1 Ocean products 2 Validation methods 3 Validation

　1. Ocean products 　2. Validation methods 　3. Validation results 　4. Summary and future plan 2003, December H. Murakami, K. Sasaoka, K. Hosoda, and Ocean PIs 1

1. Ocean products 2003, December H. Murakami, K. Sasaoka, K. Hosoda, and Ocean PIs 2

Overview of GLI ocean products (1) n Ocean atmospheric correction algorithm Developer: Hajime Fikushima (Tokai Univ. ) R. Frouin (Scripps institution of oceanography) Algorithm: Atmospheric correction (Fukushima) and Photosynthetically available radiation (Frouin) Algorithm code: OTSK 1 a (Fukushima), OTSK 14 (Frouin) Product code: NL (Level-2) NW/LA (L 2 Map, L 3 Bin, L 3 STAMap) 3 type FR(1 km)/LR(4 km)/FR_NRT(near real time) for each Parameters: normalized water-leaving radiance (n. Lw, 13 chs) aerosol radiance (4 chs) aerosol optical thickness (Tau), angstrom exp. ocean color quality flag (Bit 1 -26: quality and cloud info. ) photosynthetically available radiation (PAR) (Ein/m 32/day)

Overview of GLI ocean products (2) n In-warter algorithms Developer: B. G. Mitchell (Scripps institution of oceanography) Motoaki Kishino (Tokyo University of Marine Science and Technology) Algorithm: In-water algorithms Algorithm code: OTSK 2, 5, 7 (Mitchell), OTSK 6 (Kishino) Product code: CS (Level-2) CHLA, CDOM, K 490, SS (L 2 Map, L 3 Bin, L 3 STAMap) 3 type FR(1 km)/LR(4 km)/FR_NRT(near real time) for each Parameters: chlorophyll-a concentration (CHLA), colored dissolved organic matter absorption at 440 nm (CDOM) Mitchell attenuation coefficient at 490 nm (K 490), redtide flag, suspended solid concentration (SS), Kishino 4 in-water quality flag (Bit 27 -32: Case-2 etc. )

Overview of GLI ocean products (3) n Sea surface temperature algorithm Developer: Hiroshi Kawamura (Tohoku Univ. ) Algorithm: sea surface temperature Algorithm code: OTSK 13 Product code: ST (Level-2, 2 Map, 3 Bin) ST_daynight, ST_all (L 3 STAMap) 3 type FR(1 km)/LR(4 km)/FR_NRT(near real time) for each Parameters: Bulk sea surface temperature (SST) (K), SST quality flag (16 bit: cloud information etc. ) (cloud detection was improved by EORC and PI) 5

GLI ocean operation flow (1/2) (1 -km Resolution) L 2 Map QF_OC Level 1 B OTSK 1 a NL_ FR L 2 Mapping H. Fukushima R. Frouin CLFLG S. Ackerman K. Stamnes OTSK 2/5/6/7 L 2 Map NW CS_FR B. G. Mitchell, M. Kishino OTSK 13 L 2 Map LA L 2 Map ST ST_FR H. Kawamura L 2 Map QF_ST L 2 Map CHLA L 2 Map CDOM L 2 Map K 490 L 2 Map SS 4 -km resampling 1600 km scene Near real time operation 6 at direct receiving station

GLI ocean operation flow (2/2) (4 or 9 -km Resolution) 4 -km resampling L 2 A_OA OTSK 1 a NL_LR OTSK 2/5/6/7 OTSK 13 4 -km path CS_LR ST_LR L 3 space bin L 3 time bin (Daily, 8 -day, Monthly) L 3 bin NW L 3 bin LA L 3 bin CS L 3 Statistical Mapping L 3 STAMap NW L 3 STAMap LA L 3 bin ST 9 -km equal area grid L 3 STAMap. CHLA L 3 STAMap CDOM SS K 490 Night ST L 3 STAMap ST 9 -km 7 global map

Examples of ocean products (1/3) n. Lw 380 nm-n. Lw 680 (26 May 2003, California) 8

Examples (2/3) CHLA increase CHLA, SS, CDOM, K 490, Tau, SST… research product: Fluorescence line height 9

Examples of ocean products (3/3) CHLA and SS distributions CHLA SS GLI neural network algorithm looks to present CHLA and SS characteristics well. 10

