Radiometric investigation of different snow covers in Svalbard

Radiometric฀investigation฀of฀different฀snow฀covers฀ in฀Svalbard Ruggero฀Casacchia,฀Francesca฀Lauta,฀ Rosamaria฀Salvatori,฀Anselmo฀Cagnati,฀ Mauro฀Valt฀&฀Jon฀B.฀Ørbæk This฀ paper฀ examines฀ the฀ relationship฀ between฀ reflectance฀ and฀ physical฀ characteristics฀of฀the฀snow฀cover฀in฀the฀Arctic.฀Field฀data฀were฀acquired฀ for฀ different฀ snow฀ and฀ ice฀ surfaces฀ during฀ a฀ survey฀ carried฀ out฀ at฀ NyÅlesund,฀Svalbard,฀in฀spring฀1998.฀In฀each฀measurement฀session฀reflectance฀ in฀ the฀ spectral฀ range฀ 350฀ -฀ 2500฀ nm,฀ snow฀ data฀ (including฀ temperature,฀grain฀size฀and฀shape,฀density฀and฀water฀content),฀surface฀layer฀ morphology,฀ and฀ vertical฀ profile฀ of฀ the฀ snow฀ pack฀ were฀ recorded.฀ A฀ detailed฀analysis฀of฀reflectance฀based฀on฀the฀physical฀structure฀of฀snow฀ was฀performed.฀Field฀reflectance฀data฀were฀also฀re-sampled฀at฀the฀spectral฀ intervals฀ of฀ Landsat฀ TM฀ to฀ compare฀ the฀ ability฀ of฀ identifying฀ different฀ snow฀ targets฀ at฀ discrete฀ wavelength฀ intervals.฀ This฀ analysis฀ shows฀ that฀ reliable฀data฀on฀snow฀structure฀and฀thickness฀are฀necessary฀to฀understand฀ albedo฀changes฀of฀the฀snow฀surfaces. R.฀Casacchia,฀F.฀Lauta฀&฀R.฀Salvatori,฀National฀Research฀Council,฀Institute฀of฀Atmospheric฀Pollution,฀Via฀ Salaria฀km฀29.300฀C.P.฀10,฀00016฀Monterotondo฀Stazione฀(Roma),฀Italy;฀A.฀Cagnati฀&฀M.฀Valt,฀ARPAV฀CVA,฀ Strada฀Passo฀Campolongo฀122,฀32020฀Arabba฀di฀Livinallongo฀(BL),฀Italy;฀J.฀B.฀Ørbæk,฀Norwegian฀Polar฀ Institute,฀Polar฀Environmental฀Centre,฀N-9296฀Tromsø,฀Norway. The฀multitemporal฀analysis฀of฀the฀snow-covered฀ surfaces฀in฀polar฀regions฀may฀provide฀data฀useful฀ for฀monitoring฀Earth’s฀climate฀changes.฀Radiometric฀data฀acquired฀by฀multispectral฀satellite฀images฀ recorded฀at฀wavelengths฀between฀400฀-฀2500฀nm฀ can฀support฀this฀study,฀owing฀to฀the฀link฀between฀ the฀ reflectance฀ and฀ physical฀ characteristics฀ of฀ snow฀at฀these฀wavelengths. Snow฀reflectance฀in฀the฀visible฀part฀of฀the฀electromagnetic฀spectrum฀is฀related฀to฀water฀and฀ice฀ content,฀ absorbing฀ impurities,฀ and฀ topographic฀ effects,฀ while฀ in฀ the฀ near-infrared฀ wavelengths฀ snow฀ reflectance฀ is฀ more฀ sensitive฀ to฀ grain฀ size฀ (Wiscombe฀&฀Warren฀1980;฀Warren฀1982;฀Dozier฀ 1989).฀ These฀ parameters,฀ together฀ with฀ snow฀ density,฀ are฀ generally฀ an฀ index฀ of฀ the฀ physical฀ conditions฀ of฀ snow,฀ and฀ thus฀ can฀ provide฀ information฀ about฀ the฀ relative฀ ages฀ of฀ various฀ snowCasacchia฀et฀al.฀2001฀year:฀Polar฀Research฀20(1),฀13–22 covered฀surfaces.฀Reflectance฀data฀as฀derived฀from฀ Landsat/Thematic฀ Mapper,฀ SPOT฀ and฀ NOAA/ AVHRR฀have฀been฀used฀to฀investigate฀snow/ice฀ surfaces฀ (Hall,฀ Chang฀ &฀ Siddalingaiah฀ 1988;฀ Hall,฀ Kovalick฀ et฀ al.฀ 1990;฀ Bourdelles฀ &฀ Fily฀ 1993;฀ Thomas฀ 1993;฀ Winther฀ 1993a).฀ Comparisons฀ of฀ in฀ situ฀ and฀ satellite฀ derived฀ reflectance฀ have฀also฀been฀carried฀out฀(Hall,฀Chang,฀Foster฀et฀ al.฀1989;฀Hall,฀Bindschadler฀et฀al.฀1990;฀Winther฀ 1993a,฀ 1993b;฀ Boresjö฀ Bronge฀ &฀ Bronge฀ 1999).฀ This฀comparison฀is฀important฀for฀a฀correct฀interpretation฀of฀satellite฀data฀and฀for฀a฀detailed฀analysis฀of฀snow’s฀spectral฀behaviour฀and฀metamorphic฀ conditions.฀ Furthermore,฀ ground฀ measurements฀ of฀albedo฀have฀been฀performed฀on฀different฀snow฀ covers฀and฀ice฀surfaces฀by฀Gerland฀et฀al.฀(1999),฀ Knap฀et฀al.฀(1999)฀and฀Winther฀et฀al.฀(1999). This฀paper฀presents฀data฀acquired฀in฀a฀survey฀ 13 Fig.฀1.฀Map฀of฀Brøggerhalvøya฀ (Brøgger฀Peninsula)฀showing฀the฀ location฀of฀the฀investigated฀sites฀ (see฀Table฀1฀for฀details);฀Svalbard฀ Archipelago฀inset.฀The฀surface฀ area฀is฀about฀14฀km฀by฀12฀km. carried฀out฀in฀Svalbard.฀Field฀data฀were฀analysed฀ to฀ identify฀ quantitative฀ changes฀ in฀ snow฀ cover฀ albedo฀ as฀ a฀ function฀ of฀ snow’s฀ physical฀ characteristics.฀ This฀ work฀ is฀ also฀ aimed฀ at฀ providing฀ ground-truth฀data฀to฀support฀satellite฀monitoring฀ of฀snow/ice฀surfaces. Spectral฀characteristics฀of฀snow Snow฀is฀a฀collection฀of฀ice฀grains฀and฀air฀and฀often฀ contains฀organic฀impurities฀like฀dust,฀soot,฀pollen฀ and฀other฀plant฀materials.฀The฀optical฀properties฀ of฀the฀snowpack฀in฀the฀visible฀and฀near-infrared฀ wavelengths฀ depend฀ on฀ grain฀ size฀ distribution,฀ thickness฀of฀the฀snowpack,฀occurrence฀of฀impurities฀ and฀ liquid฀ water฀ content฀ (Wiscombe฀ &฀ Warren฀ 1980;฀ Warren฀ 1982).฀ The฀ investigation฀ of฀snow/ice฀spectral฀properties฀usually฀takes฀into฀ account฀the฀reflectance,฀defined฀as฀the฀ratio฀of฀the฀ radiant฀ energy฀ reflected฀ by฀ a฀ body฀ to฀ that฀ incident฀upon฀it.฀The฀optical฀properties฀of฀ice฀can฀be฀ very฀different฀according฀to฀ice฀types.฀However,฀in฀ the฀visible฀wavelengths,฀ice฀is฀highly฀transparent,฀ so฀ that฀ its฀ albedo฀ may฀ change฀ according฀ to฀ the฀ amount฀ and฀ pattern฀ of฀ inclusions฀ (Wiscombe฀ &฀ Warren฀ 1980).