373Journal of Oceanography, Vol. 62, pp. 373 to 393, 2006Keywords:⋅⋅⋅⋅⋅ Remote sensing,⋅⋅⋅⋅⋅ validation,⋅⋅⋅⋅⋅ ocean color,⋅⋅⋅⋅⋅ atmosphericcorrection,⋅⋅⋅⋅⋅ chlorophyll,⋅⋅⋅⋅⋅ ADEOS-2,⋅⋅⋅⋅⋅ GLI,⋅⋅⋅⋅⋅ match-up.* Corresponding author. E-mail: [email protected]©The Oceanographic Society of Japan/TERRAPUB/SpringerValidation of ADEOS-II GLI Ocean Color Products UsingIn-Situ ObservationsHIROSHI MURAKAMI1*, KOSEI SASAOKA2, KOHTARO HOSODA1, HAJIME FUKUSHIMA3,MITSUHIRO TORATANI3, ROBERT FROUIN4, B. GREG MITCHELL4, MAT I KAHRU4,PIERRE-YVES DESCHAMPS5, DENNIS CLARK6, STEPHANIE FLORA7, MOTOAKI KISHINO8,SEI-ICHI SAITOH9, ICHIO ASANUMA10, AKIHIKO TANAKA11, HIROAKI SASAKI11,KATSUMI YOKOUCHI12, YOKO KIYOMOTO13, HIROAKI SAITO14, CÉCILE DUPOUY15,ABSORNSUDA SIRIPONG16, SATSUKI MATSUMURA16 and JOJI ISHIZAKA171Earth Observation Research and Application Center, JAXA, Harumi, Chuo-ku, Tokyo 104-6023, Japan2Frontier Research Center for Global Change, JAMSTEC, Showa-machi, Kanazawa-ku, Yokohama 236-0001, Japan3School of High-Technology for Human Welfare, Tokai University, Numazu, Shizuoka 410-0395, Japan4Scripps Institution of Oceanography, UCSD, La Jolla, CA 92037, U.S.A.5Laboratoire d’Optique Atmosphérique, Université des Sciences et Technologies de Lille, Villeneuve d’Asq. Cedex, France6National Oceanic and Atmospheric Administration, National Environmental Satellite Service, Washington, D.C., 20233, U.S.A.7Moss Landing Marine Labs, Moss Landing, CA 95039, U.S.A.8Marine Science, Tokyo University of Marine Science and Technology, Kohnan, Minato-ku, Tokyo 108-8477, Japan9Fisheries Sciences and Faculty of Fisheries, Hokkaido University, Hakodate 041-8611, Japan10Tokyo University of Information Sciences, Yato-cho, Wakaba-ku, Chiba 265-8501, Japan11Nagasaki Industrial Promotion Foundation, Dejima-machi, Nagasaki 850-0862, Japan12Fisheries Agency, Kasumigaseki, Chiyoda-ku, Tokyo 100-8907, Japan13Seikai National Fisheries Research Institute, Fisheries Research Agency, Taira-machi, Nagasaki 851-2213, Japan14Tohoku National Fisheries Research Institute, Fisheries Research Agency, Niihama, Shiogama, Miyagi 985-0001, Japan15IRD UR103 CAMELIA, Centre d’Océanologie de Marseille, F-13007 Marseille, France16Marine Science Department Faculty of Science, Chulalongkorn University, Bangkok 10330, Thailand17Faculty of Fisheries, Nagasaki University, Bunkyo-machi, Nagasaki 852-8521, Japan(Received 28 June 2005; in revised form 8 January 2006; accepted 7 February 2006)The Global Imager (GLI) aboard the Advanced Earth Observing Satellite-II (ADEOS-II) made global observations from 2 April 2003 to 24 October 2003. In cooperationwith several institutes and scientists, we obtained quality controlled match-ups be-tween GLI products and in-situ data, 116 for chlorophyll-a concentration (CHLA),249 for normalized water-leaving radiance (nLw) at 443 nm, and 201 for aerosol op-tical thickness at 865 nm (Tau_865) and Angstrom exponent between 520 and 865 nm(Angstrom). We evaluated the GLI ocean color products and investigated the causesof errors using the match-ups. The median absolute percentage differences (MedPD)between GLI and in-situ data were 14.1–35.7% for nLws at 380–565 nm, 52.5–74.8%nLws at 625–680 nm, 47.6% for Tau_865, 46.2% for Angstrom, and 46.6% for CHLA,values that are comparable to the ocean-color products of other sensors. We found374 H. Murakami et al.that some errors in GLI products are correlated with observational conditions; nLwvalues were underestimated when nLw at 680 nm was high, CHLA was underesti-mated in absorptive aerosol conditions, and Tau_865 was overestimated in sunglintregions. The error correlations indicate that we need to improve the retrievals of theoptical properties of absorptive aerosols and seawater and sea surface reflection forfurther applications, including coastal monitoring and the combined use of productsfrom multiple sensors.conditions. The characteristics of GLI hardware, oceancolor products, and the version-2 algorithm are given inSection 2. In-situ observations and their processing meth-ods are provided in Section 3. The match-up statistics andcorrelation analysis are presented in Section 4. The re-sults presented in Section 4 and reasons for the errors arediscussed in Section 5.2. GLI and Standard Ocean Color Products2.1 GLI designGLI is a cross-track scanning radiometer with 36channels in the visible and near infrared (VNIR, 380 to865 nm), shortwave infrared (SWIR, 1050 to 2210 nm),and middle and thermal infrared (MTIR, 3700 to 12000nm) bands (Nakajima et al., 1998). The GLI channels aredesigned not only for the ocean but also various otherobservation targets, including bright clouds and land sur-face. Tables 1 and 2 list the sensor operation and channelcharacteristics of GLI (Kurihara et al., 2003; Tanaka etal., 2005). Channels 2, 3, 6, 9, 10, 11, 12, 14, 16, and 18are high-gain bands designed for atmospheric correctionand ocean-color detection. Channels 1, 13, 15, 17, 19,24, 25, 26, and 27 have a wide dynamic range designedfor observations over land, the cryosphere, and clouds.Channels 4, 5, 7, and 8 are piece-wise linear gain bandsdesigned for both dark and bright targets. The 250 m-resolution channels (20, 21, 22, 23, 28, and 29) were de-signed for compatibility with LANDSAT/TM data andother 1-km channels used for global observation. GLI alsohas a tilt function to prevent sun glitter, allowing accu-rate ocean-color observations in the summer at middleand lower latitudes.Features of GLI sensor design compared to OCTS,SeaWiFS and MODIS are:• Thirteen channels applicable to ocean color in vis-ible wavelengths from 380 nm to 710 nm; OCTS andSeaWiFS have six, and MODIS ten (seven high-gain chan-nels are used for the standard MODIS nLws).• Twelve bit (digital numbers from 0 to 4095)digitization; OCTS and SeaWiFS are coded over ten bits(from 0 to 1023), and MODIS is the same as GLI (twelvebits).• Tilt function (OCTS and SeaWiFS have it, butMODIS does not).1. IntroductionThe objectives of the GLI ocean science mission areto establish satellite data retrieval algorithms, to monitorphysical and biological ocean variables, and to addressissues of global climate change. Although it is difficultto use GLI data for monitoring purposes due to the lim-ited data period of only seven months, we must createsufficiently accurate GLI ocean products and describetheir