2. Validation methods 2003, December H. Murakami, K. Sasaoka, K. Hosoda, and Ocean PIs 11

2. 1 Validation methods for each product Products normalized water-leaving radiance n. Lw (13 chs) [m. W cm 2 str 1 m 1] aerosol radiance (4 chs) Target accuracy Validation methods (after 3 years) -35 +50% (offshore) loptical observations by radiometer -50 +100% (coastal) above/under the sea surface Validated through n. Lw loptical photosynthetically available radiation [Ein m 2 day] -10 +10% (10 km/ monthly) lcomparison chlorophyll-a concentration [mg m 3] -35 +50% (offshore) -50 +100% (coastal) aerosol optical thickness at 865 nm [ ] aerosol angstrom exp. 749/865 nm [ ] colored dissolved organic matter absorption at 440 nm [m 1] -50 +100% attenuation coefficient at 490 nm [m 1] -35 +50% observations (same as above) lindirectly validated through n. Lw with Sea. Wi. FS or GMS products lsolar irradiance measurements at islands lwater sampling and measurement on ship lindirectly validated through n. Lw; relations between in-water parameter and n. Lw have been investigated by ship measurements lcomparison to Sea. Wi. FS or MODIS products suspended solid concentration [g m 3] -50 +100% lcomparison bulk sea surface temperature [K] 0. 6 K with ground SST measurements by mooring or drifting buoys distributed by GTS lcomparison with other products e. g. , 12 Reynolds

2. 2 observation instruments targets PRR 600/610 under & above wavelength nm 412, 443, 490, 520, 565, 670 and PAR ldownward + MER 2040/2041 irradiance (Ed ) lupward radiance under & above (Lu ) Free Fall under & above lsky irradiance (Es ) 412, 443, 465, 490, 510, 520, 555, 565, 625, 665, 670, 683 and PAR 380, 400, 412, 443, 455, 490, 520, 555, 565, 620, 665, 683, 705 and PAR n. Lw Calculate water leaving radiance (Lw) by upward radiance profile Lu. n. Lw is normalized by Lw, Ed and F 0. ldownward TRIOS above only irradiance (Ed+) 350 -950 nm lupward radiance (Lu+) 1 nm intervals ldownward radiance (Ls+) normalize Lu by Ed, F 0 and sea surface reflection factor. 13

2. 3 available match up data Routine observations: GTS (JMA), AERONET, SKYNET, MOBY (MODIS, NOAA) Special observations: 2003. 04. 03 -04. 27 California Cooperative Oceanic Fisheries Investigations (Cal. COFI) Mitchell (SIO), Frouin (SIO), Hokkaudo Univ. 2003. 04 -. . Ocean color obs. around Japan Nagasaki Univ. , Tokyo Ocean Univ. , Nagoya Univ, Tokai Univ. , NFRI, Hokkaido Univ, JAMSTEC etc. Relatively many ground observations are provided from PI team and corroborative organizations in the early phase basing relationship until OCTS mission. The data is still increasing, and we continue 14 the data collection and match-up processing now.

2. 3 other datasets for evaluation Global distribution can be evaluated by the following datasets. l Sea. Wi. FS global dataset (n. Lw, CHLA, Tau, PAR) l Reynolds global SST data l Tohoku Univ. GMS solar irradiance dataset (PAR is converted from SW) 15

3. Validation results 2003, December H. Murakami, K. Sasaoka, K. Hosoda, and Ocean PIs 16

3. 1 vicarious calibration study (1) candidates for the ocean color processing GLI LTOA is compared with simulated LTOA which is calculated using Sea. Wi. FS n. Lw, pressure, ozone, water vapor, GLI geometry and GLI atmospheric correction look-up tables. We had many data source for the vicarious coefficients. Most of the results show the similar tendencies. 17

3. 1 vical study (2) match-up processing tests • Best choice is Sea. Wi. FS Apr-Jun n. Lw base • Possibly temporal change considering Sea. Wi. FS Feb-Mar base coefs. • MOBY coefs make lower n. Lw • cannot use Barrow and Railroad ones Five Candidates of vical coefs: Sea. Wi. FS Apr-Jun/Feb-Mar, Barrow Apr, MOBY 2 Fix/ Aeronet, Railroad Test data: MOBY and Nagasaki ferry, Cal. COFI 200304 data number Cal. COFI obs. are only at CH 4, 6, 9 bias RMSE ratio 18

3. 2 validation by ground observation match-ups GLI products (n. Lw) derived using the selected vical coefs show high correlation with ground observations. Figure: n. Lw 380 -545 x-axis: ground obs, y-axis: GLI products 19