฀ In฀ the฀ near-infrared฀ wavelengths฀ ice฀ is฀ more฀ absorptive,฀ so฀ that฀ albedo฀ depends฀ mainly฀on฀grain฀size฀(Warren฀&฀Wiscombe฀1980;฀ 14 Dozier฀ 1989).฀ Furthermore,฀ snow฀ reflectance฀ is฀ higher฀ in฀ the฀ visible฀ part฀ of฀ the฀ electromagnetic฀ spectrum,฀ decreasing฀ rapidly฀ at฀ longer฀ wavelengths,฀from฀about฀700฀nm.฀The฀increase฀of฀grain฀ size฀ gives฀ a฀ decrease฀ in฀ reflectance฀ all฀ over฀ the฀ spectral฀range฀from฀visible฀to฀shortwave฀infrared฀ (350฀-฀2500฀nm),฀particularly฀relevant฀in฀the฀infrared฀regions฀(Warren฀&฀Wiscombe฀1980;฀Warren฀ 1982;฀Warren฀et฀al.฀1986).฀The฀reflectance฀of฀both฀ wet฀and฀refrozen฀snow฀is฀usually฀lower฀than฀that฀ of฀dry฀snow฀due฀to฀the฀strong฀absorption฀coefficient฀of฀water฀and฀ice,฀especially฀at฀near-infrared฀ and฀infrared฀wavelengths. Field฀measurements Radiometric฀ and฀ snow฀ data฀ were฀ collected฀ between฀25฀April฀and฀10฀May฀1998,฀in฀four฀measurement฀sites฀on฀Brøggerhalvøya฀(Brøgger฀Peninsula),฀ in฀ the฀ area฀ surrounding฀ the฀ international฀ scientific฀ station฀ of฀ Ny-Ålesund฀ (Fig.฀ 1).฀ This฀ location฀was฀chosen฀because฀it฀offered฀the฀opportunity฀of฀measuring฀flat,฀snow/ice฀surfaces฀large฀ enough฀ to฀ be฀ sampled฀ on฀ satellite฀ images฀ and฀ far฀enough฀from฀dense฀human฀settlement฀to฀provide฀relatively฀uncontaminated฀snow฀spectral฀signatures.฀ As฀ expected฀ (Ørbæk฀ et฀ al.฀ 1999),฀ the฀ weather฀ conditions฀ in฀ this฀ time฀ period฀ did฀ not฀ Radiometric฀investigation฀of฀different฀snow฀covers฀in฀Svalbard cause฀ a฀ significant฀ melting฀ of฀ the฀ snow฀ cover,฀ while฀ some฀ snowfall฀ occurred฀ together฀ with฀ storms฀and฀clear฀sky฀days.฀All฀field฀data฀were฀collected฀ during฀ clear฀ sky฀ conditions;฀ air฀ temperatures฀were฀always฀below฀0฀°C฀(Table฀1).฀During฀ the฀ survey฀ sun฀ elevation฀ was฀ between฀ 23°฀ and฀ 27°.฀Even฀though฀this฀datum฀is฀crucial฀for฀reflectance฀measurements฀(Winther฀et฀al.฀1999),฀in฀this฀ paper฀our฀attention฀is฀mainly฀devoted฀to฀the฀physical฀structure฀of฀snow. The฀surveyed฀snow฀surfaces฀usually฀comprised฀ a฀mixture฀of฀grains฀of฀different฀classes;฀the฀snow฀ surface฀definitions฀given฀in฀this฀study฀are฀based฀ on฀the฀most฀representative฀grain฀type.฀In฀particular,฀ as฀ mentioned฀ in฀ Table฀ 2,฀ the฀ term฀ “equilibrium฀ forms”฀ is฀ applied฀ to฀ rounded฀ snow฀ grains฀ characterized฀ by฀ almost฀ no฀ growth,฀ because฀ of฀ the฀low฀thermal฀gradient.฀The฀analysed฀snow฀surfaces฀were:฀new฀snow,฀equilibrium฀forms,฀drifted฀ snow,฀and฀basal฀ice฀derived฀from฀freezing฀of฀meltwater. the฀top฀stratum.฀The฀penetration฀test฀was฀carried฀ out฀using฀a฀Swiss฀percussion฀probe฀(Rammsonde,฀ cone฀ tip฀ angle฀ 60°,฀ base฀ diameter฀ 40฀ mm,฀ tube฀ weight฀ 10฀ N/m,฀ ram฀ weight฀ 10฀ N).฀ The฀ vertical฀ profile฀led฀to฀the฀identification฀of฀different฀strata,฀ and฀ the฀ following฀ parameters฀ were฀ reported฀ or฀ estimated฀for฀each฀of฀them:฀grain฀shape฀and฀size,฀ hardness฀ (hand฀ test),฀ density฀ and฀ temperature.฀ The฀ water฀ content฀ was฀ investigated,฀ too,฀ using฀ a฀ Snow฀ Fork฀ (Toikka,฀ Finland),฀ though฀ liquid฀ water฀content฀of฀snow฀was฀null฀at฀all฀sites.฀Snow฀ description฀is฀based฀on฀the฀international฀classification฀of฀seasonal฀snow฀on฀the฀ground฀established฀ by฀the฀International฀Committee฀for฀Snow฀and฀Ice,฀ International฀Association฀of฀Scientific฀Hydrology฀ (Colbeck฀et฀al.฀1990).฀Surface฀roughness฀(furrow฀ distance฀and฀depth)฀was฀also฀measured฀in฀mm฀in฀ accordance฀with฀Colbeck฀et฀al.฀(1990).฀The฀main฀ physical฀characteristics฀of฀the฀surveyed฀snow฀surfaces฀are฀shown฀in฀Tables฀3฀and฀4. Spectroradiometric฀measurements Snow฀data A฀conventional฀survey฀of฀snow฀surfaces฀was฀carried฀out฀at฀all฀measurement฀sites,฀including฀a฀penetration฀test฀and฀a฀vertical฀profile฀with฀regard฀to฀ Snow฀ and฀ ice฀ reflectance฀ was฀ acquired฀ by฀ the฀ field฀spectroradiometer฀Fieldspec฀FR฀(Analytical฀ Spectral฀Devices฀Inc.,฀Boulder,฀CO),฀covering฀the฀ wavelength฀range฀350฀-฀2500฀nm,฀and฀calculated฀ Table฀1.฀Characteristics฀of฀the฀investigated฀sites. Surface฀type฀ ฀ ฀ Symbol฀ ฀ ฀ Site฀ ฀ ฀ Latitude฀/฀ longitude฀ ฀ Day฀ ฀ ฀ Time฀ ฀ ฀ Average฀฀ snow฀thick-฀ ness฀(mm) Air฀T฀฀ Sun (°C)฀ elev. New฀snow฀ ฀ N฀ ฀ Storvatnet฀ ฀ 78°฀55’฀28”฀N฀ 11°฀50’฀11”฀E 04฀May฀ 13:30฀ 200฀(+760)฀ -2.9฀ 27° Equilibrium฀forms฀ ฀ E฀ ฀ Storvatnet฀ ฀ 78°฀55฀‘28”฀N฀ 11°฀50’฀11”฀E 27฀April฀ 13:30฀ 600฀ -9.6฀ 24° Equilibrium฀forms฀on฀ basal฀ice฀(40฀mm)฀ E4฀ ฀ Tvillingvatna฀ ฀ 78°฀55’฀16”฀N฀ 11°฀55’฀37”฀E 28฀April฀ 15:10฀ 40฀ -5.7฀ 24° Equilibrium฀forms฀on฀ basal฀ice฀(10฀mm)฀ E1฀ ฀ Tvillingvatna฀ ฀ 78°฀55’฀16”฀N฀ 11°฀55’฀37”฀E 28฀April฀ 15:00฀ 10฀ -5.7฀ 24° Smooth฀drifted฀snow฀ ฀ SD฀ ฀ Storvatnet฀ ฀ 78°฀55’฀28”฀N฀ 11°฀50’฀11”฀E 01฀May฀ 14:00฀ 760฀ -4.3฀ 26° Drifted฀snow฀ (barchans)฀ DB฀ ฀ Storvatnet฀ ฀ 78°฀55’฀28”฀N฀ 11°฀50’฀11”฀E 01฀May฀ 13:15฀ 760฀ -4.3฀ 26° Drifted฀snow฀ (ripples)฀ DR฀ ฀ Storvatnet฀ ฀ 78°฀55’฀28”฀N฀ 11°฀50’฀11”฀E 01฀May฀ 12:00฀ 760฀ -4.3฀ 26° Drifted฀snow฀ ฀ DE฀ ฀ Midre฀Lovénbreen฀ ฀ 78°฀54’฀01”฀N฀ 12°฀02’฀56”฀E 05฀May฀ 16:00฀ >800฀ -6.3฀ 24° Melt–freeze฀crust฀ ฀ C฀ ฀ Stuphallet฀ ฀ 78°฀57’฀58”฀N฀ 11°฀37’฀52”฀E 04฀May฀ 12:00฀ 200฀ -3฀ 26° Basal฀ice฀ ฀ I฀ ฀ Tvillingvatna฀ ฀ 78°฀55’฀16”฀N฀ 11°฀55’฀37”฀E 28฀April฀ 15:30฀ –฀ -5.7฀ 23° Casacchia฀et฀al.