The costs and benefits of a satellite-based system for natural resource management

THE COSTS AND BENEFITS OF A SATELLITE-BASED SYSTEM FOR NATURAL RESOURCE MANAGEMENT? JONATHAN llm\erslty

of Cahfomla.

Department

F

BARD

of Industnal Engmeenng Hall, Berkeley, CA 94720

I Ret erred I 5 Februarv

and Operations U S 4 I983

Research, 4115 EuzheLern

)

Abstract -This paper presents the results of a two-vear study commlssloned b> the Department of Intenor to examme the costs and benefits of a remote data acqulsmon system for natural resource management Because earher expenence with a number ofexpenmental programs had been so favorable, It aas felt that a full analjsls should be undertaken Subsequendy, five areas were selected for detailed In\estlgallon agnculture land use forestry, water resources, and rangeland In each Instance, the modehng uas performed althm the framework of a case stud) Net benefits were denLed from INo sources (I) cost sa~mgs ,It the data acqulslllon level, and (2) Increased economic efficiencies at rhe operational level CJlcularlons were done parametrically lo account for technologcal dltTuslon uncertamtles In ImplementalIon and \anous assumptions concernmg discount rates and cloud co\er impacts The results ha\e bhown that d two-sdlelhte system UIII more than hkelv pay for l&elf o\er the program s planned ten bear

I

WTRODM-TION

cluslbe espenmental results, narrowed the focus somewhat to areas uhere potentIalI> large payoffs existed The estlmatlon of costs and benefits rested upon an understanding of the techniques and procedures b> which mformatlon may be denved from ERS data and used by declslon makers When posblble management strategres and mformatlon flows here simulated to gve a more reahstlc picture of the underlvmg dynamics Uncertainties Inherent In the estimates were addressed b> parametncally analyzing their response to the various Input factors For the purpose of companng investment alternatives, a baseline svstern was defined compnsmg current practices and technologes Overall measures of effectiveness rneluded econormc effiaency. economic dlstnbutlon and envlronmental and social Impacts In the sequel It WIII become clear that man) of the factors routmely affectmg the calculations were speculative m nature and thus required a mixture of Judgement and caution In their treatment Others however. inspired a much higher degree of confidence due to the presence of a famlhar operational base The results should therefore be Interpreted ds a condltlonal mdlcatlon of system performance rather than as a prescnptlon for Investment In the next section. an outline of each of the case studies IS presented along wrth the ratIondIe for their selectlon Thus IS tollowed m Section 3 b) a dIscussIon of system performance requirements and the Impact of cloud cover on the attainment of benefits As It turns out the Importance of cloud cover cannot be overstated since It directly affects the tlmehness and quality of data collected, as uell as the number of Imagmg platforms ultimately required Svstem design and costs dre highlighted m Section 4 Section 5 concludes by assemblmg the summarq statlstlcs for an overall comparison and offenng d number of observations based on the assumptions and climate m which the proJect was undertaken

Ongoing developments m remote sensing and telecommumcatlons now make It possible to provide natural resource managers wclth the type of techmcal data they need for broad-based declslon making at both the local and national levels Man) of the public systems currentI> m use can be traced back to the technology transfer programs of the least tbo decades The U S Department of lntenor (DOI) has figured promlnentlq m this regard by sponsonng and actIveI supportmg a banetj of earth resource satelhtes data networks, and advanced processmg facllIties One of their more recent programs centers on an Integrated Earth Resource Survey (ERS) system. compnsmg satelhte, aIrcraft, and ground data acqulsltlon technologies 4s envIsIoned, such a system uould offer repetitive Imagery at various scales, spatial resolutions, and spectral coverages m accordance with specific user requirements The purpose of this paper IS to present the set of costs and benefits associated with this particular program[ I] The overall motl\atlon for the underlying stud) came directly from two sources First, previous anal) tlcal work suggested that the apphcatlon of an ERS svstem to long range planning and resource allocation problems would yield slgmlicant pubhc sector benefits[Z. 31, and second, early expenence Mlth Earth Resource Technology Satellite (ERTS) data Indicated that It \\ould be techmcally and economlcally feasible to extract the t)pe of mformatlon needed for shortterm natural resource management[4-61 A case study approach provided the analytic framework ulth land use water resources. rangeland. forestry, and agnculture being smgled out for broad mvestlgatlon The time constraints of the proJect. though, coupled with the hmlted a\allablht~ of con____ tThe author wl,hes lo acknoaledge the Couan Research Fund for probiding partial funding for ths research IS

J F BYRD

16 2 Sl~hlM4RI

OF CASE STC~DIES

d~sumptions require. a parametric dpprodch ids taken elements In the Figure 2 shows the li\e pnmarj npphcatlons system The datd Interpretation model relates the raw ERS data to extractable earth resource InformatIon (ERI) such as mater supphes and vegetation type5 Operationally, ERI IS manipulated b\ management InformatIon systems to weld InLentories land use patterns. crop forecasts. and other >ummar> statlstlcs The management declslon model then uses this output for both short-term action and long-range planning FInally, potential benefit3 In terms of cost sa\lngs Improved efficlenc) dnd mcreased Income. are analyzed and combined \ilth s> stem costs to pro\ Ide dn overall picture of net cdsh

