic tides is a major lacuna in moderngeophysical knowledge. Apart from the in-trinsic interest in defining the response of theocean to the known gravitational forces andthe driving mechanism for the co-oscillatingtides in all the shallow seas, reliable tidalmaps are fundamental to the solution of a

TABLE 3-VI.—Average Position of an Objecton the Surface of the Water

Precision, m

Activity , X h

Absolute (A)or

Relative (R)

Oceanographicbuoys

Oceanographicship on station

Drifting buoys ..

Oceanographicvessel en route

Submarine cableoperation . . . .

Hydrographicvessel on station

70 —

35 —

50 —200 —

100 —

100 —

5 —70 —

RA

R

R

RA


number of other geophysical problems. Re-searchers on tides of the Earth and of theatmosphere, on geomagnetisms, and on themomentum balance of the Earth rotation areincreasingly finding their work hinging onthe oceanic tides. They turn to the ocean-ographers, who are unable to supply therequired data.

State of the Art

Approximately six maps of the M2 tide(lunar semidiurnal constituent) exist thathave been computed for most of the world'soceans in recent years by various methodsand assumptions. Assumptions vary, but themost fundamental divisions occur (1) be-tween those who force the solutions to agreewith assumed continental boundary tidal con-ditions and those who eschew the use of allempirical data, and (2) between those whoattempt to allow for crustal yielding andthose who assume a rigid Earth. (For amodern review, see Hendershott (ref. 3-55).) Two of the most authoritative and ad-vanced results are compared in figure 3-31.They are in fair agreement on certain fea-tures, such as the tidal configuration in theNorth Atlantic, but their results for theSouth Atlantic are quite different. Such dis-agreement, here and in other oceans, is typi-cal of all cotidal maps to date. Ironically, it isextremely difficult even to obtain the meas-urements necessary to find out which inter-pretation, if any, is correct. Maps of thediurnal tides are rarer, but those that arepublished show a similar lack of agreement.

Much of the controversy about global tidalcomputations centers on ocean/land interac-tion, which is rather complicated, but thereare also limitations due to the lack of properinformation to be fed into the computermodel from the boundaries bordering shallowseas. A typically crucial area for the AtlanticOcean appears to be the Patagonian Shelf,but reliable data from here and many otherplaces are sparse. Computer models wouldalso benefit from measurements across trans-oceanic sections, which would provide bound-ary conditions for separate evaluation of

smaller zones. Isolated spot measurements atcertain places where the tidal amplitude isknown to be large would also be valuable.

Seabed pressure sensors for measuringtides in the open sea have been increasing innumbers over the last decade, but they areexpensive devices with a high risk of loss andrequire expert supervision and many ship de-ployments. The sensors are supported onlyby a few nations that have well-developedoceanographic technologies; thus, the chancesof obtaining data in the foreseeable future,from most of the Southern Hemisphere atleast, are remote. To date, the only tidal datathat have been taken from the open sea(apart from islands) have been from withina few hundred kilometers of North America,Northwestern Europe, and South Australia.Monitoring the tides in all the world's oceansby direct measurement, or even obtaining theminimal data that might enable computation,is not possible at this time.

Radar altimetry from spacecraft will trans-form the measurement situation. The differ-ence between satellite range and altimeterheight above the sea surface, suitably cor-rected for various biases and errors, will varyfrom a variety of causes discussed elsewherein this report (ref. 3-45). The variables maybe divided into (1) permanent or quasi-permanent features of ocean topography,(2) features that vary in time at any givenEarth location, and (3) tracking and othererrors. Variables of type (1), including thegeoid and steady dynamic features, althoughapproximately 1 to 100 m, can in principlebe extracted either from independent knowl-edge or, in due course, by averaging from re-peated satellite passes. Variables of type (2)include the tides, up to approximately ±2 min amplitude, and also include random varia-tions caused by the weather, on the order of0.1 to 0.3 m. (Transients such as surges andtsunamis may be ignored in this context be-cause their amplitudes are appreciable onlyin shelf seas.) A large proportion of theweather influence is a direct reflection of at-mospheric pressure at 1 m of sea level perN/m2. This can be removed by means of


(b)

FIGURE 3-31.—Cotidal maps of the Atlantic Ocean using different methods and bound-ary conditions, (a) Cotidal maps computed by Pekeris and Accad, 1969 (ref. 3-56).(b) Cotidal maps computed by Hendershott, 1972 (ref. 3-55).

synoptic meteorological data computed onland. The residual signal will therefore beconsiderably tide dependent, provided vari-ables of type (3) are well below 1 m. Thisdependence is further enhanced by using thestrong coherence that exists between anyoceanic tidal variable and the perfectly calcu-lable tide-generating potential of the Moonand Sun (ref. 3-57).

Requirements

As in all applications of satellite altimetry,the only basic requirement demanded is ac-

curacy of the interpretable signal. Becausemodern trends in altimeter design enable itsaccuracy to be reduced to a few centimeters,the major requirement is in the accuracy oforbital tracking over as wide a global cover-age as possible. Stating minimal require-ments outside the context of discussion of ap-proaches to various problems is difficult, butthe situation may be summarized as "givenany reasonable accuracy, one can expect tolearn so much tidal information and nomore." With 1- to 2-m accuracy, there is somehope of gaining useful information about theoceanic regions with largest tides, such as


parts of the North Atlantic. With 0.5-m ac-curacy, one should certainly obtain good in-formation from such regions and could hope-fully consider wider oceanic fields. Completetidal definition over the world's oceans wouldrequire 5- to 10-cm accuracy. In any case,continuous signals for many months or yearsare assumed.

An altimeter "footprint" of approximately100 km should be adequate for most tidalwork in the deep ocean. To resolve the smal-ler horizontal scale of tides in most shelf seas,resolution better than 20 km would be needed.

Technical Approach

At some stage in the interpretation of thealtimeter tidal signal, one will have to allowfor the fact that the signal differs fundamen-tally from the usual land-based concept of theocean tide. If z' is the anomaly in sea levelsensed by a tide gage, that sensed by thealtimeter is z = z' + z", where z" is the tidal ele-vation of the Earth crust, approximately 0.2m. (See ref. 3-55 for a full dynamical dis-cussion.)

A large part of z" consists of a direct yield-ing of the body of the Earth to the gravita-tional forces, which is easily calculable, butthere is a smaller residue of approximately0.1 m that can only be calculated in terms .ofglobal integrals of the ocean tide itself (i.e.,the "loading tide"). Clearly, for some pur-poses, one can interpret z in terms of z'merely by approximating to z" by the simplebody tide, with small loss of accuracy. Forhigher precision, a fair approximation to theloading tide could be computed from one ofthe existing rough cotidal maps. The entiresituation procedure is, in any case, one ofsuccessive approximations.

Two possible approaches to data analysisare possible. One method relies on obtainingmany traverses of a given track zone; thatis, for example, 500 km wide. After correct-ing for the geoid (which preferably shouldnot vary too steeply in the chosen area), foratmospheric pressure, and for any otherknown anomalies, the successive altimeterreadings in any 500-km-square section are

treated as a signal that is partly coherentwith both the diurnal and semidiurnal partsof the tide-generating potential, calculatedat the appropriate times, together with arandom noise due to tracking errors. Onewould attempt to extract perhaps as manyas eight arbitrary constants to describe thecoherent signals in terms of "admittances"to the potential, with one further arbitraryconstant to absorb a residual error in thegeoid.

Applying numerical filters aimed at isolat-ing the Afj, tide is similar in principle (ref.3-58). With orbital height errors of ap-proximately 0.5 m and tidal amplitudes ofapproximately 1 m, one could obtain solutionsof reasonable accuracy from about 50 passes.Larger numbers of passes give greater ac-curacy or permit more refined definition ofadmittances. Several possible variants couldbe considered. The admittances could betreated as arbitrary functions of distancealong an entire oceanic track, or as localizedfunctions of two dimensions, constrained tosatisfy the tidal dynamical equations. Insome places, the diurnal tide could be ignored.The ultimate result would be empirical tidalmeasurements in a network of grid areascovering a chosen zone.

