ROBERT W. PEASEDAVID A. NICHOLS

University of CaliforniaRiverside, CA 92502

Energy Balance Maps fromRemotely Sensed ImageryIsarithmic maps showing urban heat islands, surface albedos,energy absorbed by the surface, and net radiation were constructedfrom scanner data acquired over Baltimore.

INTRODUCTION

C ALIBRATION AND constant gain areproperties which have rendered electro­

optical scanners essentially imaging radiom­eters. The simultaneous imaging of short­wave solar and longwave thermal spectralbands has made possible synoptic mapsshowing the spatial diversity of variousenergy-related phenomena as complex asthe radiation balance at the terrestrial sur­face known as net radiation.

rough surfaces of an urban business districtpresent thermal cavity structures the en­vironmental effects of which cannot be accu­rately assessed from observations at groundlevel or within the cavity structure itself.Calibrated multispectral scanners on air­borne or satellite platforms acquire datafrom above, rather than in, urban and otherterrain and make possible the display of thespatial diversity of these energy-related phe­nomena in synoptic map form.

ABSTRACT: Synoptic maps showing the spatial distributions ofenergy-related phenomena over the terrestrial surface have not beenfeasible with traditional instruments and observation techniques.The advent of calibrated multispectral optical scanners, however,has made possible the rapid measurement of surface energy emit­tance and reflectance. Data acquired in afew minutes with scannersfrom airborne platforms can be translated into maps showing urbanheat islands, surface albedos, energy absorbed by the surface, andnet radiation. An experiment using the ERIM M-7 scanner, andBaltimore as a test-site city, has successfullY yielded isarithmicmaps of these phenomena as they existed during an imaging over­flight. Methods for acquiring calibrated data from a scanner andthe construction of maps from image transparencies are describedand evaluated.

Surface energy-exchange phenomenahave become important aspects of climato­logical research. Much of the effort to datehas been involved with analysis of energybudgets for relatively few specific points,however, because of the elaborate and costlynature of ground radiation measuring arrays.Cost and observational time constraintshave prevented a synoptic spatial viewwhich truly relates the thermodynamic pro­cesses taking place. As an example, the

Map displays from scanner imageryrequire not only the smoothing or generaliza­tion of data points but, for complex phenom­ena such as net radiation, the combinationof data from two or more multispectral bandsas well. In the project described, this manip­ulation must be repeated for each data cellin several large matrices. The assistance ofa computer, therefore, avoids the extensiveand tedious amount of bit-by-bit mathe­matical manipulation and decision making

PHOTOGRAMMETRIC ENGINEERING AND REMOTE SENSING,

Vol. 42, No. 11, November 1976, pp. 1367-1373.

1367

,-----------------------------------------------

1368 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1976

that would be required by manual reductionof data to a map. This paper, then, is notonly concerned with the use of scanner datafor making maps of surface energy-exchangephenomena but also reports successful at­tempts to do so by automated means. With­out automation, construction of the mapsdescribed would not be feasible except insmall experimental instances. In the follow­ing sections, the maps and their- scannerrequirements are described first, followedby methods used in the map constructions.

The original imaging for data and theensuing development of mapping techniqueshave been sponsored by the Geography Pro­gram of the United States Geological Surveyin conjunction with studies of the CentralAtlantic Regional Test Site (CARETS).Funding for the aerial imaging was provideddirectly by NASA.

THE MAPS

Four maps show the synoptic distribu­tions of several energy-related surface phe­nomena for a flightpath from the northwestto the southeast across Baltimore, Maryland(Figure 1). The data for map constructionwere generated close to 1345 (1:45 PM)EDT on May 11, 1972, with the aircraftflying at 1,526 metres (5,000 feet) altitude.The image (Figure la) is that of the visualred band (0.62-0.70 !-tm) and has been in­cluded to provide geographical orientationfor the maps. On the left hand (NW) end isthe central business district (CBD) withdowntown Patterson Park the prominentrectangular dark area. To the southeast, anindustrial area with large-roofed structures,parking lots, rail lines, and tank farms sepa­rates the CBD from a residential area thatextends to an arm of the Patapsco Riverestuary. Across the estuary at the southeastend of the flightpath is Sparrows Point steelmill. In all, a considerable variety of urbansurface types is represented. The image isone of the three used to provide data for themaps. The other two are that for the nearinfrared (1.0-1.4 !-tm) and that for the ther­mal infrared (9.8-11.7 !-tm).