The dress chosen for detailed examlnatlon typ~cdllq reflected those In which ndturdl resource managers hd\t hdd experience working filth remote data acqulsItIon systems and Image-enhanced products The mimedlate goal uas to develop apphcatlons that \\ould be fdkorablb received and easy to Implement. thereby a\oldlng resistance at either the mstltutlonal or user lebels Figures I and 2 respectlbelv shoN the orgamzatlon Jnd methodologv of each case studv Figure I summarizes the flow of ke, anal) tic tasks Those on the left dedl ulth the tcchmcal aspects of the program ndmrl!, the experimental results and the ddtd systems needed for general management Comparisons were made between the wthout system uhlch IS essentlalli a collectlon of exlstlng systems, and the prosystem which IS based on an operposed w th ational ERTS-t>pe (now LANDSAT) technolog> Ta& on the nght deal wth the analysis of benefits To permit rephcation and re\ision of estimdtes ds new technical data become dvdilablc or as different

tlO\\

The assumptions and underlying rationale tar each ot the case studies are outlined belo\\ A full dercrlptlon mdq be tound In [I] In most Instances. elIstIng models Here used for the calculations, benefits mere expressed as the Incremental difference m net Income or cost sa\lng resulting from wlthlwlthout * system

Analysis of Management Use of Data

Determlnatlon of System Performance Reqdlrements

I

I

I

t

1 t

_

I _

kllth System

Wlth@ut 5vstem

J

Estlmatlon of Econnmrc Benefits

Capabllltles

I t

l-

Costs an3 System Effectiveness

Social

FIN I Caw stud\ organwauon

The costs and benefits of a sdtelhte-based system for natural resource management

17

ER3 DATA

1nterprctat1cn

Earth

Resntirces

Informatlcn

RefIneA Unlerstanjlng

Mana
t Model

Management Actwn v Benefit Models

Cress Benefits t Cost-Ssneflt Analysis

FIN 3 Management apphcauons system diagram

comparisons A ten year operatlonal hfe (1983-92) was chosen to reflect what was thought to be a reasonable penod over which the system could be Implemented and normahzed before obsolescence set m When appropriate, a learnmg curve was Imposed upon the costs and benefits prior to dlscountmg All results are presented m 1979 dollars Sensltlvmes were examined by employmg several different learnmg curves and varqlng the discount rate from 5 to 1.5”” Land use This case study focused on land use plannmg and management at the state level, and Identified mappmg Inventorying, and pattern updating as the pnmary areas of apphcatlon Secondary constderatton was g\en to facility site selectton, traffic routmg, utlhty service load planning, and the compllatlon of environmental impact statements The estlmatlon of mappmg benefits was based on the assumption that It would be possible to obtain cartographic products from ERS photographlc data SEPS

\nl

Ik

No

I--B

comparable to those obtained from con\entlonal sources, but at a substantial saving m productlon costs, and Interpretation and analysis time Owing to the passage of extenstve land use and envlronmental legslatlon, state-wide demands for these products are expected to nse rapidly over the next ten years Cost savings were calculated by companng conventlonal acqulsltlon and extraction methods with those of the ERS system As was true rn each of the case studies, though, the cost of the ERS system was treated separately due to the basic difficulty of equltablb allocatmg Its costs ober mdlvldual apphcatlon areas ‘This omlsslon biased the results against the without” system making It lmposslble to drau any conclusions before refernng to the analysis In Section 4 The calculations for Inventory benefits followed from the propoatlon that ERS-dribed mformatlon could be economlcally made available to state plannmg departments on a routme basis (see [7] for an example) Generally, states exploit land use datd at three different scales or levels of detall[g] Level I

IS

.I F BARD

I 500.000 Level II I 13 000. and Level 111. I X000 dependmg on their sophlsttcatton and planrung need5 AL\) concelLed an ERS system would satlsl\ these damands ulth the appropriate comblnatlon ot satellite Jnd dnclllary data collectton technique> SpecIficall>. Level I data would be furnished boleI> hv the ERS system. Level II data by the ERS s\\tern supplemented by high-altitude aircraft, and, LetsI 111 data pnmarll> b, low-dltltude aircraft guided bv ERS Imagerv Therefore. after suhJectl\el> classlfqmg each of the \tdtes Into one of four cdtegones based on their current planning trends the dollar figures In Table I were uled to compute cost savings This was done for each \tdte h> multlpl~lng the savings per square mile h\ the rlrea mapped as a function of lebel. and then \ubJecting the results to one of four learning curves depending on the pdrtlculdr state categor> Llslng this approach the IO yr discounted benefits \tere consenatlvelq estimated to fall between $45 I and $85 5 mllhon The upper end of the range reflects
technologlcdl coefictents to account for reduced data acqulsltlon costs and Improbed Image processing> FRES was also used to estimate benefits from Improvements In range producttvlty LJnderlylng this phase of the analysis was the assumption that an ERS system would lead to better investment declslons For the third tier a slmulatlon model[l I] kas used to measure the value of ERS-derived range feed condltlon Information a5soclated with buying dnd selhng hbestock Table ? summdnzes the benefits expected to result o\er the 104’ operattonal hfe of the system A IO”,, discount rate was used throughout 4s a consenatlve measure the benefit streams mere sharpI> reduced In the earlier )ears of operation to account for Implementation problems and user reluctance See [I I] for an esdmple