The other approach would be to compareevery traverse of an ocean, on whatevertrack, with the corresponding simultaneoussolution of a given computer model of oceanictides, again after removing the major part ofthe geoid and other variations assumedknown. The residual variance about space-located averages would be compounded be-cause of tracking errors and errors in thecomputer model. The computations wouldbe repeated for the same data with variationof frictional and other uncertain parametersin the model until the residual variance wasminimized. One could then say that themodel output was "correct" for the oceanarea considered, and a refinement of the geoidwould also result.

In any approach to tidal analysis of al-timeter signals, the accuracy of the solutionwill obviously be a function of the tracking


error variance divided by the number oftrack repetitions. Any means of reducingthis quotient is beneficial and will extendthe scope of any investigation; for example,to regions of smaller tidal range. With cur-rent practices, vertical errors can apparentlybe as little as 0.2 m in the vicinity of track-ing stations, worsening to 1 to 2 m in areasremote from these stations. Increasing thenumber of tracking stations would have obvi-ous advantages. However, oceanic zones existwhere the tide is already known to an ac-curacy of 0.5 m or better, either from pelagicmeasurements or by approximate calcula-tions where the tidal range is known to bevery small. These zones could be used as"surface truth" to identify the trackingerrors where they might otherwise be large.A program can also be envisioned of pelagictidal measurements in oceanic zones selectedfor this purpose. (The surface measurementsdo not necessarily have to be taken simul-taneously with the altimetry.) However,0.5-m accuracy is hardly sufficient to resolvetidal signals in the controversial South At-lantic (fig. 3-31), where, according to islanddata, the amplitudes are approximately 0.3 m(ref. 3-59). Such areas would have to becomputed from altimeter measurementsalong boundary zones where the tidal rangeis larger.

Applications

Finally, assuming that all errors can bereduced to useful levels and all the analysishas been done, what applications wouldknowledge of ocean tides have, both to scienceand to the practical world of mankind ? Thescientific uses have been outlined by Munkand Zetler (ref. 3-60), and their conclusionsare restated here.

1. The knowledge of oceanic tides on aglobal basis is necessary to properly interpretnearly all geophysical measurements at tidalfrequencies, including measurement of grav-ity, magnetic field, etc.

2. Measurements of oceanic tides will leadto improved tidal models and ultimately toimproved calculations of tidal dissipation in

the ocean, which, when subtracted from dis-sipation obtained from changes in the lengthof day, will yield the dissipation in the solidEarth. These measurements will provide im-portant information concerning the plasticproperties of the solid Earth.

3. Oceanic tidal currents, when coupledwith electric and magnetic field measure-ments on the ocean bottom, can probe the con-ductivity (and thus temperature) within theupper mantle as a function of frequency(depth). Horizontal gradients of tempera-ture should exist near volcanic belts, in areasof downthrusting plates, and at postulatedupper mantle "hotspots."

4. Finally, the response of the Earth tothe shifting weight of the oceans, as meas-ured by Earth strain or tilt meters, givesinformation about the local elastic structureof the Earth. This, too, is expected to varynear margins of lithospheric plates.

Knowledge of oceanic tides at the bound-aries of shallow land-bounded seas enablesbetter computation of the tides and tidalstreams in such seas for purposes of naviga-tion of ships, tidal power-generating schemes,and flood warning (see the following sec-tion). The inland Earth tide, which requiresthe oceanic tide for proper computation, isof value to geodetic leveling, surveying pro-cedures, and the accuracy of astronomicalsiting.

STORM SURGES AND WIND SETUP

Storm surges may be defined as transientanomalies in sea-surface elevation (and cur-rents) caused by storms, with time scales inthe region of 10' to 105 sec, horizontal scalesof approximately 102 to 10" km, and verticalexcursions as high as approximately 6 m.The associated currents may be regarded asdeterminable from the gradients of elevationand will not be considered here as inde-pendent variables. Storm surges are largelyconfined to shelf seas, partly because the dy-namic effect of wind stress is inversely pro-portional to water depth, and also becausecritical wave speeds in shallow water are


comparable with speeds of propagation ofweather systems, causing more efficientcoupling with the atmosphere than with thedeep ocean. Storm surges have been intenselystudied in certain sea areas where theypresent a frequent danger of flooding to ad-jacent lowland, such as in the southern partof the North Sea, the Adriatic Sea nearVenice, and the shelf seas bordering theEastern and Southern United States. Flooddanger also depends on the coincident state ofthe tide, and, to some extent, study of stormsurges and local tides are inseparable.

An important ingredient of some stormsurges is a quasi-static pileup of water towardthe coastline, caused by the onshore windstress and the radiation stress in the as-sociated waves. This effect is usually called"wind setup." In some areas, where windstrength and direction persist for many daysand coastal topography prevents the forma-tion of steady currents, wind setup may forma steady deformation of sea level of somedecimeters. However, although such a setupmay not be thought of as a storm surge, itsdynamics are not essentially different.

When tidal predictions were sufficientlywell developed for sea level to be approxi-mately partitioned into "astronomical tide"and "surge residual," researchers observedthat the latter was to some extent a directfunction of local weather and tried to ex-press it as a simple linear regression of windcomponents and local atmospheric pressure.Results varied from passable to poor, andobviously neglected the spatial as well as thetime history of the weather patterns and thetransmission of energy from other parts ofthe shelf in the form of a free wave (some-times called an "external surge"). Morecomplex response operators, which take intoaccount the spatial and temporal wind andpressure fields, have been developed withsome success, but scientific practice hastended to veer away from such empiricalapproaches, whereby "prediction constants"are estimated from data by a least-squarescalculation. The modern tendency is to pre-dict storm surges by numerical solution of

the wave equations over a network of smallareas covering the shelf sea concerned; ateach area, the three components of atmos-pheric stress are specified at regular incre-ments of time.

State of the Art

Three examples of disastrous surges areshown in figures 3-32 and 3-33. Figure 3-32shows elevations in the North Sea computedover a numerical grid from the dynamicalequations. The surge of February 17, 1962,reached more than 3 m above tide level andcaused severe flooding in the Cuxhaven-Hamburg area of Germany. Figure 3-33shows the surge height, exceeding 7 m, alongthe U.S. gulf coast during the passage ofHurricane Camille.

The principal features and causes of stormsurges are fairly well understood and, withincertain limits, can be roughly reproduced byintegrating the wave equations with drivingstresses from the wind and pressure fields.But the demand continues for further re-search and information to help understandthe finer scale features, such as their inter-action with the tide and with coastal topog-raphy, and the response to unusual weatherconditions. Because wind stress is the mostimportant factor, success in predicting orunderstanding storm surges depends on theaccuracy with which the winds over the seacan be specified. A serious hindrance to re-search is the lack of observational data ofsurface elevations offshore. With few excep-tions, nearly all the observations of surgescome from coastal tide gages, but these ob-servations can give a distorted and/or limitedpicture of their main structure in the opensea because waves and currents cause steepgradients near the shoreline. Like the oceantide problem, the logistics of offshore re-cording are difficult. Stormy shallow seaspresent a particularly hostile environmentfor recording equipment and data trans-mission systems.

Active microwave technology could helpresearch on storm surges in two ways. Winddetermination over the sea by the "scatter-


FIGUHE 3-32.—North Sea storm surge, (a) September 14,1962. (b) September 17, 1962.

ometer" technique would give a more ac-curate picture of the driving forces thanis usually obtainable from computationsbased on the pressure field, and altimetertracks over surge-ridden seas would givevaluable information on the open-sea struc-ture of the surge itself.