Data for specific maps were derived fromscanner outputs as follows. The first map(Figure Ib) shows energy emitted by thesurface and used data from only the thermalinfrared (TIR) channel. For calibration ofthis channel to surface targets, the radiationtemperatures of the target surfaces weremeasured with a Barnes PRT-5 radiationthermometer with a 10 to 12 micrometrespectral sensitivity. Radiation temperaturesrelate to the Stefan-Boltzmann equation

when the emissivity of the radiating surfaceis considered unity and are the temperaturesderived in this experiment rather than sensi­ble or kinetic temperatures. For this reason,only when longwave energy is absorbed bythe surface is emissivity of any concern tothe inquiry. The use of radiation tempera­tures is an intermediate step in the calcula­tion of radiant emittance. By definition thistemperature form implies the total energy ina black body curve whereas energy througha spectrally narrow instrument window maynot follow the fourth power law. Longwaveradiometers, therefore, are typically cali­brated in radiation temperatures.

The map of albedos (Figure lc) requiredthe combination of images made with visualred and near infrared, weighted as willsubsequently be described. The map ofenergy absorbed by the surface (Figure Id)uses the unity-complement of the albedo asa multiplicative modifier of ground­measured solar radiation (both beam anddiffuse) to ascertain the solar energy ab­sorbed. To this is added absorbed down­welling longwave atmospheric radiation.Although we are here primarily concernedwith radiation temperatures, an emissivitymust be assumed to assess that portion of thelongwave radiation which enters the surfaceenergy budget. For most terrestrial surfaces,the emissivities for wavelengths close to 10micrometres range from 0.90 to 0.98, and anaverage value of 0.95 has been used forlongwave absorption. The map of net radia­tion (Figure Ie) represents the subtractionof values of energy emitted from values ofenergy absorbed.

In brief, whereas the map of albedosrequires only the combination of data fromtwo spectral channels, that for energy ab­sorbed requires the use of the same twochannels plus the ground-truth data meas­ured at the surface. The map of net radiation,on the other hand, requires data from threechannels to be combined in data cells alongwith data from surface observations. Theways by which the data from the multiplechannels have been combined and ways bywhich congruency of data cells from morethan one spectral image has been main­tained are treated in detail in the followingsection.

Before the explanation of the map con­struction methods, it should be noted that noattempt has been made to remove the scaledistortions inherent in straight scanline geo­metry. The transverse scale factor is a non­linear 0.79 to a flightpath scale of 1.00. Thiserror is inverse to the nonlinearity of the

E ERGY BALANCE MAPS FROM REMOTELY SE SED IMAGERY

Patterson Park

1369

FIG. 1. A sequence of isarithmic maps showing various energy-related phenomena for a flightpathNW-SE across Baltimore, Maryland,. at 1345 EDT on May ll, 1972. The image of the visual red band(a) shows the diversity of urban surfaces mapped. The maps generalize (b) the energy being emittedby the surface in langleys/minute x 100, (c) surface albedos in percentages, (d) the energy beingabsorbed by the surface (Iy/min. x 100), and (e) the balance of energy flows at the surface, or netradiation (Iy/min. x 100).

tangent functions produced as the opticalpath swings with constant angular velocityaway from nadir.

METHODS USED IN MAP CONSTRUCTION

Requisites for making surface energy­exchange maps from imaged data are: (1) themultispectral scanner must have constantgain and internal calibration sources aredesirable, (2) the scanner must be calibratedto the terrestrial surface for the spectralbands or channels from which the map­making data are to be taken, (3) data from

both a thermal infrared channel and short­wave (solar wavelengths) channels appro­priate for albedo determination must beavailable, (4) data must be entered into theprocessing system in such a manner thatvalues from several spectral bands will befor congruent surface areas, and (5) datapoints must be smoothed or averaged topermit the simplification inherent in mapdisplays.

