6 1055 52 364

Updated Inflight Calibration of Hayabusa2’s Optical Navigation Camera (ONC) forScientificObservationsduringtheCruisePhaseEriTatsumi1

ToruKouyama2

HidehikoSuzuki3

ManabuYamada4

NaoyaSakatani5

ShingoKameda6

YasuhiroYokota5,7

RieHonda7

TomokatsuMorota8

KeiichiMoroi6

NaoyaTanabe1

HiroakiKamiyoshihara1

MarikaIshida6

KazuoYoshioka9

HiroyukiSato5

ChikatoshiHonda10

MasahikoHayakawa5

KoheiKitazato10

HirotakaSawada5

SeijiSugita1,11

1DepartmentofEarthandPlanetaryScience,TheUniversityofTokyo,Tokyo,Japan

2NationalInstituteofAdvancedIndustrialScienceandTechnology,Ibaraki,Japan

3MeijiUniversity,Kanagawa,Japan

4PlanetaryExplorationResearchCenter,ChibaInstituteofTechnology,Chiba,Japan

5InstituteofSpaceandAstronauticalScience,JapanAerospaceExplorationAgency,Kanagawa,Japan

6RikkyoUniversity,Tokyo,Japan

7KochiUniversity,Kochi,Japan

8NagoyaUniversity,Aichi,Japan

9DepartmentofComplexityScienceandEngineering,TheUniversityofTokyo,Chiba,Japan

10TheUniversityofAizu,Fukushima,Japan

11ResearchCenteroftheEarlyUniverse,TheUniversityofTokyo,Tokyo,Japan

6 1055 52 364

Abstract

TheOpticalNavigationCamera(ONC-T,ONC-W1,ONC-W2)onboardHayabusa2arealsobeingusedforscientific

observationsofthemissiontarget,C-complexasteroid162173Ryugu.Scienceobservationsandanalysesrequire

rigorous instrument calibration. In order to meet this requirement, we have conducted extensive inflight

observationsduringthe3.5yearsofcruiseafterthelaunchofHayabusa2on3December2014.Inadditiontothe

firstinflightcalibrationsbySuzukietal.(2018),weconductedanadditionalseriesofcalibrations,includingread-

outsmear,electronic-interferencenoise,bias,darkcurrent,hotpixels,sensitivity,linearity,flat-field,andstraylight

measurementsfortheONC.Moreover,thecalibrations,especiallyflat-fieldsandsensitivities,ofONC-W1and-W2

areupdatedfortheanalysisofthelow-altitude(i.e.,high-resolution)observations,suchasthegravitymeasurement,

touchdowns,andthedescentsforMASCOTandMINERVA-IIpayloadreleases.TheradiometriccalibrationforONC-

TisalsoupdatedinthisstudybasedonstarandMoonobservations.Ourupdatedinflightsensitivitymeasurements

suggesttheaccuracyoftheabsoluteradiometriccalibrationcontainslessthan1.8%errorfortheul-,b-,v-,Na-,w-,

andx-bandsbasedonstarcalibrationobservationsand~5%forthep-bandbasedonlunarcalibrationobservations.

TheradiancespectraoftheMoon,Jupiter,andSaturnfromtheONC-Tshowgoodagreementwiththespacecraft-

basedobservationsoftheMoonfromSP/SELENEandWAC/LROCandwithground-basedtelescopicobservations

forJupiterandSaturn.Ourcalibrationresultssuggestthatthe0.7-µmabsorptionbandtypicallyobservedonChand

Cgh asteroids at the ~3-4% level can be detected with the ONC’s signal-to-noise ratio (SNR) of ~2. We also

demonstrateadecreaseinSNRduetoCCDtemperatureincreasescausedbyradiantheatwhenthespacecraftis

closetothesurface,astheSNRismeasuredtobe150ataCCDtemperatureof20˚C(theworstcasescenario).Since

Ryugumaypossessasignificantamountofinternalvolatiles,asodiumatmospherearoundRyuguisconsideredto

behighlyplausible.WeevaluatedtheupperlimitofdetectabilityofasodiumatmospherearoundJupiterusingthe

Na-filteras100Rwith100images.ThisimpliesthattheONC-Tcandetectasodiumatmosphereofseveral10skR

basedonasingleimagesetofv-andNa-bandsandofseveral100sRbasedon100imagesets.Finally,wereportthe

first inflight observation of Ryugu by ONC-T from 1.3×106 km away on 26 February 2018. The ONC-T v-band

observationdisplaysconsistencywithground-basedobservation,whichconfirmsthecapabilityofONC-T.

6 1055 52 364

1.Introduction

TheHayabusa2spacecraftwaslaunchedbytheJapanAerospaceeXplorationAgency(JAXA)inDecember

2014andarrivedatitstarget,theC-complexasteroid162173Ryugu(formerly1999JU3)in27June2018.Its

mission is to rendezvouswithRyuguand to return samples from theasteroid’ssurface. Inadditionto the

samplingsystem,Hayabusa2alsocarriesremote-sensinginstruments,suchastheOpticalNavigationCamera

(ONC), the 3µm Near-Infrared Spectrometer (NIRS3), the Thermal Infrared Imager (TIR), and the Light

Detection andRanging laser (LIDAR). Before sampling, theasteroid surfacewill bemappedpreciselyand

accurately with the remote-sensing instrument payload, for example 0.5-2 m/pix for the whole asteroid

surfacebyONC-T,andcharacterizedfrombothamineralogicalandmorphologicalpointofview.Thispaper

describestheinflightcalibrationsofthecharge-coupledevice(CCD)forthetelescopiccamera(ONC-T)andthe

twowide-anglecameras(ONC-W1andONC-W2)onboardtheHayabusa2spacecraft.ONC’sdesignsareshown

inTable 1.1. Their exposure time settings of three cameras are different becauseONC-T engages optical

navigationsbyobserving starsandRyuguwith longexposure time,ONC-W1and-W2aremainlyused for

observingduringcloseencountertotheRyugusurfacewithshortexposuretime.Especially,ONC-W1which

needs to capture target markers with flash light, are set to have shortest exposure times. More detailed

specificationssuchasaperformanceofCCDandhardwareweightandsizeareshowninAppendixA.

Anumberofmissionsareplannedtovisitspectrallydifferenttypesofasteroids,suchasHayabusa2toa

C-typeasteroid,OSIRIS-RExtoaB-typeasteroid(Laurettaetal.,2011),PsychetoaM-typeasteroid(Elkins-

Tanton, 2018), andLucy toa set of Jupiter Trojanasteroids (Levisonet al., 2017). The comparison of the

absolute reflectance from the remote sensing observations across these missions is important for

understanding the difference between the various target asteroids, including differences in composition,

aqueous alteration, the degree of space weathering, and thermal metamorphism. Furthermore, the

compositional constraints and connections to meteorite analogs are important for understanding the

propertiesofprimitivematerialsintheearlySolarSystem.Absolutereflectancespectra,sometimesreferred

toasspectralalbedo,isoneofthemostpowerfulindicesforthemineralogicalclassification(e.g.,Johnsonand

Fanale,1973;Gaffy,1976)ofsolarsystemobjects.Also,reflectancevariationsacrossanobject’ssurfacecan

beusedtodeterminelevelsofaqueousalteration(Fornasieretal.,2014),thermalmetamorphism(Hiroietal.,

1996),grain-sizevariation(Cloutisetal.,2013),andspaceweathering(e.g.,Matsuokaetal.,2015;Lantzetal.,

2017).

Another important feature of theHayabusa2mission is theMobile Asteroid Surface Scout (MASCOT)

lander,whichhasan instrumentpayloadwithoverlappingwavelengthcoveragewiththemainspacecraft’s

payload, both of which will be observing the asteroid’s surface (e.g., Ho et al., 2017) at different spatial

resolutions. The counterpart of theONC is theMASCOTCamera (MasCam),which hasa 1024x1024 pixel

complementarymetaloxidesemiconductor(CMOS)imagesensorsensitiveinthe400-1000nmwavelength

6 1055 52 364

rangewithfourcolorLEDs.InordertocompareandconnectthespectrafromMasCamandONC-T,arobust

spectralradiometriccalibrationisnecessary.

Themainobjectiveofthisstudyistodescribetheinflightcalibrationofthreecameras,ONC-T,ONC-W1,

and ONC-W2, during the 3.5 year cruise phase from December 2014 to May 2017, and to describe the

calibrationpipelinewhichwill begeneratingproducts for JAXA’sDataArchives andTransmission System

(DARTS; https://www.darts.isas.jaxa.jp/planet/project/hayabusa2/) and NASA’s Planetary Data System

(PDS).TheresultsofpreflightcalibrationsofONC-Tandtheresultsof inflightgeometriccalibrationswere

described indetailbyKamedaet al. (2017)andSuzuki et al. (2018), respectively.However,differences in

temperature and environment between the laboratory and inflight calibration measurements requires a

validation and update to thepreflight radiometric calibration. Thus,we conduct re-calibration of theCCD

sensitivity based on inflight star observations obtained at the reference temperature (~-30℃ ). We also

measured the sensitivity temperature dependence based on inflight Jupiter observations under varying

hardware temperatures.Furthermore,we summarize the sensitivities ofONC-W1and -W2 based on both

preflightandinflightmeasurements.Thegoalofthispaperistoprovidereliablesensitivitymeasurementsfor

theONCsystemandevaluateitsuseinmineralogicalmapping.

Firstwewilldescribe the calibration flowofONC images inSec.2, andeachdetailedanalysis in the

calibrationprocessesisshowninSec.3forONC-TandSec.5forONC-W1and-W2.Detectabilityofsodium

atmospherewillbediscussedinSec.4.ONCsystemalignmentwiththespacecraftandNIRS3willbeanalyzed

inSec.6and7,respectively.Finally,applicationsinscientificanalyseswillbediscussedinSec.8,includingthe

reportofthefirstinflightobservationofRyuguinSec.8.4.

Table1.1.DesignsofONC.

ONC-T ONC-W1 ONC-W2

Effectivelensaperture 15.1mm 1.08mm 1.08mm

Focallength 120.50±0.01mmfor

wide-filter*(Suzukiet

al.,2018)

10.22mm 10.38mm

Fieldofview 6.27˚(Nadirview) 69.71˚(Nadirview) 68.89˚(Slanted~30˚

fromnadir)

Fnumber(F#) 9.05 9.6 9.6

Colorfilters 7colorbandpassfilters

(ul:0.40µm,b:0.48µm,

v:0.55µm,Na:0.59µm,

w:0.70µm,x:0.86µm,

Clearfilter Clearfilter

6 1055 52 364

p:0.95µm)and1clear

filter(wide).

Exposuretime 5.44ms,8.20ms,10.9

ms,16.4ms,21.8ms,

32.8ms,43.5ms,65.6

ms,87.0ms,131ms,

174ms,262ms,348ms,

525ms,696ms,1.05s,

1.39s,2.10s,2.79s,

4.20s,5.57s,8.40s,

11.1s,16.8s,22.3s,

33.6s,44.6s,67.2s,

89.1s,134s,178s,0s

(forsmear)

170µs,256µs,340µs,

513µm,680µs,1.03ms,

1.36ms,2.05ms,2.72

ms,4.1ms,5.44ms,8.20

ms,10.9ms,16.4ms,

21.8ms,32.8ms,43.5

ms,65.6ms,87.0ms,

131ms,174ms,262ms,

348ms,525ms,696ms,

1.05s,1.39s,2.10s,

2.79s,4.20s,5.57s,0s

(forsmear)

1.36ms,2.05ms,2.72

ms,4.1ms,5.44ms,8.20

ms,10.9ms,16.4ms,

21.8ms,32.8ms,43.5

ms,65.6ms,87.0ms,

131ms,174ms,262ms,

348ms,525ms,696ms,

1.05s,1.39s,2.10s,

2.79s,4.20s,5.57s,

8.40s,11.1s,16.8s,

22.3s,33.6s,44.6s,0s

(forsmear)

*Focallengthsforv-andNa-filtersofONC-TareinAppendix.A.

6 1055 52 364

2.ImageCalibrationFlow

Raw imagesacquiredby theONCsystemneedtobeprocessed ina sequenceof stepsto scientifically

calibratetheimagedata.Inthissection,wesummarizethecalibrationflowoftheONCimagesfromrawdata

(Level0)tohigher-level,radiometricallyandgeometricallycalibrateddata(Level2).Figure2.1displaysthe

datacalibrationflowchartfortheONCandshowstheproductlevelstobearchivedinbothDARTSandPDS

(L0:rawdata,L1:rawdatawithaheader,L2a:spacecraft informationaddeddata,L2b: flat-fieldcorrected

digitalunitdata,L2c:radiance,L2d:radiancefactor(I/F),L2e:reflectancefactorat(incidenceangle,emission

angle,phaseangle)=(30˚,0˚,30˚)).

Thedigitalnumberforeachpixelinarawimageisgivenasa8-,10-,or12-bitbinarynumber.Forhigher-

levelproducts,weusetheFITS(FlexibleImageTransportSystem)formattedfiles.AFITSformattedfileforthe

ONCsystemincludesaprimaryheaderdataunitandanimageextension.Headersincludehardwarestatus,

navigation,andotherancillaryinformation,suchasspacecraftpositionandattitudecalculatedfromtheSPICE

kernels. As described inFig. 2.1, the calibration flow requires inputparameters for each step, and those

parametersare included intheheaders.Thebiasandsmearsubtraction isconducted inflightduringhome

position (at alt. ~20 km) observations. However, for images acquired during the cruise phase and the

proximityoperationsduringtherendezvousphase,thebiasandsmearhastobecorrectedongroundthrough

thepipelineprocess.OnemorethingtonoteisthatsomestepsforONC-Thavenotyetbeenimplementedin

thecalibrationpipeline(grayboxesinFig.2.1).Darkcurrentandstraylightwillbenegligibleinmostasteroid

observations,exceptforsomeproximityobservations,suchasobservationsduringtheverycloseapproachto

thesurfaceorscanobservations.Whendelicateanalyses,suchasphotometricorspectralanalysesclosetothe

asteroid limbare conducted, point-spread-function (PSF) correctionsneed to be incorporated. However,

becausethesamePSFcorrectionmethodmaynotalwaysbeapplicable,dependingonwhereintheFOVthe

targetfallsorifitfillsoronlypartiallyfillstheFOV,wedonotapplythisstepforfirstpublicproductrelease.

Nevertheless,the innerpartof thetarget’sdiskwillbe littleaffectedbythePSF; thevaluewithoutthePSF

correctionissufficientlyaccurate.TheeffectofthePSFwillbediscussedmoreinSec.3.7.1.Thisprocesscould

beimplementedinafutureupdatetothepipelineprocess.

Each calibration step is described in detail in the following sections. Calibration campaigns were

conductedprefightandinflightduringthecruisephase.TheONC-Tincludes7colorbandpassfiltersandone

“wide”clearfilter,whereastheONC-W1andONC-W2includeasingle“wide”clearfilter.Wecarefullytestand

validatethespectralreconstructionoftheONC-Tusingmultiplespectrallywell-knownstarsandplanets,since

theabsolutesensitivityiscriticalforthespectralcharacterizationoftheRyugu’ssurface.

FileNamingRules

ONCimagingdata,Level2a–2dareproducedinFITSfileformat.ThefilenamingconventionsforONCimages

are

6 1055 52 364

“hyb2_onc_(Date)_(Time)_t(Filter)(i)_(Level).fit”forONC-T,

and

“hyb2_onc_(Date)_(Time)_(Cam)(i)_(Level).fit”forONC-W1and-W2,

where(Date)indicatestheobservationdatein8digitnumber,YYYYMMDD,(Time)indicatestheobservation

timein6digitnumber,hhmmss,(Filter)isthebandnameoffilter(u:ul-filter,b:b-filter,v:v-filter,n:Na-filter,

w:w-filter,x:x-filter,p:p-filter,andi:wide-filter),(i)is“f”for1024x1024pixelsfeardataand“b”for32x1024

pixels dark-reference data (Optical Black) (see Fig. 5 in Kameda et al. (2017)), (Cam) is “w1” or “w2”

correspondingtotheinstrumentsONC-W1andW2respectively,and(Level)indicatestheproductlevels.Note

thatthehigher-levelproductsmaybeprovideddifferentnamingrulesandformats.

Figure2.1.ONCdata calibration flowchart, starting fromraw images tohigher-levelproducts.Calibration

processes input parameters are acquired from the FITS file headers. Note that gray boxes have not been

implemented inthecurrentversionof thepipeline,butareexpected ina futureupdate.Products levels in

6 1055 52 364

bracketsareintermediateproductsandnotgoingtobearchived.SomeL2aproductshavebeensmearandbias

correctedonboard.(n:filter,t:exposuretime,TAE:electronicspackageofONCsystem(AE)temperature,TCCD:

CCDtemperatureofONC-T,TELE:theelectriccircuittemperatureofONC-T,(XPNL,YPNL):spacecraftattitude,Ssun:

solarirradiance)

6 1055 52 364

3.RadiometriccalibrationofONC-T

Forradiometriccalibration,imagedigitalnumber(DN)isconvertedtophysicalunitsofradiance(Wm-

2µm-1sr-1).Theradiometriccalibrationequationhastheform:

" =$%&(()*+,(-.*/01223,5,1678,5,196;,(/<=*)(()*+),(?*)@0A,1223,5;),(/B

C5,D∗FD(1223,5)∗A [Wm-2µm-1sr-1],

(n=ul,b,v,Na,w,x,p)

(3.1)

whereF[Wm-2µm-1sr-1]istheradiancefromthesurface, GHIJ [DN]isarawimage, GKLMN0OPPQ,1, OR$R,1; [DN]

isthebiaslevel, GSIHT0U, OPPQ,1; [DN]isthedarkcurrent, GVWXIH(GHIJ) [DN]isthesmearimage, GVY [DN]is

thestraylightlevel.L-1(I)isthelinearitycorrectionfunction,Z1,[ istheflat-fieldimageofONC-T, \[0OPPQ,1;

[(DN/s)/(Wm-2µm-1sr-1)]isthesensitivity,t[s]isexposuretime, OPPQ,1 [℃] istheCCDtemperatureofONC-

T, OR$R,1 [℃] istheelectriccircuit(ELE)temperatureoftheONC-T,O_R,1 [℃] istheAEtemperatureofthe

ONC system, and n indicates filter name. Each term is described in detailed in following subsections.

Calibrationsconductedso far forboth inflightandpreflightaresummarized inTable3.1.Notethatall the

calibrationimagesduringthecruisephasewereobtainedin12-bitformat,buttherendezvousphasedatawill

be12-bit,10-bit,or8-bitformatsfordifferentpurposes,suchas10-bitimagesforspectroscopicobservations

and8-bitimagesforgeometricobservations.Thebitdepthinformationisincludedintheheaderofeachimage.

InKamedaetal.(2017)theeffectivewavelengthswerecalculatedusingonlythetransmissionofband-

passfilters.Toobtainamoreprecisemeasurement,werecalculatetheeffectivewavelengthsusingthesystem

efficiencyΦ[(a) (Fig.3.1),includingthetransmittanceoftheband-passfilters,theneutraldensity(ND)filter,

lenses,CCDcoverglasses,thequantumefficiencyoftheCCD,andthespectrumofthelightsource.Theeffective

wavelength for each filter, using the solar spectrum (ASTM E-490) and the white light as a light source,

respectively,aretabulatedinTable3.2.Table3.2alsoshowsthebandwidth,theeffectivebandwidth,andthe

effectivesolarirradiancethrougheachfilter.ThebandwidthisdefinedinthesamewayasKamedaetal.(2017),

buthereweusedthesystemefficienciesinsteadofthetransmissionsthemselves.Theeffectivewavelengthis

thusdefinedas

aXbb,[ =∫de(d)fD(d)gd

∫e(d)fD(d)gd , (3.2)

where the system efficiency Φ[(a) = h[(a)hij(a)hk(a)hPl(a)m0a, OPPQ,1; ; h[(a) , hnQ(a) , h$(a) , and

hPl(a) are the transmissionsof theband-pass filters, theND filter, the lenses,and theCCDcoverglasses,

respectively, m0a, OPPQ,1; is the quantum efficiency, and o(a) is the light source spectrum. The effective

bandwidth ΔaXbb isdefinedas

ΔaXbb,[ ∙ Φ[0aXbb,[; = ∫Φ[(a)ra . (3.3)

Hereafterweusetheeffectivewavelengthwithrespecttothesolarspectrum.Notethatthequantumefficiency

usedhereisextrapolatedto-30℃ ofCCDtemperatureusingthetemperaturedependencefunctionprovided

6 1055 52 364

byitsmanufacturer,whichisdefinedforthetemperaturerangefrom-20℃ to40 ℃ andnormalizedat20

℃. Thequantumefficiencyat20℃ andother componentsof the systemefficiencyare shown indetail in

AppendixB.Exceptforsomeproximityoperations,theONCwillbeoperatedataCCDtemperaturearound-

30℃(thereferencetemperature).

Figure3.1.SystemefficiencyofONC-Tbandpassphotometricsystem.

Table3.1.Summaryofdatausedforcalibrationduringthecruisephase.

Preflightmeasurement Inflightmeasurement InflightObservationDate

Bias 0-sec exposures (Kameda

etal.,2017)

0-secexposures Continuouslyduringthecruisephase

EMI N/A 0-secexposures Continuouslyduringthecruisephase

Darkcurrentand

Hotpixels

Thermal vacuum test

(variationofCCD,ELE,and

AEtemperatures)

Deepskyobservationwith

variation of CCD and ELE

temperatures

17April2017,23May2017,25May2017

Sensitivity Integrating sphere

(Kameda et al., 2017),

temperature dependence

provided the

manufacturer (Suzuki et

al.,2018)

ESO standard and

Alekseeva (1996) stars.

Temperature dependence

byJupiterwithvariationof

CCDtemperature.

The sensitivities derived

fromstarsaretestedbythe

Moon, Mars, Jupiter, and

Saturn.

(Stars)19October2016,23May2017,10

October2017,12October2017

(Moon)5December2015

(Mars)24May2016,31May2016,

7June,2016

(Jupiter)16to18May2017

(Saturn)3November2017

6 1055 52 364

Sensitivity Time-

dependence

N/A Healthcheck(FFlamp) 11 December 2014, 16 April 2015, 10

September2015,8July2017,16October

2017,5December2017

Linearity Integratingsphere FF lampwith variation of

exposuretime

2December2017

Flat-Field Integrating spheres

(Kameda et al., 2017;

Suzukietal.,2018)

Star observation with

variation of spacecraft

attitude

12to14October2017

Geometric

distortion

N/A Stars(Suzukietal.,2018) 11December2017

Alignment to the

S/C

N/A Stars 11December2017

SharpPSFs Pinhole (Kameda et al.,

2017)

Stars(Suzukietal.,2018) 11December2017

BroadPSFs Smallextendedsources EarthandMars 4December2015,25May2016

Straylights N/A Radiator stray lights with

variationof thespacecraft

attitude.GhostwithMoon

andEarth.

11December2014,23June2015,12 to

17October 2015, 9November 2015, 14

November 2015, 22 December 2015, 9

February to 17 March 2016, 21 March

2016,28June to23July2016,13 to20

June2017,17to21October2017

Table3.2.CharacteristicsoftheONC-Tphotometricsystem.

ul b v Na w x p

asbbt) (nm) 397.5 479.8 548.9 589.9 700.1 857.3 945.1

aXbbu) (nm) 395.2 480.0 549.0 590.0 700.3 857.7 945.7

aXbb (nm)

Kamedaetal.(2017)390.4 479.8 549.0 590.2 700.4 858.9 949.7

Effectivebandwidth

ΔaXbb (nm)36.0 26.6 30.6 11.8 29.2 41.7 56.0

Bandwidth(nm) 34.7 26.6 27.9 11.6 28.8 42.4 57.2

Bandwidth(nm)

Kamedaetal.(2017)45 25 28 10 28 42 57

6 1055 52 364

Effective solar

irradiance(W/m2/um)1343.7 1969.1 1859.7 1788.0 1414.4 985.8 834.9

1)withrespecttosolarspectrum.2)withrespecttowhitelight.

