PLOTTINGFROMABOVE

ENHANCINGAGRICULTURALMAPPINGINASIAANDTHEPACIFIC

AnthonyBurgard;AnnaChristineDurante;PamelaLapitan;

MahinthanJosephMariasingham;ArturoY.Pacificador,Jr.;andMashalRiaz

JUNE2024

ASIANDEVELOPMENTBANK

PLOTTINGFROMABOVE

ENHANCINGAGRICULTURALMAPPINGINASIAANDTHEPACIFIC

AnthonyBurgard;AnnaChristineDurante;PamelaLapitan;

MahinthanJosephMariasingham;ArturoY.Pacificador,Jr.;andMashalRiaz

JUNE2024

ASIANDEVELOPMENTBANK

CreativeCommonsAttribution3.0IGOlicense(CCBY3.0IGO)

?2024AsianDevelopmentBank

6ADBAvenue,MandaluyongCity,1550MetroManila,Philippines

Tel+63286324444;Fax+63286362444

Somerightsreserved.Publishedin2024.

ISBN978-92-9270-774-3(print);978-92-9270-775-0(PDF);978-92-9270-776-7(ebook)PublicationStockNo.TCS240326-2

DOI:

/10.22617/TCS240326-2

TheviewsexpressedinthispublicationarethoseoftheauthorsanddonotnecessarilyreflecttheviewsandpoliciesoftheAsianDevelopmentBank(ADB)oritsBoardofGovernorsorthegovernmentstheyrepresent.

ADBdoesnotguaranteetheaccuracyofthedataincludedinthispublicationandacceptsnoresponsibilityforany

consequenceoftheiruse.ThementionofspecificcompaniesorproductsofmanufacturersdoesnotimplythattheyareendorsedorrecommendedbyADBinpreferencetoothersofasimilarnaturethatarenotmentioned.

Bymakinganydesignationoforreferencetoaparticularterritoryorgeographicareainthisdocument,ADBdoesnotintendtomakeanyjudgmentsastothelegalorotherstatusofanyterritoryorarea.

ThispublicationisavailableundertheCreativeCommonsAttribution3.0IGOlicense(CCBY3.0IGO)

/licenses/by/3.0/igo/

.Byusingthecontentofthispublication,youagreetobeboundbythetermsofthislicense.Forattribution,translations,adaptations,andpermissions,pleasereadtheprovisionsandtermsofuseat

/terms-use#openaccess

.

ThisCClicensedoesnotapplytonon-ADBcopyrightmaterialsinthispublication.Ifthematerialisattributed

toanothersource,pleasecontactthecopyrightownerorpublisherofthatsourceforpermissiontoreproduceit.ADBcannotbeheldliableforanyclaimsthatariseasaresultofyouruseofthematerial.

Pleasecontactpubsmarketing@ifyouhavequestionsorcommentswithrespecttocontent,orifyouwishtoobtaincopyrightpermissionforyourintendedusethatdoesnotfallwithintheseterms,orforpermissiontousetheADBlogo.

CorrigendatoADBpublicationsmaybefoundat

/publications/corrigenda

.

Notes:

Inthispublication,“$”referstoUnitedStatesdollars.ADBrecognizes“RepublicofArmenia”asArmenia.

Photos:AllphotosbytheAsianDevelopmentBankunlessotherwiseindicated.

Featuredonthecover,startingfromthetopleftandmovingclockwise,areimagescapturingfieldworkconductedinArmenia,theCookIslands(includingthethirdphoto),andtheLaoPeople'sDemocraticRepublic.ThefifthphotoisfromaremotesensingtrainingsessioninVietNam.

CONTENTS

TablesandFiguresv

TablesandFiguresiv

Forewordv

Introduction1

Methodology5

AnalysisofReportedandMeasuredArea16

ImplicationsforPolicyandRecommendations25

Conclusion26

Appendixes

12022CookIslandsPostEnumerationSurveyDesign27

2

DetailsofAreaSamplingFramefor2022PostEnumerationSurveyoftheCookIslands

34

3

2022CookIslandsPostEnumerationSurveyQuestionnaire

39

4

ProtocolsforGPSParcelAreaMeasurement

45

5

ProceduresforEditingGPSDataCollectedforAreaMeasurement

49

References

51

iii

TABLESANDFIGURES

Tables

Tables

1NumberofAgriculturalHouseholdsbyLevelofAgriculturalActivityfrom2011and2021Censuses9