There are two ways for calibrating a con~

stant gain scanner to the surface. That em­ployed in this study is by the use of calibra­tion targets, areas of homogeneous surface

1370 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1976

large enough to yield image elements thatcan be measured densitometrically. Severaltargets were chosen to give a range in bothtemperature and reflectance values. The useof target calibration automatically compen­sates for any effects that the atmosphere inthe optical path may have on the surfacesignal.

The second method applies only to theterrestrially emitted or longwave TIR spec­tral band. For this, a relatively simple gray­window model can be constructed whichinvolves the attenuation of the surface­emitted signal by the intervening watervapor, turbidity, and pollution in the opticalpath while at the same time these absorbingelements are by their own emission adding.strength to the attenuated surface signal(Pease, 1971). Current processing of Skylabdata, where no calibration targets are avail­able, uses the gray-window approach (Jenneret al., 1976), but targets were available andused for the Baltimore experiment.

The day of the imaging flight was cool andclear following the passage of a cold frontthe previous night. Relatively dry air mini­mized the modification of the TIR signal inthe 9.8-11.7 IJ-m spectral sensing band. Al­though the transmissivity of the relativelydry air was 0.90 (surface to sensor), a smallerror would have ensued for the surfacetemperatures of the CBD which were some20°C above the mean temperature of the aircolumn. The gray-window modification ofthe surface-emitted signal by the atmos­phere cannot be ignored in careful remotetemperature measurements through the so­called water vapor window (8.0-12.0 IJ-m)due to residuals of spectrally close watervapor absorption bands. At this time, nosatisfactory gray-window model has beencreated for reflected sunlight and targetcalibration must be used.

Reduction of data to a map form involves

the use of data-averaging cells in a matrix, aform of block smoothing, to achieve thegeneralization of information a map re­quires. Data cells can be delineated onphotographic transparencies of desiredspectral channels or on digitized versions ofscanner magnetic tapes. This inquiry ex­tracted data from transparencies. Severaltypes of map formats are available. Data­averaging cells can be printed as graymapswith a line printer or drafted as discrete datacells, in either case according to a value­intervalled hierarchy of map symbols. Themethod used for the maps of Figure 1,however, places the averaged values at thecentroids .of each data cell as control pointsfor plotting isarithmic lines. Since each cellis part of a large matrix, numerous controlpoints are available for automated plottingof the lines. The size of the averaging cellused in this study, however, reduced thenumber of cells in the matrix. This conditionfavors the use of the isarithmic format be­cause the cells are too large to produce amap with satisfactory detail when they di­rectly outline value polygons on the map.The line-printed data cell approach wascarried out by the Environmental ResearchInstitute of Michigan (ERIM) but with small­er cells which necessitated additional smooth­ing achieved by moving filters whichacted directly upon digitized data to reducespatial noise in the ensuing graymaps(Thomson and Dillman, 1973). The ERIMinquiry was also funded by NASA and theU.S. Geological Survey and was carried outin cooperation with the study reported here.

The flow of data in the construction ofthe maps is shown in Figure 2. Brieflystated, average transmittances of discretedata cell areas on the transparencies for theseveral spectral channels first were meas­ured with a custom-fabricated wide aperturedensitometer. The punching of these values

I TRANSHITTANCE I COMPUTER PROCESSING I HAP

_______ .L~A:..u~E~E~~_~ ~~D~T~ ~_P~~~~ _

: I /' ,. RADLATION TEMP.