฀2001฀year:฀Polar฀Research฀20(1),฀13–22 15 ฀ ฀ Fig.฀2.฀Reflectance฀curves฀of฀ new฀snow฀(N),฀equilibrium฀ forms฀(E),฀equilibrium฀forms฀ (40฀mm)฀on฀basal฀ice฀(E4),฀ equilibrium฀forms฀(10฀mm)฀on฀ basal฀ice฀(E1),฀bare฀basal฀ice฀(I)฀ and฀melt–freeze฀crust฀(C). as฀ the฀ ratio฀ of฀ incident฀ solar฀ radiation฀ reflected฀ from฀ the฀ snow฀ target฀ and฀ the฀ incident฀ radiation฀ reflected฀from฀a฀white฀reference฀Spectralon฀(about฀ 30฀cm฀x฀30฀cm),฀known฀as฀a฀Lambertian฀reflector.฀ This฀ratio฀gives฀the฀reflectance฀factor,฀while฀the฀ absolute฀ reflectance฀ is฀ obtained฀ by฀ multiplying฀ this฀reflectance฀factor฀with฀the฀reflectance฀spectrum฀of฀the฀panel.฀Used฀in฀the฀field฀under฀harsh฀ environments,฀the฀panel฀requires฀proper฀maintenance฀and฀calibration฀for฀correct฀calculation฀of฀the฀ reflectance฀ data฀ (especially฀ concerning฀ organic฀ impurities฀affecting฀the฀shorter฀wavelengths). The฀Spectralon฀was฀re-calibrated฀at฀the฀Norwegian฀ Polar฀ Institute’s฀ (NPI)฀ Optical฀ Calibration฀ Laboratory฀ in฀ Ny-Ålesund฀ during฀ the฀ campaign฀ by฀intercomparison฀with฀a฀primary฀standard฀reference฀Spectralon.฀The฀new฀calibration฀curve฀of฀ the฀ panel฀ was฀ thereby฀ obtained,฀ allowing฀ absolute฀reflectance฀to฀be฀calculated฀with฀an฀error฀of฀ 2฀%;฀this฀value฀is฀derived฀from฀the฀standard฀deviation฀ of฀ the฀ spectral฀ response฀ of฀ our฀ spectralon฀ (in฀ the฀ entire฀ 350฀ -฀ 2500฀ nm฀ wavelength฀ range)฀ Table฀2.฀Definition฀of฀snow-related฀terms฀used฀in฀this฀paper. New฀snow฀ Equilibrium฀ forms Drifted฀snow฀ Dry฀snow฀ ฀ Melt–freeze฀ crust฀ Basal฀ice฀ ฀ 16 Snow฀deposited฀within฀24฀hours Rounded฀crystals,฀shaped฀by฀a฀slow฀growth Snow฀deposited฀or฀altered฀by฀the฀wind Deposited฀snow฀that฀has฀not฀been฀subject฀ to฀melting฀or฀to฀infiltration฀of฀liquid฀water Hard฀and฀generally฀thin฀layer฀formed by฀recognizable฀melt–freeze฀polycrystals Ice฀occurring฀at฀the฀base฀of฀the฀snow cover฀formed฀from฀freezing฀of฀meltwater computed฀with฀respect฀to฀the฀NPI’s฀standard฀reference. Other฀sources฀of฀errors฀or฀noise฀in฀field฀spectroradiometric฀ data฀ may฀ have฀ included฀ incorrect฀ viewing฀ geometry฀ in฀ data฀ acquisition,฀ random฀ noise฀ produced฀ by฀ the฀ electronic฀ components฀ of฀ the฀instrument,฀atmospheric฀water฀vapour฀absorption฀band,฀and฀the฀low฀atmospheric฀irradiance฀at฀ wavelengths฀ of฀ 1400฀ nm฀ and฀ beyond฀ 1700฀ nm,฀ giving฀ a฀ low฀ signal฀ to฀ noise฀ ratio฀ (S/N).฀ A฀ correct฀orientation฀of฀the฀spectroradiometer฀over฀the฀ panel฀and฀the฀surface฀is฀necessary฀for฀snow฀targets,฀particularly฀in฀the฀visible฀wavelengths฀up฀to฀ 900฀ nm,฀ to฀ avoid฀ reflectance฀ values฀ that฀ exceed฀ 100฀%฀and฀reflectance฀curves฀that฀have฀an฀anomalous฀ pattern.฀ The฀ S/N฀ ratio฀ can฀ be฀ increased฀ by฀ increasing฀the฀number฀of฀measurements฀for฀every฀ radiometric฀acquisition. During฀data฀acquisition฀particular฀attention฀was฀ devoted฀ to฀ snow฀ surface฀ roughness฀ variability฀ (Table฀ 4).฀ The฀ different฀ pattern฀ and฀ size฀ of฀ surface฀furrows฀may฀cause฀the฀spectral฀response฀to฀ vary,฀due฀both฀to฀shadowing฀effects฀and฀to฀backscattering.฀ To฀ identify฀ all฀ the฀ snow฀ targets฀ to฀ sample,฀a฀detailed฀observation฀of฀all฀the฀surface฀ variations฀at฀each฀location฀was฀carried฀out฀before฀ every฀measurement฀session.฀Grain฀size฀and฀shape฀ were฀detected฀for฀each฀target.฀Snow฀observations฀ were฀performed฀on฀the฀same฀target฀immediately฀ after฀spectral฀measurements. Measurements฀ were฀ acquired฀ 500฀ mm฀ above฀ the฀target,฀with฀a฀field฀of฀view฀of฀25°,฀thus฀covering฀ground฀areas฀of฀230฀mm฀x฀230฀mm.฀Special฀ care฀was฀taken฀to฀ensure฀that฀the฀radiometer฀was฀ nadir฀ viewing฀ over฀ the฀ surveyed฀ surfaces.฀ The฀ Radiometric฀investigation฀of฀different฀snow฀covers฀in฀Svalbard Fig.฀3.฀Reflectance฀curves฀of฀ new฀snow฀(N)฀and฀drifted฀snows฀ (SD,฀DB,฀DR฀and฀DE). Fieldspec฀spectroradiometer฀allows฀the฀improvement฀ of฀ the฀ S/N฀ ratio฀ of฀ each฀ spectral฀ curve฀ by฀ selecting฀a฀specific฀number฀of฀samples฀to฀be฀averaged฀ to฀ obtain฀ the฀ final฀ spectral฀ curve.฀ According฀to฀the฀spectrometer’s฀manufacturer,฀a฀sample฀ average฀ from฀ 10฀ to฀ 150฀ is฀ sufficient;฀ a฀ larger฀ number฀of฀samples฀would฀require฀a฀longer฀acquisition฀time,฀and฀stable฀sky฀conditions.฀Because฀of฀ the฀weather฀during฀our฀survey,฀a฀spectrum฀averaging฀ of฀ 50฀ samples฀ was฀ chosen,฀ after฀ having฀ checked฀ that฀ reflectance฀ acquired฀ with฀ a฀ higher฀ number฀ of฀ acquisition฀ gave฀ the฀ same฀ results.