1

HighLatitudeDroneEcologyNetwork–MultispectralFlightProtocolandGuidanceDocumentByJakobAssmann(j.assmann@ed.ac.uk),JeffKerby(jtkerb@gmail.com)andIslaMyers-SmithTheUniversityofEdinburgh,Scotland(UK)andDartmouthCollege,NH(USA)The followingprotocol is theoutcomeof twoyearsofmultispectraldronework in theCanadianArctic. Two three-months field seasons provided opportunities to learn through trial and error.Shouldyouhaveanyfeedback,suggestions,questionsorcommentspleaseletusknow.Theinformationbelowisdistilledfromablogpost/shortcommunicationmanuscriptcurrentlyinpreparation. Oncepublished,thesewillbesharedontheHiLDENmailing list. If thisprotocolhasinfluencedyourmethodofdatacollectionitwouldbegreatifyoucouldcitetheabovepublicationsinresultingmanuscripts.ContentTimeofDay 1

Radiometriccalibration 1Pre-andpost-flightcalibration 2In-flightreflectancetargets 3CareandQualityControlofReflectanceStandardsandTargets 4

GroundControlPoints 4

FlightPattern 5

WeatherConditionsandMeta-Data 5TimeofDayAllflightsshouldbeconductedbetween3hoursbeforeandaftersolarnoon,andatmostwithin6hoursbeforeandaftersolarnoon.Youmightwanttohardenthisruledependingonthelatitudeofyourfieldside/timeofyear,alsoconsidermountainsandotherobstaclesthatmightinfluencetheilluminationofyourfieldsite.TheNOAAsolarpositioncalendarandothersimilarresourcesareusefultoolstocalculatesolarpositionatyourfieldsite.Shadows are themain reason for this, the lower the solar angle themore prominent shadowsbecome in your images. More shadows mean a lower proportion of the surface is illuminatedresulting in inaccurate estimates of surface reflectance and its derived data products, likeNDVI.Thisbecomeslessofaproblemonovercastdayswithdiffuseillumination.Thespectralcompositionof the sunlightalsochangeswith solarelevation.This canbecompensated to somedegreewithradiometriccalibration,butextremesshouldbeavoided.

Radiometriccalibration

2

Radiometriccalibrationisthemost importantstepofthisprotocol.Clouds,haze,timeofdayandsolarpositioninfluencethespectralcompositionofsunlight, it isthereforekeythatastandardofknown reflectance is used to calibrate the relative measurements of the multispectral sensor.Otherwise,comparisonsbetweenflights,sitesandsatelliteimageryarenotpossible.Pre-andpost-flightcalibrationRadiometriccalibrationisbestcarriedoutbytakingcalibrationimageryofaknownstandardpaneldirectlybeforeoraftertheflight.Theresultingimageryisthenlaterusedinthephotogrammetrysoftware(e.g.Pix4D)tocalibratethereflectancemaps.Onlyoneofthetwosetsofcalibrationimagery(pre/post)willbeusedforradiometriccalibration,butyoumightfindthatlightconditionschangeduringlongflights(e.g.cloudsmovingin)andthatconditionsattheendoftheflightaremorerepresentativeofthemajorityoftheflight.WerecommendusingatraceableSpectralontarget(orstandardofsimilarquality)ofapprox.15x15cm(6”x6”).Forvegetationsurveysastandardwithareflectancevalueof70%wouldbeideal,but40-70%shouldbesufficient.Ifyouareplanningtosurveysnowcoveredsurfacesapurewhitestandard>90%reflectancemightberequired.If youareusing aMicaSense Sequoia sensor, pleasebeaware thatMicaSense’sown reflectancepanels are not indented for scientific use! MicaSense also recommends the use of Spectralontargetsforscientificwork.Thoughexpensive,Spectralonisamaterialwithveryevenanddurablereflectancepropertiesacrossthespectrum.ItcanbeboughtthroughthemanufacturerLabsphereintheUSorretailersworld-wide.Welostallourpre-flightcalibrationdatafromthefieldseasons2016duetoapoor-qualitystandard,soit’sworthwhilespendingthemoney.Pre-/post-flightcalibrationimagesshouldbetakensothatalargenumberofpixelsarecoveredbythetargetandsothatthetargetIDisclearlyvisibleintheimage.Ensurethatthetargetislevelonthegroundandfreeofshadows. Ifyourcalibrationphotocapturesmanypixelsofthecalibrationpanel, a larger statistical sample can be used by the photogrammetry software to calibrate theorthomosaicreflectancemaps.Inthisinstance,alargersampleisabettersample.AsaruleofthumbwiththeMicaSenseSequoiasensor:Holdthedronejustbelowhipheightoverthetargetwithyourbacktothesun,takeastepasidewhileholdingthedroneinpositontoavoidcastingashadowoverthetarget.Ensurethatthetargetinthecalibrationimageryisnotoverexposed(i.e.saturatedintheupperendofthedigitalnumbers),asradiometriccalibrationwillotherwisenotbepossible.IfyouareusingaSequoiaSensor,usethe‘RadiometricCalibration’featureoftheWiFiinterfacetoobtainabracketof calibration imagery with different exposure levels. Placing the calibration target on a highreflectancesurface(suchasalandingpad)mayalsohelpreducethechanceofincorrectexposures.Finally,ifyou’reusingafilterorlensprotectorduringtheflight,ensurethatcalibrationimageryistakenwith the filter/protector inplace.Likewise, removeany filters/protectorsbefore taking thecalibrationimageryiftheyarenotusedinflight.