I I CONVERT TO I MAPSI lHERAAl IR CONVERT TO -ENERGY EHITTE~1/9.8-11.7 !JIll-RADIATION TEMPS. I ENERGY EMITIED

/ I \ I MAPSMULTISPECTRAL _VISUAL RED I CONVERT TO I

BANOS \. .62-.70 IJI'lIREFLECTIVITIES,-+-COHBINE INTO ~AlBEDO HAPS

I'\. NEAR I R I CONVERT TO r AlBEDOS7 \LI 1.0-1.1t IJ"l~ REFlECTIVITIES (1- Ib d y;+NET RADIATIONI r x a eo HAPSI I/'II I ADD Ol:M4WELlING ENERGY I

SOLAR RADIATION BSOR8EO _ENERGY ABSORBEDI I I HAPS

I ADD DOWNWEllI HG :: ATHaS.RADIATION II x EMISSIVITY :

FIG. 2. The flow of data followed in making the maps.

ENERGY BALANCE MAPS FROM REMOTELY SENSED IMAGERY 1371

FIG. 3. Diagram of the wide-aperturedensitometer used to read transparencytransmittances into an automated com­puter and mapping system. The trans­parency is attached to a stationary aux­iliary light table placed atop an H. DellFoster x - y digitizer. The light sourceand sensing cell move with the digitizercursor arm. Registration in a matrix ofaveraging data cells is maintained by thedigitizing system quantizer.

into IBM cards constituted the only manualoperations in the map constructions. Eachcard entry not only contained the averagetransmittance value for its cell but x - ylocational information as to the position ofthe cell in its larger matrix. Conversioninformation, previously entered into thecomputer as sub-routines, operated withappropriate programs to convert transmit­tances to radiation temperatures (TIRchannel) or reflectivities (shortwave chan­nels). The computer created plot tapes fora plotter to automatically draft the maps.An IBM 360-50 computer was used in con­junction with an H. Dell Foster RSS-700flat-bed vector plotter.

The wide aperture densitometer,· withwhich average transmittances for data cellswere measured, is diagrammed in Figure 3.It is a combination of a current-regulatedillumination system, a working aperture thesize of the data cell on the transparency, anda silicon cell sensor. The sensor actuates apotentiometric recorder mechanism with thevoltage developed across the cell load resis­tor, the latter with a wiper that permitscalibration control of the system. Thesmoothing or averaging aperture chosen was6.25 mm (0.25 inch) square which gave aground resolution of 381 metres (1,250 feet).The aperture size determines the degree ofblock-smoothing of the imaged data. AMacBeth densitometer and its calibratedstep-wedge were used to calibrate thecustom-fabricated instrument. The measur­ing components were attached in place ofthe cursor on an H. Dell Foster x - ydigitizer.

The preservation of congruency of valueareas between the spectral channels of agiven flightpath followed in concept a meth­od used successfully in making isarithmicmaps of radiation temperatures from dataobtained on Barbados during the BOMEXexperiment (Pease, 1971). In that instance, amatrix of lines was placed over a transpar­ency image and measurements were madeof the discrete cells so delineated. For thecurrent project, cell integrity between spec­tral bands was maintained by the x and ydisplays of the quantizer of the digitizer.The separate multispectral images were in­dexed into the locational system both byimage edges and prominent features thatappeared on each image. Locational infor­mation was punched into the cards auto­matically each time a transmittance valuewas entered.

Because all data cell values entered intothe automated system were transmittances,of necessity the conversion to energy­related values took place in the computer.An important step in the processing, there­fore, was the inclusion of the conversioninformation previously mentioned. Althoughscanner measurements recorded initiallyon magnetic tape are essentially linear, theirrepresentation upon a photographic trans­parency follows the S-shaped sensitometriccurve of the photographic film (Figure 4).Because mathematical models of the S­shaped curves proved to be poor fits, theactual graphic conversion curves were sim­ply digitized in their x and y coordinates andthen treated in the computer as a series ofvery short linear equations. The shape of thefilm sensitometric curve could be deter­mined from the incremental step-wedgesalong the edges of the transparencies. Tomake the curves, transmittances of targetimage elements were measured through the1 mm aperture of the MacBeth densitometerand plotted against ground-measured sur­face values.