3.1.Read-outsmearremoval

ONCisnotequippedwithamechanicalshutter, instead,eachexposure isdividedwithanelectronical

mechanism.Becauseof themechanicalshutter-lessdesignof theONC, theCCD is irradiatedduring frame

transferalong the vertical direction. So that, even 0-sec exposure images actually include read-out smear

duringthisveryshortexposuretime(<10ms).FortheglobalobservationsatthehomepositiontheONC-T

willtake0-secexposuresandproperexposuresalternately,andtheread-outsmearwillberemovedonboard

bysubtractingthe0-secexposuresfromtheproperexposures.However,duringthetouchdownsequencesthe

spatialcoverageofthefield-of-view(FOV)changesrapidlyovertheshorttimedurationoftheseoperations,

and appropriate 0-sec exposureswith the same spatial coverageas theproperexposure image cannot be

acquired.Inthisoperationalcase,theread-outsmearremovalprocesswillbeconductedaftertheimageshave

beendownlinkedusingthemethoddescribedbyIshiguroetal.(2010):

GNvsMw(x) =Ay25Ay25zA

∑ (G − GKLMN − GgMw})/�Äny,tÅÇÉ . (3.4)

The exposure duration during vertical charge transfer UÄP1 is 7.2µs × 1024 = 7.373ms for ONC-W2,

whichhasbeenderivedfromtheEarthobservationimagesacquiredduringtheEarth-Moonflyby.Becausethe

samemanufacturedCCDsareequippedonONC-T,-W1and-W2,thisverticaltransfertimesimilarbetweenthe

threecameras.Forexample,Fig.3.2showsanoriginalimageandthesmear-removedimagesofanONC-W2

(hyb2_onc_20151203_084458_w2f_l2a.fit)setofobservations.Theresidualof thesmear-removed image is

lessthan1%oftheintensityoftheEarth(~900DN),for95%ofpixels.ForthecaseofRyugu,theeffectofthe

smearwillbelessthanintheEarthobservationscaseduetothelongerexposuresneededforRyuguduetoits

correspondinglowerbrightness.ThisbiasremovalmethodworkssufficientlyfortheONC-W2image.Wewill

obtaintheverticaltransfertimesforONC-TandW1usingRyuguimagesduringtherendezvousphase.

6 1055 52 364

Figure3.2.An example of smear removal of anONC-W2 image (hyb2_onc_20151203_084458_w2f_l2a.fit)

usingEq.(3.4),showingtheread-outsmearwaswellremoved<1%intensityoftheobject.

3.2.BiasCurrentCorrection

TheONCsystemhasanelectronicbiascurrenttooffsetitszerolevel.Thetemperaturedependenceofthe

preflightbias levelwasassessedbyKamedaetal. (2017).Here,wealsoverify thebias levelusing inflight

observations.Weanalyzedimagesacquiredwithazero-lengthexposureandtargetedtoskyinordertoavoid

strong smear. As Kameda et al. (2017) reported, bias level changes depending on the CCD temperature

(OPPQ,1[℃])andtheAEtemperature(O_R,1[℃]).BiasleveldependenciesontheCCDandtheELEtemperatures

arealsoobserved inthe inflight imagedata(Fig.3.3). Itshouldbenotedthatdueto inflexibility intheAE

control,weinsteadchangedthetemperatureofELEthatresidesjustbehindtheCCD.Duringthepreflighttest,

theAEandtheELEtemperatureswerecontrolledsimultaneouslytoalmostsametemperature.Thus,because

theresultsofthepre-flighttestincludetheeffectofboththeELEandAEtemperatures,wecannotdirectly

compare the preflight and inflight testmeasurements. We plotted the preflight AE and ELE temperature

dependencetestwiththeinflightELEtemperaturedependencetestresultstogetherinFig.3.3.Thebiaslevel

is well described by fitting the inflight data with an empirical linear combination of the CCD and ELE

temperaturesas

GKLMN(OPPQ,1, OR$R,1) = 320.66 + 0.652OPPQ,1 − 0.953OR$R,1[ì�]. (3.5)

Figure3.4displayshowwelltheempiricalfunctiondescribesthebiasleveloftheinflightdataacquiredfrom

launchuntilDecember2017.Figure3.5showsthetime-dependentchangeofthebiaslevel,whichsuggests

anincreaseinthebiaslevelduringthefirsthalf-yearafterlaunch.Althoughtheempiricalrelationshipderived

6 1055 52 364

fromtheinflightimagedatadoesnotmatchthepre-flightexperiments,thisisascribedtotheeffectoftheAE

temperaturevariationinthepreflightdata.Thebiaslevelhasstayednearlyconstant,within1%change,inthe

3yearsofcruise.Moreover,theeffectoftheAEtemperaturecanbeestimatedfromthepreflightmeasurements

ofKamedaetal.(2017).AftercorrectingfortheeffectsofCCDandELEtemperatures,theremainingeffectis

ascribedtovariationsduetoAEtemperaturedifferences.Theempiricalcorrectionfactor î(O_R) fortheAE

temperature,derivedfromthepreflightmeasurements,isgivenby

î(O_R) = 0.987 − 0.00251O_R , for-30℃< O_R < 59℃ (3.6)

WhentheAEtemperaturechanges, thebias levelcanbeestimatedby GKLMN(OPPQ,1, OR$R,1)î(O_R).Notethat

thisrelationshipbasedonpreflightmeasurementsmayincludeerrorsof>4%,asthebiaslevelappearstohave

increasedby4%justafterthelaunch.

Figure3.3.Thebiaslevelofinflight(dots)andpreflight(triangles)0-secexposureimagesasafunctionofCCD

andELEtemperatures.TheAEtemperaturedeferencecontributesthebiaslevelsignificantly.

6 1055 52 364

Figure3.4. Bias level of thepreflight data (Kamedaet al., 2017) and the inflight data comparedwith the

empiricalfunction GKLMN(OPPQ,1, OR$R,1).

Figure3.5.BiaslevelvariationofONC-T.Redlinesindicatethelaunchdate.(a)Theobservedbiaslevelof0-

secexposuresduringthecruisephase.(b)ComparisonsoftheobservedbiasandthebiasmodelEq.(3.5).The

6 1055 52 364

biasmodelwellpredictstheinflightdatawithin1%changeoverthe3yearsofcruise.

3.3.DarkCurrentandHotPixelsCorrection

The CCD constantly responses to thermal electrons evenwhen the detector is not irradiated, which

producesadarkcurrentthatshouldbesubtractedfromobservedimagestoaccuratelyreproducereflectance

levels.However,becausetheONCdoesnothaveaphysicalshutter,wecannotdirectlyassessthedarkcurrent

atthetimeofeachobservation.Skyobservations,however,canprovideameasureofthesignalsfromthedark

current,hotpixels,EMI,stars,andstraylight.AsSuzukietal.(2018)reported,theradiatorstraylighthasa

gradualstructureoveranimage.Thestraylightcannotbecompletelyseparatedfromthedarkcurrent,which

alsohasa contributiontothe intensityoveran image.Thus,weobtainedupper limitsof thedark current

(<0.09DN/s)contributionatthereferencetemperature, OPPQ,1 = −30℃,basedonthepartsofimageswhere

thestray lightcontamination isminimized.Thisvalue iscomparabletothepreflightgroundmeasurement

(<0.05DN/sat OPPQ,1 = −30℃)byKamedaetal.(2017).Furthermore,thedarkcurrentlevelvariationasa

functionoftemperaturecanbemeasuredusingskyobservations,assumingthatthegradualstructureinthe

imagesareduetoradiatorstraylightasmeasuredatthereferencetemperature.Toobtaindarkcurrentlevels

athigherCCDtemperatures(-10℃,10℃,20℃,and25℃),weobtainedskyimageswithashortexposure

time(1.05s)foralltemperatureconditionsandlongexposuretimesof33.6sat-10℃,5.57sat10℃,and2.1

sat20℃and25℃.Wetook3imagesforeachexposurecondition,andremovedtheeffectsofcosmicraysby

takingamedianofthese3images.Afterthecosmicraycorrection,allimageswerestraylightcorrectedbased

on images taken at OPPQ,1 = −30℃, where the stray light contributionwas extracted using a fifth-order

polynomialregressionandneglectingoutlierswhicharecausedbystarsandhotpixels.Weobtainedthedark

currentdistributionexceptforstar-containingregionswithintheimageframe.Figure3.6(a)isahistogram

ofdarkcurrentlevelsfordifferentCCDtemperaturesintheONC-T.Theaveragesandthestandarddeviations

forthedistributionsarealsoshownintheFig.3.6(b).Thedarkcurrentgrowsexponentiallydependingon

theCCDtemperature,whichisconsistentwiththedarkcurrentbeingcausedbythermalnoise.Theempirical

relationshipoftheCCDtemperatureandthedarkcurrentlevelisgivenby

GgMw} = Uóòô(0.10OPPQ,1 + 0.52) [DN]. (3.7)

Carefullymonitoringthedarkimagesat-30℃,weobservedsomepixelsexhibitanomalyhighresponselevels

(hot pixels).We defined hot pixels asapixelwitha response larger than 30DN/s, or 0.3% the expected

intensityoftheasteroid.Hotpixelpositionson14October2017arelistedinTable3.3fortheCCDtemperature

conditioniscold(-30℃).ThenumberofhotpixelsincreasesasthefunctionoftheCCDtemperature(Fig.3.7).

Thisisespeciallyimportantduringthetouchdownsequences,astheCCDtemperaturecouldincreaseashigh

as20℃.Thus,theabsolutesignalatthehotpixellocationsshouldbeanalyzedcarefully.Notethat,however,

thenumberofhotpixelscanbeincreasedwithtimeduetocosmicrayirradiation.Thetemporalvariationin

6 1055 52 364

hotpixelsisshowninFig.3.8,indicatingthatthenumberofhotpixelsgraduallyincreases.Thus,weneedto

repeatthisanalysisduringtheapproachandrendezvousphasestomonitorthechangeinhotpixels.

Table3.3.Hotpixellistat OPPQ,1 = −30℃.HandVarethehorizontalandverticalpixelcoordinateinanimage.

H V Signallevel[DN/s]

24 653 47.0

52 276 89.3

296 939 55.8

407 239 37.1

505 29 31.4

582 204 75.1

Figure3.6.DarkcurrentleveldependenceontheCCDtemperatureofONC-T.(a)Darkcurrentdistribution.

The deviations of dark currentget largeras theCCD temperature increases. (b)Dark current level grows

exponentiallywithCCDtemperature,indicatingdarkcurrentlevelsarewelldescribedbythermalnoise.

6 1055 52 364

Figure3.7.TheamountofhotpixelsintheONCFOV.TheCCDtemperaturecouldbeashighas20℃ during

thetouchdowns.

6 1055 52 364

Figure3.8.Increasingnumberofhotpixels(sensitivepixels)inthebox{[492,118];[532,138]}.

3.4.CharacterizationofElectro-MagneticInterference

Figure3.9 shows thepresenceofelectro-magnetic interference (EMI)patternsembedded intheCCD

noise(stripepatterns).ThisiscausedbyinterferencewiththeactivecircuitsnearbytheONCsystemduring

thetransferoftheelectronsgeneratedattheCCDtothedigitalelectronics.Thus,theintensityoftheEMIis

6 1055 52 364

independent of the exposure time. The EMIpattern changeswith time depending onwhich systemswith

specificfrequencieswereactiveatthattimeofexposure.0-secexposureimageswereanalyzedline-by-line

basedonFourierFastTransforms (FFTs) toextract theperiodicpatterns.Figure3.10 (a) showsapower

spectrumof3imagestakenatdifferenttimes(16April2015,24May2016,16October2017),whichindicates

strongerpeaksforsomespecificfrequencies.WeevaluatedtheEMIfrominvertedFFTsoftheextracteddata

forfrequencies>0.001Hzandamplitudes>0.3DN.Figure3.10(b)showstheEMIstructurevariationduring

thecruisephase.AftertheEMIisextractedfromtheimages,theamplitudeoftheEMIisevaluatedasahalfof

thegapbetweenthemaximumandminimumsignalsintheline(Fig.3.11).ThissuggeststheEMIintensities

havenotdrasticallychangedduringthecruisephaseandthehighestEMIintensitywas~5DN,whichis<0.2%

oftheexpectedasteroiddiskintensity(~2500DN).However,wecontinuetomonitortheEMIcontributionto

accuratelyremovetheEMIfromimagesrequiredforsensitiveandadvancedspectralanalyses.

Figure3.9.Anexample imageof theEMIpattern(hyb2_onc_20171016_00056_tvf_l2a.fit).Thepresenceof

periodicalstripepatternsovertheimagewasobserved.

6 1055 52 364

Figure 3.10. (a) The results of FFT analysis of 3 images (hyb2_onc_20150416_074730_tvf_l2a.fit,

hyb2_onc_20160524_124508_tvf_l2a.fit, hyb2_onc_20171016_000056_tvf_l2a.fit). There are several strong

peakspossiblyassociatedwithothersystemcircuits.(b)TheextractedEMIsignalshavedifferentstructures

atdifferentobservations.Offsetsare5DNfromeachother.Samplingrateis3MHz.

6 1055 52 364

Figure3.11.VariationintheamplitudeoftheEMIwavesduringthecruisephase.Theamplitudeisdefinedas

ahalfofthegapbetweenmaximumandminimumsignalsofEMIwaves.Theredlineindicatesthelaunchdate.

3.5.Linearity

Inthissection,wereportonthelinearityoftheONC-TCCDresponsebasedoninflightobservationsof

Flat-Field(FF)lamp.TheCCDresponsetothenumberofphotonsshouldideallybelinear.Preflighttesting(Fig.

3.12) shows that the responseof theONC-TCCDwas linearupto3200DN,withinanerrorof0.6%.This

responsewasmeasuredusingan integratingsphereandtaking imagesoveravarietyofexposuretimesat

roomtemperature.ThelinearitywasmeasuredagaininflightbyobservingtheFFlampthroughthev-band

filteratexposuretimes0.131,0.348,0.525and0.696s.Wetook3shotsevery2secondsforeachexposure

time tomeasure the output stability of FF lamp.We alsomeasured the stray lightbefore and after these

observationsbyturningofftheFFlampinordertoexaminestraylighteffectsonthistest.Afterbeingcorrected

forbias levelsandradiatorstray light, thebright{[384,384] ; [415,415]}andthedark{[384,480] ; [415,

511]}boxes,showninFig.3.13,ineachimagewereaveraged.Inflight-testresultsareshowninFig.3.14.The

DNaccumulationrateissensitivetosmalldeparturesinlinearity.ThoughtheDNcountsupto3100DNappear

linearwithexposuretime(Fig3.14),smallchangesintheaccumulationrate(shownbythevariationsalong

thehorizontallinesinFig.3.14)indicatesmalldeviationsintheCCDlinearityresponse.Thistestshowsthat

stronglinearity(lessthan1%deviation)wasachievedupto3100DN,whereasa10-13%dropinlinearityis

seenatsignallevelsof3600DN.Thus,exposuretimesshouldbeselectedcarefullyinordernottoexceedthis

quantitative limitof3100DN.Thedecreaseof accumulation rate canbe fittedbya third-orderpolynomial

function,providing,theempiricalrelationshipbetweentheobservedintensityIobsandtheidealintensityI0of

GöKN = 1.0073GÉ − 2.9285 × 10,õGÉu − 3.6434 × 10,tÉGÉ

ú. (Iobs<3100DN) (3.8)

Usingthisrelationship,thelinearitycanbecorrectedwithin0.1%error(redsymbolsinFig.3.14).

6 1055 52 364

Figure3.12.Preflightgroundlinearitytestshowingthelinearrelationshipupto3200DN.(left)Theaveraged

intensityfromtheintegratingsphereimageswithdifferentexposuretimesisshownbyopencircles.Thesolid

lineshowslinearrelationshipof24.52DN/s.(right)Thesignalaccumulationrateovereachtimedurationis

shownbycrosses.Thehorizontallineindicatestheidealaccumulationrateof24.52DN/swithperfectlinearity.

Thedeviationfromtheidealaccumulationratewaswithin0.6%.

Figure3.13.Theredsquaresshowtheregionsofinterestintheinflightlinearitytest.Twodifferentintensity

regions,thebrightanddarkareas,oftheFFlampimagesaretested.

6 1055 52 364

Figure3.14. Inflight linearitytestdisplayingthe linearrelationshipupto3100DN.(left)Opencirclesand

trianglesareaveragedintensityinthebrightanddarkboxesoftheFFlampimageswithdifferentexposure

times,respectively.(right)Crossesandplussesrepresenttheaccumulationrateovereachtimedurationinthe

brightanddarkboxes.Bluesaretheobservedvaluesandredsarethelinearitycorrectedvalues.Beforethe

linearitycorrection,bothareasshowsimilartrendswithdecayinaccumulationrateforthelongerexposure

times.Thehorizontallinesindicatetheidealaccumulationrates6.05DN/msforbrightareaand5.82DN/ms

fordarkareawithperfectlinearity.Thedeviationsfromtheidealaccumulationratesare<1%forbeforeand

<0.1%forafterthelinearitycorrection.

3.6.CorrectionofScatteredandReflectedLightinOptics

3.6.1.ScatteredLight(BroadPSFs)

Thepresenceofscatteredlightaroundobjects(dimhalo)wasobserved(Fig.3.15)inallONCfilterbands.

Thatis,theskyclosetotheimagedobjectwasbrighterthanaccountableforfromthebiasanddarkcurrent

levelsduetothescatteringof light fromadjacentpixelscontainingtheobject image.This iscausedbythe

multiple scatteringofphotonswithin theoptical systembeforedetectionby theCCD.Thiseffectwasalso

observedinAMICA(AsteroidMultibandImagingCamera)imagesonboardHayabusa(Ishiguro,2014).This

scattered light affect has severe effects on the spectral analysis of regions near the object limb since the

scatteringbehaviorhasawavelengthdependency,whichisreportedanddescribedindetailbyIshiguro(2014)

fortheAMICAinstrument.TheyreportedthatfortheAMICAp-bandtheintensitychangedbymorethan10%

closetoobjectlimbinimagesacquiredattheHomePosition(altitudeof~7.5km).Thus,thescatteredlight

correction is especially critical for photometric and spectral analyses. Because the light is distributed

isotopicallyabouttheobject,wecanevaluatethescatteredlighteffectasapartofpointspreadfunction(PSF).

6 1055 52 364

However, to distinguish this scatteringeffect from the sharp peak of a PSF,which is usuallymeasured by

observingaroundstarsasinfinitelysmallpointlightsources,werefertothiswidespreadattenuationfunction

as the “broadPSF”.The sharppeakPSFhavebeenalreadyobtainedaspartof thepreflight calibrationby

Kamedaetal.(2017)andinflightcalibrationbySuzukietal.(2018).Therefore,hereweinvestigatethebroad

PSFbasedonthemethoddescribedinIshiguro(2014)andwillshowanapplicationtoanEarthimageobtained

duringtheEarth-Moonswing-by.

Methodology:

ThetotalPSF ZùFû isasummationofthesharppeakPSF, ZN ,andthebroadPSF, ZK:

ZùFû = ZN + ZK . (3.9)

Inprinciple,wedon’tobtainimageswithoutaPSFbutwhatweobtainareblurredimages.

GöKN = GÉ ∗ ZùFû = GÉ ∗ (ZN + ZK) (3.10)

where GÉ istheimagewithoutblurand GöKN istheobservedimage.Consideringwehavetostartfromblurred

images GöKN ,whatwedetermineisthefunction ZK′,soastosatisfythefollowingequationintheskyregion

wherecountsfrom ZN arenegligible:

GöKN ∗ Z†K = GÉ ∗ ZK (3.11)

ThenwecanevaluateblurredcountsduetoaPSFwith GöKN and ZK′. AsisdescribedinIshiguro(2014),we

alsoassumedthatthe ZK′ canbeexpressedasthesummationofGaussianfunctions:

Z′K(°) =¢£L

√2•¶Lexp ™−

°u

2¶Lu´

n

LÇt

(3.12)

where ° [pix]isthedistancefromapointsource, £L and ¶L areconstants(N=1to6),where ¶L = 2Lzu(� ≤

4),¶≠ = 110,and ¶õ = 710.Theeffectofscatteredlightcanthenbecorrectedas

GÉ ∗ ZN = GöKN − GöKN ∗ Z′K (3.13)

Assumingthatthe ZN issharpenoughtosatisfy ZN = ÆNØ(0,0),where ÆN isthecontributionduetothesharp

peakPSF,and

GÉ = (GöKN − GöKN ∗ Z†K)/ÆN where (3.14)

ÆN = ∫2•°ZNr°.

PreflightdiskobjectimageswereobtainedthroughallfilterswithdifferentsizeddisksplacedintheFOV(Fig.

3.16).We determined the £L Gaussian coefficient values sequentially, usinga bisectionmethod from the

broader components (beginningwith £õ ) to narrower component (endingwith £t ).We performed this

analysistoderivethecoefficients £L foreachfiltertoreducetheresidualsignalinskyregionfartherthan8

pixelsfromedgeoftheobjectto<1%oftheobject’ssignalbyusingEq.(3.14).SmalldiskobjectsintheFOV

wereusefulinderiving £L forsmallervaluesof ∞ andlargediskobjectsforlargervaluesof ∞.Thecoefficients

weobtainedarelistedinTable3.4andFig.3.17showsthebroadPSFonAMICAandONC-T.Thecontribution

6 1055 52 364

ofthebroadPSFonONC-TissmallerthanthatobservedwithintheAMICAimagedatasets.

Evaluation:

ThevalidityofthebroadPSFcorrectionisassessedwithapplicationstotheEarthimagesobtainedduring

theEarth-Moongravityswing-by.ThebroadPSFcorrected imagesare shown inFig.3.18withbrightness

variationsacrosshorizontalprofilesdisplayedtoshowthereduction inscattered light.Comparingthesky

closetotheobjectlimb,theresidualislessthan1%andnowtheghostcanberecognizedmoreclearlythan

thescatteredlight.

ThebroadPSFcorrectionisalsoverifiedbyusingasaturatedMarsimageobservedon25May2016.In

thistest,thestrategyisthatifapointsourceobjectisobservedwithalongexposuretime,thebroadPSFcan

beobservedmoreclearly.Marswasimagedalmostasthepointsource(~1.4pixelindiameterintheFOV).We

obtainedtheimagesusingthep-bandfilterwithtwodifferentexposuretimes,aproperexposuretimeof0.044

sandalongexposuretimeof178.26s.Thedarkcurrentandtheradiatorstraylightcorrectionswereapplied

beforetheintensityanalysisforthebroadPSF.Theproperexposureimagesareusedtomeasuretheradiation

fluxfromtheobjectbyfittingtheobservedintensitywithaGaussianfunction GvMwN ,whichisequivalentto

theMarssignalconvolvedbyonlythesharppeakPSF.Here,theFull-Maximum-Half-Width(FMHW)of the

Gaussianfunctionis1.53,whichisconsistentwiththesharppeakPSFobtainedinKamedaetal.(2017)and

Suzukietal.(2018).TheDNcountofthelongexposureimagewasestimatedbymultiplyingbytheratiooftwo

exposuretimes:

≤≥ö[¥ = ™U≥ö[¥Uµwöµ

´≤µwöµ (3.15).

Thus, the scattered light around Mars is estimated by ≤≥ö[¥ ∗ ZK′ . Figure 3.19 compares the observed

scatteredlightwiththelongexposureandtheestimationofourbroadPSF.Thus,ourbroadPSFobtainedbased

onpreflightimagesareconsistentwiththeobservedscatteredlightfromapointsource.

6 1055 52 364

Figure3.15.AnEarth image (hyb2_onc_20151204_040959_tpf_l2a.fit)before(top)andafter (bottom) the

scattered light correction shown by histogram equalization, which enhances contrast. The contour lines

correspondto0.5,1,2,4,8%ofthediskintensity(magenta:0.5%,red:1%,orange:2%,green:4%,blue:8%).