ICuier remurces Earl) reblew of water resource problems Identified d number of areas of potential apphcatton, lncludrng drls lands and lake Ice surveys, swamp area management. snoH mappmg, and runoff forecasting For reasons previously cited the latter two were chosen for detailed mvestlgatton In particular, It #as felt that ERS-denved snow maps and runoff forecasts would be extremely valuable tn resenotr management. producing benefits bv mcreastng the availJblht) of hvdro-electric power generatlon through more efficient water storage practices 4t the operdtlonal level confhcts that anse betuesn flood control and power generation Interests are generally resolved by constrammg water levels to follow a prescribed mass flow curve[ 121 Because forecasts are used to make the adJustments, mltlal oberestimates of runoff will result m excess flood control storage 4 reduction In forecast error will therefore permit a more rapid rate of refill without reducing the flood protectlon afforded by the dam This means that dddihonal water will be avallable at drawdown to meet power demands with a greater degree of rehablhty To estimate the benefits of usmg ERS-denbed runoff forecasts for adJusting releases dunng the refill period. d simulation of d multi-purpose storage facllIty was undertaken The Hungry Horse resenolr system served as the model Assuming ERS data \\ould be more timely and accurate, calculattons were performed parametncally with the percentage lmprovement m water runoff forecast accuracy’ being vdned from I to ISo, Local benefits were estimated then extrapolated to the nation as a whole bv consld-

The analbsls of management InformatIon needs m the ran&eland case study, and the ablhty of an ERS \\stem to meet these needs, resulted m the selectlon of three areas of apphcatlon (see [9] for the details) Thebe Included In\entor)lng for long-term resource re~lloc,~tlon. momtormg for mid-term mlestments, and the development of range feed condltlon reports Ior short-term hbestock management The first resource reallocatlon. Involves the determlnatlon of the tape and mtenylty of rangeland use o\er extended periods of time Benefits result either from In\enton cost sJvmgs or from a portion of the efliclencles of reallocatlon that can be attnbuted to the ERS s\htem FRES[IO], a linear programming model developed bv the U S Forest Service provided the framework after adjustments were made In Its

Table

I

Cosl

to extract land use mlorm&on (per square rnlle)

Ls\el

WIthout ERS

With ERSt

I II III

$4 93 $5 26 $14 65

$0 36 %I 62 %I? 77

tlncludes

the costs ol supplementary

Table ~~

-_-

aircraft

2 Summary

of benefits (%m~lhon)

Tier I II

Resource reallocauon Improved range

productl~lt> III Liveslock drcwons

In\enlor\

Benefit area Information cost sabmg and Improbed efficiency Reduction in cost per AUM (0 5-2 00,) Rancher Income

Total

Benefits I2 5-37 9 22 190 6 60-92

409-137

7

The COSTS and be&t>

of d satelltte-based skstem lor naturd

enng the pnnclpal regions where hydro-electnc power IS currently generated The proJected benefits In 1979 dolldrs rdnged from $28 9 to $89 9 mllhon m present value teTms Fiwesrr I The lmtlal anal>sls of apphcatlon areas concluded that benefits m forestry would result from Improvements In Inventor> practices Estimates uere based on the potential capablhty for an ERS-asslsted multi-stage sdmphng hjgtern to furnish growth and contour stdti5tics equivalent in qualit!, to those currentI> being pro\rded by comentlonal means 4t the time of the stud\ though the onI1 demonstrated ERS capdbiht~ in forest inventorying related to the estimation of timber volume--Just one of the many measurements needed for effective management However m the Judgement of the pnnclpal Investlgators. all Information products of the present svstem would ultImateI\ be aLalIable from the ERS system without an\ reduction rn qualit) (see [I?]) Therefore benefit estlmdtlon proceeded under a senes of dssumptlons concermng to perceived demand and user acceptance In particular m\entory mfomiatlon suppI> dnd demdnd Lur\es developed b> the Forest Service In cooperdtlon ulth the Bureau of Land Management mere used to compute cost savings and consumer surpluses It UJS next decided that survey frequencies should be dltered In certain dynamic growth areas to exploit the with s~steni s svnoptic cdpdbihtles Thub,. Intentor) rates \\ere slgnrficantl> Increased m the South dnd Southeast Cost sabmgs were estimated for the old level of output while a new benefit was computed for the Incremental Inventor) Oier the IO>r life of the program. a $15 mllhon-$27 mllhon gain IS expected to result

This area Included dll aspects of crop productlon from the farm level through domestic consumption, ds well as the support Industries producing fertilizer pestlcldes and machinery The case study. hoheber. focused on the potential benefits that Nould result from improvements in crop acreage forecasts Specific techmcal capdblhties are dlscussed m [14] The Ltlthout’ system ma) be thought of ds the Ll S Department of Agricultures crop reporting serbice hhlch currently uses a national sampling plan to develop monthI> productlon forecasts Data are gathered through a combmatlon of denal photographv and ground truth In contrast, a statIstIcal sampling svstem based on satellite Imagery IS expected to oKer the lolloulng Improvements l A substantially larger number of sample units to reduce gdrnphng error l 4 floating sample to permit substltutlon of fields which dre cloud free for those ahlch may be obscured l A computer-based estlmatlon program affording frequent forecast updates as ddditional Imagery IS collected. and ds ldentlficatlon accuracy IS Improved through enhanced processmg To determine the accurdcj of the ‘filthout system an analysis \sras made of the average forecast errors by crop dnd month for the years 1969-1978 The September stdtlstlcs were chosen for the com-