฀ Twenty฀ to฀ thirty฀ spectral฀ curves฀ were฀ acquired฀ for฀ every฀ target฀ in฀ order฀ to฀ have฀ a฀ statistically฀ meaningful฀ sample฀ of฀ each฀ target;฀ the฀ number฀ of฀ acquisitions฀ is฀ a฀ user฀ decision,฀ depending฀ on฀ how฀stable฀the฀signal฀to฀be฀recorded.฀Increasing฀ the฀number฀of฀spectral฀curves฀for฀each฀target฀also฀ contributes฀ to฀ the฀ reduction฀ of฀ random฀ errors,฀ resulting฀ in฀ a฀ better฀ spectral฀ characterization฀ of฀ the฀target฀itself.฀The฀curves฀discussed฀below฀are฀ absolute฀reflectances฀computed฀as฀mean฀values฀of฀ all฀the฀acquisitions฀collected฀for฀each฀target. Table฀3.฀Physical฀characteristics฀of฀the฀investigated฀targets.฀Symbols฀used฀in฀Figs.฀2,฀3฀and฀4:฀L฀=฀Layer฀thickness฀(mm);฀ρ฀=฀ density฀(kg/m3);฀R฀=฀hardness;฀T฀=฀snow฀temperature฀(°C). Surface฀type฀ Symbol฀ L฀ New฀snow฀ N฀ ฀ ฀ Equilibrium฀forms฀ E฀ ฀ ฀ ฀ ฀ Equilibrium฀forms฀ E4฀ (40฀mm)฀on฀basal฀ice฀ ฀ ฀ ฀ ฀ ฀ ฀ ฀ Equilibrium฀forms฀ E1฀ (10฀mm)฀on฀basal฀ice฀ ฀ Smooth฀drifted฀snow฀ SD฀ Drifted฀snow฀(barchans)฀ DB฀ Drifted฀snow฀(ripples)฀ DR฀ Drifted฀snow฀ DE฀ ฀ ฀ ฀ ฀ Melt–freeze฀crust฀ C฀ Basal฀ice฀ I฀ 30฀ ฀ 80฀ ฀ ฀ 40฀ ฀ ฀ ฀ ฀ 10฀ ฀ 110฀ 110฀ 110฀ 20฀ ฀ ฀ 30฀ 150฀ ρ฀ 55฀ ฀ 145฀ ฀ ฀ 135฀ ฀ ฀ ฀ ฀ n.d.฀ ฀ 385฀ 385฀ 385฀ 100฀ ฀ ฀ n.d.฀ n.d.฀ R฀ T฀ very฀low฀ -0.7฀ ฀ ฀ very฀low฀ -11.9฀ ฀฀ ฀ ฀ ฀ very฀low฀ -6.2฀ ฀ ฀ ฀ ฀ ฀ ฀ ฀ ฀ very฀low฀ n.d.฀ ฀ ฀ high฀ -5.8฀ high฀ -5.8฀ high฀ -5.8฀ very฀low฀ -5.4฀ ฀ ฀ ฀ ฀ high฀ -1.5฀ very฀high฀ n.d. Casacchia฀et฀al.฀2001฀year:฀Polar฀Research฀20(1),฀13–22 Snow฀crystals partly฀decomposed฀particles฀(1.5฀mm) and฀stellar฀dendrites฀(3฀mm) small฀rounded฀particles฀(0.3฀-฀0.5฀mm), highly฀broken฀particles฀(0.5฀mm), rare฀surface฀hoar฀crystals฀(0.5฀mm) 10฀mm:฀partly฀decomposed฀precipitation particles฀(1฀mm),฀stellar฀dendrites฀(2.5฀mm), surface฀hoar฀crystals฀(0.5฀mm); 30฀mm:฀small฀rounded฀particles฀(0.5฀mm), surface฀hoar฀crystals฀(1฀mm) partly฀decomposed฀precipitation฀particles฀(1฀mm),฀ surface฀hoar฀crystals฀(0.5฀mm) small฀rounded฀particles฀(0.3฀mm) small฀rounded฀particles฀(0.3฀mm) small฀rounded฀particles฀(0.3฀mm) 20฀mm:฀partly฀decomposed฀particles฀(1.5฀mm), stellar฀dendrites฀(3฀mm),฀surface฀hoar; 20฀mm:฀small฀rounded฀particles฀(0.4฀mm) mixed฀forms฀(0.8฀mm)฀and฀rounded฀polycrystals฀(1.2฀mm) 17 Fig.฀4.฀Reflectance฀values฀re-sampled฀in฀discrete฀wavelength฀ intervals:฀ (a)฀ equilibrium฀ forms฀ (E,฀ E4,฀ E1),฀ basal฀ ice฀ (I),฀ melt–freeze฀ crust฀ (C);฀ (b)฀ drifted฀ snows฀ (SD,฀ DB,฀ DR฀ and฀ DE).฀New฀snow฀(N)฀is฀shown฀in฀both฀plots.฀ Analysis฀of฀spectra At฀ wavelengths฀ between฀ 350฀ and฀ 2500฀ nm฀ the฀ snow฀ physical฀ characteristics฀ that฀ mostly฀ affect฀ reflectance฀are฀grain฀size,฀presence฀of฀absorbing฀ impurities,฀ water฀ content฀ and฀ surface฀ morphologic฀and฀geometric฀characteristics฀(Wiscombe฀&฀ Warren฀1980;฀Warren฀1982;฀Dozier฀1989). Figure฀ 2฀ presents฀ the฀ spectral฀ albedo฀ of฀ new฀ snow฀ (N),฀ melt–freeze฀ crust฀ (C),฀ equilibrium฀ forms฀(E),฀equilibrium฀forms฀with฀thicknesses฀of฀ 40฀ mm฀ (E4)฀ and฀ 10฀ mm฀ (E1)฀ on฀ basal฀ ice,฀ and฀ bare฀basal฀ice฀(I);฀Fig.฀3฀shows฀new฀snow฀(N)฀and฀ drifted฀ snow฀ surfaces฀ sampled฀ at฀ Midre฀ Lovénbreen฀ (DE)฀ and฀ three฀ drifted฀ snow฀ surfaces— smooth฀ (SD),฀ with฀ surface฀ barchans฀ (DB),฀ with฀ surface฀ ripples฀ (DR)—sampled฀ at฀ Storvatnet.฀ New฀ snow฀ reflectance฀ is฀ shown฀ in฀ both฀ Figs.฀ 2฀ and฀3,฀as฀it฀is฀used฀as฀a฀reference฀to฀which฀compare฀ the฀ spectral฀ response฀ of฀ the฀ other฀ snows.฀ Tables฀3฀and฀4฀report฀the฀main฀physical฀characteristics฀ of฀ the฀ surveyed฀ surfaces.฀ The฀ general฀ trends฀of฀the฀reflectance฀values฀(Figs.฀2,฀3)฀agree฀ with฀those฀reported฀by฀other฀authors฀(Wiscombe฀ &฀Warren฀1980;฀Warren฀1982;฀Warren฀et฀al.฀1986;฀ Hall,฀Chang฀&฀Siddalingaiah฀1988;฀Dozier฀1989;฀ Hall,฀ Chang,฀ Foster฀ et฀ al.฀ 1989;฀ Hall,฀ Bindschadler฀et฀al.฀1990;฀Hall,฀Kovalick฀et฀al.฀1990;฀Zibordi฀ et฀ al.฀ 1996;฀ Winther฀ et฀ al.฀ 1998):฀ reflectance฀ is฀ higher฀ in฀ the฀ visible฀ region฀ of฀ the฀ spectrum,฀ and฀ decreases฀ at฀ longer฀ wavelengths.฀ Atmospheric฀ water฀ vapour฀ absorption฀ causes฀ low฀ S/N฀ ratio฀affecting฀all฀the฀measurements฀at฀1400฀nm,฀ between฀ 1700฀ -฀ 2000฀ nm฀ and฀ beyond฀ 2300฀ nm;฀ two฀ reflectance฀ minima฀ can฀ also฀ be฀ observed฀ at฀ 1500฀nm฀and฀2000฀nm.฀Beyond฀1700฀nm฀the฀incident฀radiation฀is฀very฀low;฀this฀is฀a฀further฀source฀ of฀radiometric฀error฀affecting฀infrared฀data. The฀ new฀ snow฀ curve฀ (N)฀ pattern฀ shown฀ in฀ Fig.฀ 2฀ is฀ typical฀ for฀ this฀ surface:฀ its฀ measured฀ albedo฀ is฀ higher฀ between฀ 350฀ -฀ 700฀ nm฀ (values฀ of฀0.94฀-฀0.98),฀and฀shows฀lower฀values฀at฀wavelength฀greater฀than฀700฀nm.฀Its฀high฀reflectance฀is฀ owed฀to฀the฀shape฀and฀size฀of฀the฀new฀snow฀grains฀ (Table฀3),฀which฀have฀not฀been฀altered฀since฀dep- Table฀4.