The scatterometer data would be more fre-quently available, because the wide swathof the instrument would include a given seaarea on a large number of tracks. By thenature of altimeter tracks, the chance ofsensing a large storm surge, which may occuronly once or twice in a year, would be cor-respondingly less. However, as in the caseof tsunamis, a single fortuitous track couldprovide useful data for checking a computermodel that would be virtually unobtainableby conventional methods. Any track acrossthe sea in calm weather would naturallyprovide useful information about the tides.

Requirements

Requirements of the scatterometer wouldbe at least twice-daily coverage of a givensea area with dimensions of approximately1000 km, with interpretation to windspeedover a grid spacing of approximately 100 km.To improve on surface-based estimates, wind-speed would require precision to approxi-mately 2 to 3 m/sec. Because, in storm condi-tions, the wind field may change considerablyin a few hours, the winds measured by thesatellite would not in themselves be sufficientto follow the full time history of a stormsurge, but the data for intermediate timescould be supplied from a computed weathermodel, which would itself be improved bythe scatterometer data.

The altimeter "footprint" would need to besmaller than that required for oceanic tidalresearch. A diameter of approximately 20 kmwould be satisfactory. Vertical precision


Calculated surge maximum (large grid model)

Calculated (small grid model)

Both calculated values extrapolated to coast

60 80 100 120 140Kilometers along coast from hurricane landfall

160 180 200 220

(a)

MississippiCity v

Gulfport,

™- Long Beach'- Pass Christian

Cat Island

(b)

>u<yar^?> Ship Island Horn lsland

-— Petit Bois Island''

vGrand Bay

Dauphin IslandGulf Shores

FIGURE 3-33.—Comparison of observed and calculated surge height maximums along theGulf Coast during Hurricane Camille. (a) Surge height compared to distance alongcoast. (6) Geographical map.

of 0.5 m in the altimeter signal would beuseful for measuring a large surge, but be-cause the chances of encountering a largesurge are slight, one would like in any caseto record the more frequent small surgeswith an accuracy of 0.2 m. At the same time,orbital tracking errors and other systematicerrors that do not vary by more than a fewcentimeters in 100 km can be neglected, be-cause the seas of interest are always wellprovided with coastal tide gages and onlythe gradients of the sea surface are reallyrequired from the altimeter.

Applications

One application of these data would be toimprove computer models designed to fore-cast storm surges. These models in turn areused as a basis for warning the public ofdanger from floods and for alerting sea de-fense operators. Another use of these models,of growing importance, is in warning largeships of unusual shoaling caused by "nega-tive surges"; that is, depressions in sea levelassociated with the more familiar positivesurges.


CURRENTS

Until recently, the methods of study inoceanography were limited to the standardshipbound tools. By comparing the area thatcan be covered by a ship in a reasonableamount of time to the total ocean, trying tounderstand the entire ocean by using thestandard methods is equivalent to trying tounderstand the nature of a very large forestby painstakingly examining each tree. Manyof these methods and their shortcomings arediscussed in detail by Fomin (ref. 3-61)and Neumann (ref. 3-62). This does notimply that the traditional methods are use-less; in understanding some particularaspects of the ocean, they are actually in-dispensable. However, as far as the globalocean model or large-scale phenomena areconcerned, the traditional methods are noteffective enough to produce a synoptic view.This shortcoming can be compensated to alarge extent by the newly developed remote-sensing techniques.

Remote-sensing techniques are the onlyobservation methods that are capable of fastscanning over a vast area. Because of thisunique capability, these techniques will be-come the most important tools in large-scalephysical oceanographic research. However,because of the way data are collected, allthe information thus obtained is confined tothe surface. But the environment of theocean is a complicated one; all the physicalprocesses in the ocean are controlled by bothsurface and subsurface parameters. Theseparameters act and interact to produce thephenomena actually observed. Remote-sens-ing techniques are only effective in measur-ing those phenomena controlled by surfaceparameters. Fortunately, such phenomenainclude most of the crucial problems in physi-cal oceanography. Furthermore, some sub-surface phenomena such as internal wavescan also be inferred indirectly from micro-wave measurements of the surface.

The most severe deficiency of the tradi-tional methods is the time factor. All oceanphenomena change with time as well as space.The traditional methods, because they are

shipbound, are highly time constrained. Theycannot produce any synoptic data of globalscale. As a result, the so-called global dataare actually patched-up works from differ-ent times and different cruises. For example,the meandering of the Gulf Stream is a well-known feature of the motion. If, as shown infigure 3-34, the stream path from A to B isPI at time fi and a ship surveys the areain time interval t.2 — tlt the path changes con-tinuously to P2 in the same time interval.The result of the survey will produce a pathP that is not a path of the event at any time.

Another view of this deficiency of timefactor is the spatial restriction. To cover areasonable area in a given time interval, theship will have to move with a minimal amountof delay, thus occupying fewer stations.Therefore, the gaps will be wide. This dis-crete set of data will create problems wheninterpolation is used in processing the data.

The combined effect of the space and timedeficiencies is to eliminate the possibilityof getting field data to improve the accuracyof global condition prediction, which dependsnot only on a local system but also on theinterrelation between different systems. TheGulf Stream, for example (fig. 3-35), isdominantly controlled by the wind field of thenorthern Atlantic as a whole, rather thanby local winds. However, events in the south-

Pj • path of the current at time taP2 • path of the current at time t2P • path given by shipboard measurement

FIGURE 3-34.—Sketch illustrating problem with tra-ditional methods of data gathering. Time for aship to travel from A to B is t-—ti.


ern Atlantic will eventually feed to the GulfStream through current systems (nos. 15, 2,20, etc., in fig. 3-35).

Although severely limited, the traditionalmethod can produce data on the water columnunder the surface. Such information is im-portant in studies on interval waves andbaroclinic current, etc.

Advantages of Remote Sensing

Because remote-sensing techniques arecapable of rapidly scanning large areas, theycan be used to obtain continuous data on aglobal scale. Such data are essential to thepredictive models on weather and sea state.From remote sensing, useful data can be ob-tained concerning the mean sea level, the seastate, and the surface temperature field.(Also see fig. 3-36.)

The first method of measuring ocean cur-rents by remote sensing was by satellite

tracking of drifting buoys. This was theFrench meteorological EOLE satellite system(Cooperative Application Satellite CAS-A)that was launched from the NASA WallopsIsland Station on August 16, 1971. TheEOLE data collection and positioning systemhas been operational since early September1971. The satellite tracks and collects datafrom drifting buoys during each pass byinterrogating the buoy transponders. Thestored range and range-rate observationsare later transmitted to ground telemetrystations and processed at the Centre Nationald'Etudes Spatiales (CNES) computing cen-ter in Bretigny, France. In the future, theobservations will be processed at the NASAGoddard Space Flight Center.

Although originally designed for balloontracking, the EOLE system has proved tobe an ideal tool for applications such asdrafting objects positioning (icebergs and

E30°90°E W 0" E

70° —-

E 30° 90° 150° E E 180° 150° 90° W W60 0

70

Note; 1 = North Equatorial Current, 2 • South Equatorial Current, 3 =• Equatorial Countercurrent, 4 • Guinea Current, 5 • AntillesCurrent, 6 ° Florida Current, 7 = Gulf Stream , 8 • North Atlantic Current, 9 • Norwegian Current, 10 • I rminger Current, 11 -East Greenland Current, 12 • West Greenland Current, 13'Labrador Current, 14 • Canary Current, 15 • Guiana Current, 16'BrazilCurrent, 17 • Falkland Current, 18 = Antarctic Circumpolar Current, 19 • Agulhas Current, 20 • Benguela Current, 21 • Kuroshio,22 * North Pacific Current, 23 • California Current, 24 • Aleutian Current, 25 - Oya Shio, 26 • Peru Current, 27 • East AustralianCurrent, 28 =West Australian Current, 29 = Somali Current, 30 • Mozambique Current, 31 • (Indian) Monsoon Current (NorthernHemisphere summer).

FIGURE 3-35.—World chart of ocean currents during Northern Hemisphere winter (ref.3-62).