Figure3.16.ONC-Tpreflightimages(p-band)usedforfittingthebroadPSF.ThesizeofobjectsintheFOV

were225,468,and979pixelsindiameterfromlefttoright.Thecontourlinescorrespondto0.5,1,2,4,and

8%ofthediskintensity(magenta:0.5%,red:1%,orange:2%,green:4%,blue:8%).

6 1055 52 364

Figure3.17.ComparisonofthebroadPSFonONC-T(solidlines)andAMICA(dottedlines).

Table3.4.ThebroadPSFcoefficientsAi (10-4)andthebroadPSFcontribution ∫∂∑∏π′∫ª∏ onONC-Tand

AMICA by Ishiguro (2014). Note that the contribution of broad PSF on AMICA is evaluated by π∫† =

[∫∂∑∏πºª∏/∫∂∑∏πΩæøª∏] × π∫.

ONC-T A1 A2 A3 A4 A5 A6¿2•°Z′Kr°

ul(0.4µm) 12.0 0.7 0.7 0.5 0.7 0.3 0.11

b(0.48µm) 10.0 0.5 0.0 0.0 0.4 0.2 0.07

v(0.55µm) 9.0 0.0 0.2 0.1 0.4 0.2 0.07

Na(0.59µm) 10.0 0.3 0.4 0.0 0.3 0.2 0.08

w(0.70µm) 11.0 0.4 0.5 0.2 0.2 0.3 0.09

x(0.86µm) 10.0 1.3 0.6 0.1 0.6 0.5 0.14

p(0.95µm) 9.0 1.5 1.2 0.1 1.1 0.7 0.19

AMICA A1 A2 A3 A4 A5 A6 Contributionof ZK

ul(0.38µm) 11.4 7.6 1.1 1.0 0.8 0.7 0.23

b(0.43µm) 9.7 1.5 0.3 0.4 0.4 0.5 0.13

v(0.55µm) 9.6 1.3 0.3 0.4 0.4 0.5 0.13

w(0.70µm) 9.2 1.4 0.6 0.7 0.6 0.5 0.15

x(0.86µm) 7.9 3.1 1.8 2.4 1.9 0.4 0.20

p(0.96µm) 7.9 4.0 6.6 3.2 5.1 1.4 0.53

zs(1.01µm) 33.5 10.7 4.0 6.0 6.4 3.0 0.95

6 1055 52 364

6 1055 52 364

Figure3.18.EnhancedEarthimagesbefore(thefirstrowimages)andafter(thesecondrowimages)the

broadPSFscatteredlightcorrection.Horizontalprofilesalongredlinesofthemareshowninthethirdrow.

Figure3.19.Observed scattered lightas a functionofdistance fromMars, comparingwithourbroadPSF

obtainedpreflight.NotethatclosepartfromMarsisaffectedbythesharp-peakPSF(grayhatch).

3.6.2.GhostEffect

ThereisnoisewithintheONCsystemduetoreflectionsbetweenoptics,suchasfilters,lens,andtheCCD,

thatcreateaneffectcapturedwithinthe imagehereafterreferredtoasthe“ghosteffect”.BecausetheONC

opticsareaxi-symmetric,aghostimageusuallyappearsatasymmetricallocationabouttheopticalaxis.Here,

weinvestigatedtheintensityoftheghosteffectandthepositionoftheghostimagewithintheimageframe.

Figure3.20displaysexamplesofghostsintheimagesobtainedduringtheEarth-Moonswing-by.Thedegree

ormagnitudeoftheghosteffectismeasuredtobeweakerthan0.1%oftheintensityoftheobservedobject

forallfilters,however,thereisawavelengthdependencytothemagnitudeoftheeffect.Forexample,v-,Na-,

andw-filtersdonothaveghosteffectslargerthanotherCCDnoisesources,suchasMEInoise,darknoise,and

shotnoise,whiletheul-,x-,p-,andwide-filtersallsufferfromasmall(<0.1%)butdetectableghosteffect.From

theEarth-Moonswing-by images,wedeterminethereflectioncoefficients(the intensityratiobetweenthe

ghostandobject)forthefiltersas0.0033,0.0012,<0.0001,<0.0001,<0.0001,0.0045,0.0065,and0.0028from

shorttolongwavelengthandthewidefilter,respectively.Thelightreflectssymmetricallyaboutthesub-pixel

positionat~(496.5,501).Usingthesevalues,examplesofghostremovalareshowninFig.3.21.Itshouldbe

notedthatwhentheobjectisfarfromthecenterofFOV,thiscorrectionmethodisinsufficient(Fig.3.22).The

ghosteffectonimagesthatincludetheobjectatacornerismuchweakerthanthatonimageswhichinclude

theobjectatthecenter.Thismayindicatethattheintensityoftheghosteffectisafunctionofthedistancefrom

6 1055 52 364

theopticalaxis.Thus,theintensityassessmentisvalidforthecenterregionofFOV,whereRyuguwillbelocated

on home position observations. Moreover, when the object fills the entire field of view, such as close

approachestotheasteroid<10km,thisghostremovalmethodmayworkonlyinthearea<500pixradius

fromthecenter.

Figure 3.20. Ghost effect captured in images of the Moon with multiband filters

(hyb2_onc_20151205_115044_tuf_l2a.fit, hyb2_onc_20151205_115024_tbf_l2a.fit,

hyb2_onc_20151205_114908_tvf_l2a.fit, hyb2_onc_20151205_114952_tnf_l2a.fit,

hyb2_onc_20151205_114920_twf_l2a.fit, hyb2_onc_20151205_114940_txf_l2a.fit,

hyb2_onc_20151205_115012_tpf_l2a.fit, hyb2_onc_20151205_115056_tif_l2a.fit). The centers of images,

200x200pix,areshown.GhostsoftheMoonappearpointsymmetricallyattheopticalaxis(atlowerleftin

thiscase).ColorbarsindicatethesignalnormalizedbythediskaveragesignaloftheMoon.

6 1055 52 364

Figure 3.21. Two examples of ghost removal for Earth images taken through the p-filter

(hyb2_onc_20151126_023805_tpf_l2a.fit, hyb2_onc_20151204_040959_tpf_l2a.fit). The centers of the FOV,

200x200pixand600x600pixrespectively,areshown.

Figure3.22. Over-correction of an imagewith the object observedwithin the corner of the image frame

(hyb2_onc_20151204_045401_tpf_l2p.fit).Theghosteffectismuchweakerfortheobjectimageatacorner.

3.7.RadiatorStrayLightRemoval

The radiator stray light contamination is caused by the reflection of sunlight by apart of theONC-T

radiatorwhichwasreported inSuzukietal. (2018).Figure3.23 summarizesthestray light intensitywith

respecttothespacecraftattitude,whichwascontinuouslymonitoredwithonboardattitudesensors(i.e.,star

trackers).TherearetwowaystodescribethespacecraftattitudewithrespecttotheSun,whichare(XPNL,YPNL)

and(¡, ¬).ThedefinitionofattitudesandspacecraftcoordinatesaredescribedinAppendixC.Theintensity

6 1055 52 364

ofthestraylightwasevaluatedatthehighestintensitypartoftheimageframe,inthesamewayasSuzukiet

al. (2018),andwenormalized the intensityby thedistance from theSun.Thedefinitionof thespacecraft

attitudewasreferredtoSec.8.1 inSuzukietal. (2018).Thereferenceattitudeof thespacecraft is that the

antennaispointingtotheEarth.Therebythevalueof ¡ correspondstothephaseangle,i.e.,theEarth-Ryugu-

Sunangle,andisautomaticallydeterminedbythepositionoftheEarth,RyuguandtheSunattheobservation

time.Weexpectthat ¡ willbechangefrom-20˚to-10˚beforethefirsttouchdown.Thetrendofthestraylight

didnotchangeduringthe3.5yearsofthecruisephase.Thatis,thestrongestpartisalwaysinthe(+H,+V)

corner,andthestraylightcomponentgraduallyweakenstowardstheopposite(-H,-V)corner(seeFig.21in

Suzuki et al. (2018)). Because attitudeswith ¬ < −7° do not have strong stray light component, we are

planningtoobtainimagesofthetargetasteroidatthoseattitudeswithnegligiblestraylight,<3.8DN/sat1

AU,forglobalmappings.Ontheotherhand,fortheproximityobservations,suchasthelanderreleases,the

SmallCarry-on Impactor (SCI) crater scan observations, and touchdowns,when the observationattitudes

allowstraylightatthe ~1000DN/slevelmaycontaminatetheimages.Here,wemodelthestraylightasa

functionofspacecraftattitudetowardstheSuntoprovideamethodforremovingthiscomponentfromthe

images.

Basedoninflightobservations,thespatialpatternsoftheradiatorstraylightdoesnotchangedrastically

with the spacecraft attitude.Moreover, the stray light component is reproducible for the same spacecraft

attitudewhenthe intensity isnormalizedbythesquareof theheliocentricdistanceof thespacecraft(Fig.

3.23).Thus,weseparatelymodeledthestraylightpatternwithintensityvariation(modeledatthecenterof

FOV)andthespatialdistribution.

Investigationsofastrongstraylightcomponentforpossibleattitudesduringtherendezvousphasewas

conductedon17,19,and21October2017,especiallywiththespacecrafttwistingangleof ¬ < −5°.First,we

empiricallymodeledtheintensityofthestraylightatthecenterofFOVwithapolynomialfunction:

GN≥ = (−0.0879¡ú − 5.616¡u − 119.4¡ − 603.8)

(−0.0037¬u + 0.113¬ + 0.956) [DN/s].

(3.16)

ThecomparisonofthestraylightintensitymodelandtheobservedintensityisshowninFig.3.24.Bydividing

the stray light images by the central intensity, the spatial distributions can be examined. We bin those

normalizedpatternsinto8×8(averageof64×64pixels)imagestosmoothoutthenoise.Afternormalization

andsmoothing,weappliedprincipalcomponentanalysistothoseimagesandextractedtheaveragepattern

and the principal components of the stray light patterns. The principal component analysis extracts the

orthogonalcomponentswithhighestvariancetolowervariance.Thecomponentwithhighest

variance,thefirstprincipalcomponent,mayexplainthestraylightpatternchangemost.Thecomponentswith

smallvariancecouldbenoise.Thus,thecomponentswithhighvarianceareenoughtoexpressthevariation

instraylightpattern.Inouranalysis,Thefirstprincipalcomponent(PC1)scorecontains81%ofvarianceand

6 1055 52 364

the second principal component (PC2) score contains 18 % of variance. Thus, 99% most of the spatial

distribution variation can be explained by the PC1 and PC2 scores. The stray light patterns ƒVY can be

decomposedlinearlybytheprincipalcomponentsas

ƒVY = ƒI≈X + ∆«»tƒ«»t + ∆«»uƒ«»u, (3.17)

Where ƒI≈X, ƒ«»t, andƒ«»u aretheaveragespatialdistributionimage,thefirstprincipalcomponentimage,

andthesecondprincipalcomponentimagewhichareshowninFig.3,25. ∆«»tand∆«»u arethePC1andPC2

scores,respectively.Moreover,wefoundthatthePC1scoreforstraylightcorrelatewiththevalueof ¡ (Fig.

3.26).Basedontheintensitymodelandthespatialdistributionmodel,wecanestimatethestraylightatan

arbitralattitudeintherangeof −30° ≤ ¡ ≤ −10° and −5° ≤ ¬ ≤ 10°.Applyingthisremovalmethodtothe

observed images, the residual of the stray light averaged over the FOV is smaller than 25 DN/s/pixel,

correspondingto0.25%oftheintensityoftheasteroiddisk.Notethat,theedgeofFOVtendstohavemore

errorthanthecenterwiththismethod.

Figure3.23.(a)Typicalpatternofradiatorstraylight.(b)Theintensityclassificationofradiatorstraylight.

RadiatorstraylightoccursasafunctionoftheangleofthesuntotheXPNLandYPNL(left),andthesolarphase

angle(¡)andthetwistinganglearoundZSCaxis(¬)(right).Duringtherendezvousphase,weareplanningto

twistthespacecraftto ¬ < −7° asmuchaspossibletoavoidstrongstraylightcontributions.

6 1055 52 364

Figure3.24.Intensitymodel(blackmesh)ofstraylightasafunctionof −30° ≤ ¡ ≤ −10° and −5˚ ≤ ¬ ≤

10˚,comparingwithobservations(red).

Figure3.25.Theaverageandtheprincipalcomponentsofthestray-lightspatialdistributions.

Figure3.26.ThePC1scoresandthephaseangle ¡ ofthestray-lightspatialdistributions.

3.8.Flat-FieldCorrection

Flat-fieldcorrectionisoneofthekeycalibrationsneededforderivingreliablebrightnesspropertiesat

6 1055 52 364

anypositionwithinan image frame. After deriving the preflight flat-field correction using the integrating

sphere measurements, the front hood position of ONC-T was changed slightly during disassembly and

assemblyatthelaunchingsite.Thus,allONC-Tbandsareexpectedtohavedifferentsensitivitydistributions

(flat-fields)inflightthanthoseobtainedpreflightusingtheintegratingsphereimages(Kamedaetal.,2017;

Suzukietal.,2018).Infact,~10%differencesinthelunarbrightnesswereconfirmedbetweenimageswith

theMoonlocatedinthecenterofanimageandthosewiththeMoonlocatedinthecornerforallfilterbands

(Suzukietal.,2018).ThelunarimageswereacquiredduringHayabusa2’sEarth-Moonflybyon3December

2015,whichprovidedthegravityassisttoreachitstarget,162173Ryugu.Thelunarimageswerecorrected

withanadditional flat-field correctionderived frommeasurements takenwithaportable flat light source

conductedafterthehoodpositionwasslightlychanged.Byapplyingtheadditionalflat-fieldcorrection,the

discrepancybetween lunarbrightnessvalueswasreducedto2%inallbandsexcept forul-band,while the

relativelylargediscrepancies(~3%)stillremainintheul-band(Suzukietal.,2018).

Forevaluatingthevalidityoftheflat-fieldcorrectionsandinvestigatingthedetailedstructureofthecurrent

sensitivitydistributioninul-band,astarobservationcampaignwasconductedfrom12to14October2017.

DuringthiscampaignfourbrightstarswereobservedunderfivedifferentattitudeconditionsoftheHayabusa2

spacecrafttoplacethestarswithindifferentlocationswithintheONC-TFOV.Thefourbrightstarsobserved

werePhi,Sigma,Tau,andZetaSagittariiwhoselocationsarecloseenoughtoeachothertobeobservableat

thesametimewithintheFOV.StarlocationsinONC-Timageframeduringtheobservationcampaigncover

almostthewholeFOVoftheimageframe(Fig3.27(a)).Starobservationswereconductedwithexposuretimes

of16.7and22.8secondsforv-band,and16.7and67.2secondsforul-band,toobtainsufficientcountsfrom

thedifferentstarbrightnesstomeasuretheCCDsensitivity.

TheexpectedfluxfromastarJstar,nthroughn-filtercanbedeterminedfrom

oVÕIH,[ =∫de/Œ*)(d)fD(d)gd

∫dfD(d)gd, (3.18)

whereJstar[W/m2/µm]istheirradiancefromthestar.Thisvalueshouldbeproportionaltothetotaldigital

countrateoftheobservedstarobtainedbyONC-T:

(∑œ–,—“–,—

–,— )/A

∫ aostar(a)Φ’(a)ra= Æ(÷◊’ÿU∆’U), (3.19)

whereIi,jistheDNcountsandfi,jisacoefficientofflatfieldcorrectionat(i,j)pixel.Thisindicatesthatlefthand

sideofEq.(3.19)shouldhavethesamevalue foranystarandanyexposuretime, if flat fieldcorrection is

perfectatalllocations.Inotherwords,ifastellarobservationhasasmallervaluefortherighthandsideofEq.

(3.19)thanotherstars,theflatfieldcorrectionatthepositionisinsufficient,andthusthecorrectedsensor

6 1055 52 364

sensitivityissmallerthanexpected.

To evaluate sensor sensitivities at the star positions, we performed flat field correction with those

proposed inSuzukietal(2018)afterperformingdarksignalsubtractionandhot-pixelreduction.Thenwe

measured the total DN counts for each star by integrating theDN counts in a 20-pixel radiusaround the

brightnesscenterofthestar.TheonlyexceptiontothiswasforTauSagittariilocatedat(H,V)=(86.7,619.7)

withanexposuretimeof67.2sec.Inthiscaseweuseda15-pixelradiusforintegratingtheDNcountstoavoid

includinghugecountsfromacosmicrayhitlocatednearbythestarintheFOV.Inthesecalculations,weused

the star flux from Alekseeva et al. (1996), which is available in the database VisieR (http://vizier.u-

strasbg.fr/viz-bin/VizieR).

Table3.5summarizesthevaluesofnormalizedCwhicharecalculatedfromasCvaluescalculatedfrom

Eq.(3.19)ineachobservationarenormalizedtothemeanCvalueforeachfilter.Examinationofthestandard

deviationsofCvaluesshowsthatthev-bandimageframeachievesuniformsensitivitywithadeviation<2%

withnosignificantoutliers.Consideringthatthestandarddeviationincludeserrorsinstarirradianceandin

noisereductionprocesses,the2%deviationcanbeconsideredasanupperlimitofthesensor’suniformityin

sensitivity.Inaddition,gradualsensitivityvariationsacrosstheimagecanbeobservedinthev-bandFOV(Fig.

3.27(b)),indicatingthatadditionalflatfieldcorrectionscouldbeperformed,thoughitishardtoderiveahigh-

accuracy flat-field correction only from this star observation campaign due to the sparseness of the star

positionsacrosstheFOV.

Theul-bandimageframehasalargersensitivitydeviation(3.9%)acrosstheimagescenecomparedto

thesensitivitydeviationinthev-band.AgradualsensitivityvariationacrosstheFOV(lowsensitivityarounda

lower-leftimagecornerandhighsensitivityinupperregion)canalsobeobservedinul-band(Fig.3.27(c)).

Especiallyatpositionsof(H,V)=(85.8,903.8)and(38.6,958.1),Cvalues(denotedbyunderlines)are6-9%

smallerthanameanvalue,indicatingadeclineinsensorsensitivityuniformityneartheimageedges.From

theseresults, thebrightness inconsistencyseen inthe lunarul-bandobservation,reportedbySuzukietal.

(2018),couldbeexplainedbytheworseuniformityneartheimagecorners.Thiscouldaffectmappingand

photometricanalysisespeciallyintheperipheralregionsintheFOV.Thisdecreaseduniformityintheul-band

flat-fieldmaybefundamentallycausedbythepreflightmeasurements.Duringthepreflightintegratingsphere

measurementsstraylightcontributionsfromthegapbetweenthefilterwheelandbaffleweredetected(see

“FWroundabout”straylightdescriptioninSuzukietal.,2018).TheFWroundaboutstraylighthasastronger

contributionattheshorterwavelengths,thuspreferentiallycontaminatingtheul-bandimages.Itshouldbe

notedthatifwefocusonlyonthestarsobservedinsideoftheradiusof400pixfromtheimagecenter,where

the standard deviation of C values is 2.2% in the ul-band, indicating we may have sufficient sensitivity

uniformitywhenweuseonlythecenterofFOV.

6 1055 52 364

Figure3.27.(a)Starpositionsobtainedduringthestarobservationcampaignfrom12to14October2017.

Bluecrosses,greencrosses,redtriangles,andorangediamondsindicateobservedpositionsofSigma,Phi,Tau,

andZetaSagittarii,respectively.(b)Deviationsofthemeasuredsensorsensitivitiesatstarpositionsfromtheir

meanvalueforv-bandafterperformingtheflat-fieldcorrectionproposedinSuzukietal(2018).(c)Samefor

ul-band.

Table3.5.ObtainedCvaluesfromEq.(3.19)andtheirstandarddeviationsfromstarobservationsforul-and

v-bands.EachCvalueisnormalizedbyanaveragedCvalueofallstarstogetherforeachband.

Starname

Position(pixel) NormalizedC(=normalizedsensitivity)

H Vv-band ul-band

16.8sec 22.3sec 16.7sec 67.2sec

Phi

836.7 430.3 0.968 0.997 1.043 1.019

837.6 146.3 0.995 0.965 0.985 1.003

837.1 713.5 0.974 1.000 1.019 1.003

594.3 430.2 0.989 0.984 1.046 1.031

Sigma

480.3 350.8 1.004 1.002 1.045 (saturated)

480.5 66.8 1.016 1.014 1.058 (saturated)

480.5 633.4 0.991 0.995 1.042 (saturated)

238.0 350.4 1.025 1.020 0.983 (saturated)

804.5 350.2 0.994 0.994 1.039 (saturated)

Tau

86.7 619.7 0.990 1.020 (toolow) 0.973

87.0 336.4 1.023 1.010 (toolow) 1.004

85.8 903.8 0.995 0.985 (toolow) 0.938

6 1055 52 364

3.9.RadiometricCalibration

Inthissection,theconversionfactorfromcountperseconds(DN/s)toradiance(W/m2/µm/sr),referredto

as“radiometriccalibration”,isdiscussed.First,thetemporalvariationinsensitivityisdiscussedinSec.3.9.1.Suzuki

etal.(2018)usedthesensitivitycorrectedbythetypicaltemperaturedependenceprovidedbythemanufacturer

(E2V). However, the temperature dependence had not been evaluated inflight at that time. Thus, we second

measuredtheeffectofCCDtemperatureonthesensitivitybasedoninflightfluxmeasurementstakenofJupiter

(Sec.3.9.2).Theobjectiveinthissectionistoupdatethesensitivityconversionbasedoninflightmeasurements.

Weevaluatethesensitivityconversionintwoways;1)basedonthehardwarespecificationsofcomponents,such

asCCD,filters,andlensoftheONC-T(Sec.3.9.3),and2)basedonstarobservations(Sec.3.9.4).Finally,theupdated

conversiontoradianceisvalidatedbasedonONCobservationsoftheMoon,JupiterandSaturnasdiscussedinSec.

3.9.5.

3.9.1.Time-DependentSensitivityChange

TheFFlampisdesignedtomonitorthesensitivity(absolutevalueandspatialpattern)oftheCCDduringthe

mission.Table3.6summarizestheFFlampobservationstakenduringthecruisephase.FFlampobservationsare

included in the default routine of the health-checkout sequence of operations. Health-checkouts have been

conducted four times over the cruise phase after the initial post-launch checkout. An extra observation on 5

December2017wasconductedtomonitorthelinearitypropertiesoftheCCD(seeSec.3.5).Onlyv-bandwasused

inthislinearitymeasurement,andonlyasection(128×128pixels)oftheFOVwasdownlinkedforexamination.

Figure3.28 showsanexample theONC-T images takenof theFF lampoutputon16October2017.The

brightness gradient in the vertical direction of the ul- and b-band images are characteristic of the FF lamp

properties,due to theFWroundaboutstray light.Figure3.29examines thebrightnessgradient in thevertical

directionintheFOVforallfiltersbydisplayingtheratiooftopquarteroftheimage(#1)andthebottomquarterof

theimage(#4).Thereasonforthestronggradationattheshortwavelengthscanbeexplainedasfollows.TheFF

lampradianceisclosetoa2000Kblackbodyradiatorthathaspeakemissionatlongwavelengths.Forexample,

411.8 618.9 1.022 1.030 (toolow) 1.008

Zeta

282.4 957.0 0.985 1.005 0.999 0.971

283.3 672.4 1.003 1.016 1.015 1.002

38.6 958.1 0.993 0.996 0.921 0.911

607.4 956.2 0.993 1.009 0.972 0.974

Standard

deviation(%) 1.6 3.9

6 1055 52 364

theenergyfluxfromtheFFlampatthep-bandwavelengthrangeisabout500timeslargerthanthefluxattheul-

bandwavelength. Therefore, the FW roundabout stray light is a negligible percentage of the transmitted light

throughthelongerwavelengthfiltersbutisameasurablepercentageofthetransmittedlightthroughtheshort

wavelength (e.g., ul- and b-band) filters. The difference between segments #1 and #4 are due to the spatial

distributionbiasoftheFWroundaboutstraylight.ThisFWroundaboutstraylight isnotexpectedtoaffectthe

observationsofRyugu,sincethereflectancespectrumofRyuguisrelativelyflatandthesolarradiancespectrum

hasapeakatthev-bandwavelengthrangeofonly~2timeslargerthanul-bandwavelengthrange.