resource

management

19

panson because the) uere sufficlentlk far enough Into the growing season to have permItted several satellite passes The analysis. Itself. was aimed at determmmg what benefits aould be achlebed If dn ERS system. m fact Improved crop acreage and production forecasts An economic model developed by Hayaml and Peterson [ 151 and Bullock [ 161 and subsequentI> modified for case study purposes. provided the analltlcal tool In d general sense. the model calculates changes In gross economic welfare from improbements m crop productlon forecasts Benefits were estimated pardmetncallv to reflect a range of values for crop elastlcrtles and pnces. reduction In forecast error dnd degree of government mtervention The calculations led to projected benefits ranging from $80 mllhon to $130 mllhon In present \ dlue terms 3 SbSTECl PERFORMG’KE QND CLOUD CO\ER

REQllIREhlENTS 9NALL1SlS

The next step m the project was to determine the mlmmum data acqulsltlon charactenstlcs of the system so that the benefits discussed m the previous section \nould be achieved with a reasonable degree of confidence Based on a comblnatlon of dndl> tlcal results and Department of lntenor guldehnes operational requirements were debeloped for each of the followlng parameters geographic regon of Interest. frequencv of co\er,ige time of year coverage, time of ddi cotcrage maximum allowdble cloud LoLer per photograph. resolution dnd spectral co\erdge Flelclbllrt> m speclfjlng performance measures for resolution and spectral coverage WJS somewhat hmIted bq the constrdlnts of aLalIable technolog) Nommal Lnlues were therefore chosen to conform nlth current LANDSAT and high altitude aircraft sensor designs (see Section 4) The other parJm2ters tver2 fix2d bj detsrmmmg the mmlmum performance lebels required to realize mdlbldual case stud> benefits Table 3 ~ummanze~ the rebults Because processing costs and interpretation problems medsurablv Incredse with sun angle Lanation d sun-synchronous orbit at IZ 00 p m local solar time was established for the time of day coverage It should be understood though that this choice does not ImpI) d fixed sun angle between the local berncal and the sun hne Both the latitude of the object being Imaged dnd the time of bear also affect the dlrectlonal ianatlon Flnallv each area of apphcatlon required coverage on dn annual basis except rangeland and forest mventor)lng hhlch folio\\ a 5-15 jr cycle Gtwerol ca~eruge rtq4wenletIIs Llrban land use pldnnmg tends to be governed bv mdlvldual proJects at the local or regonal level. so It IS difficult to set a fixed time table for data collectlon It Hds theretore felt that at least one complete image per annum would be needed during a non-winter season This requirement serves to maxlmlze the chances of clear geographic ldentlficatlon For non-urban tracts, the frequencv of coverages vanes dlrectlv with the mtenslt) of use unpopulated areas ma> require imagmg only once elerq lew jears while Jgncultural lands must be sampled teberal times each season m order to assess grov.mg condltlons and crop extent In each Instance though it uas

J F BARD

20

determrned that enhanced recogmtlon of geographic features through shadou mmlmlzatlon could be ,lchle\ed with a solar noon orbit This line of reasonmg extended to the other case studies as well The achleiement of rangeldnd benefits In each of the three areas of apphcatlon IS dependent upon seasonal colerage acrass the western and southern portions of the country Expenmental results have Indicated that an ERS system could be designed to Integrate satellite and aircraft sen$mg Qstems with current ground samphng procedures for mventorylng range dlstnbutlon and stdnd[6] Multi-stage techniques would call for cocerage from March through August to provide data on plant development during the dltlerent gro\+mg penod5 Smce satelhte Imager> uould be used for lmtlal stratlficatlon and allocatton of addItIonal samples one photogrdph In the spring and summer \hould be adequate For first level strdtlficdtlon It nas felt that no more than JO’,, cloud co\er per frame could be tolerated In addition the case study dnalvsls suggested that dn ERS system could provide a unique means ol monitoring range resource5 Forage development, h\estocb turnout dates, range conchtlon and trend, and to some extent patterns of utlhzatlon can all be Inferred from ERS Imagery acquired dunng the grazing season Mtmmum requirements dictate 70”,, cloud-tree coterdge once each spring and summer Range feed condltlon reports on the other hand.

Table 1 Summon

CASE

STUDI

LAND USE Morlltcr1ng

of performance

TIHE OF YEAR FOR ALLOWABLE IMAGING

Iequlre monthly update\ except perhaps during the winter Inputs would Include data relebdnt to soil. Legetdtlon. and motsture content In the hater resource5 case stud>. It was determined that the tlmlng of ERS overfltghts IS cntlcal In obtammg useful hydrologic InformatIon Generally the first flight should take place soon dlter the light IOU I)lng snow has melted (as lndlcated b> the mltldl nse In streamtlow) This permits the measurement ot the wedge extent dssoclated with the remamlng snoupack. however LANDSAT In\estlgators habe reported that Imagery with more than 30”, cloud co\er I\ of little Ldlue m denbmg forecasts (e g see [I711 In addltron. a sun angle vdnatton of onlb d ten degrees IS lIkeI\ to make It lmposslble to construct mosdlcb from photographs taken during different orbital tllghts High altitude alrcrdft would therefore. hd\e serious difficulty satlsfjmg rnls5lon requirements For forestry the vearly coverage Indicated In Table 3 was predicated on benefits estimated for m\entor\ cycles ranging from 5 qr In parts of the south and southeast to I5 jr m Alaska The reasons for thlb difference relate to timber usage and growth rates The most dynamic forests dre the commercial comlers of the Fouth and southeast. and ma) produce mature trees In twenty to thlrtl years Using a cycle Irequencq of 5 kr near complete coverage (95”,) 15 needed for one-fifth of this acreage each lear to achieve the proJected benefits Depending on the