ofPopulationandDwellingCookIslands

2ComparisonBetweenthe2011CensusofAgricultureand2022ADBPostEnumerationSurvey10

NumberofAgriculturalHouseholdsbyLevelofAgriculturalActivityBasedonCensusof

AgricultureDefinitions

3EstimatedTotalAreainSquareMetersandAcresofAgriculturalHoldings16

4ComparisonofEstimatedTotalAreaofHoldingsbyReportedandGPS-AssistedMeasurements17

A1.1DescriptiveStatisticsofSelectedEnumerationAreaCharacteristics(Rarotonga,CookIslands2021,32

78EnumerationAreas)

A1.2CorrelationMatrixBetweenDesign-TestVariables(A1,A2)andStratificationVariables(A3,A4)32

A1.3MeanandCoefficientofVariationValuesofDesign-TestVariablesbyStrata.33

A1.4CalculatedSampleSize(NumberofEnumerationAreas)atDifferentLevelsofTargetedPrecision34

Figures

1LandUseandLandCoverClassificationsofRarotonga7

2AgriculturalIntensityonRarotongaIsland,CookIslands7

3DefinitionsforLevelofAgriculturalActivityin2011CensusofAgricultureandthe2022Post9

EnumerationSurvey,CookIslands

4EffectsofGPSErroronAreaMeasurement11

5CookIslands—ParcelAreasCapturedbyHandheldGPS,GarmineTrex32x12

7CookIslands—ParcelAreasCapturedbyDigitizationonSatelliteImageMethod13

6DigitizinganAgriculturalParcelBoundaryinSurveySolutions13

8DigitizinganAgriculturalParcelBoundarybyWalkingAroundtheParcelinSurveySolutions14

9GPSMeasurementErrorsinIrregularShapes14

10DigitizationofAgriculturalParcelonSatelliteImagebySelectionofParcelCorners15

11CookIslands—Farmer-ReportedandDigitizedAreasVersusGPSAreaMeasurement18

12QuantileRegressionCoefficientsComparingFarmer-ReportedandGPSAreaMeasurement19

13DifferencesinthePerceptionofScaleasCapturedbyWalking(Green)andDigitization(Red)Methods20

14ArmeniaFarmer-ReportedandDigitizedAreaVersusGPSAreaMeasurement21

15GPSParcelAreasCapturedinPakPokVillage,VangVieng,VientianeProvince,LaoPeople’s22

DemocraticRepublic

iv

TablesandFigures

16LaoPeople’sDemocraticRepublicFarmer-ReportedandDigitizedAreaVersusGPSAreaMeasurement23

A1SampleEnumerationAreasSelectedforthePostEnumerationSurvey35

A4.1SampleSketchofParcel48

A4.2CheckingSatelliteAccuracyontheGarmineTrex32x49

A3.3MarkingaWaypointontheGarmineTrex32x49

A4.4MarkingaWaypointontheGarmineTrex32x50

A4.5SavingtheParcelArea50

A4.6WorkflowofGPSAreaMeasurementProtocol51

v

FOREWORD

ThisreportPlottingfromAbove:EnhancingAgriculturalMappinginAsiaandthePacificprovidesacomprehensiveoverviewoftheapplicationofamethodologyforagriculturalareameasurementandinsightsgainedfromits

implementationinthesethreecountries.

TheAsianDevelopmentBank(ADB)haslaunchedatechnicalassistanceprojecttostrengthenthecapabilities

ofnationalstatisticsofficesandotherministries,equippingthemwiththenecessaryskillstomeettheSustainableDevelopmentGoals’increasingdatademands.Apioneeringaspectofthisprojectistheuseofgeospatial

technologies,whichhavebeenemployedtocreatemethodologicaltoolsforagriculturalareameasurement.

Thesetoolsaredesignedtoevaluatethediscrepanciesinagriculturalareaestimatesbetweentraditionalfarmer-reportedmethodsandmoremodernapproachesusingGlobalPositioningSystem(GPS)devicesforobjective

measurements.