AveragesignalvaluesoftheFFlampimagesareusedtomonitor thetimevariationsintheCCDresponse.

Figure3.30showsthetimevariationofthecountfluxoftheFFlampimagesobtainedduringthecruisephase.The

FFlamphastwomodesofvoltagesettingswhichisselecteddependingonthesensitivityoffiltersusedtoimage

thelamp.Figure3.30(a)showsthe“Highvoltagemode”data(usedfortheshorterwavelengthfilters:ul-toNa-

bands),andFig.3.30(b)showsthe“Lowvoltagemode”data(usedforthelongerwavelengthfilters:w-top-,and

wide-bands).Toshowrelativesensitivitychanges,themeasuredoutputoftheFFlampisnormalizedtotheoutput

measuredon16April2015.Notethatthedataacquiredon8July2017areexcludedfromtheplot linesdueto

possiblecontaminationbyradiatorstraylightaffects(Sec.3.7).

Figure3.30(a)indicatesthatthesensitivityoftheshorterwavelengthfilters(ul-toNa-bands)decreasedafter

theinitialcheckout.Figure3.30(b)alsoindicatesthatthesensitivityofthelongerwavelengthfilters(w-through

p-andwide-bands)decreasedaftertheinitialcheckout.However,wecannotconcludethedegradationisrealdue

to the unstable emission from the “Low voltagemode” of the FF lamp used for the preflight longwavelength

measurements.Sincethe“Highvoltagemode”wasmorestableduringthepre-flightmeasurements,wearelimiting

ourdiscussiontotheul-to-NabandswhichwerecharacterizedusingtheFFlamp“Highvoltagemode”.Thereare

twopossibleexplanationsforthedecreasingtrendinthe“Highvoltagemode”data;1)degradationoftheFFlamp,

e.g., low electric current and 2) degradation in the transmittance of the filters due to exposure to the space

environment. During the initial checkout, the instrument temperatures were slightly higher than those of

subsequentobservations(Table3.6).Duetothesetemperaturedifferences(FFlampandCCD),weinter-compare

theobservationsacquiredafter16April2015,whichwereobtainedundersimilartemperatureconditions.After16

April2015,theFFlampfluxismorestable,decreasing~1%fortheb-to-Nabandsanddecreasing~2.5%fortheul-

band. This demonstrates that the ul-bandhas a steeper decrease in sensitivity than the other three bands. To

investigatethereasonforthedecreaseinsensitivity,wecomparethelocalmeasuredoutputchange.Figure3.29

showsthemeasuredoutputchangeforsegments#1and#4.Althoughsegment#1ismorecontaminatedbytheFW

roundaboutstraylight,thedecreasingrateinthemeasuredoutputisalmostsame.Thus,thisdecreaseinsensitivity

ismorelikelycausedbydegradationoftheFFlamp.

Sincethemainpurposeofthis5December2017observationwastoexaminethelinearityresponse,onlythe

ONC-Tv-filterwasused.Additionally,toreducethedatadownlinksize,onlyaportion(128x128pixels)oftheFOV

6 1055 52 364

was downlinked, using the region of interest (ROI) function form the on-board processing algorithms. Time

variation in the FF lamp output in this area of the CCD are plotted in Fig. 3.31. Although the temperature

environment (Table 3.6) during this observation is very similar to the environment of 16 October 2017

observation, there isa small increase (~0.4%) fromthe16October2017observation. It is suspected that this

randomtrendmaybeduetoslightstabilitiesintheFFlampoutput.

Insummary,weconcludethatthecountfluxfromtheFFlamphavealong-termdecreasingtrend(~1–2.5%)

between16April2015to5December2017.Therearetwoplausiblecandidatesforthedecreasedcountflux:(1)

degradationoftheFFlamp,and(2)degradationofthefiltersunderthespaceenvironment.Toresolvewhetherthe

sensitivitydegradationintheCCDisrealornot,additionalobservationsareneeded;especially,repeatcalibration

observationsusingstarsthathavealreadybeenobservedbytheONC.

Table3.6.SummaryoftheFFlampobservationsheldafterlaunch.

DateObservation

purposeNote Temperature[℃]

ONC-AE ONC-TCCD

Start End Start End

11December2014Initial

checkout7band+wideband -4.49 -4.10 -14.79 -14.79

16April2015Health

checkout7band+wideband -6.46 -6.20 -29.56 -29.54

10September2015Health

checkout7band+wideband -6.95 -6.46 -29.46 -29.43

8July2017Health

checkout

7band+wideband.

SolarStraylightwas

contaminated

-7.05 -7.02 -28.77 -28.77

16October2017Health

checkoutONC-T(7band+wide) -5.77 -5.77 -28.59 -28.91

5December2017Linearity

observation

ONC-Tvband.Asmall

partoftheFOViscutout.-5.77 -5.77 -28.72 -28.72

6 1055 52 364

Figure3.28.ExamplesoftheFFlampimages.Theseobservationsweretakenon16October2017.

Figure3.29.VerticalbrightnessgradientsinanFFlampimage.Theobservationwastakenon16October2017.

Theimagewasdividedinto4segmentsintheverticaldirection,andtheaveragesignalratiobetweensegment#1

(topquarterofanimages)andsegment#4(bottomquarter)asafunctionofwavelengthisshown.TheONC-Twide

banddataisplottedat550nmforreference.

6 1055 52 364

Figure3.30.ThetimevariationintheFFlampcountfluxduringthecruisephase.(a)ul-toNa-banddatawere

obtainedinthe“Highvoltagemode”.(b)w-top-,andwide-banddatawereobtainedinthe“Lowvoltagemode”.

Figure3.31.TimevariationoftheFFlampcountfluxinthev-band.Square:FullFOVdata(sameasFig.3.30).

Cross:ROIsub-frame.Thedatapointsarenormalizedtothe16April2015observation.

3.9.2.SensitivityDependencefromCCDtemperatureBasedonJupiterObservations

ThespectralsensitivityoftheONC-Twasmeasuredusinganintegratingsphereatroomtemperaturepriorto

launch (Kameda et al., 2017). The sensitivity at cold temperatureswas estimated using the typical sensitivity

dependencefromCCDtemperatureprovidedbythemanufacturer(E2V),andwasdemonstratedtobeconsistent

6 1055 52 364

with the SP/SELENE lunar reflectance model (Suzuki et al., 2018). In this section, we examine the Jupiter

observationstakenbytheONC-TwithdifferentCCDandELEtemperaturestoprovideamodelofthesensitivity

dependencefromhardwaretemperatureseachfilterband.

WeobservedJupiterusingONC-TfromMay16–17,2017.TheCCDtemperaturewascontrolledbytheheater

torangebetween-29℃to+25℃andtheELEtemperaturefrom-10℃to0℃(Table3.7).BecauseofJupiter's

non-uniformity of reflectance, wemeasured the Jupiter’s brightness over 4 rotation periods under a different

temperatureforeachperiod,foratotalof4rotations.Figure3.32showsthenormalizedfluxdependencytothe

CCDtemperature.TheeffectoftheCCDtemperatureismuchlargerthanthatofELEtemperatureespeciallyforthe

longwavelengthfilterbands.AstheCCDtemperatureincreases,thesensitivitiesoftheul-tow-bandsdecrease,

whilethesensitivitiesofthex-andp-bandsincrease.Weassumedthatthesensitivity,Sn,canbeexpressedasa

functionoftheCCDtemperature,TCCD,T;

\1 = \É,[0∆PPQ,[0OPPQ,1 + 30; + 1;, (3.20)

whereS0,n is thesensitivityforn-bandat the referencetemperature OPPQ,1 = −30℃ andan is the temperature

dependencefactordeterminedbyχ2-fitting.ThesensitivityofeachbandatanytemperaturesofTCCD,T(-30℃ to

+25℃)canbecalculatedbyEq.(3.19)andthevalueslistedinTable3.8,wherethe2-σerrorisbasedondeviations

ofthemeasurementsfromEq.(3.19).Similarly,theELEtemperaturedependencecanbeexpressedas(Table3.9):

\′1 = \É,[0∆R$R,[(OR$R + 10) + 1;. (3.21)

ThereferenceELEtemperatureis-10℃.Notethat,theELEtemperaturedependencesforul-to-Nabandsarenearly

0withinthe2-σerrors,showingnodependenceontheELEtemperatureintheCCDandfilterresponse.However,

w-to-pbandsshownoticeabledependencesfromELEtemperature.TheCCDtemperaturehasthedominanteffect

ontheCCDandfilterresponses.Thus,weuseEq.(3.19)toquantifytheCCDsensitivityatCCDtemperatureranging

from-30℃to25℃.AlthoughthedependencefromELEtemperaturemightbeobservedforlongerwavelengths,we

do not apply ELE temperature correction for first products becausewe do not have physical explanations of

dependencedifferenceinfilters,whilethedependencefromCCDtemperatureforeachbandshouldbedifferent

duetothetemperaturedependenceofquantumefficiencyindifferentwavelengths.ToapplytheELEtemperature

correction,furtherinvestigationisneeded.Forexample,wewillbeabletotestthecorrectiontoimagesobtained

during close approaches. While this does not account for sensitivity variations caused by other hardware

temperatures,forexample,theerrorintroducedbytheELEtemperatureeffectissmall(<0.1%/℃fortheul-to-Na

bandsand<0.3%/℃forthew-to-p-band).

6 1055 52 364

Figure3.32.ObservationsofJupiterusingtheONC-Tforeachband(ul-p).Thesymbols,�,△,□,and×,indicate

thenormalizedsignalrate(DN/s)forrotationalperiods1,2,3,and4,respectively.Theblacklinesshowtheresults

ofχ2-fittingwiththegraylinesof1-σerrors.

Table3.7.CCDandELEtemperaturesduringtheONC-TJupiterobservations.

Phase# CCDTemp.

[℃]

ELETemp.

[℃]

DateinMay,2017 Starttime,UT Endtime,UT

1 21.04 -8.84 16 12:30 12:33

2 21.05 -8.32 16 15:00 15:03

3 -27.94 -9.89 16 17:30 17:33

4 -28.20 -9.89 16 20:00 20:03

6 1055 52 364

1 -27.94 -9.89 16 22:30 22:33

2 -28.06 -9.89 17 1:00 1:03

3 -8.66 -9.37 17 3:30 3:33

4 -9.18 -9.37 17 6:00 6:03

1 11.93 -8.84 17 8:30 8:33

2 11.41 -8.84 17 11:00 11:03

3 25.11 -8.32 17 13:30 13:33

4 25.22 -8.32 17 16:00 16:03

1 -29.24 -3.63 17 18:30 18:33

2 -28.46 -3.63 17 21:00 21:03

3 -28.72 0.19 17 23:30 23:33

4 -28.77 -0.03 18 2:00 2:03

Table3.8.SensitivitydependencefromCCDtemperatureforallband-pass-filters.Theerrorscitedarethe2-σ

errors.

band ∆PPQ,[[/℃]

ul(0.40µm) −0.001449 ± 0.000244

b(0.48µm) −0.000968 ± 0.000108

v(0.55µm) −0.000814 ± 0.000090

Na(0.59µm) −0.000866 ± 0.000154

w(0.70µm) −0.000355 ± 0.000112

x(0.86µm) 0.001771 ± 0.000158

p(0.95µm) 0.004201 ± 0.000202

Table3.9.SensitivitydependencefromELEtemperatureforallband-pass-filters.Theerrorscitedarethe2-σ

errors.

band ∆R$R,[[/℃]

ul(0.40µm) −0.00077 ± 0.00084

b(0.48µm) −0.00055 ± 0.0078

v(0.55µm) −0.00085 ± 0.0072

Na(0.59µm) −0.00049 ± 0.00058

w(0.70µm) −0.00118 ± 0.00034

x(0.86µm) −0.00113 ± 0.00086

6 1055 52 364

p(0.95µm) −0.00288 ± 0.00094

3.9.3SensitivityPredictionBasedontheHardwareSpecifications

TheDNderivedfromtheONC-TCCDcanbemodeledbyusingthehardwarecomponentspecifications,such

astheCCDquantumefficiency,filterandlenstransmittances.SuchmodelequationsfortheONC-W1and-W2are

describedinEqs.(2)–(4)ofSuzukietal.(2018).Inthissection,wequantifytheONC-Tsensitivitybasedonthe

specificationsofthehardwarecomponents.

WeformulatethattheexpectedsignalIsys,n[DN/s]fromONC-Tisgivenby:

GN¤N,[ =‹

›fifll‡ µ·

ûopt·‚∫

t

‹o(a)„£ì"(a, ∞, ó, ‰)aΦ’0a,OÆÆì,O;ra, (3.22)

whereFopt is the effectiveF-number of the optics (= 9.05),p is the CCDpixel size (=1.33×10-5 [m]), J(λ) is the

irradianceoftheincidencelight[W/m2/μm]ontheobservationtarget,RADF(λ,i,e,α)istheradiancefactorI/F

(dimension-less)oftheobservationtargetataspecificsetofilluminationandviewinganglegeometries(i,e,α;

incidence angle, emission angle, phase angle). The two π symbols are not canceled in this equation to avoid

confusionaboutthephysicalunits,sincethelatterπhasaunitof[sr].

Ontheotherhand,themodelradianceJnforthen-bandcanbeformulateasfollows:

o[ =∫aÂe(d)Ê_Qû(d,L,s,Á)Φ’0a,OÆÆì,O;gd

∫ aΦ’0a,OÆÆì,O;gd. (3.23)

FromEqs.(3.22)and(3.23),thepredictedmodelsensitivitySsys,ncanthenbedescribedas

\N¤N,[ = GN¤N,[/o[ . (3.24)

Table3.11showsthevaluesofthepredictedmodelsensitivityderivedfromEq.(3.24).Toderivethesevalues,we

assumethesolarirradiancespectra(ASTM-E490)forJ(λ),aflatI/FspectraRADF(λ,i,e,α)equalto1andtheCCD

TemperatureTCCD,Tequalto-30[℃].

3.9.4SensitivityCalibrationBasedonStarObservations

AmbiguitiesbetweenthepreflightCCDsensitivity(measuredinthelaboratorybyKamedaetal.(2017))and

theinflightCCDsensitivity,duetotemperaturedifferencesbetweenpreflightandinflightmeasurements,prompted

furtherCCDsensitivitycharacterizationsunderinflighttemperatureconditions.Tomeasuretheabsolutesensitivity

inflight,weobservedeightstars,withV-magnituderangingfrom2.02to4.73,duringthecruisephase.Theobserved

targetsaresummarizedinTable3.10.Thosestarswereselectedbasedonspacecraftattitudeconstraints.The

sensitivitycanbeobtainedfromthelinearrelationshipbetweenthephotonfluxandthedigitalcountobserved

withineachfilter.Forthereferencestarspectra,weusedcatalogsbyHamuyetal.(1992,1994),Alekseevaetal.

(1996),andBurnashev(1996)sincewecouldnotfindasinglecatalogthatincludedallourobservablestars.μAqr,

6 1055 52 364

φSgr,andτSgrhavenotbeenobservedinwavelengthslongerthan750nm,thus,thesensitivitiesofx-andp-filters

areconstrainedwithmeasurementsfromonlyfivestars.

Although these different catalogs used the same calibration target α Lyr (Vega), they used a different

reference spectrum for α Lyr, which contributes non-negligible systematic errors. Thus, we first derive the

conversion factors to cancel the systematic errors caused by the reference spectrum differences. Using these

conversionfactors,thestarenergyfluxbyAlekseevaetal.(1996)weremodifiedtobeconsistentwiththatofHamuy

etal.(1992,1994)inallfiltersotherthanthep-filter.Notethatweusetheoriginalfluxvaluesforthep-filterfrom

allcatalogssincethep-filterbandpasspartlycontainsthetelluricwaterfeaturebetween930–970nm,causing

particularlyhigherspectralerrorsasdiscussedinHamuyetal.(1992,1994).Burnashev(1996)usedthesame

referencespectrumofαLyrasHamuyetal.(1992,1994)andthereforethestellarspectrafromthiscatalogdidnot

requirere-standardizationtoacommonreferencespectrum.Usingthesespectra,theexpectedfluxJstar,nthrough

eachofthefiltersisobtainedbyusingEq.(3.9).ThedigitalcountsfromthestarobservationsobtainedbyONC-T

arecalculatedasdescribedinSec.3.8.Therangeofsensitivitiesderivedfromthestarobservationscorrelateto

within 5%, except for ul- and p-bands (Fig. 3.33).Note that random scatter between the stars fromdifferent

catalogssuggeststhatthesystematicerrorbetweenthecatalogshasbeenproperlyremoved.Figure3.34displays

therelationshipbetweenthestarfluxesanddigitalcounts.Thesensitivitiesforthefilterswereobtainedusinga

weightedleastsquarelinearregressionfitofthe5or8pointslistedinTable3.11.Theerrorsarederivedbasedon

the 95% confidence level. These deviations are consistent within the errors in the star observations of ~1%

(average)givenbyHamuyetal.(1992,1994)and<3%(90%confidencelevel)fortheb-tox-bandsbyAlekseeva

etal.(1996).Thelargedeviationinthep-filtersensitivitymaybeduetotheloweraccuracyofreferencespectra,

whichhas5–10%errorintheHamuyetal.(1992,1994)and<5%errorintheAlekseeva(1996)spectra.

Forcomparison,thesensitivitybasedonthehardwarespecificationsfromSec.3.9.3isshownalsoinTable

3.10.Eventhoughthereareabsolutedifferencesbetweenthetwosensitivities,ontheorderof~17%,therelative

sensitivitiesbetweenthedifferentfiltersareconsistentexceptforthep-filter(Fig.3.35).Theabsolutesensitivity

differencebetweenthehardwarespecificationandstarcalibrationmeasurementscanbeascribedtosomefluxloss

intheoptics,buttheconsistencyintherelativevaluesindicatethatthetransmittancesforthefilterswerewell

definedpreflight.Thiscomparisonalsosuggestsanotherradiometriccalibrationisneededforthep-filter.Inthe

next section, we update the p-filter sensitivity based on ONC lunar observations. We conclude that the CCD

sensitivityderivedfromthestarobservationsisconsistentwiththepredictionfromhardwarespecificationsexcept

forthep-filter.

Table3.10.Listofobservedstarsduringthecruisephase.

HRNo. Observed

Date

Vmaga) Spectral

Typea)

Filters Referencespectra

6 1055 52 364

ζPeg 8634 2016-10-19 3.40 B8V ul,b,v,Na,w,x,p Hamuyetal.(1992,1994)

θCrt 4468 2017-05-23 4.70 B9.5Vn ul,b,v,Na,w,x,p Hamuyetal.(1992,1994)

εAqr 7950 2017-10-10 3.77 A1V ul,b,v,Na,w,x,p Hamuyetal.(1992,1994)

μAqr 7990 2017-10-10 4.73 A3m ul,b,v,Na,w Burnashev(1996)

σSgr 7121 2017-10-12 2.02 B2.5V ul,b,v,Na,w,x,p Alekseevaetal.(1996)

ζSgr 7194 2017-10-12 2.60 A2III+A4IV ul,b,v,Na,w,x,p Alekseevaetal.(1996)

φSgr 7039 2017-10-12 3.17 B8III ul,b,v,Na,w Alekseevaetal.(1996)

τSgr 7234 2017-10-12 3.32 K1+IIIb ul,b,v,Na,w Alekseevaetal.(1996)

a) HoffleitandJaschek(1991)

Table3.11.Sensitivityfactorsatthereferencetemperature(TCCD,T=-30℃).

Filtername SensitivityatTCCD,T=-30℃(DN/s)/(W/m2/µm/sr)

starobservations hardwarespecifications

ul(0.40µm) 439.1±2.2 547.3

b(0.48µm) 969.3±7.9 1104.6

v(0.55µm) 1175.0±10.0 1371.9

Na(0.59µm) 546.9±2.0 642.3

w(0.70µm) 1515.0±19.3 1728.4

x(0.86µm) 1499.8±24.6 1736.1

p(0.95µm) 1033.0±71.1* 1023.1

*forthep-filterwewillusethevalue961.2±28.8(DN/s)/(W/m2/µm/sr)fortheradiometriccalibrationpipeline

basedonthelunarobservations(SeeSec.3.9.5).

Figure3.33.Thesensitivitydeviationmeasurementsfromthestellarobservations.Thesensitivitiesarenormalized

6 1055 52 364

bytheaveragesensitivitywithineachband.

Figure3.34.StarfluxesanddigitalcountsobservedbytheONC-Tfilters.Thedottedlinesindicatelinearregression

linesandtheslopescorrespondtothesensitivitiesofthefilters.

Figure3.35.Sensitivitycomparisonsbetweenthosebasedonthehardwarespecifications(dottedline)andthose

basedonthestellarobservations(solidline).

3.9.5.Jupiter,Saturn,andMoonObservations

Here,weevaluatetheCCDandfiltersensitivitiesbasedontheONC-TobservationsofJupiter,Saturn,andthe

Moon. Comparisons with well-known spectra are useful for evaluating whether the derived sensitivities are

accurate.AlthoughthelunarobservationshavealreadybeenanalyzedbySuzukietal.(2018),thesensitivitieshave

beenupdatedandrequirere-evaluation.Thus,inordertoevaluatetheupdatedsensitivities,werevisittheONC-T

lunar observations. Comparisonwith lunar observations is awell-established technique for evaluating camera

radiometriccalibrations.TheJupiterandSaturnobservationsareanalyzedinasimilarmannerasdescribedinSec.

3.9.2forthecharacterizationoftemperaturedependencies.

6 1055 52 364

Figure 3.36 shows the comparison between spectra obtained by the ONC-T on 5 December 2015 with

SP/SELENE(Kouyamaetal.,2016)andWAC/LROC(Satoetal.,2014)spectralmodels.Notethatbothmodelsare

calibratedtofittheRoboticLunarObservatory(ROLO)spectralmodel(KiefferandStone,2005).Comparisonsof

theabsoluteradiancecomparisonshowtheradiancebasedontheupdatedONC-TCCDsensitivityexceedsthat

predicted from the lunarmodels.On the other hand, the relative spectral radiance is consistent (within 2.5%)

betweenthemodelsandtheONC-TobservationsTheROLOmodelhasbeenshowntohavea5-10%uncertainty

initsabsolutevaluedetermination,whereastherelativespectraluncertaintyisontheorderoflessthan3%.

TherelativeuncertaintyisbasedontheROLOmodelcomparisonwithlunarbrightnessmeasurementstaken

with Moderate Resolution Imaging Spectroradiometer, which is a well calibrated Earth observing sensor

onboardAquaandTerra(Stone,2008).Thus,theupdatedsensitivityreproducesthelunarradiancespectrumto

withintheambiguityinthemodels.Nevertheless,wecanupdatethesensitivityconversionofthep-filterbasedon

thelunarobservationssincethelunarreferencemodelsaremoreaccuratethanthestellarobservationsatthis

wavelength.Thesensitivityderivedfromthelunarobservationsforthep-filter’snormalizedradiancesareidentical

in both the SP/SELENEmodel and theONC-Tmeasurement.The sensitivity for the p-filter based on the lunar

observations is 961.2±28.8 (DN/s)/(W/m2/µm/sr). This valuewill be used for product generation during the

mission’srendezvousphase.