requirements for each apphcatlon

ALLOrlABLE CLOUD COVER FREWENCY

March-Novemoer

1 non-winter

PER

ERTS

FRAME

ared

REGIONAL DESCRIPTION

ID%

All

western half of ” s

II S

photn:year AANGELAND Honltoring

Inventory

March-Hay.

1 pnoto, Seas""

30%

July-August March-Yay,

1 photo'

30%

June-August

Fee9

Condltlan

SeaSon

April-September

1 pncto/

rnC”Ltl HATER REWURCES Regwn I

Apr11

15-4lay I5

15-May

Reglcn

?

April

Regnn

7

tlarcn 15April 15

FQRESTRi Inventory

Hay-acrcber f1eciJuous All

tiestern half of u 3

15

areas)

Year

i c0nlrerou3 area31

AGRICULTURE Crcp Fnrecascs

All

yaar

All

II j

The costs Jnd benclits of

d

satelhte-based system for naturdl resource management

lnstrtutmnal arrangements. partwxlar tracts may be specified m advance as current practice dictates, or ma\ be Larted as a function of Image avallabthty However, because the dynarmc areas exhlblt relatlvel!, constant spectral slgnatures. photographs gathered oker an entlre year may be used to produce time composite mosaics This means that any gaps caused b> cloud cover or mlscahbratlon may be filled m from Imager> collected In previous years For dectduous forests winter defollatlon Implies that only Apnl through October data ma) be used Once again though. photographs taken m previous years may be Interpolated to produce the desired 95”, cloud-free mosdic Requirements for achlevlng agncultural benefits mere based on the multi-stage sampling scheme outlined aboLe It a fall forecast IS to be obtamed, a mlmmum of one photograph over the growmg fields would be needed by September Since a floating pomt sample IS called for cloud cover from 50 to 80”, could be tolerated The exact figure depends upon the partrculdr crop, Its spatial extent, and the surrounding blologcal phenomena

Because the sensor systems considered m this stud) are designed to gather data m the vlslble and Infrared portlons of the electromagnetic spectrum, their performance IS Imuted by the presence of clouds The ablhth of a glben system to operate effectively, therefore, IS a function not only of the mdlvldual case study coverage requirements, but of the temporal and spatial cloud dlstnbutlons as well From a desrgn pomt of Lien. the number of lmagmg platforms and their rrspectlbe l%ght patterns are the Independent karlnbles In the equation The statIstIcal nature of the problem though, precludes an optlmal solution Despite the fact that cloud dlstnbutlons have been extensltelq developed over the last I5 yr [l8, 191, It wds dltlicult to speclfq their exphclt relatlonshrp to benefit3 on a per frame basis As an alternative. the mdslmum percent of a LANDSAT frame that could be obscured and still not matenally affect the proJected benefits’ wds used as a system performance requirement For each apphcatlon, the probablhty of achreblng at least one photograph with equal to or less than the maximum allowable cloud cover over the region of Interest Has thus determined Benefits were subsequent11 reduced (see below) by the corresponding proportlons for each Investment alternatiles under conslderatlon The stdtlStlCS used in the calculations were denved from d NASA supported study[ll] by adJustlng previous measurements for LANDSAT’s field of blew Depending on the area of apphcatlon, one of two methods was used to obtain the final probabIlltIes The first centered on the attainment of a single cloud-free Image’ and Involved the derermlndtlon of the mlmmum number of o\erl%ghts necesgarl to achlebe the desired degree of \lslblhty m d single frame Graphs deplctmg the parametric relatlonshlps between number of passes relative trequency of clear skies and probablhty of observing cloud-free are,ls here constructed for each of the I9 homogeneous cloud regions Identified for the U S m [I91 This method IS most appropnate for apphca-

?I

tlons in Hhlch the required data are highly time dependent as m the water resources case study The second method m\olved the constructlon of a ’ time composite mosaic” from sequentially acquired LANDSAT frames Here, cloud-free sections of the same area are pieced together from mdlvldual photographs Each section IS categorized,,“› separately anaIqzed and they merged to form a composite This approach IS most appropriate when the physlcal or blolo@cal phenomena of Interest are slob to change Tree growth m the northNest IS a good example Once the probablhtles of satlsfymg the performance requirements m any gtven year had been computed, tradeoffs were made between addItIonal lmagng platforms. attending costs, and the Increased hkehhood of achieving the proJected payotrs The followlng equation was used to compute net benefits (excluslce of ERS system costs) m year t