Remote-sensingtechniques

Data

Activemicrowave

Directinformation

Applications

Tsunami warnings

Bottom topography

Density structure

Global currents

Prediction model

FIGURE 3-36.—Uses of remote-sensing techniques.

buoys). Routine operation can provide posi-tions with accuracies of approximately 2 to3km.

The Virginia Institute of Marine Sciencesstarted drifting buoy experiments using theEOLE positioning system in September 1972.The main objective is a study of currentsalong the coast of Virginia, outside the mouthof the Chesapeake Bay. The buoys are rounddisks carrying the radio and EOLE antennas,under which is attached a cubic case contain-ing the electronics. A 5- to 20-m cable linksit to a submarine crosslike sail.

Although the drifting buoy system usessatellites for data acquisition, it has manyof the disadvantages of the traditionalmethods discussed previously. In this sense,this system does not offer many of the realadvantages achieved by remote sensing.

The most promising method for measure-ment of ocean-surface currents has been de-veloped from increasingly accurate radar oractive microwave altimeters. The basis forcurrent measurement using satellite-bornealtimeters is that most important ocean cur-rents are driven by changes in the ocean-surface elevation.

By measuring the surface slope alone, thesurface-current velocity can be determined.

This calculation is not new, but the practicalimplications of this approach have only re-cently been realized through the possibilityof accurate measurements of ocean-surfaceslope by satellite-borne radar altimeters.The accuracy to which the slope is measureddirectly determines the accuracy of the sur-face current.

Measurement of surface current seemssimple; however, difficulties arise becausedata obtained from satellite altimeters stillcontain substantial amounts of noise. Toanalyze the data and to calibrate the instru-ment, the satellite measurements must berelated to models of the ocean-surface relief.To do this, one must rely on in situ hydro-graphical data available in the literature.The most useful data are measurements thatlead to surface relief by means of dynamicheight.

Requirements

The expected change in surface elevationdue to geostrophic currents for areas of well-defined current integrity is approximately0.5 to 1.0 m change in elevation over distancesof 25 km or larger. Current features in theseregions, such as eddies or weaker subcurrentsystems, cause a height variation of approxi-


mately 25 cm. To detect the transient topo-graphic signatures of these magnitudes, thepermanent topographic features must first bedetermined to somewhat higher resolution.On the basis of defining permanent featuresto within 10 cm with 80 percent confidence(over a 25-km spatial extent), using nominalsatellite groundtrack velocities, the requiredresolution translates to 12 cm rms for alti-tude samples that are statistically inde-pendent for a 1-per-second data rate. Insome special situations, the height changemay occur over distances less than 25 km,in which case a resolution (footprint) ofapproximately 1 to 3 km is recommended.

Technical Approach

The present technical approach is to meas-ure the surface elevation and use this meas-urement to compute surface currents. Com-puted currents can be compared withcurrents from in situ measurements of dy-namic height or measured elevations com-pared with measured values of dynamicheight. At places where the density profileis known, the subsurface currents may alsobe computed. The accuracy of the surfacecurrent computed by this method depends onthe accuracy of the measured height and thegeostrophic assumption. In most situations,the geostrophic assumption is good and theaccuracy of the surface elevation is the onlyparameter affecting accuracy.

If subsurface currents are desired, thenthe density profile is also required. Thedensity profile cannot be measured remotely;therefore, any computation of subsurfacecurrents will require local measurements.

The present approach is based on data ob-tained solely from an altimeter. The in-creased development of remote sensing willpermit the measurement of many parts of thewave spectrum. Presently, the high-fre-quency or high-wave number part of thespectrum can be measured by active micro-wave backscatter as shown by Krishen (ref.3-63), Daley (ref. 3-64), and Valenzuelaet al. (ref. 3-65). In addition, part of thelow wave number spectrum can be obtained

for an altimeter by the measurement of sig-nificant wave height Hl/3 or by imagingradar. Huang et al. (ref. 3-66) have shownthat there is a relation between nondirec-tional wave spectrum, surface wind, and cur-rent. Moreover, they have determined thechange in wave spectrum with surface cur-rent using windspeed as a parameter. Thiswas done by using the Pierson-Moskowitz-Kitaigorodskii (ref. 3-67) spectrum for thezero current condition. Figure 3-37 showsthe wave number spectrum as a function ofsurface current u for a windspeed W of10 m/sec. If the surface wind and the spec-trum can be obtained, this would provide analternate method of current measurement.

It may not be possible, at least in the nearfuture, to remotely sense the entire wavespectrum. Because of this, Huang et al. (ref.3-66) have computed the relation betweenrms surface roughness and current. Surfaceroughness also depends on surface wind, andfigure 3-38 shows the rms slope as a functionof windspeed with current speed u as a pa-rameter. In principle, surface roughnesscould be used for current measurements.Figure 3-39 contains the same informationas figure 3-38, but with windspeed as theparameter.

W -10 m/sec

.05 .1 .15Wave number, nr1

.2 .25

FIGURE 3-37.—Changes of wave number spectra fora 10-m/sec windspeed under different current con-ditions.


u • -0.3 m/sec

I6

15

7 2c(DO>

E 1

§Of

1 .3 .5

5 10Windspeed, m/sec

15 20

FIGURE 3-38.—Variation of rms surface slope withwindspeed, using current speed as a parameter.

The surface roughness or the energy in thehigh wave number waves increases quiterapidly with wind (see figs. 3-38 and 3-39).Thus, an increase in the energy in the highwave number region can be caused by eitheradverse currents or the wind. At high wind-speeds, the effect caused by the wind willprobably overshadow any change in wavespectra caused by currents. Therefore, it ispostulated that only in low winds can sur-face roughness be used to measure current.At this time, however, additional efforts mustbe made to separate the two effects. Thisalternate method of current measurement, atleast at low windspeeds, can be used as acheck (also obtained by remote sensing) onthe currents obtained from the satellitealtimeter.

The ultimate goal must be remote sensingof the complete directional wave-height spec-trum. This is slightly different from thepresent approach of measuring the high-frequency spectrum by means of backscatterand of inferring wind by a relation betweenwind and the wave-height spectrum at spe-cific (high) wave numbers. If, by usingvarious remote-sensing systems and possiblymodels of the spectra, the total wave-heightspectrum can be obtained, then the high-frequency portion could be used to inferwind; and the rest of the spectrum, includingthe high-frequency portion, would be used

o .5 iCurrent speed, m/sec

FIGURE 3-39.—Variation of rms surface slope withcurrent speed, using windspeed as a parameter.

to measure currents and other parameters ofinterest (e.g., wave energy, significant waveheights, etc.).

Applications

Virtually all ocean waves are influenced byocean currents and, thus, information is re-quired on these currents. Any activity usingships will be able to use surface current in-formation to reduce time en route, to lowercosts or fuel consumption, and to avoid badweather. Some examples are shipping andnaval operations, fishing operations, andsearch-and-rescue operations. Access tomeasured or predicted ocean-surface cur-rents will be of immediate benefit to all theseusers.

Long-range weather forecasting and/orweather models also require a global pictureof the surface currents. The proposed cur-rent information will benefit both. As theocean becomes more crowded, pollution andthe ability to predict pollution becomes in-creasingly important. Surface currents arealso needed for predictions of pollution fromany known source. Moreover, slicks detected


by a measurement of the high wave numberspectrum can reveal pollution or salt in-trusion.

GLOBAL WAVE STATISTICS

Any dynamic phenomena on the oceansurface, such as ship motion and forces onanchored or stationary platforms, are gov-erned by the actual and/or expected sea state.Because the ocean surface is random, theonly way to define it is by statistical descrip-tion. The usual method used to describe wavemotion is by the wave-number or wave-fre-quency spectrum and the expected variationof this spectrum under different conditions.Therefore, models to predict the expectedwave spectra for given geographical andmeteorological conditions have received muchattention.