Figure3.37displaysthenormalizedradiancespectraofJupiterandSaturnatTCCD,T=-29to-27℃comparedto

ground-based observations in 1995byKarkoschka (1998). Karkoschka (1998) observed Jupiter and Saturn at

phaseanglesof6.8˚and5.7˚,respectively,whiletheONC-Tobservedtheseobjectsat3.9˚and4.9˚,respectively.The

JupiterspectrumobtainedbytheONC-Tshowsoverlapwiththeground-basedobservationwhichhas~4%oferror.

Moreover,althoughthesensitivitybasedonthestellarobservationcanreproducetheradiancespectrumtowithin

theambiguityintheobservation,thespectrareproducedbythesensitivitybasedonthelunarobservationsshows

betterconsistencywiththeJupiterground-basedobservation.However,theSaturnspectraacquiredbytheONC-T

shows non-negligible differences between the x-band observations and the ground-based measurements by

Karkoschka (1998). Because the x-band wavelength coincides with a known methane absorption, a higher

reflectancefromtheringscomparedwithSaturn’sdisk(e.g.,Irvine,1971)isexpected.Dependingonthearearatio

oftheringsrelativetothediskintheONC-Timages,thex-bandcouldbebrighterthanthedisk-onlyspectraof

Karkoschka (1998). This explains the variations between the ONC-T x-band and the ground-based Saturn

observations.Moreover,thebluerspectralpropertiesoftheringsmayalsoexplainthedarkerbrightnessseenin

theul-bandcompared to theground-basedmeasurements.Thesespectralcomparisonsreasonably confirm the

ONC-T’sspectralcalibration.

6 1055 52 364

Figure3.36.ThesegraphscomparetheMoon’sradiancebetweenthatobservedbytheONC-Tandthosepredicted

bylunarspectralmodels.(left)Thedisk-averagedradiance(i<60˚ande<45˚),(center)radiancenormalizedtothe

v-band,and(right)ratioofthenormalizedspectra(showninthecentralgraph)totheONCnormalizedobservation.

GraysymbolsaretheresultsbasedonthesensitivityinSuzukietal.(2018).Thesensitivitiesforallbandsderived

fromthestellarobservationsareshownbythedottedredline.Theupdatedsensitivityforthep-bandderivedfrom

comparisonsbetweenthenormalizedradianceofONC-TandtheSP/SELENEmodelradianceisshownbythesolid

redline.

Figure3.37.NormalizedspectralcomparisonsofJupiterandSaturnobservations.(left)Comparisonofnormalized

radiance between spectra obtained byONC-TandKarakoschka (1998).Dotted lines indicate spectra produced

based on the sensitivity from the stellar observations, and solid color lines indicate those based on theMoon

observations.(right)TheJupiterandSaturnspectraofONC-TdividedbythespectraofKarakoschka(1998)are

shown.TheEarth-basedJupiterspectraandthosefromtheONC-Tareingoodagreement.Thebrightnessinthex-

bandandthedarknessinul-bandbytheONC-TinthespectraofSaturncanbeexplainedbythedifferencesinthe

spectralpropertiesbetweentheringandthedisk(e.g.,Irvine,1971).

6 1055 52 364

4.EvaluationoftheUpperLimitofaDetectableSodiumAtmosphere

Sodium(Na)isoneofthemostcommonelementalspeciessurroundingsolarsystembodies,duetoits

highabundanceandvolatility.Forexample,sodiumwasthefirstmetalliccomponentdetectedaroundMercury

byground-basedtelescopes(PotterandMorgan,1985).SodiumwasalsodetectedaroundtheMoon(~5kR)

andcomets(1kR~900MRdependingonitsdistancetoSun)throughremoteobservationusingtheresonant

scatteringofsolarlightatthesodiumemissionwavelengthsof589.594nm(D1)and588.997nm(D2)(Rahe

etal.,1976;PotterandMorgan,1988;Leblancetal.,2008).Sodium,ifpresentaroundRyugu,isexpectedtobe

observablebytheONCusingtheNa-bandfilter.Thedensitydistributionanditsseasonalvariationasafunction

ofphaseangleororbitalposition(season)isanimportantvariableinunderstandingtheformationofC-type

asteroidsandthegenerationprocessesofplanetaryatmospheres.Thissectionevaluatestheupperlimitfor

thedetectionofsodiumbytheONC-T,basedonitsobservationsofJupiter’ssodiumnebula.

Methodoutline

ThesodiumemissioncanbedetectedwiththeNa-bandfilteronONC-T.Inordertoeliminatethestray

lighteffectandfluctuationsinbackgroundnoise,animagetakenconcurrentlywiththev-bandfiltershouldbe

subtracted from the Na-band image. This eliminates the stray light component, as its contribution is

independentof theobserving filter.Therefore, thesodiumobservationsequence includesthe imagestaken

alternatelybetweentheNa-andv-bandfilters.Inaddition,thebackgroundnoisecanbeeliminatedbytaking

theratioofbackgroundcount(CNa/v) imagesof“sky-data” taken inboththeNa-bandandv-band,acquired

betweentheNa-(INa(H,V))andv-band(Iv(H,V))imagesofthetarget.Then,thereducedNa-bandimageofthe

targetiscalculatedasfollow:

Target(H,V)=INa(H,V)–Iv(H,V)×CNa/v (4.1)

Datareduction

ItisknownthatJupiterissurroundedbysodiumtorusoriginatingfromitsmoonIo.Thebrightnessofthe

D1andD2linesfromthissodiumtorusareobservedtobe20-80R(e.g.Yonedaetal.,2015).During18to20

May2017,ONCsetitstargetforthewestsideofJupiter’ssodiumtorusandacquired183setsofimages(Na-

and v-bands) continuously. Each imagewas takenwith a 178 s integration time period. The geometrical

configurationoftheobservationsisshowninFig.4.1.Eventhoughwedidnotdetectsignificantcountsfrom

sodiumaroundJupiter,theupperlimitevaluationcanstillbederivedusingthisdataset.

UpperlimitforNadetection

Asafirststep,weidentifyandremoveanyhotpixelsusingthe“darkimage”derivedinSec.3.4.Anypixels

withhighdarkcount� rates(>1-σ)areeliminatedfromthesodium-detectionprocessing.Furthermore,pixels

with higher counts (>3-σ)are alsoeliminated from theprocessing aspossible cosmic ray hits. The image

analysisprocessdescribedintheEq.(4.1)isthenadoptedforthe183imagesets.

TheratioCNa/viscalculatedtobe0.9961usingthemedianfromtheprocessedskyimages(Na-andv-

6 1055 52 364

band)takenon28April2017.NineROIsarethendefinedasshownintheFig.4.2foreachimage.Inorderto

evaluatesizeeffects,threetypesofROIs(setA,B,andC)aredefined.Thestandarddeviationincountsforeach

pixelintheROIiscalculated.ThisvalueisdefinedastheupperlimitofcountsforsodiumdetectionbytheONC

(1-σ).Integratingoverthemultipleimagesstatisticallyreducestheupperlimit.Asaresult,anupperlimitto

theNabrightnessdetection,asafunctionofthenumberofintegratedimages,canbecalculatedandisshown

inFig.4.3.Ifmorethan100imagesetsareintegrated,thentheupperlimitisforsodiumdetectionisaround

100R. In this calculation the count-to-brightness conversion factor is assumed to be 21R/DN/pix (from

Kamedaetal.,2017).

TheONC-Tobservationsduring18to20May2017didnotdetectasignificantcountfromsodiumaround

Jupiter.This isnot surprising since thebrightnessof thesodiumcloud isknowntobe lessthan80R(e.g.

Yonedaetal.,2015),whichislowerthanthedetectionlimitfortheONC-T.

ThebrightnessaroundRyuguisexpectedtobe10-10skR,althoughtheestimationisstronglydependent

ontheoutgasrateassumptionbasedonthesurfaceandthesodiumg-factor.Whenweassumetheoutgasrate

atRyugutobesameasEarth’smoonorMercury,thebrightnessisextremelylow(10-100R)whichisdifficult

todetectbytheONC-T.Ontheotherhand,ifweassumetheoutgasratetobethebrightnessofseveral10skR

whichevencorrespondstomuchloweroutgasrateoncomets,thesodiumatmosphereisdetectablebythe

ONC-Twithasingleimageset.Itshouldbenotedthatthebrightnessestimationalsodependsontheorbitof

theasteroidbecauseoftheDopplershiftinthesolarlightspectrum,whichisthesourceofthesodiumresonant

scattering(Killenetal.,2009).

Figure 4.1. This image shows the geometry for the Jupiter’s sodium cloud observations

during18to20May2017(Na-bandimage).Jupiteranditsnorthpoleisindicatedbythewhite

6 1055 52 364

circle and yellow allow, respectively. The known stars are indicated by the white broken

circles.Thecolorgradationiscausedbytheradiatorstraylight.

Figure4.2.AnexampleoftheROIregiondefinitionsforeachset.Thestandarddeviationsfor

countsineachROIwereevaluatedforthe183imagesets.

Figure4.3.TheupperlimitofthebrightnessdetectionforsodiumbytheONCasafunction

ofnumberoftheintegratedimagesets.NocleardependencyontheROIsizeisseen.

6 1055 52 364

5.InflightCalibrationoftheONC-W1and-W2

Bothpreflightandinflightcalibrationsoftheflat-fields,PSFs,sensitivities,andstraylightcontributions

aresummarizedinthissectionfortheONC-W1andONC-W2.Hayabusa2proximityoperations,suchastouch-

downs,lander-releasing,andgravitymeasurements,willbeconductedprimarilywiththeONC-W1sincethe

longfocallengthoftheONC-Twillbeoutoffocusatpositionsclosertothesurface.Duringtouch-downs,ONC-

W1willtakesequentialimagesuptoaspatialresolutionof~1.5mm/pixel.Thus,thecalibrationsoftheONC-

W1andW2areasimportantasthecalibrationoftheONC-T.Thecalibrationsconductedbothpre-flightand

inflightaresummarizedinTable5.1.

Table5.1.SummaryofthedatausedforthecalibrationsoftheONC-W1and-W2.

Groundmeasurement Inflightmeasurement InflightObservationDate

Bias 0-secexposures 0-secexposures Continuouslyduringthecruisephase

DarkcurrentandHot

pixels

Thermal vacuum test

(variation of CCD, ELE,

andAEtemperatures)

Deep sky observation with

variationofCCDtemperatures

30November2017,2December2017

Sensitivity( ) Integratingspheres StarsforONC-W2andJupiterfor

ONC-W1

9to15February2016,

19Februaryto21March2016,

28June2016,30June2016,

2July2016,17October2017,

30November2017

Flat-Field Flatpanel Star observationwith variation

ofS/Cattitudes

12March2017

Same as images for Sensitivity

calibration

Smear 0-secexposures EarthforONC-W2 4December2015

Geometricdistortion Dotted pattern (Suzuki

etal.,2018)

Stars(Suzukietal.,2018) W1:19February2015

W2:12March2017

AlignmenttotheS/C N/A Stars W1:19February2015

W2:11December2014

SharpPSFs Pinhole Stars, and Mars, Jupiter, and

Saturn

31May2016,9to15February2016,

19Februaryto17March2016,

28June2016,30June2016,

2July2016,30June2017,

6 1055 52 364

17October2017,30November2017

Straylights N/A Straylightswithvariationofthe

S/Cattitudes.

11December2014,

19February2015,

26February2015,23June2015,

12to17October2015,

9November2015,

14November2015,

22December2015,

9Februaryto17March2016,

21March2016,7June2016,

28Juneto23July2016,

12March2017,13to20June2017,

17to21October2017

5.1BiasCorrection

Weevaluatedthebias levelof theONC-W1and-W2as functionsofCCDandONC-AEtemperatures.We

analyzed0-secondexposureimagestakenduringpreflightthermalvacuumtestsandthecruisephase.Most

ofthedataacquiredduringthecruiseweretakenatthenominalCCDtemperature(-25℃),butattheendof

2017’s, we also acquired 0-sec exposure images at a CCD temperature of +20℃ to test the temperature

dependency.Figure5.1showsthebiaslevel(theaverageforallpixels)oftheONC-W1andW2asfunctionsof

theCCDandONC-AEtemperatures.Thebiaslevelobtainedduringtheinflightobservationsincreasedwith

CCDtemperature,similartotheinflightobservationresultsforONC-T(Fig.3.3)andpreflighttestsforONC-

W1and-W2.ThebiaslevelcanbeempiricallydescribedasafunctionoftheCCDtemperature, OPPQ,(ËtöwËu)

(℃),andAEtemperature, O_R (℃)usingtherelation:

GKLMN = (∆É + ∆tO_R + ∆uO_Ru )OPPQ,(ËtöwËu) + (ÈÉ + ÈtO_R + ÈuO_R

u ). (5.1)

OptimizedvaluesofthesixcoefficientsinthisequationarelistedinTable5.2.Figure5.2showscomparisons

betweenthisempiricalrelationshipandtheactualbiasmeasurements.WecanpredictthebiaslevelfromEq.

(5.1)within10%error.NotethataswasdiscussedinSec.3.2fortheONC-T,thebiaslevelmayalsodependon

thetemperatureoftheELEoftheONC-W1and-W2.WhentheELEtemperaturechangesdrasticallyduring

proximityobservations,theerrormightbegreaterandneedtobecorrectedforbydeductivemethodsbased

ontheONC-T.

Figure5.3showstime-dependentchangeofthebiaslevelatCCDtemperaturesaround-25℃ usingthe

inflightdataacquiredafterlaunch.Wefoundthatthebiaslevelsforbothcamerasstaynearlyconstant,within

1%changeover3years,similarlytotheONC-T’sbiaslevel(Fig.3.5).

6 1055 52 364

Figure5.1.BiaslevelsfortheONC-W1(left)andtheONC-W2(right)asafunctionoftheCCDtemperature.

Figure5.2.Comparisons between theactual biasmeasurementsand thosepredicted fromEq. (5.1). Left

figureisfortheONC-W1andrightfortheONC-W2.

6 1055 52 364

Figure5.3.Thetime-dependentchangeinthebiasleveloftheONC-W1andONC-W2.

Table5.2.CoefficientsforEq.(5.1)fortheONC-W1andONC-W2.

ONC-W1 ONC-W2

ƃ 0.680 0.573

∆t −5.74 × 10,ú −2.95 × 10,ú

∆u 1.06 × 10,› 8.92 × 10,≠

ÈÉ 260 288

Èt −1.72 −1.65

Èu 8.50 × 10,ú 6.29 × 10,ú

5.2.DarkCurrentandHot-PixelCorrection

On30Novemberand2December2017,weconducteddeepspaceobservationsusingtheONC-W1and-

W2inordertoevaluatetheirdarkcurrentlevels.TwodifferentCCDtemperatures, −25℃ and +20℃, were

usedforbothcamerasduringtheseobservations.Wetook4-imagesetswithexposuretimesof5.57secatboth

CCD temperatures for ONC-W1, and 44.56 sec at O»»j,Íu = −25℃ and 16.8 sec at O»»j,Íu = +20℃ for

ONC-W2.Afew0-secexposureimageswerealsoobtained,whichwereusedtoanalyzethebiaslevels(seeSec.

5.1).Brightspotsduetocosmicrayswereremovedbytakingamedianofthe4images,andthebiaslevelwas

removed by subtracting the 0-sec exposure images. Signals due to stars were omitted by comparing two

medianimagestakenfromthedifferenttwodays.Unfortunately,wecannotdistinguishstraylightanddark

current from the images takenat O»»j,(ÍtÎHÍu) = −25℃.Thus,weestimatedanupper limit for thedark

currentusingasectionofthemedianimage,wheretheeffectsofthestraylightappeartobesmall(rectangle

area{[200,200];[800,600]}pixelsonONC-W1images,and{[100,100];[900,900]}pixelsonONC-W2images).

At O»»j,(ÍtÎHÍu) = +20℃,weevaluatedthedarkcurrentontheentireimagesbysubtractingthestray-light

componentsestimatedfromtheimagesat O»»j,(ÍtÎHÍu) = −25℃.

Figure5.4(a)showsahistogramofthedarkcurrentfortheONC-W1at O»»j,Ít = −25℃ and +20℃.

TheaveragesandthestandarddeviationsareshowninFig.5.4(b),asfunctionsoftheCCDtemperature.We

definehotpixelsaspixelswherethedarkcurrentislargerthan30DN/s.Figure5.4(c)showsthenumber

ratioofhotpixelstototalpixelsintheanalyzedareaoftheimage.Weconfirmedoneofthepixelsaregreater

than30DN/sat O»»j,Ít = −25℃.SimilardatafortheONC-W2areshowninFig.5.5.Weconfirmedthreeof

the pixels are greater than 30 DN/s at O»»j,Íu = −25℃ . The hot pixel ratio of the ONC-W1 at the CCD

temperatureof +20℃ seemstobelessthanthatoftheONC-W2,andcomparablewiththatoftheONC-T(Fig.

3.7).

WeareplanningtoobservetheasteroidsurfacecloselyusingtheONC-W1,downtoafewmetersaltitude

6 1055 52 364

duringthetouch-downsequence.WecanunderstandthedetailsofthephysicalpropertiesofRyugu’ssurface,

suchasregolithgrainsizeandmorphology,withanimageresolutionofafewmm/pixelatthetouch-down

sites.Duringtheproximityobservations,theCCDtemperatureispredictedtoriseuptoafewtensofdegrees

Catmost,leadingtohighlargedarknoiselevels,estimatedtobeabout10DN/sat O»»j,Ít = 20℃.Fortunately,

however,theaveragedarkcurrentisnegligiblecomparedwiththeexpectedsignalof~50,000DN/sfromthe

reflectionfromtheasteroidsurface.Nevertheless,thehotpixelsshouldbecarefullyexaminedduringanalyses,

since1%of the imageareahasdark currentvalueshigher than100DN/s,~0.2%of theasteroid surface

brightness(themaximumvalueisashighas680DN/s,~1.3%ofasteroidsurfacebrightness)atthehigher

CCDtemperature(Fig.5.4(c)).Figure5.6showsamapofsuchhotpixels,whichneedstoberecognizedand

accountedfor,especiallyinexaminationsofpixel-scalemorphology.

6 1055 52 364

Figure 5.4. (a) Histogram of the dark current level, (b) the averaged values as a function of the CCD

temperature,and(c)numberratioofthehotpixelsfortheONC-W1.

Figure 5.5. (a) Histogram of the dark current level, (b) the averaged values as a function of the CCD

temperature,and(c)numberratioofthehotpixelsfortheONC-W2.

6 1055 52 364

Figure5.6.Hotpixelmapof(a)theONC-W1and(b)ONC-W2attheCCDtemperatureof20℃.Thehotpixels

inthisimagearedefinedasthepixelswherethedarkcurrentislargerthan100DN/s.

5.3.CharacterizationofStrayLightintheONC-W1and-W2

WeakstraylightseeninimagestakenbytheONC-W1and-W2wereobservedduringthecruisephase.

SimilartothetestswiththeONC-T,wehavetakenimagesunderavarietyofspacecraftattitudeswithrespect

totheSuninordertoexaminethepatternsandtheintensityofthestraylight.Wefoundthatthepatternand

intensityofstraylightintheONC-W1and-W2dependsonspacecraftattitudeinasimilarfashionasseenin

theONC-T.

Figure5.7showsthetypicalpatternsofstraylightobservedintheONC-W1.StraylightintheONC-W1

haveaspotlight-likepattern,unlikethegradualintensitychangeintheFOVoftheONC-T.Asthephaseangle

increases(XPNL),theobservedstraylightalsoincreases.Theintensityofthestraylightisevaluatedusingthe

medianofan11x11pixelboxinthespotlight-likezoneinordertoremovetheeffectsduetocosmicraysand

stars.BecauseoptimalexposuretimesfortheONC-W1duringRyuguobservationswillbe<50ms,straylight

darkerthan20DN/scannotbedetected.TheareaaffectedbystraylightisdefinedastheregionintheFOV

withgreaterthan20DN/sintheimage.Figure5.8displaystherelationshipbetweentheintensity,thearea

affectedbystraylight,andthespacecraftattitude.Thespacecraftattitudeisdefinedinthesamemannerasin

theONC-Tdiscussion.Exceptforoneattitude,thestraylightwasweakerthan30DN/s,whichwillresultina

1.5DNfor50-msexposures.AsseenintheONC-T,thespot-likestraylightissmallerandweakeratlarger Ïùn$ .

Moreover,theintensityofthestraylightisusuallyweakerfortheONC-W1thantheONC-Tat Ìùn$ < 50° and

isnegligiblewhenthestraylightfortheONC-Tareavoidedbytwistingthespacecraftaroundthe ÓFP axis.

6 1055 52 364

Figure5.9showsthetypicalpatternsofstraylightobservedbytheONC-W2.StraylightintheONC-W2

changesgraduallyinFOV.NotethatbrightpartsareseenattheupperandlowerrightcornersoftheONC-W2

FOV.Theyare thepartsof the frameof theONC-W2reflectingthe light source.Weevaluated theaverage

intensityofstraylightfortheONC-W2,excludingtheframeparts(redsquareinFig.5.9).Figure5.10displays

therelationshipoftheintensityofthestraylightandthespacecraftattitude.Optimalexposuretimesforthe

ONC-W2observationsofRyuguisestimatedtobe<20ms,shorterthanfortheONC-W1.Thus,thestraylight

<50DN/swillnotbedetected in suchRyugu images.TheONC-W2 is foundnottohave severestray light

contributionswhen Ìùn$ > 0°, theSun light illuminatesthe+XSCplane,butstraylightcontributionsoccur

whenthe -XSCplane is illuminated.Duringtherendezvous-phaseobservations,Sun lightwillradiateonthe

+XSCpanelatthenominalattitude.Thus, stray light in theONC-W2 isexpectedtobenegligible inmostof

rendezvous-phaseimagesofRyugu.

Figure5.7.Stray lightpatternsat Ïùn$ = 0° withachanging Ìùn$ = 15°, 30°, and48° (fromlefttoright).

Thestraylightisobservedatthelowerleftcornersintheimageswithlarge Ìùn$ .

6 1055 52 364

Figure5.8.(left)TheintensityclassificationoftheONC-W1straylight.Thestraylightisweakerthan20DN/s

(blueandgreen)andwillnotbedetectedduetotheshortexposuretimesfortheONC-W1,<50ms.(right)The

arearatioclassificationoftheONC-W1straylight.

Figure5.9.StraylightpatternoftheONC-W2at (Ìùn$, Ïùn$) = (0°, 0°).Theredsquareindicatestheregion

wherethestraylightintensitywasassessed.

6 1055 52 364

Figure5.10.The intensityclassificationof theONC-W2stray light.Thestray light isweakerthan50DN/s

(blueandgreen)andwillnotbedetectedduetoshortexposuretimefortheONC-W2,<20ms.

5.4.Flat-fieldCorrection

The flat-fields of theONC-W1and -W2weremeasured by using the portable self-emissive flatpanel

(CABINCL-5300L),whichwasalsousedtomeasurethe flat-fieldofONC-Tpriorto launch inTanegashima

(launchsite)(seeSuzukietal.,2018).Inthissection,first,weevaluatetheuniformityinreflectanceoftheflat

panelusedforthepreflightmeasurement.Then,theeffectofemissionangle(e)fromtheflatpanelisdiscussed

andtheemissionangleeffect-correctedflatfieldsoftheONC-W1and-W2areshown.