NB(t)=~~[f,,(r)B,~r)+(I I

-f,,(lWB,,(t)l

I

n here P,,(r) IS the probablhty of sdtlsfymg the 1th applrcatlon requirement m the /th cloud ober region in year t, as a function of the number dnd type of lmagmg platforms B,,(t) IS the (net) benefits from the Ith apphcatlon m the /th cloud cover regon after allowance for processmg. dlstnbutlon and mterpretatlon costs m year t. and DB,,(t) IS the dlsbenefits (costs) Incurred for not meeting the lth apphcatron requirement In the Jth cloud cober regon in qear t 4 key element m the above equation IS the dlsbenefits that result when performance requirements are not met These usually take the form of supplementdry data collectlon costs arising when the ERS system IS unavailable In such Instances the declslon to use a backup system to gather mlssmg data would depend upon the differences between the Incremental costs of that system and the expected gains to be realized Once the net benefits were computed for each time penod. their present value was determined ds follovhs p1

= NB(I) (I+r)+(l

NB(2) +r)‘+

YB( IO) +(I

+r)“’

where r IS the discount rate, and the first time period corresponds to 1983 Table 4 summanzes the results by case studv after adJustments aere made for cloud cober One and two satellite systems Here considered for the IO yr operdtlonal penod (cost are presented m the next section) As can be seen, net discounted benefits m 1979 dollars range from a loi+ of $71 6 mllhon to a high of $389 7 mllhon When two satellites are deployed almost all the benefits will be realized blth onlj one satellite benefits shrink b) approx 65””

4 S\STEM

DESIGN AND COSTS

An operatlonal ERS system expressly designed to meet case study objectives and related DOI requlrements can be readllq assembled from a banet) of er;lstmg hardware and telecommumcatlon networks Accordingly. a baseline system was defined from current mventones and engneermg data gathered from earlier NASA studles[2 31 Its maln features Include d LANDSAT-type spacecraft carrlmg a

Net Application

AgrIz-IJltur~ Fwestry rlster

one

23

Present

Sat~llitte

6 -

32

9 ? - I6 rcesolrrce

‘,5-

Value

0

T*o

million) sace11ites

I

80 'Cl-

13" 0

11

15 0 -

.?7 ‘I

14

9’:,-

354

Rangelard InVWlt@ry Honltnrlng Fee4 Land

4 2 ,9 -

'r~ndlt~@ns

Use

10 6 20 Y

8.1302

multi-spectral scanner subsystem IMSS). three return beam LIdIcon (RBV) cameras tbo wldeband video tape recorder5 and d data collectlon system as palIodd The spacecralt ~‘111 be placed m a clrculdr. >un-synchronous orblt at an altitude of 491nm1 (866 km) This ~111 allow the bensors to obtain repetItl\c IO0 n ml (I80 km) wide lmager\ oier the U S e\er1 I6 days On consecutive dd>s the satellite subtrdct will be displaced b> 86 n ml ( IS5 km) gwng the swaths 14”” overlap at the equator Ah on-bodrd orbit ddJustment svstem HIII trim the traJectory as necessar! to achlebe the I6 da) repeat cycle The launch phase and mltlal check-out will be handled b\ NASA s spdcefllght center dt Houston Thereafter. the responslblht\ for on-orbit trdckmg and datd acqulsltlon ~111 fdll to the DOI statlons at SIOUS Fdlls S D dnd Falrbdnkb Alaska The data procelslng facility will be locdted at SIOUX Fdlls. and will support the same functions as the NASA data processing fdclhtv (DPFI locdted at Goddard Basic product\ WIII be ldentlcal to those probided b> LANDSAT lncludmg system-corrected and scenecorrected 70 mm film ,md computer-compdtlble tapes Datd obtained at Fairbanks will be transmitted to SIOUX Fdllc for processing All mission dcti\lties ii111 be controlled dnd scheduled b> the operatlonb control center (OCC) located dt SIOUX Fdlls The lollouing parameters chdrdctenze the system 5 remote sensing pdckage spatidl resolution. spectral cokerage areal extent. and frequency of coverage N4SA engineers ha\e tentntlbely frozen the values of these parameters at the level> sho\\n m Table 5 The coverage cvcle no\\ set at I6 dais hohe\er can be halIed b) flwng A becond bdtelhte Because of the adtdntdges accruing to the users from Increased co\erdge both d one- and a tuo-satellite svstem habe heen cobted o\er the proposed operntlonal hfe of the problem

Strdlghtformdrd Jccountlng methods were used to estlmdte m\estment and operatlonal costs for each of the folIowIng categones spacecralt sensors oper-