The primary mechanism by which wavesare generated is the surface wind. Moreover,in the open ocean (i.e., without boundaries)with no currents, the wind uniquely deter-mines the wave spectrum. However, the re-lationship between wind conditions and thesea state is still not completely resolved.When boundaries or currents are present, theprediction of wave statistics becomes in-creasingly difficult.

Because of the inability to uniquely pre-dict the wave spectrum, the approach mustbe to measure the wave spectrum, or wavestatistics, so that mathematical models maybe refined by observations and thus yield thedesired data. The measurements must bemade on a global scale because some of thefactors (e.g., storms, geostrophic currents,etc.) that affect the spectrum extend forhundreds of kilometers. Because of theselarge-scale processes that influence the seastate, measurements at a single point or atseveral points at different times are difficultto interpret. Therefore, remote sensing offersa great many advantages in developing globalwave statistics. Such statistics have twoimmediate and widely useful purposes:

1. To establish reference conditions to beused for design and planning, and for evaluat-ing global wave models.

2. To monitor any changes from the refer-ence condition, such as those caused bystorms, currents, topography, pollution, etc.The change in wave spectra can be used toevaluate the size and nature of these phe-nomena.

A less-immediate goal is the generation ofimproved methods of predicting and model-ing ocean-wave spectra.

State of the Art

The first work on a model to predict ocean-wave characteristics was undertaken byCornish (ref. 3-68). Little work followedthis until Sverdrup and Munk (ref. 3-40),motivated by the necessity of wave informa-tion for naval operations, developed a modelto predict wave characteristics. Their analy-sis was concerned with a relation betweenwave heights and periods and did not dealwith the concept of a wave spectrum. Neu-mann (ref. 3-69) was the first to deal withthe concept of a wave-frequency spectrumfrom which statistical properties could beobtained. Using Neumann's concepts, Pier-son et al. (ref. 3-70) developed a model forthe prediction of wind-driven sea and oceanswell spectra. The model has experiencedconsiderable modification and improvementand still remains the most widely used modelfor forecasting ocean wave characteristics orstatistics. The concept of an equilibriumspectrum was introduced by Phillips (ref.3-71). This provided additional insight intothe parameters that must be included in thegravity wave spectra and on the shape of thehigh-frequency part of the spectrum. A re-cent summary of wave spectra models for acomplete range of wave numbers has beengiven by Pierson and Stacy (ref. 3-15).

Generally, wave spectra models have beenquite successful. For example, a NOAA Na-tional Data Buoy Center report (ref. 3-72)provides analyses of ocean-wave spectra asdetermined by hindcasts. The development ofthese models has required the acquisition ofa large amount of information about waves.Data concerning waves are routinely col-


lected by ships at sea. Wave heights, periods,and dominant direction of travel are recordedand radioed ashore in a special code by manyships at 00:00, 06:00,12:00, and 18:00 G.m.t.These measurements are collected, stored atnational data centers, and used to producestatistical summaries of the waves for thevarious areas of the oceans. The cost of col-lecting, storing, and analyzing data of thiskind amounts to millions of dollars annually.

Also, a very small number of weatherships have wave recorders that measure therise and fall of the sea surface as a functionof time at a point. These spectra have beenstudied and used to develop wave-forecastingtechniques and to aid in the design of ships.Wave records of this nature are obtainedroutinely at four or five locations in theNorth Atlantic and at one location in theNorth Pacific.

A consortium of oil companies has usedwave recorders on their drilling rigs to col-lect a large amount of wave data duringhurricanes in the Gulf of Mexico. The sub-stantial effort to obtain wave data in theseways is a measure of its present value. How-ever, this level of effort is inadequate forthe future needs of the United States and theother nations of the world.

The reports from ships are estimates, notmeasurements, of the wave conditions. Theyare also concentrated along shipping lanesand do not adequately describe conditions ona uniformly spaced grid over the oceans.Moreover, they tend to be substantially inerror, and the data obtained before a fewyears ago had built-in biases that tended togroup reported wave heights near 5 and 10 mand to report high waves inadequately. Inone study, in which these estimates of waveheights were compared with measured waveheights, the estimates frequently differedby a factor of 2 (either too high or too low)from the measured values. Approximately1200 ships report wave conditions in thisway every 6 hr, primarily in the NorthernHemisphere.

The shipborne wave recorder data are ofhigh quality and have proved invaluable for

many scientific investigations. However, thenumber of ships is too few and their loca-tions too random to provide an adequatedata base.

In addition, predictions of wave spectraare based on local conditions, primarily be-cause this is all the information that isavailable. However, very-large-scale phe-nomena (storms, global currents, etc.) alteror determine the wave spectrum. For thisreason, recourse must be continually madeto observations of measured wave character-istics typical of those given by Hogben andLumb (ref. 3-73). Moreover, in coastalregions, the open ocean waves must be con-sidered in combination with local topographyto develop realistic estimates of the wavespectra.

If global meteorological data were avail-able, more accurate models for the predictionof wave spectra could possibly be developed.Such models would need to be verified byglobal wave statistics, however. Therefore,remote sensing appears to not only offer theopportunity to evaluate and improve wavespectra but to be the only way presentlyavailable to do so.

Requirements

The complete wave frequency or numberspectrum need not be measured. Variousparts of the spectrum must be measured,however, so that present models can be usedto infer the complete spectra. In additionto the measurements of wave spectra or wavestatistics, data on meteorological conditionsare also required. These data are needed notonly at the point of wave measurement butalso globally, because of the large-scale phe-nomena that govern local wave spectra.

Technical Approach

The acquisition of sufficient data to de-velop a global picture of the local wavespectrum and its expected extreme is re-quired. By using presently available models(e.g., Pierson et al. (ref. 3-70) and Piersonand Stacy (ref. 3-15)), a complete wave


spectrum can be developed from the sig-nificant wave height H,/:< or the surface wind.Both together permit some checking of themodels. The significant wave height can beobtained by the present satellite altimeter,although the accuracy may not be as great asdesired. Surface wind can be obtained frommicrowave backscatter, but a higher accuracymay be needed.

Imaging radar can also be used to generatewave statistics. Imaging radar gives a pic-ture of wave patterns that is at present aqualitative picture of the wave field. There-fore, the fact that waves can be seen makesthe imaging radar a powerful tool for theoceanographer. The most exciting prospectis that of mapping wave patterns and wavebuildup during large storms. This wavebuildup occurs under cloud cover that onlyactive microwave sensors can penetrate. Itis in these large storms that the bigger, moredamaging waves are generated.

In addition to mapping storm waves, themapping of swell over continental shelveswill permit the study of wave refraction incoastal regions. The continental shelves willbe valuable real estate in the next fewdecades as a place to put man's machinery,such as offshore airports, powerplants, andoil rigs. The ability to predict wave climatesin these areas and to verify these predictionsby observations is another important contri-bution that imaging microwave sensors canmake to oceanography. Again, the ability ofactive microwave sensors to penetrate cloudsduring storms is a most important attribute.

The capability for global monitoring of seastate and ocean waves, which will be pro-vided by spaceborne imaging radar, is also avery important contribution. This is par-ticularly true for oceans in the SouthernHemisphere, where wave measurements inthe open ocean are so widely scattered thatmany areas of the ocean are essentially un-known. Little shipping and almost no landobservation points are present in the South-ern Hemisphere oceans between latitudes20° and 40°, so that the interaction between

wind and waves and the buildup of wavesis practically unmonitored.

All of this information can be used togenerate a global picture of the expectedwave spectra or the wave statistics char-acteristic of these spectra. These data wouldestablish a reference state for the ocean.This state could be considerably more ac-curate and detailed and with better resolu-tion than the present tabulation by Hogbenand Lumb (ref. 3-73) or those from the U.S.Department of Commerce (ref. 3-72).

With the acquisition of a reference con-dition, continued monitoring of the oceanscan be used to reveal any changes that occurbecause of storms or other large-scale phe-nomena. These two goals, the developmentof more accurate global wave statistics andthe monitoring of the magnitude of theirchanges, should be ample justification for theprogram.