Conversion information for the TIR chan­nel is shown as the curve in Figure 4a. Thetarget types for which radiation temperatureswere recorded at the time of the imagingoverflight readily delineate the curve. Con­version information for plotting albedos,however, posed a more complex problem.The work of Dirmhirn (1968) suggests thatthe "silicon sensitivity" which peaks at0.8 micrometres provides good measures ofalbedos due to its distribution of sensitivityabove and below the rapid rise in plant(chlorophyll) reflectance that centers at awavelength of 0.725 micrometres. Reflec-

POTENTIOMETRICR,ECORDEIl

SMOOTHING APERTURE

Tll;AHSPAROICYSILICON CELL _--"'''''''''

1372 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1976

• The TOPO program was written by RaymondPostma as a part of a land-planning tools softwarepackage for the Environmental Systems ResearchInstitute of Redlands, California.

THERMAL IR

98-117}.1m blackt~

m:40~

~ )0 /119ht

soil

grass~ 20~ ater

cells in the matrix and interpolating isarith­mic lines.

Although a variety of contouring programscould have been used for drafting the mapson the plotter, that used in this project isthe Tapa' program with a pre-processorlinear interpolation routine which has thedual purpose of scaling the matrix size andsmoothing the surface.

EVALUATION OF RESULTS

Because the dual technologies of multi­spectral scanning and automated isolinemapping are not new, it is felt that theimportant contributions of this project are(1) the successful creation of maps showing

~ , /

60

i,O-max. /

~ ////

w NEAR IR .... /

;: 7°~"" ///g)O L,.~ 20 t:~:::::, v,:~::e:ED

.62-.70 JJm

I~ rass a-min,:water

oO~~.':"'0-.2~0-.""'JO::---.""~-.~50::----:.6':-0-.~70:--'b. TRANSPARENCV TRANS"ITTANCE

FIG. 4. Curves to convert (a) the thermal in­frared image to surface radiation temperaturesand (b) two shortwave images to reflectivitieswhich when combined give albedos. Groundcalibration targets are indicated. The S-shapematches sensitometric curve of the transpa­rency film.

.10 .20 .~ .~ .~ .60 ·WTRANSPARENCY TRANSHI TIANCE

o 0

~ 10

tance curves for common substances such assoil, blacktop, and concrete, tend to changegradually over the solar wavelengths andtherefore their albedos are less prone toerror when measured by spectrally selectivemethods than are those for plant targets.Pease and Pease (1972) have demonstratedthat albedo measurement by spectrally selec­tive methods or by spectral samples shouldbe weighted with 60 percent of the sensitiv­ity below the 0.725 micrometre midpoint inthe plant reflectance rise and 40 percentabove or at longer wavelengths. The twospectral bands used in this study to measurealbedos are the visual red (0.62-0.70 JLm) forwavelengths shorter than 0.725 micrometresand the near infrared (1.0-1.4 JLm) for wave­lengths that are longer. Curves convertingtransmittances to reflectivities for these twosample bands (Figure 4b) were entered assub-routines and combined with the above60/40 weighting to yield albedos. Again,congruency of data cells is imperative toaccuracy of results.

After all data had been read from thetransparencies and punched into cards, amaster tape file was prepared. The data,now in machine-compatible form, were sub­jected to manipulation. For this purpose,two programs were written. The first utilizessub-routines to convert transmittance datafrom the TIR channel to radiation tempera­tures. The second program contains severalsteps. (1) It determines an albedo by com­bining the reflectivities of the two samplespectral bands according to the 60/40 weight­ing. (2) With the factor (1 - albedo), theprogram converts solar radiation (entered asa constant in the program for this imagingflight) into solar energy absorbed. (3) To thisis added the downwelling longwave atmos­pheric radiation (also a constant programvalue for this flight) after reduction by thepreviously explained standard 0.95 emis­sivity to achieve total energy absorbed bythe surface. (4) To determine net radiation, acell radiation temperature was entered fromthe first program, converted to langleys perminute with the appropriate form of theStefan-Boltzmann equation, and added as anegative quantity to energy absorbed as apositive quantity. This sequence of dataprocessing was performed for each data cellin the flightpath matrix. The procedureemphasizes the importance of maintainingperfect registry of data cells between themultispectral bands used. At different stagesof processing, maps showing various energy­related surface phenomena were made byapplying data values to the centroids of data

ENERGY BALA CE MAPS FROM REMOTELY SENSED IMAGERY 1373

distributions of energy-related surface phe­nomena other than simple displays of ther­mal state and (2) the symbiotic integrationof scanner and automated cartography tech­niques to achieve such maps.