-735

1

12 5 173-

‘7 Y 710

It-

31

rrb

I - 34 6

atlons control center data processing. trdcking and data acquisition dnd launch Lehlcle The dpproach tdken was similar to that developed for the original ERTS program wherein the costs to duplicate existing equipment dnd fdclhtles were determined ACcordmglq. Informdtlon \+a~ bollcIted lrom the prlmark contractors GE. RC4 dnd Hughe> dnd fornnrded to Goddard’s Resource Pldnmng Ofice for analqsls Reports uere then prepared for each cost categor} tracking dnd data dcqulsltlon 9s background costs rare associated with building the t4ltles bu)Ing and Integrating the hardware. dnd providing operational support OCC costs Include evpendltures tar mlsllon plannmg flight schedulmg record mdmtendnce. Image annotatIon and tape preparation The first data column In Table 6 summarizes the IO \r costs of the one-sdtelhte ERS system It IS assumed here that 5 \ehlcles (no spares) UIII be built As Indicated the spacecrdtt dnd pallodd account for about tbo-thirds of the $129 3 mllhon total Investment costs Fdctorlng in operation5 costs @es d dolldr figure In the nelghborhood of %I77 4 mllhon NOW adding saldries. travel expense3 dnd overhead places the grand totdl at $18X 4 milhon If a nominal IO”” discount rate 13 assumed. the correspondmg 1983 base year present bdlue IS roughly $136 mllllon 4 procurement program which Includes d spare spacecralt and pa\lodd hdb dlso been considered As emlsloned the addltlonJl satelhte uould be purchased midway through the program and launched in 1988 The present value costs for this scenano hd\e been estimdted at %I 52 I mllhon The EROS stud! [2] cdlculdted the need lor d spare in probdhihstic terms to be less than Y,, Calculations for a t\to-satelhte blstem proceeded along the same line5 ulth the folloH]ng dddltlondl ashumptlons (I) harduare costs for the spacecraft, pallodd and launch kehlcle uould double nhlle support costs would remain the bdme. (2) miestment dnd operations costs for the OCC and TD4F uould \lmlldrly remain the same (3) DPC costs \\ould nse b\ JO”,, dnd (41 cl\11 serilce costs uould Increase b\

The costs and benefits of a satelhte-based system for natural resource mdnapement IO”, Accordtngly.

the present

tlve IS seen m Table 6 Taktng mto account an addttlonal launch becomes $245 2 mllhon spares accompamed

value of this altema-

through the program elevates the system’s dlscounted cost to $266 3 mllhon

to be $222 7 nulhon a spare set of hardware with m 1988, the total cost now Fmally, the purchase of two by two launches nudway

5 DISCUSSION An examlnatlon 4 and 6 mdlcates

Table 5 4 comparison oFvanous

SATELLITE LANDSAT-TYPE SKYLAB

3.800.000 to

1

of the results presented m Tables that a basic one-satelhte system

ERS swtem parameters REnOTE

PARAMETER

SENSING

PLATFORMS

GROUND

AIRCRAFT

to

1 1

1 2.800.000 to 1 25.000

1 s00.000 to 1 500

I a0

1 125.000

260 It to (100 ft

260 <50

20 l-t <2 rt

<

Resolution

coverage/ Frame

13.000 miles

1800 sq miles to 10.270 sq miles

400 sq miles to 100 sq yards

100 sq to
16 days

Sporadic

As Needed

As Needed

Visible. Infrared 5-l 1 urn) U Bands

Vislsble Infrared. flicro-wave

Vlslble. Infrared. Micro-wave, Radar

Vlslble. Infrared, Radar

Manual.

rlatWe1. ADP

HellEll. ADP

Mellllel. ADP

Relative Scale

Relative

Frequency coverage

of

spectra1 coverage

Data

sq

(

Analyses

ADP

procedures

Table

It rt

to

6 Summary costs for altemawe

CATEGORIES

No Spares

Investment costs spacecrarc Payload

5D

23

occ DPF TDAF Launch

Vehicle

Investment

Tow1

Operarions

Costs

Operations

Costs

Subtntal ClVil

Service

Grand

Total

Present

' In

1973

1

5 rt

ft rt

one

TWO SATELLITES spare

61

6

29 6

No Spares

104 6

56 2

40 56

40 56

4i) 73

31 9

13 6

13 6 37 3

13 0 63 9

129 2

151 7

21 0 75 19 7

21 0 75 19 7

249

5

One

spare

115

2

49 2 40 7 3

13 6 68 2 277

1

iost

occ DPF TDAF Total

9

to

satelhte systems (.%mtlhon)