The many advantages provide motivationto attempt to sense more completely the totalwave spectrum. At present Krishen (ref.3-63), Valenzuela et al. (ref. 3-65), andDaley (ref. 3-64) have used microwave back-scatter to measure the high wave number(above 0.1 cm-1) part of the wave spectrum.The use of the longer wavelength radar, orperhaps imaging radar, to sense the lowerfrequency portions of the spectrum shouldbe considered. This is because the ultimategoal must be a tabulation of the completewave-number spectrum and, in the future,possibly the directional wave-number spec-trum, on a global basis. Active microwavesystems are now the only sensing techniquethat can be used to measure wave spectradirectly.

Applicability

The use of remote sensing to establish areference picture of global wave statisticswould provide data needed to plan shippingroutes, design ships and/or offshore struc-tures, and aid in the planning of future ports.This reference condition is especially neededin costal regions where the influence of thelocal topography on incoming open ocean


waves makes prediction of sea state quitedifficult.

The monitoring of local variations from anestablished reference condition probably hasits most immediate applicability in the rout-ing of ships. The financial benefits obtainedby routing ships around areas of heavy seasare quite large and can be easily accom-plished with present remote-sensing technol-ogy. Other advantages would be in themonitoring of large-scale storms and their in-fluence and the tracing of various surfacepollutants, such as oil spills or the intrusionof saltwater into freshwater. Over longerperiods, the compilation of wave statisticsand their variation can reveal any unusualor significant changes in topography. Inaddition, there are a number of specificcoastal problems that these data can helpresolve.

When the depth of the water becomes lessthan half the wavelength of a spectral com-ponent in a wave spectrum, the speed ofthe component is affected by the depth. Theshallower the water, the slower the wavetravels. Complex offshore submarine topog-raphy turns the waves away from deeper re-gions and focuses them at shallower regions.Shallow areas off the coasts of parts of thecontinents, called continental shelves, are asmuch as several hundred kilometers wide.The depths in these shallow areas are oftencomplex, and the waves are refracted in waysboth difficult to describe and to compute.

All structures to be built in such shallow-water areas require design wave data sothat the structures can withstand the forceson them produced by the high waves during astorm. Also, the continued action of thelower waves can erode the material aroundthe structure base and cause it to collapse.With the many proposed offshore structuresaround the coasts, the problem of adequatedesigns for them will become increasinglymore pressing during the next few decades.

One area studied in the past and underrenewed theoretical analysis is the NewYork Bight, especially as affected by theHudson Submarine Canyon. Offshore sites

just 10 or 20 km apart in this area can, attimes, be exposed to waves 10 times higherat one site than at another.

During selected conditions with appropri-ate offshore deepwater waves, data can beobtained that will provide verification forwave refraction studies in shoal water. More-over, potential sites all over the world canbe surveyed concerning wave refraction ef-fects for preliminary, and perhaps even final,design considerations. The actual imageswill have to be recovered because the wavesbecome shorter and change direction as thewater becomes shallower, so that the con-cept of a spectrum representative of an en-tire area is not applicable.

A scientist familiar with the needs of theuser community could rather easily select ap-proximately 500 sites on a global basis forobtaining daily data, some over the deepocean and others at coasts, so that in thecourse of each year a global data base ofapproximately 180 000 coastal refractionpatterns and deep-water wave-numberspectra could be obtained. .

Finally, advantages will be gained by im-proved predictive models to forecast wavespectra. The present numerical spectral waveforecasting models are quite good. Theyincorporate many physical features of thegeneration and propagation of waves but notall the variously proposed theoretical fea-tures of waves. However, any numericalmodel of a physical phenomenon of the oceanscan be improved because there are alwaysincreasing levels of complexity that need tobe modeled. At present, the problem is lackof an adequate verification data base onwhich to build improved numerical spectralforecasting models. The measurements to bemade by satellites would provide the kind ofdata needed to develop greatly improved nu-merical wave-forecasting models by provid-ing a means to determine the errors in thepresent models. The need will always existfor forecasting wave conditions for as manydays as possible into the future. The fore-cast problem involves wave spectra and


weather, but improvement in wave forecastsis a necessary requisite for improvement inweather forecasts.

POLAR ICE FIELDS

The polar regions are a fundamental partof the Earth heat engine, yet the way inwhich they interact with the other parts ofthe atmosphere/ocean/ice/land system ispoorly known because they have existed be-hind what has been called an "observationalbarrier." Most polar geophysical studies arebased on a paucity of data; relatively short-time series of measurements acquired at afew points spread over vast distances. Thestate and behavior of the Earth surface isone aspect of the interaction-exchange prob-lem that can be said to be the most importantin polar regions, where changes of phaseof water into snow and ice occur over largeareas at small times scales, and where largepermanent floating and grounded ice massesare involved in a complex feedback processwith atmospheric and oceanic circulation.This problem can be studied only by usingsatellite remote-sensing techniques.

In the vast array of remote-sensing tech-niques available for polar studies, microwaveremote sensing, both passive and active,promises to be the most useful tool. Thepolar regions are in the dark for a largepart of each year and when they are in theirsunlit periods they are normally cloudcovered. Therefore, sensors that can observethe polar surface at all times, such as micro-wave sensors, are needed to provide the se-quential synoptic imagery needed for elucida-tion of the complex cause and effect processesof these vast regions.

Scope

Polar ice includes three distinct forms ofice that have widely divergent properties:

1. Sea ice, which covers large parts ofpolar oceans and some subpolar seas, containsbrine, averages one to several meters inthickness, and ranges in age from hours toseveral years.

2. Icecaps and glaciers, which are com-posed of freshwater ice that averages severalkilometers in thickness for Antarctica andGreenland and is tens of thousands of yearsold.

3. Lake ice and estuary ice, which is freshand brackish water ice of a wide variety ofthicknesses and ranges in age from hours toseveral months

In the following section, the three forms ofice are discussed in relation to active micro-wave remote sensing.

Sea Ice

Of all the forms of frozen water that existon Earth, the one about which least is knownis the sea ice that covers vast areas of theoceans; 10 percent in the Northern and 13percent in the Southern Hemisphere. A goodexample of the paucity of knowledge is that,for the International Geophysical Year(IGY) period, essentially nothing is knownabout the extent and morphology of thewinter sea ice surrounding Antarctica.

Individual ice floes have been observed tomove 50 km in a day, and speeds of 10 to20 km/day are common. Leads (cracks) andpolynyas (large irregular openings) open andclose at all times. The seasonal variations inareal extent of the ice canopies are large;approximately 15 percent for the Arctic and80 percent for the Antarctic. In short, seaice is the most rapidly varying solid on theEarth surface.

Although the feedback mechanism thatexists within the air/ice/water system is notwell denned, intuition and the few factsavailable suggest that the variation of theedge of the icepack must be of prime im-portance. On the time scale of months, air/sea interaction at the edge of the pack wouldcause an intensification of the atmosphericbaroclinicity, which in turn would alter thesurface ocean regime.

To monitor the seasonal patterns of sur-face heat exchange over polar oceans, onemust systematically observe the location andextent of pack ice and the location and dura-


tion of large polynyas within the pack. Al-though meteorological satellites such asTiros, Itos, and NOAA have given (and willcontinue to give) highly useful ice data, analltime, all-weather capability of observingsea ice did not exist until the launching ofNimbus 5 in December 1972.

The electronically scanning microwaveradiometer (ESMR) on Nimbus 5, operatingat a frequency of 19 GHz (1.55 cm), is pro-viding the first daily synoptic view of polarsea ice distribution. The NASA remote-sensing flights during the 1971 and 1972Arctic Ice Dynamics Joint Experiment(AIDJEX) pilot experiments, which flew anESMR identical with that aboard Nimbus 5,have provided data that allow the interpreta-tion of ice types shown on the microwavebrightness temperature maps obtained fromspace. The difference in brightness tempera-ture between seawater and ice at 19 GHzis approximately 120 K; therefore, the ice .edge on the maps shows up clearly and can bepositioned geographically with an accuracyof 30 km, the nadir resolution of the 1.4°beamwidth of the ESMR.