Agriculturallandisacrucialassetforfarmers,servingasthefoundationoftheireconomiclivelihood.Itfacilitatesvariousactivitiessuchascropcultivation,animalhusbandry,fisheries,andforestry.Historically,obtainingaccurateandunbiasedmeasurementsofagriculturallandshasbeenachallengingaspectofagriculturalstatistics.However,byidentifyingandaddressingbiasesinthesemeasurements,policymakerscangainamorepreciseunderstanding

ofagriculturalproductivity.Theadventofgeospatialtechnologieshasmadethistaskmoreaccessibleand

economical,revolutionizingthewayagriculturalareasaremeasured.Thistechnologicalshifthasmadeobjectiveareameasurementnotonlymorefeasiblebutalsomorecost-effective.

Thisreportfeaturesanin-depthanalysisofanareaframeapproachimplementedintheCookIslands.Theapproachusesnon-overlappingandrelativelyfixedgeographicalunitsfromwhichasamplemaybedrawn,insteadofthe

moretime-consumingtraditionallistframecomprisingagriculturalholdingscompiledduringanagriculturalcensus.

Geospatialdatacanbeintegratedwithasampleofpolygonsdrawnfromanareaframe,whichcanbefurtherstratifiedbyvariouscharacteristicssuchastopography.

Additionally,thereportdelvesintotheuseofgeospatialtechnologiestoassessbiasesinagriculturallandreporting

inArmenia,theCookIslands,andtheLaoPeople'sDemocraticRepublic(PDR).Thestudyemployedarangeof

geospatialtechniquesforanunbiasedmeasurementofagriculturalland.Oneofthekeystrengthsofthisstudyisitsexplorationofthefeasibilityandapplicabilityoftheapproachacrossthreecountriesindifferentregionswithdistinctcharacteristics,agroclimaticconditions,andsocio-politicalcontexts.

TheADBprojectteamwouldliketotakethisopportunitytothanktheimplementingagenciesfromtheStatisticalCommitteeoftheRepublicofArmenia,theCookIslandsMinistryofAgriculture,andtheLaoPDRMinistryof

AgricultureandForestryfortheirinvaluablecontributionsandofferingcriticalinsightsintothediverseagriculturalpracticesacrossAsiaandthePacific.TheteamisgratefultoMr.GagikAnanyan,DeputyofthePresidentofthe

StatisticalCommitteeoftheRepublicofArmenia;Mrs.TemaramaAnguna-Kamana,SecretaryofAgriculture,CookIslandsMinistryofAgriculture;andMs.KhamvayNanthavong,DirectoroftheCenterforAgriculturalStatisticsof

vi

Foreword

theLaoPDRMinistryofAgricultureandForestryforspearheadingtheimplementationoftheprojectinthepilot

areas.SinceregratitudeisofferedtoArsenAvagyan,WilliamWigmore,TearoaIorangi,PunaKamoe,Angeylie

Ngaoire,andSengphachanKhounthikoummane,forprovidingtechnicalandlogisticalsupportintheconductofallprojectactivities.Further,theteamwouldliketoexpressappreciationtofieldandtechnicalstafffortheirdedicationtocollecting,processing,andanalyzingthedatafromwhichthisstudywouldnotbepossible.

ThereporthasbeenproducedbytheStatisticsandDataInnovationUnitwithintheEconomicResearchand

DevelopmentImpactDepartmentatADB,undertheoveralldirectionofElaineS.Tan.TheprojectandreportteamswereledbyMahinthanJosephMariasingham,withvaluableresearchandtechnicalsupportfromAnthonyBurgard,AnnaChristineDurante,PamelaLapitan,ArturoY.PacificadorJr.,andMashalRiaz.MelanieKellehercopyeditedthefinalmanuscript,EdithCreustypesetthereport,andClaudetteRodrigopreparedthecoverdesign.

Itishopedthatthisstudyaidsintheevolutionofmethodologiesformeasuringagriculturalareas.Thismethodologicaladvancementholdsthepotentialtoenhanceaccessibilityandaccuracyinagriculturaldatacollection.Itisanticipatedthatthisreportwillhaveapositiveinfluenceonthefutureofagriculturalstatistics.