5.4.1Flat-FieldsBasedonthePreflightCalibrationData

Figure 5.11 shows the radiance distribution of the panel taken by the calibration camera (Orion

StarShootG3(Monochrome))whentheemissionangleissettobezerotoshowthereflectanceuniformityof

theflatpanelcalibrationsource.Anedgeofluminousareawithadimensionof350x280mmisclearlyseen

intheleftfigure.TheONC-W1and-W2tookimagesofthispanelfroma30cmdistanceduringthepreflight

calibrationmeasurements.TheapproximateFOVoftheONC-W1and-W2,duringthecalibration,areshown

bytheareaenclosedbywhiterectanglesinthefigure.Sincethereflectanceuniformityofthepanelwithinthis

areaisessentialforaflatfieldevaluationoftheONC,thecolorscaleinbothfiguresshowsvaluesnormalized

by themean radiancewithin this area.Figure 5.11(b) is the same as Fig. 5.11(a) but the color scale is

expanded to enhance the reflectance uniformity of the panel within FOV. The standard deviation of the

radiancewithintheFOVismeasuredbythecalibrationcameratobe1.7%.UnliketheONC-T,anemissionangle

dependenceoftheirradiancefromthepanelneedstobetakenintoaccountsincetheONC-W1and-W2have

wider FOVs. Since the flat plane is a self-emitting luminous plane, only the emission angle should be of

6 1055 52 364

concerned in this experiment.Figure5.12 shows the configuration for themeasurementsof theemission

angledependenceoftheradiancepropertiesofthepanel.Inthismeasurement,theradiancesfromthepanel

weremeasuredbyacalibrationcameraate=0˚,10˚,20˚,30˚,and35˚.Figure5.13showstheemissionangle

dependenceoftheradiancemeasuredbythecalibrationcamera.Theradianceisdefinedasthemeancounts

withintheareaofFOV(enclosedbywhiterectangleinFig.5.11)normalizedbytheequivalentvalueate=0˚.

Thisshowsthattheradianceofthepanelgraduallydecreasesasemissionangleincreases.Forexample,the

radianceis14%darkerate=35˚whichcorrespondstotheedgeoftheFOVoftheONC-W1.Duetothisrange

inemissionangleacrosstheONC-W1and–W2FOV,asinglerawimageoftheflatpaneltakenate=0˚bythe

ONC-W1or-W2cannotbeusedasaflatfieldimage,unlikethecaseofthenarrowangleFOVONC-T.Thus,a

correctiontoaccountforthiseffectisnecessarytoacquiretheactualflatfielddatafortheONC-W1and-W2

fromasingleflatpanelimage.Thecountdistributionsintheflatpanelimagestakenate=0˚bytheONC-W1

and-W2containboththeactualflatfieldspecificationofthecameraandtheemissionangledependencyof

radiance across the FOV. From the pre-flight measurement, we calculated the emission angle dependent

radiancemodelbyfittingthedatainFig.5.13(redline).AnactualemissionanglefromthecenteroftheFOV

toanypixellocatedinanimagecanbecalculatedbyusingthedistortioncoefficientsforbothcameras(Suzuki

etal.,2018).Therawcountsintheimageoftheflatpanelarethendividedbythenormalizedradiancemodel

ateachangleshowninFig.5.13.Byapplyingthiscorrectiontotherawcountsforallthepixels,theactual

sensitivityfortheflatfieldimagesfortheONC-W1and-W2areobtained.Figure5.14showsthesensitivity

flat fields of the ONC-W1 and -W2 calculated using this methodology. The values in these images are

normalizedbyameanvaluewithin50pixelsfromthecenteroftheimage(thecenterarea).Theprecisionof

these flat fields isdefinedbythespatial inhomogeneityof the flatpanel(~1.7%)sinceasensitivityof the

calibrationcamerawasconfirmedtobemuchmorestablethanthisvalue.Theabsolutesensitivityofthecenter

areameasuredbyusingtheintegratingsphereispresentedforbothwideanglecamerasinSec.5.5.

6 1055 52 364

Figure5.11.Radiancedistributionoftheself-emissiveflatpaneltakenbythecalibrationcamerawhenthe

emissionangleis0˚.Thetwoimagesaresameexceptforthecolorscaleinordertoenhancetheinhomogeneity

inimage(b).Thewhitesquareinimage(a)indicatestheapproximateFOVoftheONC-W1and-W2inthepre-

flightcalibration.

Figure5.12.Theconfigurationforthemeasurementoftheemissionangledependenceoftheradiance

acrossthewide-anglecameraFOVs.

Figure5.13.Therelationshipofnormalizedradianceandtheemissionangletakenbythecalibrationcamera.

Figure5.14.ThesensitivityflatfieldfortheONC-W1(left)andONC-W2(right).

6 1055 52 364

5.4.2.Flat-FieldEvaluationBasedonInflightStarandPlanetObservations

ThevalidityofthesensitivityacrosstheFOVoftheONC-W1isexaminedusing35Jupiterimagesobtained

fromFebruarytoMarch2016,approximatelytwoyearsafterHayabusa2’slaunch.Inthisobservationperiod,

Jupiter is locatedatvariouspositionswithintheONC-W1’sFOV.Sameasdescribed inSec.3.6,wemeasure

Jupiter’sbrightnessbyintegratingthedigitalcountsaroundJupiter(toincludebrightnessdistributedoutside

thephysicaldiskbythePSF).Adarkimageforthisobservationperiodiscreatedbyaveragingthe35images

where bright spots, suchas Jupiter, bright stars, and counts from cosmic rays hits are excluded.Flat field

correctionbasedonthepre-flightmeasurementsfortheONC-W1wasnotperformedfortheJupiterimages

duetodifferenceinthebrightnessdependenceontheincidentangleintheopticsbetweenapointlightsource

andasurfacelightsource,whichissignificantatalargeincidentangles.ThoughtheJupiterobservationsmay

notbeusedforvalidatingtheflatfield,theycanbeusedtomonitortheconsistencyintheONC-W1’ssensitivity

acrossitsFOV.

Since Jupiter’sbrightnessvarieswithphaseangle,weperformaphaseangle correctionbasedon the

JupiterbrightnessmeasurementswiththeCassini/ImagingScienceSubsystem(ISS)(Mayorga,etal.,2016).

Here,weusethecoefficientsforthegreen-bandoftheISS,whichhasasimilarwavelengthcoverageasthe

ONC-W1.Thepossibleerrorfromthephaseanglecorrectioniswithin~1%(Mayorga,etal.,2016).Thephase

anglevaluebetweentheSun,JupiterandtheHayabusa2spacecraftdecreasedfrom4°to2°,andthenagain

increased to 4° during this observation period. Finally, we corrected the brightness variation due to the

distance between the Sun and Jupiter and between Jupiter and the Hayabusa2 spacecraft. After these

corrections,themeasuredbrightnessofJupiterbytheONC-W1isstablewithinastandarddeviationof1.8%

(Fig. 5.15(a))withaweak incident angle dependencewith cos0.5i,where i is the incident angle.An ideal

pinholeopticshasacosidependenceforapointlightsource.NotethattheCCDtemperaturewas-25.1±0.3℃

in this observation period, and the sensitivity variation due to CCD temperaturewas not expected to be

significant.

ThesensitivityvalidationoftheONC-W2wasexaminedwithaseriesof60starimagestakenbetween

February2016andNovember2017.Onlystarsbrighterthan0.5V-magnitude(Achernar,Canopus,Procyon,

Rigel,Sirius,andAlphaCentauriAandB)wereusedsincetheyprovidedlargerDNsignalsthanmanyofthe

hotpixelsandcosmicrayhitsduetolongexposuretimeoftheONC-W2observations.Figure5.15(b)shows

thedistribution in the sensitivitydeviationsdeduced from thestarobservations, inwhicheach sensitivity

deviationismeasuredwiththesameapproachdescribedinSec.3.8.Thestandarddeviationofthesensitivity

intheFOVofONC-W2is4.4%,which issomewhat largerthanthoseof theONC-TandONC-W1.The large

standarddeviationcouldbefromthelargernumberofhotpixelsinONC-W2,whichmayoccasionallyoverlap

starlocations.Ifthelargest(15%)andthesecondlargest(9%)deviationsareexcludedfromthecalculations,

6 1055 52 364

thenthestandarddeviationreducesto3.7%.

Insummary,ithasbeenconfirmedthatboththeONC-W1and-W2donothaveanyclearsensitivitydefects

overtheirFOVs.Thedeviationsinsensitivity,afterflat-fieldcorrection,arewithinafewpercentinlargeparts

oftheirFOVs.Inaddition,wehavenotobservedanyclearsensitivitydegradationwithintheONC-W2detector

duringthe1.5-yearstarobservationperiod.Thevalidityoftheflatfieldcorrectionforasurfacelightsource

willbere-evaluatedwithclose-upobservationofRyugu.

Figure5.15.Sensitivitydeviationsin(a)ONC-W1FOVand(b)ONC-W2FOVasmeasuredfromJupiterand

starobservations,respectively.

5.5RadiometricCalibrations

5.5.1SensitivityCalibrationBasedonthePreflightData

TheradiometricsensitivitiesoftheONC-W1and-W2weremeasuredbyusinganintegratingsphere

priortolaunch.Thedetailsoftheintegratingsphereandtheprocedurestoquantifythesensitivityforthe

ONC-TaredescribedinKamedaetal.(2017).Thus,thedifferencesbetweentheprocedureusedtoquantify

thesensitivityoftheONC-TandthoseusedtoquantifythesensitivityoftheONC-W1and-W2aredescribed

here.ThemajordifferencescomefromthewiderFOVsoftheONC-W1and-W2(~70deg)relativetonarrow

FOVoftheONC-T(~6.5deg).DuetothisFOVdifference,theluminousplaneoftheintegratingspherecould

notfullyfilltheFOVsoftheONC-W1and-W2.Figure5.16displaystheimageoftheintegratingspheretaken

bytheONC-W1,whichshowsabrightcirculararea(theluminousplane)surroundedbyadarkarea.

Althoughitisdifficulttoderivethesensitivityforallpixelsbyusingthissingleimage,theaveraged

6 1055 52 364

sensitivityofthepixelswithina50-pixelradiusfromthecenteroftheimagecanbemeasured.Bycombining

thesesensitivityvaluesandthenormalizedsensitivityflat-fieldsshowninFig.5.14,thesensitivityofall

pixelsacrosstheFOVcanbecharacterized.SincetheONC-W1and-W2arepanchromaticcameraswitha

broadbandpass,themeanradianceoftheintegratingsphere, "(F,IWÒÒÒÒÒÒÒÒÒ [W/m2/sr/μm]iscalculatedby

"(F,IWÒÒÒÒÒÒÒÒÒ =∫ ûœÚÛ.Ùıˆ<Û.˜ıˆ< (d)f¯*<(d)Sd

∫ f¯*<Û.Ùıˆ<Û.˜ıˆ< (d)Sd

, (cam=W1 or W2) (5.2)

where "(F(a) istheknownspectralradianceprofileoftheintegratingsphere(seeFig.7ofKamedaetal.,

2017)and ΦIW(a) isthesystemefficiencyfunctionoftheONC-W1or-W2(SeeFig.2ofSuzukietal.

2018).Thecurrentoftheintegratingspherewassetto2.75Aforthesemeasurements.Theradiometric

sensitivities, \IWÒÒÒÒÒÒ [DN/s]/[W/m2/sr/μm],forthecenterareaoftheFOVsareobtainedby

\IWÒÒÒÒÒÒ =(

ûœÚ,¯*<ÒÒÒÒÒÒÒÒÒÒ , (5.3)

where G isthesignal[DN/sec]averagedoverpixelswithina50-pixelradiusfromthecenteroftheimage.

The \IWÒÒÒÒÒÒ valuesobtainedfortheONC-W1and-W2are 1.38 × 10ú and 3.84 × 10ú

[DN/s]/[W/m2/sr/μm],respectively.Itisnotedthatthesecoefficientsarereliableonlywhenthespectral

radiationofanobjecthasacomparablecolortemperaturewiththeintegratingsphere.Thus,themean

radianceobtainedbyusingthesecoefficientsisreliableonlywithinanorderofmagnitudeestimation,in

general.TheresultsofthisradiometriccalibrationaresummarizedinTable5.3.

5.5.2SensitivityCalibrationBasedontheInflightJupiterandStarObservations

ThesensitivityoftheONC-W1wasmeasuredbasedontheJupiterobservations,usingthesameimages

described in Sec. 5.4.2. The sensitivity was evaluated by comparing observed Jupiter digital counts and

expectedJupitercountsbasedontheONC-W1’sspecification(Suzukietal.,2018),whichwasdeterminedprior

tolaunch.WeusedtheJupiterimagessinceevenbrightstars(0V-magnitude)provideonlylessthan200DN

intotalsignalinanONC-W1image.Thesestellarobservationsarecomparabletocountsfromhot-pixelsand

typicalcosmicrayhits,duetotheshortexposuretimesoftheONC-W1(upto5.57sec),whileJupiterprovides

morethan2000DNintotalsignal.

Sameaswithstarcalibrations,thedigitalcountsexpectedfromJupitercanbecalculatedbyusingEqs.

(3.18)and(3.19).Unlikestars,however,thedisk-integratedJupiterbrightnessobservedbytheONC-W1has

ageometricaldependenceontheobservationgeometry.Theexpectedirradiance(W/m2/μm)fromJupiter,FJ,

canbeexpressedas

"e(a) = £e(a, ‰) eeÚ(d)

‹‡t_˚

QÚ%¸‚u

, (5.4)

whereαisthephaseangleatthetimeofobservation,AJ(λ,α)isthedisk-equivalentreflectanceofJupiter,ΩJis

6 1055 52 364

the solid angle of Jupiter as seen from the Hayabusa2 spacecraft, JS(λ) is the modeled solar irradiance

(W/m2/µm)at1AU,andDS-JisthedistancebetweentheSunandJupiterinAU.Thedisk-equivalentalbedois

calculatedfrom

£e(a, ‰) = £e(a, ‰É)˝(d,Á)

˝(d,ÁÛ) , (5.5)

whereΨ(a, ‰) representsthephaseangledependenceofAJ(λ,α),andα0isthereferencephaseangle.Inthis

study,weusedasolarirradiancemodelfromGueymard(2004),theJupiterreflectancedatafromKarkoschka

(1998)takenin1995ataphaseangleof6.8°(i.e.,α0=6.8°),andthephaseangledependencefortheISSgreen-

bandfromMayorgaetal(2016),asdescribedinSec.5.4.SincetheapparentsizeofJupiterwasmuchsmaller

than1-pixelintheONC-W1FOV,thesolidangleofJupiterinsteradianisapproximately

e = •w ˇw !Q¸%"·

, (5.6)

whererJeandrJparetheequatorialandpolarradiiof Jupiter,respectively,andDJ-H is thedistancebetween

JupiterandtheHayabusa2spacecraftatthetimeofobservation.BothDS-JandDJ-Hweremorethan5AUduring

theobservationperiod.

Fromthe16JupiterimagesinwhichJupiterwaslocatednearthecenteroftheFOV,wherevariationsin

the sensitivity are not significant (Sec. 5.4.2), the sensitivity of the ONC-W1 was measured to be

(1.44 ± 0.02) × 10ú [DN/s]/[W/m2/sr/μm],whichishighlyconsistentwiththesensitivitymeasuredpriorto

launch.Theerrorrange isevaluatedfromthestandarddeviationofmeasuredsensitivities from35Jupiter

observations. It should be noted that the sensitivity value from the Jupiter observations has errors from

possibletemporalvariationsinJupiter’sreflectance(forexample,acoupleofpercent(Karalidietal.,2015)),

uncertainties inmeasuring Jupiter’s brightness in theONC-W1 images (2%based on standard deviation),

errorsintheJupiterreflectance(4%inabsolutevaluebyKarkoschka(1998)),anderrorsduetothephase

angledependence(upto1%byMayoruga,etal.(2016)).Inaddition,Jupiterandtheintegratingspherewe

usedinthepre-flightmeasurementshavedifferentspectralprofiles.Thesecouldprovidedifferencesbetween

thesensitivitiesderivedfrompre-flightandinflightmeasurements.

ThesensitivityoftheONC-W2wasmeasuredbasedonstarobservationsconductedfromFebruary2016

toNovember2017.StarsimagesusedforthismeasurementaresamefortheevaluationoftheONC-W2flat-

fielddescribed inSec.5.4. Comparing theobservedDN to theexpectedDNbasedon star irradiances, the

sensitivity of the ONC-W2 in 6 star observations was evaluated as (4.03 ± 0.07) × 10ú

[DN/s]/[W/m2/sr/μm] at the center of FOV, which is consistent with the sensitivity from the preflight

measurement.Inaddition,thisvaluecorrespondstoa47%lowersensitivitythanexpectedfromtheONC-W2

specification,whichwas commensuratewith the results from the Earth observations,where theONC-W2

measuredEarth’sreflectancetobe40%darkerthanthereferencevalue(Suzukietal.,2018).

6 1055 52 364

ItcanbeconcludedthatboththeONC-W1and-W2basicallyretainedtheirpreflightsensitivitiesduring

thecruisephase.Continuoussensitivitymonitoringwithbrighttargetswillprovideacontinuingmeasureof

thestabilityofthesensitivityinflightforthesecameradetectors.

Figure5.16.AnimageoftheintegratingspheretakenbytheONC-W1.

Table5.3.Summaryof the radiometric calibrationof theONC-W1and -W2basedonpreflight integrating

spheremeasurements.

Camera Date Current

settingofthe

integrating

sphere.

\IW

[DN/s]/[W/m2/μm/sr]

ONC-W1 27March2014 2.75A 1.38 × 10ú

ONC-W2 27March2014 2.75A 3.84 × 10ú

5.5.3.PointSpreadFunctions(PSFs)

The PSF of theONC-W1 is evaluated from observations of bright targets, i.e. Mars, Jupiter, and stars

brighterthan1V-magnitude(Betelgeuse,Capella,andProcyon).Marsimagesweretakenon31May2015,and

Jupiter imagesarethesameimagesusedforthe flat-fieldandsensitivityevaluation.Thestar imageswere

takenon12March2017.ThePSFoftheONC-W2wasalsoevaluatedfromstarsbrighterthan1V-magnitude,

takenbetweenFebruary2016toNovember2017.Figure5.17showsasummaryoftheFWHMofeachbright

target for theONC-W1and-W2.TheFWHMvaluesareevaluated froma two-dimensionalGaussian fitting

method,sameasthatusedformeasuringthePSFoftheONC-T(Suzukietal.,2018).Duetotheshortexposure

timeoftheONC-W1(upto5.57sec),onlyalimitednumberofstarimageswereused,while142starsamples

wereavailableformeasuringthePSFoftheONC-W2,whoseexposuretimewas44.56sec.

6 1055 52 364

AroundthecenteroftheFOV,theFWHMwas~1.8pixels,similartotheONC-TforbothONC-W1and-W2,

andtheyarealmostsameas(slightlybetterthan)theFWHMdeterminedfromthepre-flightmeasurements.

ThelargerFWHMattheimagecornerintheFOVoftheONC-W2couldbecausedbyavignettingeffectfroma

baffle,sameasfortheONC-T(Suzukietal.,2018).WhiletheFWHMoftheONC-W1showsanincreasingtrend

withdistancefromthecenteroftheFOV,theFWHMvaluesinlargepartsofboththeONC-W1and-W2FOVs

are smaller than each FHWM value at the image corner from the pre-flightmeasurements, indicating no

significantdegradationofthePSFafterlaunch.Itshouldbenotedthatsincethestarsare20-50timesdarker

thaneitherJupiterorMarsintheONC-W1,duetotheshortexposuretimeoftheONC-W1stellarobservations

(5.57sec),starobservationsprovideratherlargeerrorsintheFWHMoftheONC-W1.

Figure5.17.ThemeasuredFWHMof(a)theONC-W1and(b)theONC-W2PSFasafunctionofthepixel

distancefromthecenterofFOV.TrianglesanddottedlinesindicatePSFvaluesatthecenterandcornerofthe

FOVmeasuredinpre-flightexperiments.

6 1055 52 364

6.ONCSystemAlignmentwiththeSpacecraft

UnderstandingthealignmentoftheONCsystemwithrespecttothespacecraftisessentialforcomparing

themeasurementswithinthedifferentFOVsormatchingfootprintswithotherinstruments,suchastheTIR

andNIRS3,orevenwhenwecompareimagestakenfromtheONC-TandONC-W1.Moreover,thealignment

informationisnecessarytocalculatetheaccurateattitudeandpositionofthespacecraftfromimagesduring

operation.Inthissection,wepresenttheresultsofmeasurementsderivingtheviewingdirectionoftheONCs

relativetothespacecraftattitude.AsmentionedinSuzukietal.(2018),theFOVsoftheONC-W1andONC-

Tweredesignedtoaligninthe−ZSCdirectionwithinthespacecraftcoordinatesystem,whilethatofthe

ONC-W2wasslantedbyapproximately30˚fromthe−ZSCdirection.However,actualalignmentsmayhave

small offset from thesedesignedvaluesdue to an ineluctable error in assemblyproceduresprior to

launchor offsets createdby the strongvibrationsduring launchof theHayabusa2 spacecraft. These

offsetsaremeasuredpreciselybyusingstarfiledimagestakenduringthecruisephase.Starfieldimages

usedinthedeterminationandverificationofthedistortioncoefficients(seeSuzukietal.,2018)ofthe

threecamerasarealsovalidatedinthismeasurement.InSecs.3and6ofSuzukietal.(2018),celestial

coordinateswerefittedtothestarfieldimagestoverifytheestimateddistortioncoefficients(SeeFig.5

fortheONC-W1and-W2,andFig.12fortheONC-T).Fromtheseresults,rightascensionanddeclination

anglescorrespondingtothecenteroftheFOVattheobservationtimeforeachcameracanbedetermined.

Ontheotherhand,informationofthespacecraftattitudeattheseobservationperiodsarealsoavailable

fromthetelemetrydata.ThisallowsustopredictpointingdirectionsandthecentersoftheFOVsofthree

camerasduringtheobservationperiodsbyassumingthatthecameraswereperfectlyassembledtothe

spacecraftasdesigned.Thus,theoffsetvaluesbetweenthepredicteddirectionofanopticalaxisofeach

cameraandthatestimatedfromtheactualstarimagecanbedefinedasalignmentinformation.Table6.1

summarizestheobservationdatesandtimesofthestarfieldimages,theobservedcenteroftheFOVin

thecelestialcoordinatesystem,thecenteroftheFOVpredictedbytheattitudeinformation,andtheoffset

betweenthem(i.e.,thealignmentinformation)intermsofanglesandinimagecoordinate(inpixels).The

lastcolumnofTable6.1givesthealignmentinformationintheimagecoordinate(fordefinitionsseeFig.

1ofSuzukietal.,2018)definedas

Δx = xöKN − 511.5[pixels] (6.1)where,the xöKN isobservedlocationofthepredictedcenter(e.g.,thedirectionof−ZSCfortheONC-W1

andONC-T)andthevalue511.5pixelsisthelocationofimagecenter.Wealsocanexpressalignmentby

Eulerangleswithrespecttothespacecraftcoordinatesystem.Thematrixtransformingfromthe

spacecraftcoordinatesystemis

ƒ = %cosΘ) sinΘ) 0− sinΘ) cosΘ) 0

0 0 1*%

cosΘ+ 0 − sinΘ+0 1 0

sinΘ+ 0 cosΘ+*%

1 0 00 cosΘ, −sinΘ,0 sinΘ, cosΘ,

*, (6.3)

6 1055 52 364

where Θ, ,Θ+ , andΘ, are Euler angles from the spacecraft coordinate system to the instrumentcoordinatesystem.TheEulerangelsforONCarealsolistedinTable6.1.

*Suzukietal.(2018)

Table6.1.Alignmentinformationforthethreecamerasestimatedbyusingstarfieldimages.

7.NIRS3

Hayabusa2 payload includes the near infrared spectrometer NIRS3, whose purpose is to detect and

measuretheabsorptionaround3µm,associatingwiththehydrationofmineralspecies(Iwataetal.,2017).

NIRS3isdesignedtoobservewavelengthsfrom1.8to3.2µm.Itisimportanttoconnectthespectrafromthe

ONC-TandNIRS3instrumentstomeasurethespectralcharacteristicsoverabroadrangeofwavelengths,from

theultraviolet (UV) through thevisible (vis) to thenear infrared (NIR).Thiswill allowus toexamine the

mineralogyofRyuguusingtheintegratedspectrafrombothinstruments.Inthissection,weshowtheONC-T

Camer

a

Observation

time for the

starimages.