฀Surface฀roughness฀of฀the฀investigated฀targets. Surface฀type฀ New฀snow฀ Equilibrium฀forms฀ Equilibrium฀forms฀ (40฀mm)฀on฀basal฀ice Equilibrium฀forms฀ (10฀mm)฀on฀basal฀ice Smooth฀drifted฀snow฀ Drifted฀snow฀(barchans)฀ Drifted฀snow฀(ripples)฀ Drifted฀snow฀ Melt–freeze฀crust฀ Basal฀ice฀ 18 Symbol฀ Surface฀roughness฀ N฀ E฀ E4฀ small฀irregular฀ripples฀ smooth,฀little฀ripples฀ smooth Furrow฀distance฀ Furrow฀depth E1฀ smooth SD฀ DB฀ DR฀ DE฀ C฀ smooth snow฀barchans฀ irregular฀ripples฀ irregular฀furrows฀ concave฀furrows฀ I฀ smooth 7฀mm฀ 0.5฀-฀1฀mm฀ 2฀mm 0.1฀mm 5฀mm฀ 5฀-฀10฀mm฀ 4฀mm฀ 5฀mm฀ 0.5฀-฀1฀mm 1฀-฀1.5฀mm 0.5฀-฀1฀mm 1฀mm Radiometric฀investigation฀of฀different฀snow฀covers฀in฀Svalbard osition.฀ The฀ equilibrium฀ forms฀ (E)฀ reflectance฀ curve฀ is฀ close฀ to฀ that฀ of฀ new฀ snow฀ (N)฀ in฀ the฀ visible฀ part฀ of฀ the฀ spectrum฀ (within฀ 3฀ %฀ up฀ to฀ 900฀nm).฀For฀wavelengths฀greater฀than฀900฀nm,฀ albedo฀of฀E฀is฀considerably฀lower฀than฀that฀of฀new฀ snow฀ (N),฀ probably฀ due฀ to฀ the฀ presence฀ of฀ surface฀ hoar฀ and฀ to฀ different฀ grain฀ size฀ and฀ shape.฀ The฀“equilibrium฀forms”฀grains฀are฀more฀rounded฀ than฀ new฀ snow฀ grains,฀ giving฀ a฀ reduced฀ reflectance฀at฀these฀wavelengths฀(Wiscombe฀&฀Warren฀ 1980;฀ Warren฀ 1982;฀ Zibordi฀ et฀ al.฀ 1996).฀ The฀ “equilibrium฀forms”฀(40฀mm)฀on฀ice฀curve฀(E4)฀ shows฀high฀reflectance฀(0.8฀-฀0.89)฀up฀to฀900฀nm,฀ and฀is฀about฀10฀%฀lower฀than฀new฀snow฀(N)฀and฀ equilibrium฀ forms฀ (E)฀ curves;฀ the฀ comparison฀ between฀the฀curves฀of฀equilibrium฀forms฀(E)฀and฀ of฀ equilibrium฀ forms฀ on฀ basal฀ ice฀ (E4)฀ reveals฀ that฀the฀increased฀grain฀size,฀the฀presence฀of฀surface฀hoar฀and฀of฀ice฀below฀the฀snow฀cover฀determine฀ a฀ lower฀ reflectance฀ of฀ E4฀ with฀ respect฀ to฀ E.฀In฀the฀infrared,฀beyond฀1600฀nm฀the฀opposite฀ case฀is฀observed,฀as฀E4฀reflectance฀is฀higher฀than฀ reflectance฀ measured฀ on฀ E,฀ because฀ E4฀ surface฀ snow฀grains฀are฀less฀rounded,฀as฀revealed฀by฀snow฀ field฀data.฀The฀equilibrium฀forms฀(10฀mm)฀on฀ice฀ (E1)฀reflectance฀is฀considerably฀lower฀than฀previous฀ ones,฀ owing฀ to฀ the฀ presence฀ of฀ ice฀ close฀ to฀ the฀ surface฀ and฀ of฀ surface฀ hoar.฀ This฀ curve฀ pattern฀is฀similar฀to฀the฀basal฀ice฀curve฀(I)฀between฀ 350฀ -฀ 1200฀ nm;฀ reflectance฀ values฀ of฀ the฀ two฀ curves฀ are฀ very฀ close฀ up฀ to฀ 550฀ nm฀ and฀ appear฀ more฀separate฀as฀wavelength฀increases.฀Between฀ 550฀and฀600฀nm,฀the฀E1฀curve฀shows฀a฀reflectance฀ maximum฀of฀0.73,฀followed฀by฀a฀sharp฀decrease฀ beyond฀ 600฀ nm,฀ although฀ it฀ never฀ falls฀ to฀ zero,฀ as฀basal฀ice฀albedo฀does.฀We฀can฀therefore฀infer,฀ according฀ to฀ the฀ measurements฀ carried฀ out,฀ that฀ 10฀ mm฀ of฀ snow฀ considerably฀ affect฀ ice฀ albedo,฀ mainly฀ in฀ the฀ infrared฀ wavelengths.฀ The฀ basal฀ ice฀reflectance฀(I)฀is฀lower฀than฀any฀other฀surface฀ at฀ wavelengths฀ greater฀ than฀ 450฀ nm.฀ Ice฀ reflectance฀increases฀from฀350฀nm฀up฀to฀550฀nm,฀where฀ a฀ maximum฀ of฀ 0.68฀ occurs,฀ and฀ then฀ rapidly฀ decreases฀ to฀ zero฀ because฀ of฀ the฀ high฀ spectral฀ absorption฀at฀1200฀nm.฀Ice฀spectral฀behaviour฀has฀ been฀investigated฀by฀Wiscombe฀&฀Warren฀(1980),฀ Warren฀ (1982),฀ Dozier฀ (1989);฀ in฀ particular,฀ the฀ minimum฀in฀the฀ice฀absorption฀coefficient฀curve฀ reported฀by฀Dozier฀(1989)฀seems฀to฀correspond฀to฀ the฀ reflectance฀ maximum฀ of฀ curve฀ (I)฀ in฀ Fig.฀ 2.฀ According฀to฀these฀reflectance฀data,฀the฀presence฀ of฀ a฀ snow฀ cover฀ on฀ an฀ ice฀ stratum฀ significantly฀ affects฀its฀reflectance.฀At฀visible฀and฀NIR฀waveCasacchia฀et฀al.฀2001฀year:฀Polar฀Research฀20(1),฀13–22 Fig.฀5.฀Reflectance฀values฀(Rs)฀re-sampled฀in฀discrete฀wavelength฀intervals฀and฀normalized฀with฀respect฀to฀new฀snow฀reflectance฀ (Rn):฀ (a)฀ equilibrium฀ forms฀ (E,฀ E4,฀ E1),฀ basal฀ ice฀ (I),฀melt–freeze฀crust฀(C);฀(b)฀drifted฀snows฀(SD,฀DB,฀DR฀and฀ DE).฀ lengths฀up฀to฀1350฀nm฀there฀is฀a฀wide฀gap฀between฀ ice฀with฀a฀snow฀cover฀thickness฀of฀40฀mm฀(E4),฀ ice฀covered฀by฀10฀mm฀of฀snow฀(E1)฀and฀basal฀ice฀ (I);฀beyond฀1350฀nm฀the฀differences฀are฀reduced.฀ Moreover,฀comparing฀new฀snow฀(N),฀melt–freeze฀ crust฀(C)฀and฀basal฀ice฀(I)฀curves,฀we฀observe฀that฀ increasing฀metamorphism฀and฀age฀of฀grains฀cause฀ a฀ progressive฀ decrease฀ in฀ reflectance฀ at฀ wavelength฀between฀350฀and฀1400฀nm. The฀albedo฀measurements฀of฀new฀snow฀and฀of฀ drifted฀snows฀measured฀respectively฀at฀Storvatnet฀ (N,฀SD,฀DB,฀DR)฀and฀Midre฀Lovénbreen฀(DE)฀are฀ shown฀in฀Fig.฀3.฀At฀Storvatnet฀drifted฀snow฀surfaces฀ have฀ the฀ same฀ physical฀ and฀ granulometric฀ characteristics,฀ but฀ different฀ surface฀ roughness,฀ according฀to฀which฀three฀kinds฀of฀drifted฀snows฀ have฀been฀defined฀and฀described฀in฀Tables฀3฀and฀ 4.