3. 2 validation by match-ups (in-water products) CHLA ground-obs locations SS CDOM K 490 scattered points are coastal obs. 20

3. 2 validation by match-ups (SST) Available data number: 4225 Bias: 0. 09 K (D: 0. 06, N: 0. 17) RMSE: 0. 67 K (D: 0. 67, N: 0. 69) 22 days sample: 2003/ 3/19, 3/20, 4/8, 5/28, 6/4, 6/5, 6/10, 6/19, 6/20, 6/30 7/2, 7/5, 7/10, 7/15, 7/19, 7/20, 7/25, 7/30, 8/4, 8/10, 8/19, 9/19 21

3. 3 comparison with other dataset (CHLA) GLI monthly 200304 CHLA distribution agrees with Sea. Wi. FS one Sea. Wi. FS monthly 22 200304

3. 3 comparison with other dataset (PAR) GLI v. s. Sea. Wi. FS PAR GLI monthly 200309 Sea. Wi. FS monthly 200309 PAR distribution and data range agree well. 23

3. 3 comparison with other dataset (PAR) GMS GLI v. s. SW PAR Sea. Wi. FS PAR GLI monthly 200304 GMS monthly 200304 (Kawamura et al. , 1998) PAR distribution and data range agree well. 24

3. 3 comparison with other dataset (SST) AVHRR + buoys (one-degree grid) GLI-SST Level-3 STA Map Reynolds-SST 4096 x 2048 (0. 088 -degree grid) 2003/04/03 -22 mean SST • Distribution and data range agree well. • GLI describe fine structure of the ocean current. GLI SST-Reynolds SST 25

4. Summary and future validation plan 2003, December H. Murakami, K. Sasaoka, K. Hosoda, and Ocean PIs 26

4. 1 summary of GLI validation all ocean products can be open today Param code Param name Final target Today’s accuracy NWLR normalized water leaving radiance 35 +50% (offshore) 50 +100% (coastal area) 16 47% (CH 01 -09) 84 284% (CH 10 -12) PAR photosynthetically available radiation 10 +10% (10 km/monthly) 11% CHLA chlorophyll-a concentration 35 +50% (offshore) 50 +100% (coastal area) 130% CDOM colored dissolved organic matter absorption at 440 nm 50 +100% (82%) SS suspended sediment concentration 50 +100% K 490 attenuation at 490 nm SST bulk sea surface temperature Notes Open l. Select vical coef. of Sea. Wi. FS Apr-Jul base. l. Bad results longer the 565 nm are due to low (<1/20) n. Lw over the ocean. llooks better than the early OCTS lproblem under absorptive aerosol lagree well with Sea. Wi. FS one llarge scatter is caused by coastal OK OK points lcomparable quality to OCTS OK linsufficient data number (now increasing) OK (34%) linsufficient data number OK 35 +50% (78%) linsufficient data number OK 0. 67 K lcloud detection problem l. Electric noise on MTIR image 27 OK

4. 2 Future validation plan date until GLI work shop (2003/03) items l. Continue match-up process and analysis (all products) l. Evaluation quality of each ground observation data (n. Lw and CHLA) l. Compare with ground SW observations not only other satellite data (PAR) l. Reanalyze using reprocessed (geo/radiometric) L 1 B (all products) l. Consider vical coefficients and their temporal change (Atmos. Corr. ) l. Improve absorptive aerosol frequently appeared over Japan (Atmos. Corr. ) l. Further study of in-water optical model in the coastal area (absorption, until Version 2 (2004/06) scattering, sea-surface reflection coefficients) l. Improve cloud detection (SST) l. Improve EORC MODIS processing applying the latest GLI code (all) l. Improve sunglint correction using AMSR wind speed (Atmos. Corr. ) l. Switching saturation alternative bands, and try to use 710 nm (Atmos. Corr. ) Ground l. Consider diurnal cycle of cloud amount (PAR) observations l. Evaluate SS and CDOM algorithm in global coastal area (in-water) are still l. Improve water vapor correction for SST needed for l. Improve absorptive aerosol (Atmos. Corr. ) the items l. Investigate Atmos. Corr. and CHLA estimation using 250 m data after 2004/06 (Atmos. Corr. CHLA) for the l. Further study of in-water optical model parameters in the coastal area mission goal l. Coding Level-3 processing of PAR (PAR/research product) l. Evaluate primary production algorithms (research product) l. Study of fluorescence algorithm and its application (research product) 28