YYYY-MM-

DDTHH:mm:

ss

Observed

center of the

FOV in (RA,

DEC)[˚].

Predicted

center of the

FOV in (RA,

DEC).[˚]

Offset

in

angle.

[˚]

Offset of

the center

of FOV in

the image

coordinate

(alignment

informatio

n)

(ΔH, ΔV)

[pixels].

Euler

angles

(

Θ),Θ+ ,Θ,)[Degree]

Distortion

parameter

s*(-t,-u),

°†

= ° + -t°ú

+ -u°≠.

Focal

length*

(mm)

W1 2015-02-

19T09:50:10

(138.880,6.15

0)

(138.789,6.47

5)

0.313 (-3.5,-2.5) (0.0, -

180.2728,

0.1948)

(3.134×10-

7, -

1.716×10-

13)

10.22

W2 2014-12-

11T13:38:12

(136.500,6.02

0)

(136.484,6.05

1)

0.051 (0.5,-0.5) ( -270, -

121,

0)

(2.893×10-

7, -

1.365×10-

13)

10.38

T 2014-12-

11T14:52:53

(77.388,21.96

1)

(77.532,

21.996)

0.151

0

(9.5,-22.5) (0.0, -

179.9415,

0.1386)

(-9.28×10-

9,0)

120.50±0.0

1

6 1055 52 364

andNIRS3spectraoftheMoonobservedalmostsimultaneouslyon5December2015(Kitazatoetal.,2016).

TheconsistencyofthespectraisdiscussedbasedonthecomparisonswiththeSP/SELENEmodel(Yokotaet

al.,2011;Kouyamaetal.,2016)forvis-NIRspectra,sincethereisnooverlapinthewavelengthrangebetween

the two instruments. In this comparison, we used a modified SP/SELENE model which covers longer

wavelengths(~2μm)basedontheupdatedreflectancedatabyYokotaetal.(2012)thanthemodelinYokota

etal.(2011)andKouyamaetal.(2016).ThereflectancespectrumfromtheSPmodelwascorrectedwiththe

ROLOreflectancemodelsincetheSPmodelshowssomewhatdarkerandbrightenertendenciesinshorterand

longerwavelengthranges,respectively(Ohtakeetal.,2013).

First,thealignmentoftheONC-TandNIRS3arecalculatedbasedontheSPICEInstrumentKernel.The

NIRS3alignmenthasbeendeterminedthroughscanobservationsof theEarthcombinedwithslewsofthe

spacecraftattitude.Theerrorinalignmentiswithin+/-0.005˚,correspondingto~5%ofthesizeoffieldof

view.Figure7.1showsthealignmentoftheNIRS3intheFOVoftheONC-T,showinglittleoffsetfromthecenter

oftheFOV.ThefootprintoftheNIRS3is{[476.5,454.3];[492.7,472.1]}pixelsintheONC-Timagecoordinate

system.NIRS3onlyobtainedspectraofpartoftheMoon’ssurfaceduetoitsnarrowfootprint(Fig.7.2).The

comparisonsofthelunarspectrafromdifferentpartsofthesurfaceareshowninFig.7.3.Wealsodisplaythe

modelspectrafromSP/SELENEtoreasonablyextrapolatetheONC-TspectrumtotheNIRS3spectrum.There

seemstobelittleoffsetbetweenthespectraoftheONC-T,NIRS3,andmodelspectra.Notethattheradiometric

calibrationof theNIRS3wasprimarilybasedonpreflight calibration testdata.TheSPmodelspectramay

underestimatetheradianceforlargeincidenceangleregionsasshowninKouyamaetal.(2016),whichisof

importancesincethefootprintsoftheNIRS3arelocatedathigherlatituderegionswheretheincidentangleis

large.Thus,weshifttheSP-modelspectratofittheONC-Tv-bandradianceandtheNIRS3spectratofittheSP-

modelintherangebetween1819to2038nm.WecalculatedtheoffsetfactorbetweentheONC-TandNIRS3

as0.87(Fig.7.4).Usingthisvalue,wecanevaluateslopeoftheintermediatewavelengthfrom0.95to1.8µm.

Thiswavelengthrangewasobservedbyground-basedtelescopes(Moskovitzetal.,2013;LeCorreetal.,2017)

andisreflectiveofthemineralogicalcompositionofthesurface.Moreover,thereisasignificantdifferencein

theslopesinthiswavelengthrange,whichwasobservedbyMoskovitzetal.(2013)andLeCorreetal.(2017).

Thus,wemayobservesomeheterogeneityintheslopeinthiswavelengthrangeoverthesurfaceofRyugu.

6 1055 52 364

Figure7.1.ThefootprintoftheNIRS3intheFOVoftheONC-T.

Figure7.2.SquaresindicatetheROIsoftheNIRS3observationsinanONC-Timage(red(ROI1):11:35:14on

5December2015,green(ROI2):11:42:27on5December2015,yellow(ROI3):11:45:29on5December2015).

6 1055 52 364

Figure7.3.Radiance comparisonof the lunar spectrumobtainedby theONC-TandNIRS3.Thegray lines

indicatetheSPmodelspectrabyKouyamaetal.(2016)andthegraysymbolsindicatetheWACmodelspectra

bySatoetal.(2014).NotethatradiometriccalibrationsoftheONC-TandNIRS3wereconductedindependently.

Figure 7.4. Corrected radiance of the Moon observed at 11:35:20 on 5 December 2015. The NIRS3 and

SP/SELENEmodelspectraareshiftedbythecoefficientsinthelegendtofittheONC-Tradiance.

8.ApplicationstoScientificAnalysesonRyugu

Wehaveevaluatedthebias level,EMInoise,dark level,hotpixels, linearity, flat-fields, stray light,and

sensitivityofthedetectorsandfiltersfortheONCsystem.Thissectionaddressestotheevaluationofaccuracy

thekeyscientificanalysestoanswerimportantnatureoftheasteroid.Morespecifically,oneistodetectthe

presenceofhydratedminerals,whichisanindicatorofthepresenceofwater(pastorpresent)ontheasteroid,

and the another is to characterize the real radiance variations on its surface, even when temperature

conditionsarechanging.Weevaluatetheaccuracythatisexpectedbasedonthecalibrationstepswhichare

6 1055 52 364

describedinprevioussections.

8.1.Signal-to-NoiseRatio

The signal-to-noise ratios (SNRs) of the ONC-T, -W1, and -W2 are important for establishing the

detectabilityofradiancevariationovertheasteroidsurface.Thescientificqualityoftheimagesischangedby

temperatureconditions.Unlikeorbitingmissions,theHayabusa2spacecraftwillexperiencealargerangeof

temperatures,becausethechangeinthedistancebetweenthespacecraftandtheasteroidincludesradiative

heatchangewhenatlowaltitudes.Especially,duringtouchdownstheCCDtemperaturescanbeashighas20

℃.Thereby,understandingtheeffectsofthetemperatureonthedetectorperformanceisnecessaryinorder

toanalyzetheclose-approachimages.Ourcalibrationresultsshowthetemperaturecauseslargeeffectson

darknoise.Here,wediscusstheSNRsdependenceontheCCDtemperature.

Thetotalnoise ∆G canbedefinedasthecombinationofread-outnoise ∆Gwö ,EMInoise ∆GR/( ,shotnoise∆GNfiöA70G[DN] ∙ 3[e, DN⁄ ] 3[e, DN⁄ ]5 whereg=20.95[e-/DN]isthegainfactor,anddarknoise ∆GgMw} :

∆G = 0∆Gwöu + ∆GR/(u + ∆GNfiöAu + ∆GgMw}u [DN], (8.1)

where,onlythedarknoiseisafunctionoftime,buthasanegligiblecontributionatthereferencetemperature

(-30 ℃).BecausethethreecameraswithintheONCusethesamemanufacturedCCDswiththealmostsame

characteristics,theONCcamerasallimagewithaSNR~200atthereferencetemperature.However,oncethe

CCDtemperaturegetshighas20 ℃,thedarknoisecontributionwilldominantinthetotalnoiseandyieldsa

SNR~150fortheONC-Tv-band,whiletheONC-W1andW2willbelessaffectedduetotheshorterexposure

timesduringoperation.Thesevaluesuggestthatthevariationacrosstheasteroidsurfaceoverasingleimage

canbemeasuredwithinerrorsof0.7%attheworstcase.Moreover,theremaybeunavoidableeffectofhot

pixels(>100DN/s),sothat~1%oftheFOVareawillnotbeuseable.

8.2.Detectabilityof0.7-µmAbsorptionBand

One of themost important inquiries for the spectralmapping on Ryugu is the determination of the

presenceofhydratedminerals,whichareindicatorsofaqueousalterationprocessesonprimordialasteroids.

AsteroidsareconsideredtobeapossiblesourceofwatertotheinnerSolarSystem,includingourownEarth

(e.g.,O’Brienetal,2014;Altweggetal.,2014).Moreover,thedegreeofaqueousalterationalsoprovidesinsight

intotheheatinghistoryofasteroids.Here,wediscussthetotalerrorrevealedbythecalibrations,especially

sensitivity,andtheirimpactonthepossibledetectionofhydratedmineralsfromONC-Tmultibandimages.

OurtargetasteroidRyuguiscategorizedasaC-complexasteroid,whicharetheorizedtohaveformedin

the volatile-rich region in theSolar System, in the Bus andBinzel taxonomy (BusandBinzel, 2002). The

presence of hydrated minerals is usually discussed by the most prominent and unambiguous 3-micron

absorption(e.g.,Lebofsky,1980;TakirandEmery,2012).However,the3-µmbandisdifficulttoobserveon

6 1055 52 364

smallasteroidsfromtheEarth.Thereby,theattributedironcharge-transfer0.7-µmabsorption,whichiseasier

to observe, is used as the reliable proxy for hydrated minerals (e.g., Vilas et al., 1994). Ground-based

observations have established the correlation between the 0.7- and 3-µm absorptions among C-complex

asteroids(e.g.,Howelletal.,2011).ManyobservationsofRyuguhavebeenmadetodatetoobtainmineralogy

informationpriortoHayabusa2’sencounter(Binzeletal.,2001;Vilas2008;Lazzaroetal.,2013;Moskovitzet

al.,2013;Sugitaetal.,2013;Pernaetal.,2017).Mostofthesespectradisplayveryflatspectralshapes,while

only one of them showsevidence of a 0.7-µmabsorption (Vilaset al., 2008). This suggests theremay be

variations in the compositionacrossRyugu’ssurfacethatare characterizedby thepresenceorabsenceof

the0.7-µmabsorptionfeature.Thus,itisimportanttomapoutthe0.7-µmabsorptionbytheONC-Tbasedon

v-,w-,andx-bandobservations.Thedepthof0.7-µmabsorptioncanbemeasuredby

rÉ.6 = 1 −ú.tÊ7

t.õÊ8zt.≠Ê9. (8.2)

Kamedaetal.(2015)reportedthatthe0.7-µmabsorptiondetectiononCM2chondrites,suchasMurchison

andNagoya,bytheflightmodelofONC-T.Fromourupdatedsensitivity,theambiguityinthesensitivitiesfor

the v-, w-, and x-bands are 0.85%, 1.3%, and 1.6%, respectively. The error of 0.7-µm absorption depth

calculatedfromerrorpropagationofEq.(8.2)are1.6%.Thereby,thetypicalabsorptionofserpentine(3-4%)

canbedetectedbyaSNR~2.Thisvaluecanbeimprovedbydecreasingthestatisticalerrorsbasedonmore

starobservations.

8.3.UV-SlopeEvaluation

Asisnotedabove,mostoftheground-basedobservationsofRyugudonotreportanabsorptionaround

0.7-µm. Rivkin (2012) reported that one third of C-complex asteroids in Sloan Digital Sky Survey, which

includeslargenumberofasteroidsdowntosmallobjects,showthe0.7-µmfeature,whereastwothirdsdonot.

Moreover, the0.7-µm feature seen inCMchondrite spectraeasilydisappearswhendehydratedbyheating

(Hiroietal.,1996)andspaceweathering(Matsuokaetal.,2015).Thus,thereispossibilitywewillnotfind0.7-

µmfeatureonRyugu.However,thisdoesnotindicatethattherearenohydratedmineralsonthesurface.Other

than0.7-µmabsorptionfeature,thestrengthofUV-absorptionshortwardof0.55µm,whichiscausedbyFe2+

andFe3+insilicates,alsocorrelatespositivelywiththe3-µmfeature(e.g.,Freierbergetal.,1985).Vilasetal.

(1993)investigatedtherelationshipof0.7-and0.43-µmabsorptionsandsuggestedthatasteroidswith0.7-

µmabsorptionalsodisplaythe0.43-µmabsorption.Moreover,asteroidspectrathatdidnotdisplaythe0.43-

µmfeaturealsodidnotshowthe0.7-µmfeature.Thus, theUVabsorptioncanbealsoaproxyforaqueous

alterationintheONC-Twavelength,suchasaspectralturnovernearthev-band.Thus,wecanusetheUVto

blue spectral slope index instead of the 0.7-µm absorption. TheUV slope can be calculated by ln(Rul/Rv).

AccordingtoHiroietal.(1996),therearelargediscrepanciesofthisvaluefrom-1to0amongC-,G-,B-,andF-

typeasteroidsintheTholentaxonomy(Tholen,1984),andthisvaluecorrelatesnearlylinearlywiththe3-µm

6 1055 52 364

bandabsorption.Althoughtheyused0.34µmfortheshorterwavelength,ul-bandwavelengthrange,0.4-µm,

canserveasthesameproxy.Basedonoursensitivityambiguity,theerrorforthisvaluewillbe~1%inthe

centeroftheFOV.However,itshouldbenotedthatthespatialerrorcouldbelargebecauseofthedeviationin

ul-bandflat-fieldcausedbytheround-aboutstraylightonthepreflightflat-fieldmeasurement.Forthatreason,

thespatialuncertaintyofmeasuringtheUVslopewillbedominatedby flat-fielduncertaintyof~4%.This

errorestimationstillsuggeststhatwecandistinguishunheatedCMsof ln(Rul/Rv)~-0.6to-0.3 fromhighly

dehydrated(700-1000 ℃)CMsof~0quiteeasily,assumingthatRulandR0.34µmarelinearlycorrelated.

8.4.FirstInflightObservationofRyugu

ThefirstinflightobservationsofRyuguwereconductedon26February2018whentheanglebetween

theSun-Ryugu-Hayabusa2was~1.1-1.6˚.Takingadvantageof the illuminateconditionclosetoopposition,

RyuguwasapparentlybrightandwesucceededinobservationwiththeONC-Tbylongestexposuretime(178

sec).Weconductedtwodifferentobservationsfordifferentpurposes;1)lightcurveobservationand2)Ryugu

multi-band imaging.Onthe lightcurveobservation, thewide-bandwasused inordertostackthesignalas

much as possible. On themulti-band observation, we observed different rotational phases.We took four

sequentialimagesetsforeachbandatonerotationalphase.ThesummaryofobservationsarelistedinTable

8.1.Afterdarkandflatcorrections,invalidimages,whichincludecosmicrayhitsclosetoRyuguduetolong

exposureswereomittedfromtheanalyses.Thebackgroundnoise,suchasdarkcurrentandhotpixels,were

maincauseoferrorsinmeasurements.Thelightcurveisobtainedbytakingthreemovingaveragesovertime

(Fig.8.1).

Onthemulti-bandobservations,theimageswereacquiredevery¼rotationalphasefor7sets.Duetothe

very small signals,we employ two analysismethods to obtain robust results.Onemethod is theaperture

analysis,which isusuallyappliedtostar fluxanalysesasdiscussed inSec.3.Anothermethodistherough

estimationfromthepeaksignalofRyugu.Morespecifically,weobtainedtherelationshipbetweenthepeak

signalandtotalsignalfromstarobservations,inwhichsignalsaremuchlargerthanthebackgroundsignals.

However,thesignalratioofpeaktototalmaychangefrom0.15to0.33withtheRyugudetectedareainone

CCDpixel.Becausethesignalsoftheul-andp-bandsaresmallerthanthedarklevel,<0.05DN/s,thespectral

shapeisdiscussedwithoutthesetwofilters.Figure8.2displaysnormalizedreflectancespectraofRyugufrom

thetwomethodsfromb-tox-bands.Wefoundthatthiscasewithverylowsignalrate,theroughestimategives

smaller errors. Although, the shapes of reflectance spectra are not perfectly reproducible at two similar

rotationalphase,suchasobservationsetof2and6inFig.8.2,theslopeofreflectancespectraforallrotational

phasesvariesto include0(flat)consideringerrors inmeasurements.Suchflatspectraareconsistentwith

ground-basedobservations(Binzeletal.,2001;Vilas2008;Lazzaroetal.,2013;Moskovitzetal.,2013;Sugita

etal.,2013;Pernaetal.,2017).Theremaybedarkening intheb-bandcomparedwiththev-bandformost

6 1055 52 364

rotationalphases.Furthermore,wecomparethemagnitudeofthev-bandobservedwiththeONC-Tandthe

correspondingV-magnitudeoftheEarth-basedobservationsfromIshiguroetal.(2014).Becausewestackthe

signal fromall 28 v-band images to reduce theerror, the v-bandmagnitude here corresponds to a global

average,ordisk-integratedvalue.Figure8.3showsthecompiledphasecurveofRyugu,displayingconsistency

betweenONC-TandEarth-basedmeasurements,althoughtheerrorbarisstilllarge~10%.Thisresultalso

supportstherobustnessofthecalibrations.

Table8.1.SummaryofRyugufirstinflightobservations.

ObservationTime(UTC) NumberofImages

Lightcurveobservation(wide-band) 26February201803:01–10:28 100

7-bandobservation(1) 26February201810:39–12:01 28

7-bandobservation(2) 26February201812:33–13:56 28

7-bandobservation(3) 26February201814:28–15:50 28

7-bandobservation(4) 26February201816:22–17:45 28

7-bandobservation(5) 26February201818:15–19:39 28

7-bandobservation(6) 26February201820:11–21:34 28

7-bandobservation(7) 26February201822:04–23:28 28

Figure8.1.First inflight lightcurveobservationofRyugutakenon26February2018.Everypointshowsa

threepointsmovingaverageanderrorsareevaluatedbythedeviationbetweenthreepoints.Duetothechange

indistancebetweenthespacecraftandRyugu,thesignalsarenormalizedbythesquareofthedistance.

6 1055 52 364

Figure8.2.5-bandobservationsofRyuguwiththeONC-Ton26February2018.Thecolorlinesindicatethe

reflectanceobtainedbyapertureanalysesandgraylinesindicateroughestimatesbasedonpeaksignalrate.

6 1055 52 364

Figure8.3.ThephasecurveofRyugu,compilingtheEarth-baseddatafromIshiguroetal.(2014)(black)and

theinflightmeasurementsfromthisstudy(red).TheinflightONC-TmeasurementoverlapswithEarth-based

measurements.

9.Summary

Inthisstudy,wehavepresentedthecalibrationsofthethreeONCcamerasduringthecruisephaseover

thepastthreeyears;includingcorrectionsforread-outsmearnoise,bias,EMInoise,dark,hotpixels,linearity,

flat-field,straylight,andCCDsensitivity.NotethattheCCDsofONC-T,-W1,and-W2havesamemanufacturer

productnumber.Thus,thebasicbehaviorisexpectedtobesimilarforallthreecameras.

ONC-T

l Read-outsmearcanbeomittedusinginflight0-secexposureimagesubtraction,orbymodelingusingthe

empiricalrelationshipinEq.(3.4),similartothemethoddescribedbyIshiguroetal.(2010).Weshowed

thevalidityofthissmearcorrectionusinganONC-W2imagesincetheverticalchargetransfertimehas

notbeenmeasuredfortheONC-T.WewillmeasuretheverticalchargetransfertimeoftheCCDforthe

ONC-TduringtheapproachphasetoRyugu.

l Biasleveldependsonthetemperatureandcanbedescribedbyanempiricalbiasmodel.Basedonthe

modelthebiaslevelwasshowntobestablewithin1%deviationsoverthepastthreeyears.

l EMInoisecauseswave-likepatternsonimages.Theamplitudeshavebeen<5DNforthecruisephase.

l DarkcurrentandhotpixelsaregreatlychangedbytheCCDtemperature.Atthereferencetemperature,

thedarkcurrentissmallerthan0.09DN/s,whichisconsistentwithpreflightmeasurementsandlessthan

10hotpixels>30DN/sarefound.AttheCCDtemperatureof20℃,thedarkcurrentis20DN/sandhot

pixels>100DN/soccupy~1%oftheFOV.

l ThelinearityoftheCCDwasfoundtobeslightlydegradedfromthe0.6%errorpreflightto1%inflight.

6 1055 52 364

Theempiricallinearitycorrectionmodelcorrectsthelinearityto<0.1%.

l Flatnessof the flat-fields forv-bandandul-band isevaluatedbasedonstarobservationswithslightly

differentattitudes.Thesensitivityvariationsinthev-bandflat-fieldandtheul-bandflat-fieldareabout

2%and4%,respectively.Theul-bandflat-fielddiscrepancyislargerthanthev-bandbecausetheround-

aboutstraylightmightcontaminateandmakelittlegradationintheFOVduringpreflighttests.

l Scattered light in theopticsarecharacterizedandthePSFmodelsarederivedforallcolor filters.The

effectofabroadPSFintheONC-TissmallerthanthatontheAMICA/Hayabusa.Theeffectcanbecorrected

inawaysimilartothatdescribedbyIshiguro(2014).

l Theghosteffectwasobservedandismostapparentinthep-filter,withasignalasmuchas0.65%the

intensityoftheobjectobserved.

l Suzukietal.(2018)reportedtheradiatorstraylightdependenceonspacecraftattitudewithrespectto

theSun.Theradiatorstraylighthasbeenintensivelyobservedduringthecruisephase.Theattitudeswith

negligiblestraylightwerefoundandreportedinSuzukietal.(2018).Inthisstudywepresentthestray

lightmodel based on PCA. Thismodel reduces the radiator stray light contamination to 25DN/s/pix,

0.25%theexpectedintensityoftheasteroiddisk.

l CCDsensitivitywascarefullyexaminedbasedonstar,Jupiter,Saturn,andMoonobservationsinaddition

to preflightmeasurements.We characterized the sensitivity in twoways; 1) based on the hardware

specificationsoftheONC-T,and2)basedonstarobservations.Moreover,thesensitivityofthep-bandis

moreaccuratelydeterminedbasedonthelunarobservationsthanstellarobservationsduetotheinherent

ambiguityinthestellarreferencespectra.Sincethesensitivitybasedonstellarobservationsfortheul-to

x-bandsandthelunarobservationsforthep-bandreproducedthespectraoftheMoon,Jupiter,andSaturn,

currentcalibrationsusethischaracterizationofthesensitivity,summarizedinTable.3.10.Moreover,we

measured the temperature dependency of the sensitivity using Jupiter observations acquired with

differentCCDandELEtemperatures.Usingtherelationshipderivedfromtheseobservations,weobtained

a method for determining the sensitivity for different temperature circumstances, such as during

touchdowns,androverandlanderreleaseoperations.

ONC-W1and-W2

l EmpiricalbiasmodelsoftheONC-W1and-W2werederivedbasedonthepreflightmeasurements.The

biaslevelduringthecruisephasewasfoundtobestableoverthepastthreeyears.

l DarklevelandthenumberofhotpixelsaresimilartotheONC-T.

l Theflat-fieldsfortheONC-W1and-W2werederivedbasedonthepre-flightmeasurements.However,

becauseasurface lightsource isneededtoassesstheaccuracyof these flat-fields,weareplanningto

evaluatetheflatnessusingwholediskimagesofRyuguduringtheapproachphase.

l CCDsensitivitiesfortheONC-W1and-W2arenewlyderivedbasedonthepreflightintegratingsphere

6 1055 52 364

measurementsandinflightstarandJupiterobservations.Theinflightsensitivitiesareconsistentwithina

10%difference.Thedifferenceinspectralshapesbetweenthesensitivitiesderivedfromtheintegrating

sphereandtheinflightreferenceobjectsaresuspectedtobethemajorsourceofthedifferences.

l StraylightintheONC-W1and-W2wascharacterized.Thestraylightcontaminationisbasicallyweaker

thanthatseenintheONC-T.Moreover,shorterexposuretimesforONC-W1and-W2yieldsquitesmall

straylighteffects.