฀In฀the฀visible฀part฀of฀the฀spectrum฀all฀curves฀are฀ similar฀ and฀ show฀ reflectance฀ values฀ higher฀ than฀ 0.93;฀at฀700฀nm฀they฀appear฀more฀distinctive฀and฀ from฀1000฀up฀to฀2500฀nm฀a฀progressive฀decrease฀ in฀reflectance฀can฀be฀noticed,฀from฀new฀snow฀(N),฀ followed฀by฀drifted฀snow฀with฀irregular฀furrows฀ (DE),฀ and฀ drifted฀ snows฀ characterized,฀ respec19 tively,฀ by฀ barchans฀ (DB),฀ ripples฀ (DR)฀ and฀ a฀ smooth฀ surface฀ (SD).฀ Due฀ to฀ the฀ occurrence฀ of฀ surface฀hoar,฀at฀infrared฀wavelengths฀reflectance฀ of฀ drifted฀ snow฀ (DE)฀ is฀ lower฀ than฀ that฀ of฀ new฀ snow฀(Fig.฀3).฀The฀lower฀albedo฀measured฀at฀Storvatnet฀ (SD,฀ DB฀ and฀ DR)฀ with฀ respect฀ to฀ that฀ acquired฀at฀Midre฀Lovénbreen฀(DE)฀is฀due฀to฀the฀ occurrence฀ of฀ smaller฀ and฀ more฀ rounded฀ snow฀ grains.฀Moreover,฀the฀main฀differences฀in฀albedo฀ among฀the฀drifted฀snow฀surfaces฀at฀Storvatnet฀can฀ be฀seen฀at฀about฀1100฀and฀1300฀nm,฀where฀probably฀ drifted฀ snows฀ DR฀ and฀ DB฀ surface฀ roughness฀lead฀to฀higher฀reflectance฀than฀that฀of฀smooth฀ drifted฀snow฀(SD). Discussion The฀ analysis฀ of฀ the฀ reflectance฀ curves฀ (Figs.฀ 2,฀ 3)฀has฀shown฀the฀possibility฀of฀identifying฀different฀kinds฀of฀snow/ice฀surfaces฀based฀on฀the฀spectral฀ response฀ between฀ wavelengths฀ of฀ 350฀ and฀ 2500฀nm.฀The฀same฀curves฀have฀been฀analysed฀in฀ discrete฀ spectral฀ intervals฀ covered฀ by฀ the฀ Landsat฀5฀Thematic฀Mapper,฀with฀the฀purpose฀of฀comparing฀field฀reflectance฀with฀satellite-derived฀data฀ (Fig.฀4).฀In฀addition,฀we฀observe฀in฀our฀data฀(Figs.฀ 2,฀ 3)฀ significant฀ reflectance฀ differences฀ even฀ at฀ 1030฀ -฀ 1085฀ nm฀ and฀ 1260฀ -฀ 1350฀ nm.฀ Satellite฀ data฀ represent฀ an฀ integration฀ of฀ the฀ reflectance฀ recorded฀ over฀ a฀ broad฀ spectral฀ band;฀ field฀ data฀ have฀to฀be฀expressed฀in฀the฀same฀way฀to฀allow฀a฀ proper฀comparison.฀Therefore,฀in฀Fig.4฀the฀reflectance฀values฀of฀the฀surveyed฀surfaces฀are฀shown,฀ averaged฀ over฀ the฀ considered฀ spectral฀ intervals.฀ These฀ data฀ allow฀ the฀ quantitative฀ estimatation฀ of฀ the฀ reflectance฀ of฀ the฀ different฀ snow-covered฀ surfaces฀ in฀ defined฀ spectral฀ intervals,฀ providing฀ useful฀ information฀ for฀ satellite฀ data฀ interpretation.฀ Considering฀ instrument฀ errors,฀ it฀ has฀ been฀ estimated฀ that฀ reflectance฀ variations฀ exceeding฀ 5฀ %฀ could฀ be฀ great฀ enough฀ to฀ detect฀ significant฀ differences฀among฀the฀surveyed฀surfaces. At฀ wavelengths฀ corresponding฀ to฀ TM1฀ (450฀ -฀ 520฀ nm),฀ TM2฀ (520฀ -฀ 600฀ nm)฀ and฀ TM3฀ (620฀ -฀ 690฀ nm)฀ bands,฀ new฀ snow฀ (N),฀ drifted฀ snows฀ and฀ equilibrium฀ forms฀ (E)฀ have฀ similar฀ and฀very฀high฀albedo฀(>฀0.94;฀Fig.฀4a,฀b).฀At฀these฀ wavelengths฀ equilibrium฀ forms฀ on฀ ice฀ (E4)฀ and฀ melt–freeze฀ crust฀ (C)฀ reflect฀ the฀ solar฀ incident฀ radiation฀with฀similar฀intensity.฀In฀the฀same฀spectral฀bands฀all฀the฀other฀surfaces฀are฀progressively฀ less฀reflective,฀as฀surface฀hoar฀increases.฀Between฀ 20 1030฀-฀1086฀nm฀and฀1260฀-฀1350฀nm,฀E4฀reflects฀up฀ to฀17฀%฀more฀radiation฀than฀a฀melt–freeze฀crust฀ (C),฀ though฀ its฀ particles฀ are฀ slightly฀ larger฀ and฀ hoar฀ covers฀ the฀ surface.฀ New฀ snow฀ (N),฀ drifted฀ snows฀ (SD,฀ DB,฀ DR฀ and฀ DE)฀ and฀ equilibrium฀ forms฀ (E)฀ begin฀ to฀ be฀ distinguished฀ from฀ TM4฀ (760฀ -฀ 900฀ nm)฀ wavelength฀ interval,฀ where฀ new฀ snow฀ can฀ be฀ clearly฀ distinguished฀ from฀ smooth฀ drifted฀snow฀(SD)฀and฀snow฀with฀surface฀barchans฀(DB).฀At฀1030฀-฀1085฀nm฀the฀grain฀and฀morphologic฀ characteristics฀ of฀ new฀ snow฀ (N)฀ make฀ it฀ more฀ reflective฀ than฀ drifted฀ snows฀ (DB,฀ DR,฀ DE฀ and฀ SD).฀ Between฀ 1260฀ and฀ 1350฀ nm฀ new฀ snow฀(N)฀reflectance฀is฀higher฀than฀that฀of฀drifted฀ snow฀ (DE),฀ which฀ has฀ the฀ same฀ grain฀ size฀ but฀ appears฀ covered฀ by฀ surface฀ hoar.฀ Still฀ at฀ wavelengths฀longer฀than฀1000฀nm,฀new฀snow฀(N)฀can฀ be฀ distinguished฀ from฀ equilibrium฀ forms฀ (E),฀ characterized฀by฀rounded฀grains฀and฀surface฀hoar฀ (Fig.฀4a).฀Reflectance฀differences฀of฀drifted฀snows฀ with฀the฀same฀thickness฀and฀grain฀characteristics฀ are฀due฀to฀surface฀roughness:฀smooth฀snow฀(SD)฀ can฀ be฀ distinguished฀ from฀ 5฀ -10฀ mm฀ furrowed฀ snow฀ surfaces฀ (DB฀ and฀ DR)฀ at฀ 1030฀ -฀ 1085฀ nm฀ and฀1260฀-฀1350฀nm฀wavelength฀ranges฀(Fig.฀4b).฀ Between฀1260฀and฀1350฀nm฀it฀appears฀also฀possible฀to฀distinguish฀drifted฀snows฀considering฀the฀ different฀ grain฀ shape,฀ as฀ for฀ drifted฀ snow฀ (DE)฀ and฀the฀surface฀roughness฀for฀drifted฀snows฀SD,฀ DB,฀DR,฀while฀the฀presence฀of฀surface฀hoar฀seems฀ to฀be฀less฀decisive.฀In฀TM5฀(1550฀-฀1750฀nm)฀and฀ TM7฀(2080฀-฀2350฀nm)฀bands฀the฀low฀albedo฀of฀ equilibrium฀ forms฀ (E)฀ with฀ respect฀ to฀ the฀ other฀ surfaces฀ is฀ due฀ to฀ the฀ presence฀ of฀ surface฀ hoar.฀ The฀ basal฀ ice฀ reflectance฀ (I)฀ is฀ lower฀ than฀ those฀ of฀the฀surveyed฀snows฀and฀is฀therefore฀easily฀distinguished฀in฀the฀first฀four฀spectral฀intervals฀considered.