AlbertFrancisPark

ChiefEconomistandDirectorGeneral

EconomicResearchandDevelopmentImpactDepartmentAsianDevelopmentBank

vii

INTRODUCTION

ATechnologicalShiftinAgriculturalStatistics

Thegrowingaccessibilityofgeospatialtechnologiesisreshapinghowagriculturalstatisticsaregathered,processed,anddisseminated.Advancedtechnologieslikeremotesensingusingsatelliteimagery,GlobalPositioningSystem

(GPS),andunmannedaerialvehicles(UAVs)offerthepotentialformoreefficientmethodstomonitorchangesinagriculturewithgreaterprecisionandfrequency.

Oncecost-prohibitiveforlarge-scalestatisticaldatacollection,geospatialtechnologiesarebecomingincreasinglycommonplace,efficient,andaccessibleinofficialstatistics.Consumer-gradeequipmentisnowmorecapable

andlessexpensive.Ascomputer-assistedinterviewingdatacollectionmethodsonceusheredinthedigitizationofstatisticaldatacollections,thesamemoderntechnologiesintabletcomputersprovideameanstosimplify

geospatialdatacollection.Itisnowmorecommonplaceforagriculturaldataproducerstoregularlycollect,forexample,thepointlocationofagriculturalhouseholdsandboundaryareasofagricultureandcroplands.

Similarly,thecostofremotesensingsatelliteimageryhasdecreased,largelyduetoinitiativesbyorganizationsliketheEuropeanSpaceAgency(ESA),theJapanAerospaceExplorationAgency,andtheNationalAeronauticsandSpaceAdministration,whichhavemadehigh-resolutionsatelliteimagerymorereadilyavailableasafreeglobal

publicgood.

GreateraccessibilityofUAVsandultra-high-resolutionimageryhaveempowered,forexample,thePacificislandcountries—whichhavetraditionallybeensusceptibletoclimatechange—toregularlymapandmonitorland

changeswithgreaterdetailandtimeliness.Theseadvancementspavethewayforthebroaderadoptionandutilizationofgeospatialtechnologies,significantlyenhancingagriculturalstatistics.

OneoftheobjectivesoftheAsianDevelopmentBank(ADB)technicalassistanceprojectistointegrategeospatialtechnologiesintotraditionalsurveydesignandbuildthecapacityofdataproducersintheregiononitspotential

usecaseapplications.Despitetheincreasingrelevanceofsuchtechnology,nationalstatisticalofficesandofficialproducersofagriculturalstatisticsoftenfacechallengesinitsutilization.Thisincludeshiringstaffskilledin

workingwiththesedatatypesandupgradinginformationtechnologyinfrastructuretomanagethelargerdatasets.Whilemanyorganizationshaveestablishedgeographicinformationsystem(GIS)unitswithintheiroffices,their

applicationisprimarilylimitedtocreatinganddistributingmaps.Therehasbeenlimitedprogressbycomparisoninintegratinggeospatialtechniquesintosurveydesignanddatacollectionprocesses.

Oneinnovation,however,usesgeospatialtechniquesforsurveydesignandimprovesuponthetraditional

samplinglistframe.Constructionofthetraditionallistframe—acomprehensivelistofagriculturalholdingsina

country—isacostlyundertakingoftencompletedonceevery5to10yearsduringagriculturecensusesand—ifwellmaintained—updatedthroughasystemofintercensalagriculturalsurveys.Thelistframeisapainpointformany

1

PlottingfromAbove

officialdataproducersasitisnotoriouslydifficultandexpensivetobuild,update,andmaintain.Whilecountries

intheregionhavetrendedtopursueafarmer-registry-basedapproachtotheconstructionoftheseframes,they

arecurrentlylimitedbylowfarmervolunteerratesandmaylackkeyauxiliarydatatoserveasaneffectivesamplinglistframe.Acomplementaryapproachhasbeenforofficialstatisticsproducerstoadoptamixed-frameapproachincorporatingtheuseofanareaframe.Anareaframeconsistsofnon-overlappingandrelativelyfixedgeographicalunitsfromwhichasamplemaybetaken.