Inadditiontotheglobalspectralmapping,weareplanningtoconducthighlightedobservations,suchas

sodium atmosphere observations, hydratedmineral observations, and high-resolution imaging associated

withtouchdowns.

AsodiumatmosphereonRyuguishighlypossible,asRyuguisexpectedhavebeenformedinavolatile-

rich region in the early Solar System. We evaluated the detectability of a sodium atmosphere by testing

observationsofJupiter’ssodiumtorus.Mainlyduetothedarknoise,thesodiumatmosphereonJupiterwas

notdetected,buttheupperlimitfordetectionwasevaluated.WeexpecttodetectNaatmosphereofseveral

10skR,similartocomets(Leblancetal.,2008),byasingleimageset(vandNa)andofseveral100Rbyusing

100ofsets.

Both0.7-µmabsorptionandsteepUVslopearepossible indicatorsofHydratedmineralsonasteroids.

Our calibration results suggested that theerror inmeasurement of 0.7-µm absorption is 1.6% and the in

measurementinUVslopeis<4%usingONC-T.Thesecalibrationaccuraciesimplythatwecandetect0.7-um

absorptionbandwithSNR~2,assumingthetypicalserpentineabsorptionof3-4%andalsowemaydistinguish

unheatedandhighlyheatedCM-likematerialsbasedonUVslope.

Duringthetouchdownsuptoafewmetersofaltitude,theradiativeheatfromRyuguisexpectedtoraise

theCCDtemperatureasmuchas20 ℃.ThishightemperaturemaydegradetheSNRsofthecamerasto150at

most. However, this may still be sufficient to detect the surface radiance variations of 3% caused by

illuminationconditionsoralbedovariation.Thus,quantitativeimagingwithveryhighresolutionuptoafew

mm/pixelispossible.

OurfirstinflightobservationofRyuguon26February2018showsverygoodagreementwiththeEarth-

basedobservationbyIshiguroetal.(2014).Thisconfirmsthatourradiometriccalibrationisrobust.Finally,

drasticdegradationintheONC-Tsystemhasnotbeenobservedduringthe3.5yearcruisephase.However,

someobservations suchas charge transfer timemeasurement, flat-fieldevaluation,and round-about stray

lightinvestigationbasedonRyugudiskobservations,andupdatingsensitivitybymorestarobservationsare

desiredtoimproveandmonitorourcalibration.Andalsotimedependentchangeshouldberecordedatregular

intervalstomonitorthehealthoftheONCsystemforoneandahalfyearsoftherendezvousperiod.

Acknowledgement

6 1055 52 364

We would like to show our greatest appreciation to Hayabusa2 team members who gave enormous

contributiontoachieveobservations.WewouldalsoliketoexpressourgratitudetoDr.D.Dominguewhogives

insightful and constructive comments.We alsowish to acknowledge the constructive reviews from Dr. C.

Güttlerandananonymousreviewer.ThisresearchissupportedbyJapanSocietyforthePromotionofScience

(JSPS)Core-to-Coreprogram“InternationalNetworkofPlanetarySciences”.

6 1055 52 364

AppendixA.SpecificationsofONC

DetailedspecificationsofONCthreecamerasaresummarizedhere.Mostoftheinformationarealready

publishedinKamedaetal.(2017)andSuzukietal.(2018).

Table A1. Performance of CCD (E2V CCD47-20 (AIMO)). Note that three cameras of ONC use same CCD

products.

ONC-T ONC-W1 ONC-W2

Format 1056(H)pixels×1024(V)pixels

(16×1024pixelsonbothsidesareOpticalBlackpixels)

CCDpixelsize 13µm

Gainfactor(measuredvalue) 20.95e-/DN 20.86e-/DN 20.11e-/DN

Read-outnoise(measuredvalue) 38.5e- 36.3e- 37.0e-

A/Dconversion 12bit

Full-well(measuredvalue) 91,000e- 84,000e- 96,000e-

Pixelsamplingrate 3MHz

TableA2.Weight,size,andelectricconsumptionofONChardwarecomponents.Thebasicdesignincluding

CCD,electronics,andopticsofW1andW2arethesame.Thus,theirelectricconsumptionisthesame,buttheir

dimensionandweightareslightlydifferentbecauseofdifferenceinradiatorandhooddesign.

Weight[kg] Size[mm] Electricconsumption[W]

ONC-T 2.1 430×155×134 8

ONC-W1 1.355 249×155×95 8

ONC-W2 1.295 222×155×95 8

ONC-AE 1.285 200×220×50 14

DE 2.67 95×221×170 15

TableA3.Focallengthanddistortioncoefficientforwide-,v-,andNa-bandsofONC-T.

Band Focallength[mm]* Distortioncoefficientε1

[1/pix2]*

Note

wide 120.50 −9.28 × 10,:

(RMSerror0.1pixels)

Suzukietal.(2018)

v 120.50 −6.766 × 10,:

(RMSerror0.3pixels)

Newly derived from the star image

hyb2_onc_20151105_065349_tvf_l2a.fit.

6 1055 52 364

Na 120.49 −1.024 × 10,;

(RMSerror0.3pixels)

Newly derived from the star image

hyb2_onc_20171015_000153_tnf_l2a.fit.

*DetailsforthedistortionmodelaredescribedinSuzukietal.(2018).

Itisnotedthatthefocallengthslistedinthistablearevalidonlywhencombinedwitheachdistortioncoefficient.

AppendixB.ComponentsofONC-TPhotometricSystem

ThequantumefficiencyofCCD,originallymeasured inroomtemperature(20℃), isshowninFig.B1.

Moreover,theCCDtemperaturecanchangethisfunctionshape.Anexampleoffunctionsofquantumefficiency

withtwodifferenttemperaturesbasedonthemanufacturer(E2V)derivedvalues.Thetransmissionsofband-

passfilters(sameasKamedaetal.,2017),theNDfilter,lensesandtheCCDcoverglassaredisplayedinFig.B2.

FigureB1.FunctionsofquantumefficiencyindifferentCCDtemperature.HighCCDtemperatureshowsthe

higherefficiencyinlongerwavelengthrange.

6 1055 52 364

FigureB2.(Top)Transmittancesoftheband-passfilters.(bottom)TransmittancesoftheNDfilter,lenses,and

theCCDcoverglass.

6 1055 52 364

AppendixC.CoordinateSystemandDefinitionofAttitudeoftheSpacecraft

FigureC1displaysthespacecraftcoordinatesystem.ThespacecraftattitudewithrespecttotheSuncouldbe

definedbyXPNLandYPNL,whichcorrespondtotheanglesoftheSunto+XSCand+YSCplaneofthespacecraft.

º ∙<FP = sinÌùn$ , º ∙ =FP = sinÏùn$,wheresisaunitvectorfromthespacecrafttotheSun.Theangles ¡ and ¬ arethespacecrafttwistingangle

describedinSuzukietal.(2018).TherelationshipbetweenXPNLandYPNLarederivedas

−sin¡ cos ¬ = sinÌùn$ , sin¡ sin ¬ = sinÏùn$.

FigureC1.Thespacecraftcoordinatesystemandtheimagecoordinatesystem.(afterSuzukietal.(2018))

References

Alekseeva,G.A.,Arkharov,A.A.,Galkin,V.D.,Hagen-Thorn,E.I.,Nikanorova,I.N.,Novikov,V.V.,Novopashenny,

V.B.,Pakhomov,V.P.,Ruban,E.V.,Shchegolev,D.E.(1996).ThePulkovoSpectrophotometricCatalogof

BrightStarsintheRangefrom320to1080nm.BalisticAstronomy,5,603-838.

Altwegg,K.,Balsiger,H.,Bar-Nun,A.,etal.(2014).67P/Churyumov-Gerasimenko,aJupiterfamilycometwith

ahighD/Hratio.Science,1261952.

Binzel,R.P.,Harris,A.W.,Bus,S.J.,Burbine,T.H.(2001).Spectralpropertiesofnear-Earthobjects:Palomar

and IRTF results for 48 objects including spacecraft targets (9969) Braille and (10302) 1989

ML.Icarus,151(2),139-149.

Bus,S.J.,andBinzel,R.P.(2002).PhaseIIoftheSmallMain-BeltAsteroidSpectroscopicSurvey.AFeature-

BasedTaxonomy.Icarus,158,146-177,https://doi.org/10.1006/icar.2002.6856.

Burnashev,B.I.(1996).VizieROnlineDataCatalog:Spectrophotometryof1588stars(Brunashev1985).

6 1055 52 364

Cloutis,E.A.etal.(2013).SpectralreflectancepropertiesofHEDmeteorites+CM2carbonaceouschondrites:

ComparisontoHEDgrainsizeandcompositionalvariationsandimplicationsforthenatureoflow-

albedofeaturesonAsteroid4Vesta.Icarus,223(2),850-877.

Lauretta,D.S.,OSIRIS-RExTeam.(2012).AnoverviewoftheOSIRIS-RExasteroidsamplereturnmission.43rd

LunarandPlanetaryScienceConference,#1659.

Elkins-Tanton, L. T. (2018). Asteroid 16 Psyche:NASA’s 14th DiscoveryMission.Elements: An International

MagazineofMineralogy,Geochemistry,andPetrology,14(1),68.

Feierberg, M. A., Lebofsky, L. A., Tholen, D. J. (1985). The nature of C-class asteroids from 3-μm

spectrophotometry.Icarus,63(2),183-191.

Fornasier,S.,Lantz,C.,Barucci,M.A.,Lazzarin,M.(2014).Aqueousalterationonmainbeltprimitiveasteroids:

Resultsfromvisiblespectroscopy.Icarus,233,163-178.

Gaffy,M.J.(1976).SpectralreflectancecharacteristicsoftheMeteoriteClasses.JournalofGeophysicalResearch,

81,905-920.

Gueymard,C.A.(2004).Thesun’stotalandspectralirradianceforsolarenergyapplicationsandsolarradiation

models.Solarenergy,76(4),423-453.

Hamuy, M., S. R., Walker, Suntzeff, N. B., A. R., Gigoux, P., Heathcote, Phillips, M. M. (1992). Southern

spectrophotometricstandards,AstronomicalSocietyofthePacific,Publications(ISSN0004-6280),vol.

104,no.677,pp.533-552.

Hamuy, M., Suntzeff, N. B., Heathcote, S. R., Walker, A. R., Gigoux, P., Phillips, M. M. (1994). Southern

spectrophotometricstandards,2,AstronomicalSocietyof thePacific,Publications(ISSN0004-6280),

vol.106,no.700,pp.566-589.

Hiroi,T.,Zolensky,M.E.,Pieters,C.M.,Lipschutz,M.E.(1996).ThermalmetamorphismoftheC,G,B,andF

asteroidsseenfromthe0.7μm,3μm,andUVabsorptionstrengthsincomparisonwithcarbonaceous

chondrites.Meteoritics&PlanetaryScience,31(3),321-327.

Ho,T.M.,Baturkin,V.,Grimm,C.,Grundmann,J.T.,Hobbie,C.,Ksenik,E.,Lange,C.,Sasaki,K.,Schlotterer,M.,

Talapina,M.,Termtanasombat,N.,Wejmo,E.,Witte,L.,Wrasmann,M.,Wübblels,G.(2017).MASCOT—

themobileasteroidsurfacescoutonboardtheHAYABUSA2mission.SpaceScienceReviews,208(1-4),

339-374.

Hoffleit,D.,andJaschek,C.(1991).VizieROnlineDataCatalog:BrightStarCatalogue,5thRevisedEd.

Ishiguro,M.,Nakamura,R.,Tholen,D.J.,Hirata,N.,Demura,H.,Nemoto,E.,Nakamura,A.M.,Higuchi,Y.,Sogame,

A., Yamamoto, A., Kitazato, K., Yokota, Y., Kubota, T., Hashimoto, T., Saito, J. (2010). The hayabusa

spacecraftasteroidmulti-bandimagingcamera(AMICA).Icarus,207(2),714-731.

Ishiguro,M.(2014).ScatteredlightcorrectionofHayabusa/AMICAdataandquantitativespectralcomparisons

ofItokawa.PublicationsoftheAstronomicalSocietyofJapan,66(3).

6 1055 52 364

Iwata,T.,Kitazato,K.,Abe,M.,Ohtake,M.,Arai,T.,Arai,T.,Hirata,N.,Hiroi,T.,Honda,C.,Imae,N.,Komatsu,M.,

Matsunaga,T.,Matsuoka,M.,Matsuura,S.,Nakamura,T.,Nakato,A.,Nakauchi,Y.,Osawa,T.,Senshu,H.,

Takagi,Y.,Tsumura,K.,Takato,N.,Watanabe,S.,Barucci,M.A.,Palomba,E.,Ozaki,M.(2017).NIRS3:the

nearinfraredspectrometeronHayabusa2.SpaceScienceReviews,208(1-4),317-337.

Irvine,W.M.,Lane,A.P.(1971).MonochromaticalbedosforthediskofSaturn.Icarus,15(1),18-26.

Ishiguro,M.,Kuroda,D.,Hasegawa,S.,etal.(2014).Opticalpropertiesof(162173)1999JU3:inpreparation

fortheJAXAHayabusa2samplereturnmission.TheAstrophysicalJournal,792(1),74.

Johnson,T.V.,Fanale,F.P. (1973).Opticalpropertiesof carbonaceous chondritesand their relationship to

asteroids.JournalofGeophysicalResearch,78(35),8507-8518.

Kameda,S.,Suzuki,H.,Cho,Y.,etal.(2015).DetectabilityofhydrousmineralsusingONC-Tcameraonboardthe

Hayabusa2spacecraft.AdvancesinSpaceResearch,56(7),1519-1524.

Kameda,S.,Suzuki,H.,Takamatsu,T.,Cho,Y.,Yasuda,T.,Yamada,M.,Sawada,H.,Honda,R.,Morota,T.,Honda,

C.,Sato,M.,Okumura,Y.,Shibasaki,K.,Ikezawa,S.,Sugita,S.(2017).Preflightcalibrationtestresultsfor

opticalnavigationcameratelescope(ONC-T)onboardtheHayabusa2spacecraft.SpaceScienceReviews,

208(1-4),17-31.

Karalidi,T.,Apai,D.,Schneider,G.,Hanson,J.R.,Pasachoff,J.M.(2015).Aeolus:AMarkovchainMonteCarlo

code for mapping ultracool atmospheres. An application on Jupiter and brown dwarf HST light

curves.TheAstrophysicalJournal,814(1),65.

Karkoschka,E.(1998).Methane,Ammonia,andTemperatureMeasurementsoftheJovianPlanetsandTitan

fromCCD–Spectrophotometry.Icarus,133(1),134-146.

Kieffer,H.H.,Stone,T.C.(2005).TheSpectralIrradianceoftheMoon,TheAstronomicalJournal,129,2887-

2901.

Killen,R.,Shemansky,D.,andMouawad,N.,ExpectedemissionfromMercury’sexosphericspecies,andtheir

ultraviolet-visiblesignatures(2009),TheAstrophysicalJournalSupplementseries,181,351-359.

Kitazato,K.,Iwata,T.,Abe,M.,Ohtake,M.,Tsumura,K.,Ishikawa,T.,Takato,N.,Nakauchi,Y.,Arai,T.,Senshu,H.,

Hirata,N.,Takagi,Y.,Hayabusa2NIRS3Team(2016).Near-InfraredSpectroscopyoftheEarthandMoon

DuringtheHayabusa2EarthSwing-By,47thLunarandPlanetaryScienceConference,#1903.

Kouyama,T.,Yokota,Y., Ishihara,Y.,Nakamura,R.,Yamamoto, S.,Matsunaga,T. (2016).Developmentof an

application scheme for the SELENE/SP lunar reflectance model for radiometric calibration of

hyperspectralandmultispectralsensors.PlanetaryandSpaceScience,124,76-83.

Lantz,C.,Brunetto,R.,Barucci,M.A.,Fornasier,S.,Baklouti,D.,Bourçois,J.,Godard,M.(2017).Ionirradiation

ofcarbonaceouschondrites:Anewviewofspaceweatheringonprimitiveasteroids.Icarus,285,43-57.

Lazzaro,D.,Barucci,M.A.,Perna,D.,Jasmim,F.L.,Yoshikawa,M.,Carvano,J.M.F.(2013).Rotationalspectraof

(162173)1999JU3,thetargetoftheHayabusa2mission.Astronomy&Astrophysics,549,L2.

6 1055 52 364

Leblanc,F.,Fulle,M.,Ariste,A.L.,Cremonese,G.,Doressoundiram,A.,Dalda,A.S.,andBelly,B.,CometMcNaught

C/2006P1:observationof thesodiumemissionbythesolartelescopeTHEMIS(2008),Astronomy&

Astrophysics,482,293-298.

Lebofsky, L. A. (1980). Infrared reflectance spectra of asteroids-A search for water of hydration.The

AstronomicalJournal,85,573-585.

LeCorre,L.,Sanchez,J.A.,Reddy,V.,Takir,D.,Cloutis,E.A.,Thirouin,A.,Becker,K.J.,Li,J.-Y.,Sugita,S.,Tatsumi,

E. (2018) Ground-based characterization of Hayabusa2 mission target asteroid 162173 Ryugu:

constrainingmineralogicalcompositioninpreparationforspacecraftoperations,MonthlyNoticesofthe

RoyalAstronomicalSociety,475,Issue1,21,614–623,https://doi.org/10.1093/mnras/stx3236

Levison,H.,Olkin,C.,Noll,K.,Marchi,S.,andtheLucyTeam(2017).Lucy:SurveyingtheDiversityoftheTrojan

Asteroids:TheFossilsofPlanetFormation.58thLunarandPlanetaryScienceConference,#2025.

Matsuoka,M., Nakamura, T., Kimura,Y., Hiroi, T., Nakamura, R., Okumura, S., Sasaki, S. (2015). Pulse-laser

irradiation experiments of Murchison CM2 chondrite for reproducing space weathering on C-type

asteroids.Icarus,254,135-143.

Mayorga,L.C.,Jackiewicz,J.,Rages,K.,West,R.A.,Knowles,B.,Lewis,N.,Marley,M.S.(2016).JUPITER’SPHASE

VARIATIONSFROMCASSINI:ATESTBEDFORFUTUREDIRECT-IMAGINGMISSIONS.TheAstronomical

Journal,152(6),209.

Moskovitz,N.A.,Abe,S.,Pan,K.S.,Osip,D.J.,Pefkou,D.,Melita,M.D.,Elias,M.,Kitazato,K.,Bus,S.J.,DeMeo,F.

E.,Binzel,R.P.,Abell,P.A.(2013).RotationalcharacterizationofHayabusaIItargetAsteroid(162173)

1999JU3.Icarus,224(1),24-31.

O’Brien,D.P.,Walsh,K.J.,Morbidelli,A.,Raymond,S.N.,Mandell,A.M.(2014).Waterdeliveryandgiantimpacts

inthe‘GrandTack’scenario.Icarus,239,74-84.

Ohtake,M.,Pieters,C.M.,Isaacson,P.,Besse,S.,Yokota,Y.,Matsunaga,T.,Boardman,J.,Yamomoto,S.,Haruyama,

J.,Staid,M.,Mall,U.,andGreen,R.O.,OneMoon,ManyMeasurements3:Spectralreflectance,Icarus,226,

364-374,2013.

Potter,A.,Morgan,T.H.(1985),DiscoveryofsodiumintheatmosphereofMercury,Science,229,651-653.

Potter, A. E., Morgan, T. H. (1988). Discovery of sodium and potassium vapor in the atmosphere of the

Moon.Science,241(4866),675-680.

Rahe,J.,McCracken,C.W.,Donn,B.D.(1976).Monochromaticandwhite-lightobservationsofCometBennett

1969i/1970II.AstronomyandAstrophysicsSupplementSeries,23,13-35.

Rivkin, A. S. (2012). The fraction of hydrated C-complex asteroids in the asteroid belt from SDSS

data.Icarus,221(2),744-752.

Sato,H.,Robinson,M.S.,Hapke,B.,Denevi,B.W.,Boyd,A.K.(2014).ResolvedHapkeparametermapsofthe

Moon.JournalofGeophysicalResearch:Planets,119(8),1775-1805.

6 1055 52 364

Sugita,S.,Kuroda,D.,Kameda,S.,et.al.(2013).Visiblespectroscopicobservationsofasteroid162173(1999

JU3)withtheGemini-Stelescope.44thLunarPlanetaryScienceConference,#2591.

Suzuki,H.,Yamada,M.,Kouyama,T.,Tatsumi,E.,Kameda,S.,Honda,R.,Sawada,H.,Ogawa,N.,Morota,T.,Honda,

C.,Sakatani,N.,Hayakawa,M.,Yokota,Y.,Yamamoto,Y.,Sugita,S.(2018).Initialinflightcalibrationfor

Hayabusa2opticalnavigation camera(ONC) for scienceobservationsofasteroidRyugu. Icarus,300,

341-359.

Stone,T.C.,Radiometriccalibrationstabilityandinter-calibrationofsolar-bandinstrumentsinorbitusingthe

moon.Proc.SPIE7081,EarthObservingSystemsXIII,70810X(20August2008);doi:10.1117/12.795227

Takir,D.,Emery,J.P.(2012).Outermainbeltasteroids:Identificationanddistributionoffour3-μmspectral

groups.Icarus,219(2),641-654.

Tholen,D.J.(1984).Asteroidtaxonomyfromclusteranalysisofphotometry.TheUniversityofArizona,Ph.D.

Thesis.

Yokota,Y.,Matsunaga,T.,Ohtake,M.,Haruyama,J.,Nakamura,R.,Yamamoto,S.,Ogawa,Y.,Morota,T.,Honda,C.,

Saiki,K.,Nagasawa,K.,Kitazato,K.,Sasaki,S.,Iwasaki,A.,Demura,H.,Hirata,N.,Hiroi,T.,Honda,R.,Iijima,

Y.,Mizutani,H.(2011).Lunarphotometricpropertiesatwavelengths0.5–1.6μmacquiredbySELENE

SpectralProfilerandtheirdependencyonlocalalbedoandlatitudinalzones.Icarus,215(2),639-660.

Yokota,Y.,Matsunaga,T.,Yamamoto,S.,Ohtake,M.,Haruyama,J.,Nakamura,R.,Ogawa,Y.,Morota,T.,Honda,

C.,Saiki,K.,Nagasawa,K.,Kitazato,K.,Sasaki,S.,Iwasaki,A.,Demura,H.,Hirata,N.,Hiroi,T.,Honda,R.,

Iijima,Y.,Mizutani,H.(2012).LunarPhotometricPropertiesatWavelengthover1.7MicronsAcquired

bySELENESpectralProfilerNIR-2Sensor.Proc.43rdLunarandPlanetaryScienceConference.

Yoneda,M., Kagitani,M., Tsuchiya,F., Sakanoi, T., andOkano, S., Brighteningevent seen in observations of

Jupiter’sextendedsodiumnebula(2015).Icarus,261,31-33.

Vilas,F.(2008).SpectralcharacteristicsofHayabusa2near-Earthasteroidtargets1621731999JU3and2001

QC34.TheAstronomicalJournal,135(4),1101.

Vilas,F.,Hatch,E.C.,Larson,S.M.,Sawyer,S.R.,Gaffey,M.J.(1993).Ferricironinprimitiveasteroids:A0.43-

μmabsorptionfeature.Icarus,102(2),225-231.

Vilas,F.,Jarvis,K.S.,Gaffey,M.J.(1994).Ironalterationmineralsinthevisibleandnear-infraredspectraof

low-albedoasteroids.Icarus,109(2),274-283.