The ice-edge positions shown in ESMRimages differ considerably from the averagesshown for winter months in the U.S.S.R. andU.S. Antarctic atlases. The actual ice edgesare more irregular than the ones deducedfrom aircraft and ship reports shown in theatlases and are more extensive (farthernorth) in several areas, such as the RossSea. Of special note are the large polynyasrevealed in the Ross and BellingshausenSeas, which are not shown at all in theatlases.

Gloersen et al. (ref. 3-74) have used air-craft microwave images to show that it ispossible to distinguish first-year sea ice(1.15 m thick) from multiyear sea ice (1 to3 m thick) and to estimate amounts in mix-tures of the two. Campbell et al. (ref. 3-75)have compared ESMR and aircraft micro-wave images of the Arctic ice canopy todelineate its gross morphology.

Although ESMR is a prime tool to studygross characteristics of sea ice morphology

and dynamics, it does not provide the kind ofhigh-resolution data needed for a widevariety of scientific and commercial pur-poses. To test the existing and developingnumerical models for sea ice dynamics andthermodynamics, high-resolution sequentialimagery of select areas is badly needed. Theoptimum sensors for this program wouldappear to be active microwave sensors.

American radar imaging of the polar seaice dates back to the early 1960's (ref. 3-76).In a flight by a military aircraft from theNorth American Continent to the vicinity ofthe North Pole, a long strip of imagery wasobtained with an X-band real aperture sys-tem. The ability of the radar to distinguishvarious classes and ages of ice was firstshown in Anderson's analysis of theseimages. In 1967, the NASA Earth ResourcesAircraft Program (ERAP) P-3A aircraftflew north of Point Barrow with the 13.3-GHz scatterometer. Although ground partieswere on the ice to make thickness measure-ments, their lack of mobility prevented meas-urements from being more than of marginalutility, and the ice types had to be determinedby analysis of aerial photographs madesimultaneously with the scatterometer runs.Unfortunately, the more interesting icemorphologies occured some miles from the iceparty's location. By comparing the photo-graphs with ice atlas photographs and infor-mation gleaned from conversation with iceobservers, Rouse (ref. 3-77) was able tomake tentative identification of ice types andto show that the multiangle scatterometer ob-servations could be well correlated with icetype.

In 1970, another NASA ERAP missionwas conducted north of Point Barrow, withthe ice party restricted to the uninterestingfast ice within a few miles of Point Barrow.In this mission, a series of scatterometerlines were flown with both 13.3- and 0.4-GHzinstruments; the same area was imaged withthe 16.5-GHz DPD-2 multipolarized imagingradar by flying the imager in a box aroundthe area covered by the scatterometer.Parashar et al. (ref. 3-78) were able to make


excellent identification of ice types by de-tailed stereoscopic analysis of photographsof the area. The classification by ice typewas then converted into effective ice thick-ness, using the average ice thickness appro-priate to each kind of ice. The results werethen interpreted in terms of ability to meas-ure ice thickness with the various radars,but most of the work concentrated on whatcould be done with individual incident anglesappropriate to an imaging radar that mightbe used operationally. Figure 3-40 shows theencouraging result at 13.3 GHz. Note thatthere is an ambiguity between the verythinnest ice category (new ice, less than5 cm thick) and ice approximately 1 m thick.Fortunately, this ambiguity can be resolvedby image interpretation, for the new ice hasa characteristic texture quite different fromthat of the older first-year ice. It was foundthat the 0.4-GHz data could resolve thisambiguity if needed, but that the lower fre-

30 60 90 120 150 180Ice thickness, cm

210 240 270

FIGURE 3-40.—Experimental a° compared to ice thick-ness for different incident angles (frequency of13.3 GHz, vertical transmit/vertical receive polari-zation).

quency by itself could not distinguish be-tween the thicker multiyear ice and ice ap-proximately 0.5 m thick. Indications fromimage analysis of the DPD-2 were that fourcategories (open water, ice less than 18 cmthick, ice 18 to 90 cm thick, and thicker ice)could be readily identified on the images,even though the images were somewhatsaturated. The identification was an indica-tion that the ambiguity in intensity betweennew ice and meter-thick ice did not exist onthe cross-polarized image, but saturation onthe film prevented assurance that this con-clusion was correct. Parashar et al. (ref.3-78) have also developed a theory thatseems to explain the observations.

The Arctic and Antarctic Institute ofLeningrad has engaged in imaging of sea icefor at least 6 yr (ref. 3-79). The Sovietsbelieve that they can clearly distinguish thinice, thick first-year ice, and multiyear ice.In the late spring of 1973, an ice map of theentire shipping region across the north ofthe Eurasian Continent was prepared usingthe Toros 16-GHz real aperture radar im-ager. The system is presumably being usedoperationally in connection with directingshipping convoys along this route. In fact,the Soviet participants in the 1973 jointBering Sea passive microwave experimentused the Toros images for "ground truth" toidentify whether the passive microwave wasable to distinguish water and ice and one icetype from the other. The Soviets have statedthat the radar definitely distinguished moredifferent ice types than did the passive micro-wave sensors.

The first American active microwave ex-periment on sea ice in which sequential im-ages of ice were obtained took place northof Point Barrow during April and May1973 when a joint U.S. Geological Survey(USGS)/Cold Regions Research and Engi-neering Laboratory (CRREL) team used anX-band side-looking airborne radar (SLAR)mounted in a Mohawk aircraft. In 2 weeks,nine flights were made over selected largeareas of sea ice (200 by 50 km) to the east,north, and west of Point Barrow.


FIGURE 3-41.—An X-band SLAR image from theApril 28, 1973, mission west of Point Barrow.

Figure 3-41 shows a portion of the April28 mission data from west of Point Barrow.The coast can be seen with Wainwright atthe upper left and Icy Cape at the upperright. The zone of shorefast ice shows clearly,as does the coastal lead. The pack ice isclearly seen to be composed of numerousfragmented floes and leads and has under-gone strong recent deformation. The packwas made up of mostly first-year ice withsome multiyear ice.

Figure 3-42 shows the same area 5 dayslater on May 3. The shorefast ice remainedthe same during this period, but the pack icecan be seen to have undergone strong defor-mation. Major lead changes occurred while

the pack was translated only a few kilometersto the west.

A comparison of figures 3-41 and 3-42with figure 3-43 reveals that SLAR showsthe difference in ice types. Figure 3-43shows a segment of the pack ice on May 2approximately 190 km north of Harrison Bay(240 km northeast of Point Barrow). Inthis image, numerous undisturbed large icefloes can be seen, and the ridges appear aswhite linear features. The ice in this areawas far less deformed than the ice west ofPoint Barrow.

Although much work remains to be done onthese data, SLAR remote sensing of sea iceprovides discernment of the followingfeatures.

1. Leads and polynyas can clearly be dis-tinguished from ice.

2. Thin ice (pancake, frazil, gray) can

FIGURE 3-42.—An X-band SLAR image from the May3, 1973, mission west of Point Barrow.

FIGURE 3-43.—An X-band SLAR image showing Arc-tic Ocean pack ice.


be distinguished from open water in leadsand polynyas.

3. Ridges can generally be distinguishedfrom leads, especially when the leads arelarge.

4. Ice floe shape and size can clearly be ob-served.

5. Land (permafrost) can clearly be dis-tinguished from shorefast ice, water, andpack ice.

6. Ice-type distinction is good enough topermit distinctions between water, thin ice,thicker first-year ice, and the thicker multi-year ice.

Icecaps and Glaciers

Approximately 85 percent of the fresh-water on Earth exists as ice in Antarcticaand Greenland. Glaciological research inthose areas has been aided greatly by thedevelopment of radio echo sounding tech-niques, which has made possible the delinea-tion of bedrock topography and the measure-ment of ice thickness. Both these variablesare needed to test recently developed nu-merical models. However, if the complexinteraction between icecaps and glaciers andclimate are to be understood, the ice/airinterface events must be understood, espe-cially the accumulation rate.