3

In-flightreflectancetargetsIn-flight reflectance targetsserve twopurposes:a)as surfaceswithknown/calibratedreflectancetheyallowforqualitycontrolintheoutputsofphotogrammetrysoftware(suchasthereflectanceandNDVImapsproducedbyPix4D),andb) theyareaback-up if for somereason thepre-/post-flightcalibration imagery fails (e.g.deteriorationof standardover time,overexposureof imageryandunintendedshadowscoveringthetarget).In-flight targets shouldhaveeven reflectance valueswithin thebandsof your sensor andmatchroughly thereflectanceof thesurfaceof interest.Forvegetationsurveyswerecommend40-60%reflectance.Additionaltargetsofhigherandlowerreflectancemaybeemployedtocoveralargerrangeofreflectancevalues,butthismightnotbepracticalinallsituations.In-flighttargetsshouldbeatminimum5x5theGroundSamplingDistance(GSD)(25Pixels),but10x10GSD(100Pixels)isdesirable,ifnotmore.Again,thelargerthestatisticalsampleofpixelsthebetter thequalitycontrol.For reasonsofpracticality (transportandweight) large targetsizes forhighaltitudeflightsmightnotbeachievable.The material of the in-flight target should be as durable and Lambertian (even reflectanceindependentonangleofview)propertiesaspossible.Spectralonwouldbeideal,butislikelynotacost-effectiveasan in-flight target. Inourexperience, it ismoredifficult tokeep in-flight targetsawayfromenvironmentalinfluencessuchaswatersplashes,dirtanddust.InsteadofSpectralonwetherefore suggest selecting from the following options. They are ordered by most desireablereflectanceproperties,butwerealizeyouwillneedtomakeyourdecisionbasedonintendeduses,budget,andtimeforpreparationandavailability.

• Calibrated felt targets produced by the finish companyMosaicMill , 44% reflectance forvegetationsurveys;standardsize50cmx50cm,approx.350€perpiece,customsizesalsoavailable.

• Grey-cardsmadeoutof‘Kodakcard’cardboardtypematerial,obtainablefromphotographysuppliers.OftenonlyavailableinsmallersizesaroundpaperA4orletterformat,unsuitablefor largeGSD.Tendtocome inapprox.18%reflectance (grey)and80%(white).Warning:Spectralpropertiesmayvarystronglyacrossthespectrum-testinlabbeforeuse!

• Sail-clothandcanvassheetsmaybealternatives,particularlyfor largeGSDofknownevenspectral reflectance. Ensure the material is 100% opaque and test reflectance andlambertianpropertiesinlab!

Foruncalibratedtargets,wehighlyrecommendmeasuringreflectancepropertiesbeforegoingoutintothefieldthefirsttime.However,intheendabadin-flighttargetisbetterthannotarget.Alsosee‘CareandQualityControl’Sectionbelow.Finally,wesuggestthatatleasttwoin-flighttargetsshouldbeemployed;oneateitherextremeofthe flight paths. For larger sites (> 1 ha) more targetmight be desirable, but only if practicallyfeasible.Positionthetargetssothattheaircraftpassesdirectlyoverit.