As regards accuracy of the maps, the fol­lowing observations can be made. Radiationtemperatures closely correspond to targetground measurements which may not besurprising since these measurements wereused to make the conversion curves. Albedovalues, less directly derived, agree with theauthors' own experience in measuring avariety of surfaces by standard pyranometricmethods. Vegetated grass surfaces such asparks, for example, have albedos of 20 to 22percent which match ground truth measure­ments and are in a generally accepted rangeof values for this type of surface (Sellers,1965; Kondratyev, 1969; Kung, Bryson, andLenschow, 1964). Net radiation values, whichrequired the most complex series of process­ing manipulations and thus potentially weremost prone to error, also appear reasonablein light of the authors' experience in micro­climatology. The ground-measured netradiation of the blacktop surface of one ofthe prime calibration targets at the time ofoverflight was 0.81 langleys per minute. Onthe generalized map of the calibration run(not included), the value at the target loca­tion interpolated as 0.77 ly/minute, a loworder of error. It should be noted that perfectagreement cannot be obtained betweenaveraged data and point measurements. Thevalue of 1.20 ly/min. over water reflects thelatent energy sink of evaporation.

Variations in values entered as constantsin the data manipulations could have causedserious errors in the mapped distributions.Incoming solar radiation, for example, mightwell have varied over the flightpath due tourban pollution, especially that arising fromthe steel mill. For this reason, a pyrano­metric reconnaissance to detect diversity insolar input was made over the area at thehigh sun diurnal period just before theimaging flight. Values varied less than 0.01ly/min. from the nominal 1.42 ly/min. usedin the calculations. In a like manner, thenature of the airmass was conducive tospatially homogeneous distribution of atmos­pheric down radiation. The ideal conditionsof the imaging day, however, do not rule outthe necessity for ascertaining possible in­consistencies in ground-measured energyinputs.

Regarding the utility of the maps, onepurpose is to supply spatially comparabledata to be used in modeling in such areas as

urban planning and the new field of landuse climatology. Possibilities in this direc­tion are still to be explored, but certain factsare already apparent. For example, albedoappears to be the factor controlling netradiation which is often considered a broadmeasure of the injection of natural energyinto a terrestrial thermal system. Due torelatively high albedos, parks in an urbanbusiness district appear as poor absorbers ofsolar energy. They may seem cool, therefore,for reasons other than their being latentenergy sinks, the commonly held cause.Other examples could be cited.

Current experience with Skylab S-192data is advancing techniques herein de­scribed. Use of the gray-window model isbeing advanced to eliminate the need forcalibration targets for thermal mapping. Theutility of satellite platforms for acquiringdata to make surface energy-exchange mapshas been tentatively proved (Jenner, et al.,1976).

Future energy-exchange mapping with theaid of multispectral scanners depends to agreat extent upon the availability of appro­priate data. The forthcoming Heat CapacityMapping Mission (HCMM) could well pro­vide more satellite data with a thermalinfrared channel and the necessar'y calibra­tion capability. Use of a high-flying aircraftas a scanner platform should be consideredas a compromise between the low-flyingaircraft used in this project with its narrowscan path and expensive satellite platforms.A scanner at 20,000 metres (65,500 feet)would provide an intermediate spatial cov­erage with greater detail than the satelliteand could be used at specific times whichbest fit observational needs. Although futureinstrument platforms are not well defined atthis time, it can be assumed that, if synopticmaps of surface energy-related phenomenaare of value, the necessary scanner data willbe forthcoming.

REFERENCES

Dirmhirn, Inge, 1968, "On the Use of SiliconCells in Meteorological Radiation Studies,"Journal of Applied Meteorology, Vol. 7, pp.702-707.