Alimitationofareaframesisthattheyarenotalwaysoptimizedforstatisticalpurposesandareconstructedwith

broaderusecases,suchasthedemarcationofadministrativeboundaries.Thisoftenresultsininsufficientauxiliaryinformationtoenhancesurveysampleefficiency.Withthegrowingaccessibilityofgeospatialdata,opportunities

existtointegratediversegeospatialdatasources,includinglanduseorlandcovermaps,topographicalmaps,and

thelocationsofnaturalandhuman-madefeatures.Usinglocationasareferencepoint,thesedatacanbeintegratedtoenhancesamplingmethods.Thisintegrationallowsformoreefficientor“smart”samplingapproaches—suchas

stratifyingbydifferingagriculturalcharacteristics—therebyimprovingtheaccuracyandefficiencyofdatacollectionprocesses.Thisstratificationcould,forexample,bebasedonthedensityoftheestimatedagriculturalareaor

distinctagroecologicalzones,thusimprovingtheprecisionofthesurveysample.

InseveralAsiancountries,nationalstatisticalofficesandagriculturallineministriesaretransitioningtoprecisionagricultureusingdigitalrecordsandgeospatialinformationtomapagriculturalareas.Thisincludesdigitizing

agriculturalparcelstoenhanceproductionstatisticsestimationasseeninthelastagriculturecensusofthe

People’sRepublicofChina,theSmartFarmInitiativeintheRepublicofKorea,andtheagriculturaladministrativerecordsysteminSriLanka.Additionally,theseplatformsofferthepotentialforasystemofgroundtruthvalidationpointsrequiredforothertechnicalestimationsuchasremotesensing-basedcropestimation.

ThispaperwillexploreacasestudyforimplementinganareaframeapproachintheCookIslandsusingalandcovermapdevelopedwiththeassistanceofESAasanauxiliarydatasource.Thelandcovermap—createdfromhigh-

resolutionsatelliteimageryofRarotongaIslandin2021—enabledtheclassificationofareaswithahighprobabilityofagriculturalproduction.Thisclassificationwasinstrumentalindeterminingthesampleallocationforapost

enumerationsurveyconductedafterthe2021AgricultureCensus.

Thepaperwillfurtherinvestigateusinggeospatialtechnologiestoevaluatebiasesinreportingagriculturalland

inArmenia,theCookIslands,andtheLaoPeople’sDemocraticRepublic(LaoPDR).Thestudyapplieddiverse

geospatialmethodsforanobjectivemeasurementofagriculturalland,includingspecializedhandheldGPSdevicesandtablet-basedsoftwarefordigitizingtheboundariesofagriculturalparcelsonhigh-resolutionsatelliteimagery.

AgriculturalLandisaKeyFactorinEconomicProduction

Agriculturallandisakeyproductiveassetforfarmers,formingthebaseoftheireconomiclivelihood.Itisakeyfactorofproduction,enablingthegrowingofcrops,raisinganimals,fisheries,andforestryactivities.Thesizeofagricultural

landisacriticalstatisticfromapolicyperspectiveasithelpspolicymakersbetterunderstandhowfarmingis

structuredinacountry.Accuratemeasurementsofagriculturalareaareimportantforevaluatinghowproductivefarmsare,planningforagriculturalgrowth,andmakingeffectiveagriculturalpolicies.

Fromamacroeconomicperspective,agriculturallandareafeedsintothecriticalcalculationofpotentialeconomicoutput.Inmanycases,cropproductionstatisticsarederivedbasedonreportedagriculturalareamultipliedbyanaverageyieldestimateforthelocality.Anybiasesinlandareaestimatesreportedbythefarmerwillsignificantly

compoundtheestimatefortotalagriculturaloutput.

2

Introduction

Biasesinestimatingagriculturalareasmayalsonegativelyimpactresourceallocationstogovernmentsupport

programs.Foodsecurityisaprimaryconcernformanygovernments,especiallythoseinclimate-vulnerableareassuchasthePacificislandcountries.Anoverestimationorunderestimationofagriculturallandarea—andtherebyproduction—canaffecttheabilityofacountrytomeetitscaloricandnutrientneedssustainably.Andwherefoodsecurityisachallenge,improvedstatisticsontheagriculturalareahelpgovernmentsimproveplanstoimportkeyagriculturalcommoditiestomeettheserequirements.Thesedatamaythenbeusedtosupportcropandinput

subsidies,cropinsuranceschemes,andadditionaltechnicalsupportthroughagriculturalextensionservices.

Finally,agriculturesignificantlyimpactstheenvironment,contributingtogreenhousegasemissionsand

encroachmentonnaturalforests,leadingtobiodiversityloss.Improvedestimatesofagriculturallanduseareessentialforurbanandrurallanduseplanningandmanagingtheseimpactsinthelongterm.