3NE SATELLITE CJST

Costs

Total*

dollars,

discounted

48 2

48 2

177 4

199 9

11 0

11 0

21 0

21 0

98

98

14 7

19 7

50 5 299

0

I2 1

50

5

328 2 12

1

198 4

210

9

311

1

340

I

I36 0

152

1

222

7

245

2

at

23

10 percent

2-I

J l-

cannot full] be Justified by return on Investment Jlone CJnder the most optlmlstlc circumstances the benefit-cost dlfferentldl of $18 4 mllhon barely exceed> the break even point When a two-satelhte skstem IS considered, though. the conclusions are a bit more posltlbe The correspondmg values range tram a low of - $41 9 mllhon to a high of $167 0 mllhon. suggesting perhaps that the program might pd\ lor Itself But even If this were true, Issues relatmg to user acceptance, dlstnbutlonal equltablhty, opportumt\ costs and community Impacts must first be explored During the course of the case studies, an attempt MS made to address edch of these pomts from a normdtlbe perspectlbe When possible, a full technolop\ dssessment spanning social, pohtical. mstlfutlonal dnd en\lronmental dimensions was conducted Ie g sx [91) The general consensus emergmg from these zsessments tended to support the contentlon th,lt Ihe program s non-technicdl Impacts would be mlrnmdl while Its mno\atlve range would be well \tlthln the scope 01 existing practices In addltlon, It was felt that because government agencies would be the prlmdrb system users the dlstnbutlon of costs ~md benefits would not fall dlsproportlonatelq on any one rector QuestIons concerning alternatlbe mvestment strateges and opportunity costs could not be answered directly, however It should be noted thdt onI1 d limited number of apphcatlons were examined In alI hkehhood many external economies and unconsidered mternatlondl benefits Nould result once the tull skstem \tas Implemented The final question though. Should the system be built” 15 still unanswered, dnd probably ~111 remam so m stnctlb economic terms Nevertheless If past performance IS to probide an) clues the declslon to proceed IS lIkeI to be grounded m the more abstract arguments ol national goals. percelted needs, and technolo@cal opportumtles When the first ERTS \~as ldunched in the early 1970s. there was d great hope ulthln the space communltj and among their congressional allies thdt the benefits of remote sensing would soon be made avdllable on a commercial scale Since that time pohtlcal conslderatlons have largeI\ Influenced the techmcal debate to the extent that d number of more promlsmg but less vlslble projects hd\e been temporarrlq pushed aslde The launching of the fourth LANDSAT spacecraft dunng the bummer 01 1982 seems to mdlcate the dIrectIon rn \\hlch the program IC noi\ heddmg

REFERENCES J F Bard 4lternale ERS bystems efkcri~eness anahsls Trtlr Rep 8-3504 prepared for the U S Department of Intenor IX Booz 4llen & HamIlton Bethesda MD

1I’)hOb EROS rlppllcdtlons pared tor NatIonal

benefit analbsls Rep I1695 preAeronauucs and Space Admmls-

BARD

3

3

5

6

7

8

9

IO

II

13

l-l

IS

I6

trdtlon by Westmghouse Defense and Spxe Center Baltimore MD (1971) R Krzyczkowskl D N Powell and E S Putnam Rnvew and 4ppralsal Cosr-Benehr 4nal,sry of Earrh Resoutxes SWT~I Sarellrre Swems Prepared lor the National Aeronautics and Space 4dm&nstratlon by Interolan Corn Contract No NASW-2084 Santa Barbara’ C4 (19j3) C R Frank and K P Hew, Cosr-Ben&r .Slud\ ol rhr Earrh Resourtes Obserralron Sarellrre Swem &a-my Land Cfanagemenr RCA Defense Electromc Products Astro Electromcs Dwwon, Pnnceton NJ (1973) P T Tueller G Loram and R Haholsen Natural resource InLentones and management apphcatlons In the Great Basis Prot 3rd ERTS Sump pp 107-I 18 GSFC Greenbelt MD (1973) R G BenlIe) Llsefulness 01 ERTS-I satelhte Imagerb as a datd gathenng tool b\ resource mdnagers in the bureau of land management Proc 3rd ERTS Slmp pp 67-75 GSFC. Greenbelt MD (1973) J R Wilson C Blackum and G W Spann Land use change detection from computer processed L4NDSAT data Proc 14th Con/ Urban and Reg In/iwm S~srems pp 4X&69 Atlamd G4 (1976) J R 4nderson E E Hard) J Roach and R E I4”vllmer 4 land use dnd land cober clawficauon v,stem for use wth remote sensor data ProJessronal Paper 964 U S Geologcal Suney Washington D C (1976) J F Bard dnd A Watkins Improbed rangeland management wth an edrth resource surve) system Teethnologrtal Forecasrmg and Sotral Change 1983 The ndturr s range resources 4 forest-range enblronment stud) Rep I9 U S Department of ;2gnculture Forest Senlce. WashIngton D C I 1975) J F Bard The application of a remote ddtd acquisition skstem to Il\estock management benef% estlmdtlon Socro-Ecun P/an Str 17(Z) 49-56 (1983) \’ Klemes Storage mass cur\e dndl>slb in a bjstemU’arer Re,ourte\ Res 152~ and11 tic perspective 359-370 (1979) S J Walsh. Coniferous tree species mapping using LANDSAT data Remore Sewng of Enrwonmenr 9( I ) II-26 (1980) hl E Bduer J E Clpra P E 4nurd and J B Ethendge. Identlficdtion dnd ared estimdlton of agnculturd1 crops bq computer classlficatlon of LANDSAT MSS data Rcmore Sensmg o/ Enrrronmenr 8( I ) 7’-92 t 1979) Y Hdvaml and W Peterson Social Returns IO public lnlormatlon senices statistical reporting ol L1 S farm commodltles Am Eton Rev 62 119-130 (I97?) J B Bullock Social costs caused by errors In agnculturd1 productlon lorecasts -Im J 4qruulrural Econ 9 76-80(1976) E Schandra and R Hoier Microwd\e multispectral In\sstlgatlon of snou Proc I Irh Inr Simp on Remore Senwrg ol Emwonmenr pp 601-607 Ann 4rbor hll (19771 J R Grease% S Sheer and R Glaser Cloud cober btdtwcs and their use in the plannmg ot remote sensing miwons Rtvnore Sensmg o/ Enwonmenr I 95 I-1001 11969) C Marlm and B Llleb ERTS cloud co\er srud\ Rep SD 7431 I prepared for Ihe National 4eronautics and Space 4dmmlstratlon b) North 4mervzan Rockuell Contract No N4S SI 1234 Los 4npeles CA (197-I)