Jenner, C. B., J. E. Lewis, S. I. Outcalt, and R. W.Pease, 1976, "Land Use Clir.1atology ... ,"Chapter 3 of R. H. Alexander et al., Appli­cations of Skylab Data to Land Use and Cli­matological Analysis, U.S. Geological Survey/Geography Program, Final Report Skylab/EREP Investigation No. 469.

Kondratyev, K. Ya., 1969, Radiation in the Atmos­phere, Academic Press, New York.

1374 PHOTOGRAMMETRIC ENGINEERING & REMOTE SENSING, 1976

Kung, E. C., R. A. Bryson, and D. H. Lenschow,1964, "Study of a Continental Surface Albedoon the Basis of Flight Measurements andStructure of the Earth's Surface Cover overNorth America," Monthly Weather Review,Vol. 92, No. 12, pp. 543-564.

Pease, R. W., 1971, "Mapping Terrestrial Radia­tion Emission with a Scanning Radiometer,"Proceedings of the Seventh InternationalSymposium on Remote Sensing of Environ­ment, Vol. 1, pp. 501-510, University of Mich­igan, Ann Arbor.

Pease, S. R. and R. W. Pease, 1972, PhotographicFilms as Remote Sensors for MeasuringAlbedos of Terrestrial Surfaces, TechnicalReport V, U.S. Geological Survey Contract14-08-0001-11914.

Sellers, W. D., 1965, Physical Climatology, Uni­versity of Chicago Press, Chicago.

Thomson, F. J. and R. D. Dillman, 1973, Balti­more Maryland Radiation Balance Mapping,Technical Report, Environmental ResearchInstitute of Michigan (unpublished).

ErrataIn the article "Applied Photo Interpretation for Airbrush Cartography" by Jay L. Inge and

Patricia M. Bridges in the June 1976 issue Figures la and lc (page 751) have been inter­changed; and the sentence beginning on the fourth line, first column on page 754 should read"The horizontally high-pass filtered version (Figure la) and the vertically high-pass version(Figure Ib) depress surface coloration and enhance high frequency contrast."

In the July 1976 Yearbook issue the winners of the Autometric Award for the last five years(page 962) should read

1972-Dr. Edward Yost and Sondra Wenderoth1973-Kenneth R. Piech and John E. Walker1974-M. R. Specht, N. L. Fritz, and D. Needler1975-Kalman N. Vizy1976-Dr. Robert G. Reeves

The July 1976 Yearbook issue failed to list the following Emeritus Members (page 966):

Ray Anderson Pliny Gale Gilbert B. MillerGen. Benjamin Arcoverde Douglas E. Henriques William H. MoorhouseStanley B. Ash Jack J. Ingram Richard B. Soloman

If other names were left out, please inform the Society.In the article "The Brock Brothers and the Brock Process" by Harry Tubis in the August

1976 issue, the first line, second column ofpage 1029 should read "(Figures 3 and 4).", and thefirst sentence of the last paragraph in the first column on page 1032 should begin "Thiscompletes ...".

In the article "Compilation Base Orientation by Graticule" by Richard H. Duncan in theSeptember 1976 issue, the 14th to the 6th lines from the bottom in the left-hand column ofpage 1158 should be removed; the 16th line in the right-hand column ofpage 1158 should read"Xg, Yg, and E g of points ..."; the second of the three equations toward the bottom of theright-hand column of page 1158 should read

Ymdc = btXg + b 2Yg + baXl + b4XgYg + bsYg2 + b6Xg3 + b 7Xg2Yg + b 8XgYl + bgYl;

and the second reference at the end of the article should read "Bendix Research Laboratories,Handbook for AS-llB-l Automated Analytical Stereoplotters, Vol. 1, December1969."

Cover Photos NeededPhotographs suitable for the cover of Photogrammetric Engineering and Remote Sensing

are needed. Either black-and-white or color may be used; however, because color reproduc­tion is costly, we request that the donors ofcolor material ifat all possible cover the additionalcost (approximately $750). Please submit cover material to the Editor, American Society ofPhotogrammetry, 105 N. Virginia Ave., Falls Church, VA 22046.