RecallBiasesinEstimatingAgriculturalLand

Thecommonmethodforestimatingagriculturallandincensusandsurveysinvolvessubjectiverecallbyfarmers,

askingthemtoreporttheextentoftheiroperatedland.Thisapproachpresentsseveralchallenges.Itassumes

farmers’understandingofthetermoperationalagriculturalland,aconceptthatcanintroducenon-samplingbiasesifunclear.Theseissuesarecompoundedbydifferencesinagriculturalpracticesthroughouttheregion,wherein

somecountries,itmaybecommontosharecommunalareasforagriculturalproductionandliveandworkinareaswherelandtenureisnotclearlydefined.

Thismethodassumesfarmershaveaccurateknowledgeofthesizeoftheirland.Thisknowledgeistypicallybasedoninformationfromformallandtitlesordeeds,providingofficialdocumentationofprecisemeasurementsofthe

land.Withoutthesedocuments,farmers’estimatesmaybelargelyspeculative.InArmenia,forexample,astrong

landcadastralsystemexistsinwhichlandareaislinkedwithpropertytaxandgovernmentplanningsystems.Insuchcases,farmerknowledgeoftheirareaisbasedonhowupdatedthesesystemsareandtheextentofpublicaccesstothesedocuments.

Thereisalsotheissueofreportedareameasurementunits.Familiaritywithstandardunitsofareameasurementisnotalwayscommonamongfarmers.Forinstance,intheCookIslands,afarmermightdescribetheirlandintermsofarugbyfieldsizeorreferencenaturallandmarkslikealargetree.IntheLaoPDR,localunitssuchas“l(fā)ai”and“ngam”arecommon,complicatingaccuratereportinginstandardizedunitslikeacresorhectares.Farmersmayusedifferentareaunitsfordifferentlandfeaturesincertaininstances.Forexample,totalareamightbereportedinacres,while

individualplotsaredescribedinsquaremeters(m2).Differencesinareaunitsreportedmayburdenthefieldstaffwhenperformingqualityassurancechecks.

Thelackoffamiliaritywithstandardizedunitsofmeasurementaddstotheresponseburdenforfarmers,leading

themtoeitherinaccuratelyconverttheunitsorgivespeculativeanswers.Toaddressthis,agriculturalsurveys

increasinglypermitrespondentstospecifytheunitsusedforeacharea-relatedquestion.However,inconsistencies

inreportedareasremainacommonissue.Theserequirecarefulvalidationandadjustmentduringthedataprocessingtoensureaccuracyandconsistency.

Finally,thephysicalcharacteristicsofthelandcanhinderaccurateareaestimation.InmanyAsianandPacific

countries—especiallysmallholdermixedcroppingsystems—agriculturalparcelsoftenhaveirregularshapes.

Theymaybesituatedinsteepmountainousterrain,affectingfarmers’perceptionoftheirsize.Thiscomplexitycanleadtoinaccuraciesinreportedlandarea,significantlyimpactingthequalityofagriculturalstatistics.

3

PlottingfromAbove

IntroducingObjectiveMeasurementstoAssessAgriculturalLand

Acknowledgingthesubjectivebiasesassociatedwithfarmerrecalldata,oneapproachtomitigatethisisbyintroducing

objectiveareameasurement.Traditionally,objectiveareameasurementinvolveslandsurveyingtodelineateagriculturalparcelsusingthetape-and-compassapproach(FAO,1982).Thismethodentailsmeasuringeachsideoftheparcel

withatapemeasureandusingacompasstodeterminetheanglesbetweensides.However,thistechniqueistime-consumingandresource-intensive.Italsodemandshighlyskilledworkerstoperformprecisemeasurementsandcalculatetheareausingcomplextrigonometricfunctions.Whendonecorrectly,however,thetapeandcompass

methodisconsideredthe“goldstandard”foragriculturalareameasurement(Carlettoetal.,2016).

In2018,ADBconductedapilotstudytoexploreusingGPSandsatellitedataastechnologicalalternativesfor

objectivelymeasuringagriculturalparcels.Thestudyfindingsindicatedthatthesemethodsalignedwellwiththe

“goldstandard”ofmeasurementand,onaverage,weremor