The ESMR images of the Arctic and Ant-arctic from Gloersen et al. (ref. 3-80) showvery interesting signature variations on ice-caps. For Greenland, brightness temperaturedifferences of 323 K occur across the icecap,with the highest brightness emittance cor-responding roughly to the highest elevationof the icecap. Although the cause of thesevariations is not fully understood, they areapparently not due to variation of the physi-cal temperature of the ice, but to the emis-sivity variations probably induced by varia-tions of the small-scale structure. Suchcrystal metamorphosis could be caused notonly by temperature variations but also bypressure gradients within the ice.

In the ESMR images of the Antarctic ice-cap, no such correlation with elevationoccurs; instead, the lowest brightness tem-

peratures occur near the center of the conti-nent, which would be expected if thevariation in brightness temperatures waspredominantly due to the variations in thephysical temperature of the ice. Thawingand recrystallization of the surface ice doesnot occur on the Antarctic cap as it does inGreenland.

As interesting as these passive microwavemeasurements of icecaps are, they will beuseful only for gross structure studies,whereas an urgent glaciological need existsfor detailed measurements of the surfacestructure of icecaps and glaciers. It is pre-cisely in this area that SLAR techniquescould prove highly useful.

The Jet Propulsion Laboratory (JPL) hasrecently obtained L-band sounder profiles onAlaskan glaciers and on Greenland. The datashow signal returns from structural featuresbelow the surface. The Greenland profilesshow a tilted, layered structure that looksas if it had been formed by surface flow.

The only two-dimensional SLAR imagesof glaciers and icecaps are those obtained byV. V. Bogorodsky and V. S. Loshchilov of theArctic and Antarctic Research Institute ofLeningrad.2 These Soviet scientists have ob-tained numerous SLAR (S-band) images ofSiberian glaciers. All were obtained througha dry snow cover over nontemperate glaciers(those having temperatures below 273 Kwithin the ice mass). In the great majorityof images, the foliation planes formed by theannual accumulation processes were clearlyvisible. Most of the images were of glacierswith a snow cover thicker than severalmeters.

From the glaciological point of view, thesedata are exciting because one of the mostdifficult things to measure in glaciers is therate and pattern of accumulation and abla-tion. As a glacier flows, the deposited layersof snow metamorphose into ice, which slowlyflows downslope. Therefore, the pattern offoliation planes observed is due to two proc-esses : accumulation/ablation and flow. Thus,to deduce the recent accumulation history of

- Personal communication, J. W. Campbell.


a glacier from the foliation planes, one mustallow for the dynamics. This may not be toodifficult for slow-moving nontemperateglaciers.

Bogorodsky and Loshchilov believe thatSLAR measurements of glaciers can be usedfor accumulation studies. They have alsoobtained images of Arctic ice islands (piecesof icecaps that have calved off an ice shelfand drifted about within the matrix of seaice floes). These images all showed the iceto have a parallel banded structure, with thebands having a spacing on the order of sev-eral hundred meters. They do not knowwhat these bands are, but surmise that theywere formed by the accumulation processexisting on the icecap where the island wasformed. The structural differences betweenice island and sea were so great that dis-tinguishing between the two was alwayspossible.

Lake and Estuary Ice

Several of the world's major shippingroutes are covered by lake or brackish ice forseveral months a year (i.e., Gulf of St.Lawrence and Baltic Sea), and great effortsare being made to extend the navigable seasonin these areas. To do this, two uses of se-quential radar imagery are necessary: (1)as input into numerical models of ice dy-namics and thermodynamics, and (2) asrouting maps to direct ship traffic throughthe thinnest ice in an area.

A considerable part of the Asian and NorthAmerican tundra is lake covered, and duringthe winter these lakes freeze both fully andpartly (not to the bottom). Measurement ofthe ice cover on these lakes is importantfor many biological studies.

During the Skylab 4 overflights, SLAR im-agery was obtained of the ice covers in LakeOntario and the Gulf of St. Lawrence. Thesedata show that radar remote sensing oflake and brackish ice can make it possibleto distinguish the following features:

1. Leads and polynyas can clearly be dis-tinguished from all forms of ice.

2. Rafted ice can be distinguished fromundisturbed ice.

3. Shorefast ice can be distinguished frommoving ice.

4. Ice floe size and shape can clearly beobserved.

5. The approximate age of ice can be ob-served in terms of gray, gray-white, andwhite ice.

6. Ridges can generally be distinguishedfrom small leads.

Bogorodsky and Loschilov, who have madeSLAR X-band observations of lake ice in theSoviet Union, agree with these six tentativeconclusions.

A very interesting phenomenon was ob-served on the tundra lakes east of PointBarrow during the USGS/CRREL experi-ment. In figure 3-44, a SLAR image of aseries of thaw lakes to the east of PointBarrow is shown. Some of the lakes appearwhite (strong return), whereas others ap-pear dark (weak return). After comparingthis and other SLAR images with surfacemeasurements, it was concluded that thelakes that appear white in the images arefrozen all the way to the bottom (to the

FIGURE 3-44.—An X-band SLAR image showing fro-zen thaw lakes east of Point Barrow.


permafrost table) and the lakes that appeardark have a water layer beneath the icecover. This finding is important from severalpoints of view. Such images could provideinformation to local communities concerningwhich lakes to tap for a winter water supply.The biological regime between completelyand partly frozen lakes is different, the latterhaving far more fish.

Conclusions

Satellite-borne active microwave sensorsoffer the only means of obtaining high-resolu-tion imagery of polar ice during all times,thus penetrating the "observational barrier"that has hampered much-needed research.Sequential synoptic active microwave im-agery of sea ice is urgently needed to testand help develop numerical models of air/ice/ocean interaction. Such data will greatly en-hance the usefulness of ESMR imagery ofthe gross characteristics of the sea icecanopies.

The only remote-sensing means capableof observing subsurface structures of ice-caps and glaciers is active microwave. TheSLAR images of glaciers clearly delineatethe accumulation foliation planes throughseveral meters of snow cover. Those of ice-caps reveal a subsurface parallel banded

structure. Such images promise to provideneeded data on accumulation amounts, andpatterns may help in determining surfaceflow patterns. Satellite radar is the best wayto track and study the metamorphosis of iceisland (bergs).

Active microwave imaging of lake andestuary ice will provide high-resolution, se-quential, synoptic data needed to extend thenavigable season in busy waterways such asthe Gulf of St. Lawrence and the Baltic Sea.

The SLAR can be used to determine whichtundra lakes are frozen to the bottom andwhich are not, an important logistical andecological technique.

Although the potential of active microwavesensors that currently exist has been demon-strated as ice sensors, neither U.S. norU.S.S.R. scientists yet know what frequency(or combination of frequencies) or whatpolarization (or combination of polariza-tions) is best for monitoring either sea iceor lake ice. Although earlier theories mightbe used to permit some extrapolation or in-terpolation, these too need improvement andfurther verification. Hence, in addition tofurther flights over the ice, both theoreticalresearch and research with a microwavespectrometer that can be transported ontothe ice seems to be needed if the optimumsystem parameters are to be established.

PART E

11821f TECHNICAL APPROACHES

INSTRUMENTATION

Radar altimeters, scatterometers, andimaging radar are the major instrumentsneeded to accomplish the applications dis-cussed earlier in this chapter. This sectiondescribes the instrumentation in terms ofits functions, future developments, con-straints, and applications.

Altimetry

Introduction.—Satellite altimetry has dem-onstrated the capability for determiningmean sea level. Although past efforts havenot achieved the ultimate goal in spatial andheight resolution, it is attainable in the nextdecade. On attaining this goal, satellitealtimetry will enable improved knowledge of

active microwave workshop report - nasa · 192 active microwave workshop report part d large-scale...

Documents