฀However,฀when฀10฀mm฀of฀snow฀cover฀the฀ ice,฀an฀appreciable฀change฀in฀albedo฀is฀observed฀ (E1),฀ mainly฀ at฀ wavelengths฀ beyond฀ 600฀ nm;฀ at฀ TM5฀and฀TM7,฀the฀E1฀curve฀coincides฀with฀that฀ of฀E฀because฀of฀the฀effect฀of฀backscattering฀due฀to฀ the฀presence฀of฀snow,฀an฀effect฀particularly฀strong฀ at฀these฀spectral฀channels. To฀better฀emphasize฀mutual฀reflectance฀changes,฀ the฀ reflectances฀ of฀ the฀ surveyed฀ surfaces฀ were฀ normalized฀to฀new฀snow฀and฀shown฀in฀Fig.฀5a฀and฀ b:฀in฀these฀plots฀higher฀values฀signify฀reflectance฀ similar฀to฀that฀of฀new฀snow.฀The฀distance฀between฀ normalized฀reflectance฀values฀at฀each฀wavelength฀ interval฀ reveal฀ the฀ possibility฀ of฀ detecting฀ different฀snow฀surfaces,฀even฀though฀their฀original฀ reflectance฀ values฀ were฀ very฀ similar.฀ The฀ infor- Radiometric฀investigation฀of฀different฀snow฀covers฀in฀Svalbard mation฀retrieved฀from฀this฀figure฀is฀similar฀to฀that฀ obtained฀from฀Fig.฀4,฀particularly฀concerning฀the฀ first฀ four฀ TM฀ spectral฀ channels฀ and฀ the฀ 1030฀ -฀ 1085฀ range.฀ In฀ the฀ 1260฀ -฀ 1350฀ nm,฀ TM5฀ and฀ TM7฀wavelength฀intervals฀differences฀are฀markedly฀higher฀than฀those฀in฀Fig.฀4. Conclusions Field฀ spectroradiometric฀ data฀ of฀ different฀ snow/ ice฀surfaces฀confirm฀the฀necessity฀of฀examining฀ thoroughly฀the฀relation฀between฀snow฀reflectance฀ and฀ snow฀ metamorphic฀ state.฀ Although฀ ice฀ and฀ snow฀ behaviour฀ is฀ generally฀ well฀ known฀ in฀ the฀ spectral฀ interval฀ 350฀ -฀ 2500฀ nm,฀ a฀ better฀ understanding฀ of฀ snow/ice฀ reflectance฀ under฀ natural฀ conditions฀would฀substantially฀improve฀the฀interpretation฀ of฀ their฀ physical฀ features฀ based฀ on฀ reflectance฀ data.฀ Fieldwork฀ also฀ revealed฀ that฀ snow฀grain฀size฀is฀often฀larger฀than฀that฀used฀in฀ the฀snow฀spectral฀models฀proposed฀(Wiscombe฀&฀ Warren฀ 1980):฀ snow฀ is฀ made฀ up฀ of฀ four฀ to฀ five฀ kinds฀of฀grain฀differing฀in฀size฀and฀shape,฀decisively฀affecting฀the฀snow฀spectral฀response,฀particularly฀in฀the฀infrared. It฀ has฀ also฀ been฀ observed฀ that฀ significant฀ changes฀ in฀ surface฀ reflectance฀ properties฀ are฀ related฀to฀the฀occurrence฀of฀hoar฀and฀that฀a฀detailed฀ characterization฀ of฀ the฀ sub-surface฀ snow฀ (snow฀ layering,฀metamorphism)฀is฀also฀highly฀important฀ for฀data฀interpretation.฀Surface฀roughness,฀which฀ increases฀as฀furrow฀distance฀decreases฀(Table฀4),฀ contributes฀to฀increasing฀reflectance฀particularly฀ at฀wavelength฀higher฀than฀1000฀nm.฀This฀is฀especially฀evident฀in฀the฀case฀of฀drifted฀snows฀where฀ the฀ smoother฀ the฀ surface฀ the฀ lower฀ the฀ reflectance.฀Moreover,฀reflectance฀increases฀when฀snow฀ crystals฀are฀not฀rounded฀(regardless฀of฀their฀size)฀ and฀ surface฀ roughness฀ is฀ relatively฀ high,฀ while฀ reflectance฀decreases฀as฀rounded฀particles฀became฀ larger฀in฀size฀and฀the฀presence฀of฀surface฀hoar฀and฀ impurities฀is฀higher.฀However,฀it฀is฀not฀always฀possible฀to฀identify฀precisely฀the฀feature฀that฀mainly฀ determines฀ snow฀ field฀ reflectance,฀ which฀ is฀ the฀ sum฀ of฀ the฀ above-mentioned฀ factors.฀ This฀ has฀ important฀ implications฀ when฀ field฀ spectroradiometric฀ data฀ have฀ to฀ be฀ used฀ to฀ support฀ satellite฀ data฀interpretation,฀unless฀a฀broad฀description฀of฀ snow฀ physical฀ characteristics฀ is฀ required.฀ Based฀ on฀the฀data฀shown฀in฀Figs.฀4฀and฀5,฀better฀results฀ concern฀ the฀ discrimination฀ between฀ new฀ and฀ drifted฀snows฀with฀respect฀to฀melt–freeze฀crust,฀ Casacchia฀et฀al.฀2001฀year:฀Polar฀Research฀20(1),฀13–22 bare฀ basal฀ ice฀ and฀ basal฀ ice฀ covered฀ by฀ a฀ thin฀ snow฀layer.฀A฀snow฀layer฀40฀mm฀thick฀over฀basal฀ ice฀show฀reflectance฀values฀close฀to฀those฀of฀the฀ melt–freeze฀crust฀in฀the฀visible฀and฀the฀near-infrared฀ and฀ is฀ close฀ to฀ that฀ of฀ equilibrium฀ forms฀ at฀ wavelengths฀ greater฀ than฀ 1000฀ nm.฀ Reflectance฀ differences฀between฀new฀snow฀and฀drifted฀snows฀ are฀difficult฀to฀detect,฀unless฀the฀1030฀-฀1085฀and฀ 1260฀ -฀ 1350฀ nm฀ wavelength฀ ranges฀ are฀ considered. The฀use฀of฀a฀number฀of฀spectral฀channels฀larger฀ than฀those฀presently฀available฀on฀the฀TM฀sensor฀ would฀ help฀ in฀ snow฀ and฀ ice฀ monitoring,฀ particularly฀ if฀ the฀ 1030฀ -฀ 1085฀ and฀ 1260฀ -฀ 1350฀ nm฀ wavelength฀ranges฀were฀considered฀in฀designing฀ future฀satellite฀sensors.฀And฀our฀knowledge฀of฀the฀ interaction฀between฀snow฀and฀solar฀radiation฀can฀ be฀enhanced฀by฀improving฀field฀data฀acquisition฀ techniques,฀ devoting฀ more฀ attention฀ to฀ features฀ like฀ surface฀ roughness฀ and฀ grain฀ size฀ assemblages. It฀ is฀ important฀ both฀ to฀ intensify฀ snow฀ field฀ surveys฀in฀different฀glacial฀environments฀and฀to฀ repeat฀ the฀ same฀ measurements฀ under฀ different฀ climatic,฀ atmospheric฀ and฀ solar฀ conditions.฀ It฀ is฀ also฀of฀importance฀to฀collect฀accurate฀data฀about฀ snow’s฀physical฀and฀textural฀characteristics฀along฀ with฀ spectroradiometric฀ data,฀ because฀ a฀ correct฀ interpretation฀of฀snow฀spectral฀response฀would฀be฀ extremely฀difficult฀without฀this฀information. Acknowledgements.