4

CareandQualityControlofReflectanceStandardsandTargetsHandle reflectance standards and in-flight targetswithasmuch careaspossible!Avoid touchingthesurface,exposuretorain,dirtandwatersplashesparticularlyoftheSpectralonpre-/postflightcalibration targets). Prolongedexposure to sunmayalsohave anegative impacton the surface.Returnthetargetstoaprotectivecontainerassoonaspossibleafteruse.All targets should be calibrated before and after each field season, then cleaned according tomanufacturer’s recommendations and re-measured ahead of the next field season.We suggestdetermining theabsolutereflectanceof the targetsusingaspectroradiometer inadark room. Inthepast,wecarriedoutfivecontactmeasurementspertarget,oneinthecentreandfourtowardseachcorner(butnotintheverycorner),averagereflectancevaluescanthenbecalculatedforthetargetandbandaveragesdetermined.GroundControlPointsInadditiontofollowingthegeneralgroundcontrolpoint(GCP)recommendationsoftheHiLDENafew particular considerations apply to GCPs used with multispectral sensors. GSDs for themultispectralsensorsare largecomparedtohighresolutionRGBsensorsandthemonochromaticnatureofsinglebandimagerycanthereforemakeidentifyingGCPsdifficult.WerecommendtheuseoflargeGCPs(8x8to10x10GSD,absoluteminimum5x5GSD)withacheckboardpattern, inourexperiencetheseworkbest (Figure1).Thecheckboardpatternworkswell, as the squares show up as clearly defined surfaces in themonochromatic imagery. BeforeheadingoutintothefieldwithnewGCPs,ensurethattheyshowupwellinallbands.Notallwhitesare white and not all blacks are black in the near-infrared (NIR). Furthermore, make sure thatsurfacereflectancepropertiesdonotvarystronglywithorientationtothesunandsolarangle.AgoodwayoftestingthesuitabilityofnewGCPs istaking imageswiththemultispectralsensoronthe ground and then conducting tests flights. We have made life difficult for ourselves by nottestingGCPs in thepast,assumingthatsurfacesthatshowupwell in thevisiblespectrumwould

Figure1SequoiaNIRimageof60cmx60cmGCPongrass.TheGCPismadefromself-adhesivetilesobtainedinalocalhardwarestore.Groundsamplingdistance:approx.7.4cm/pixel.ImagecurtseytoTomWadeandCharlieMoriarty,TheUniversityofEdinburgh.

5

showupwellintheNIRaswell.FlightPatternFlight plan and pattern will varydepending on the site. However, if alawn mower flight pattern of parallelflight lines is used we recommendintroducing some flight lines diagonallyoracrossthepattern.SeeFigure2forasuggested flight plan. This will likelyimprove the outputs of thephotogrammetry process, particularly ifvegetationsurfacesaremappedthatarenotveryvaried in theirnature. Wewillsend out amore detailed post on flightplanningearlynextweek.WeatherConditionsandMetaDataInadditiontousualflightdata,itisworthwhiletologweatherconditions,aswellasthestandardsandtargetsused.Forweather,windandskyconditionsareparticularlyusefulformultispectraldatacollection,astheyinfluencesurfacebehaviourandsolarirradiation.Forskyconditionswesuggestascalefrom0-9(Table1).In addition temperature and information about vegetation surface wetness/dew should berecorded.PleasealsoseeHILDENmultispectralflightlogspread-sheetattached(maybefilledinonpaperordigitalonatablet).Table1SkyCodes-adaptedfromNERCFieldspectroscopyFacility’sLogSheets

0 Clearsky1 Haze2 Thincirrus–sunnotobscured3 Thincirrus–sunobscured4 Scatteredcumulus–sunnotobscured5 Cumulusovermostofsky–sunnotobscured6 Cumulus–sunobscured7 Completecumuluscover8 Stratus–sunobscured9 Drizzle

Figure2Lawn-moverflightpatter(blue)withdiagonalflightlinesaftertake-offandbeforelanding.