—This฀ work฀ has฀ been฀ supported฀ by฀ the฀ CNR฀ (National฀ Research฀ Council฀ of฀ Italy)฀ Arctic฀ Strategic฀ Project฀and฀by฀the฀National฀Research฀Programme฀in฀Antarctica฀ (Project฀ Remote฀ Sensing,฀ GIS฀ and฀ Hydrography).฀ Special฀thanks฀go฀to฀Mr.฀Roberto฀Sparapani฀for฀logistic฀support฀ during฀ the฀ field฀ survey.฀ The฀ authors฀ also฀ wish฀ to฀ thank฀ Dr.฀ J.-G.฀ Winther฀ and฀ Dr.฀ S.฀ Gerland฀ from฀ the฀ Norwegian฀ Polar฀Institute฀(the฀latter฀now฀with฀the฀Norwegian฀Radiation฀ Protection฀ Authority)฀ for฀ their฀ availability฀ to฀ discuss฀ snow฀ reflectance฀acquisition฀methods,฀and฀Dr.฀B.฀Johnsen฀from฀the฀ Norwegian฀Radiation฀Protection฀Authority฀for฀valuable฀optical฀calibration฀experience. References Boresjö฀ Bronge,฀ L.฀ &฀ Bronge,฀ C.฀ 1999:฀ Ice฀ and฀ snow-type฀ classification฀in฀the฀Vestfold฀Hills,฀East฀Antarctica,฀using฀ Landsat-TM฀and฀ground฀radiometer฀measurements.฀Int.฀J.฀ Remote฀Sens.฀20,฀225–240. Bourdelles,฀B.฀&฀Fily,฀M.฀1993:฀Snow฀grain-size฀determination฀from฀Landsat฀imagery฀over฀Terre฀Adélie,฀Antarctica.฀ Ann.฀Glaciol.฀17,฀86–92. Colbeck,฀S.,฀Akitaya,฀E.,฀Armstrong,฀R.,฀Gubler,฀H.,฀Lafeuille,฀ 21 J.,฀Lied,฀K.,฀McClung,฀D.฀&฀Morris,฀E.฀1990:฀The฀international฀classification฀for฀seasonal฀snow฀on฀the฀ground.฀International฀commission฀on฀snow฀and฀ice฀report.฀Wallingford,฀ UK:฀International฀Association฀of฀Scientific฀Hydrology. Dozier,฀ J.฀ 1989:฀ Spectral฀ signature฀ of฀ alpine฀ snow฀ cover฀ from฀Landsat฀Thematic฀Mapper.฀Remote฀Sens.฀Environ.฀28,฀ 9–22. Gerland฀S.,฀Winther฀J.-G.,฀Ørbæk฀J.฀B.,฀Liston฀G.฀E.,฀Øritsland฀N.฀A.,฀Blanco,฀A.฀&฀Ivanov,฀B.฀1999:฀Physical฀and฀optical฀properties฀of฀snow฀covering฀Arctic฀tundra฀on฀Svalbard.฀ Hydrol.฀Process.฀13,฀2331–2343. Hall,฀D.฀K.,฀Bindschadler,฀R.฀A.,฀Foster,฀J.฀L.,฀Chang,฀A.฀T.฀C.฀ &฀Siddalingaiah,฀H.฀1990:฀Comparison฀of฀in฀situ฀and฀satellite-derived฀reflectances฀of฀Forbindels฀Glacier,฀Greenland.฀ Int.฀J.฀Remote฀Sens.฀11,฀493–504. Hall,฀ D.฀ K.,฀ Chang,฀ A.฀ T.฀ C.,฀ Foster,฀ J.฀ L.,฀ Benson,฀ C.฀ S.฀ &฀ Kovalick,฀W.฀M.฀1989:฀Comparison฀of฀in฀situ฀and฀Landsat฀ derived฀reflectance฀of฀Alaskan฀glaciers.฀Remote฀Sens.฀Environ.฀28,฀23–31. Hall,฀D.฀K.,฀Chang,฀A.฀T.฀C.฀&฀Siddalingaiah,฀H.฀1988:฀Reflectance฀ of฀ glaciers฀ as฀ calculated฀ using฀ Landsat-5฀ Thematic฀ Mapper฀data.฀Remote฀Sens.฀Environ.฀25,฀311–321. Hall,฀D.฀K,฀Kovalick,฀W.฀M.฀&฀Chang,฀A฀.T.฀C.฀1990:฀Satellitederived฀ reflectance฀ of฀ snow-covered฀ surfaces฀ in฀ northern฀ Minnesota.฀Remote฀Sens.฀Environ.฀33,฀87–96. Knap,฀ W.฀ H.,฀ Reijmer,฀ C.฀ H.฀ &฀ Oerlemans,฀ J.฀ 1999:฀ Narrowband฀ to฀ broadband฀ conversion฀ of฀ Landsat฀ TM฀ glacier฀ albedoes.฀Int.฀J.฀Remote฀Sens.฀20,฀2091–2110. Ørbæk,฀J.฀B.,฀Hisdal฀V.฀&฀Svaasand฀L.฀E.฀1999:฀Radiation฀climate฀variability฀in฀Svalbard:฀surface฀and฀satellite฀observations.฀Polar฀Res.฀18,฀127–134. Thomas,฀R.฀H.฀1993:฀Ice฀sheets.฀In฀R.฀J.฀Gurney฀et฀al.฀(eds.):฀ 22 Atlas฀of฀satellite฀observations฀related฀to฀global฀change.฀Pp.฀ 385–400.฀Cambridge:฀Cambridge฀University฀Press. Warren,฀ S.฀ G.฀ 1982:฀ Optical฀ properties฀ of฀ snow.฀ Rev.฀ Geophys.฀Space฀Phys.฀20,฀67–89. Warren,฀S.฀G.,฀Grenfell,฀T.฀C.฀&฀Mullen,฀P.฀C.฀1986:฀Optical฀ properties฀of฀Antarctic฀snow.฀Antarct.฀J.฀U.฀S.฀21,฀247-248. Warren,฀S.฀G฀&฀Wiscombe,฀W.฀J.฀1980:฀A฀model฀for฀the฀spectral฀albedo฀of฀snow.฀II:฀Snow฀containing฀atmospheric฀aerosols.฀J.฀Atmos.฀Sci.฀37,฀2734–2745. Winther,฀J.-G.฀1993a:฀Studies฀of฀snow฀surface฀characteristics฀ by฀Landsat฀TM฀in฀Dronning฀Maud฀Land,฀Antarctica.฀Ann.฀ Glaciol.฀17,฀27–34. Winther,฀J.-G.฀1993b:฀Landsat฀TM฀derived฀and฀in฀situ฀summer฀ reflectance฀of฀glaciers฀in฀Svalbard.฀Polar฀Res.฀12,฀37–55. Winther,฀J.-G.,฀Gerland,฀S.,฀Ørbæk,฀J.฀B.,฀Ivanov,฀B.,฀Blanco,฀ A.฀&฀Boike,฀J.฀1998:฀Spectral฀reflectivity฀and฀physical฀properties฀of฀snow฀and฀sea฀ice฀near฀Ny-Ålesund,฀Svalbard.฀Proceedings฀ of฀ the฀ 4th฀ Ny-Ålesund฀ Seminar฀ the฀ Arctic฀ and฀ Global฀Change.฀Ravello,฀Italy,฀5-6฀March฀1998.฀NySMAC฀ Publ.฀7,฀113–116.฀Longyearbyen:฀Ny-Ålesund฀Science฀Managers฀Committee. Winther,฀J.-G.,฀Gerland,฀S.,฀Ørbæk,฀J.฀B.,฀Ivanov,฀B.,฀Blanco,฀ A.฀&฀Boike,฀J.฀1999:฀Spectral฀reflectance฀of฀melting฀snow฀in฀ a฀high฀Arctic฀watershed฀on฀Svalbard:฀some฀implications฀for฀ optical฀ satellite฀ remote฀ sensing฀ studies.฀ Hydrol.฀ Process.฀ 13,฀2033–2049. Wiscombe,฀ W.฀ J.฀ &฀ Warren,฀ S.฀ G.฀ 1980:฀ A฀ model฀ for฀ the฀ spectral฀ albedo฀ of฀ snow.฀ I:฀ Pure฀ snow.฀ J.฀ Atmos.฀ Sci.฀ 37,฀ 2712–2733. Zibordi,฀G.,฀Meloni,฀G.฀P.฀&฀Frezzotti,฀M.฀1996:฀Snow฀and฀ice฀ reflectance฀spectra฀of฀the฀Nansen฀Ice฀Sheet฀surfaces.฀Cold฀ Reg.฀Sci.฀Technol.฀24,฀147–151. Radiometric฀investigation฀of฀different฀snow฀covers฀in฀Svalbard