Full product set validation over coastal Case II

Скачать презентацию Full product set validation over coastal Case II

7465d86b622ee77c819ca63482da3a81.ppt

Количество слайдов: 26

Full product set validation over coastal Case II waters Dominique Durand, Are Folkestad, Jo Høkedal, Kai Sørensen NIVA Samantha Lavender Univ. Plymouth , Page 1 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Scope • The FPS dataset • DDS for coastal case II waters • Validation against in-situ data – ”global” assessment – Regional and water type approach – Specific results at diagnostic sites • Relative validation against pre-merged data • 10 -year time series • Conclusions Page 2 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

FPS dataset • 10 -year of global observations – Sea. Wi. FS: 1997 - 2006 – MODIS: 2002 - 2006 – MERIS: 2002 - 2006 • Use of an extended in-situ dataset to try to ensure a robust and statistically sounded validation – OBPG, NOMAD, Boussole – Several coastal DDS Page 3 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Level-2 Characterisation Chl-a Results Achievement: Statistical uncertainties have been derived and are used for merging Page 7 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Types of In Situ Data • Chlorophyll (HPLC and in-vitro fluorometric) • Fully normalised water-leaving radiance primarily at 412, 443, 490, 510, 531, 555, 620, 670 nm – Computed when Lw, Ed, Es, Chl, and solar and viewing geometry available. • Diffuse attenuation coefficient at ~490 nm • Dissolved Coloured Organic Material Page 8 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Identification of Case 2 Validation Sites • Selection of sites for Case 2 validation based on general knowledge Page 9 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

DDS used for validation of coastal case 2 waters • • • Venice Tower (Adriatic Sea) Sopot (Poland) Helgoland (Germany) Skagerrak (NIVA-Ferrybox) Helsinki Light House (Gulf of Finland) Gustav Dahlén Tower (Baltic Sea) • CALCOFI (California) Gulf of Maine (USA Canada) MVCO (Rhode Island, Chesapeake Bay) COVE • Benguela (South Africa) A priori DDS for coastal Case 2 waters Page 10 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Identification of Case 2, subclassification • Selection of sites for Case 2 validation based on general knowledge • Identification of Case 2 -S using the EL 555 product • Case 2 -Y flag is not active yet in MERIS RR products Page 11 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Number of match-ups per validation site Venice Tower Rhode Island 8 - - - - 90 98 2736 239 282 262 1237 - 4756 Kd 490 3 - - 19 3 - 25 CDM - - - 15 2 - 17 Chl hplc 8 - - 157 165 Chl fluor 2952 301 307 332 1305 - 5197 L 412 3023 285 250 302 1315 9 5184 L 443 3024 314 347 334 1319 9 5347 L 490 42 136 347 334 453 9 1321 L 510 9 - - 4 2 9 24 L 555 2781 239 281 279 1234 4 4818 L 670 Page 13 Chesap. Bay Chl fluor AVW Helsinki Light. House Chl hplc GSM Gustav Dalen Tower 584 313 347 317 1317 9 2887 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST NIVA Ferry. B Skagerrak Global

Validation of n. Lw over coastal case II waters r 2 RMS BIAS SLOPE OFFSET 0. 35 0. 36 0. 13 0. 62 0. 10 r 2 RMS BIAS SLOPE OFFSET 0. 57 0. 31 0. 01 0. 69 0. 23 r 2 0. 72 RMS 0. 27 BIAS -0. 01 SLOPE 0. 72 OFFSET 0. 20 Page 14 r 2 0. 40 RMS 0. 46 BIAS -0. 01 SLOPE 0. 60 OFFSET 0. 41 r 2 RMS BIAS SLOPE OFFSET 0. 55 0. 09 0. 02 0. 73 0. 03 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Validation of Chl products over coastal case II waters r 2 RMS BIAS SLOPE OFFSET r 2 0. 29 RMS 1. 12 BIAS -0. 59 SLOPE 0. 48 OFFSET 0. 69 Page 15 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST 0. 22 1. 33 0. 88 0. 35 2. 37 r 2 RMS BIAS SLOPE OFFSET 0. 29 0. 92 0. 02 0. 38 0. 94 0. 63 0. 81 0. 48 0. 72 1. 83

Validation of CHL in turbid/ non-turbid coastal waters r 2 0. 42 RMS 0. 55 BIAS -0. 05 SLOPE 0. 62 OFFSET 1. 06 r 2 RMS BIAS SLOPE OFFSET Page 16 r 2 RMS BIAS SLOPE OFFSET 0. 24 0. 55 0. 39 0. 37 0. 83 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST 0. 27 1. 09 0. 11 0. 37 0. 83 r 2 RMS BIAS SLOPE OFFSET 0. 21 1. 89 1. 35 0. 56 0. 32

Regional validation of CHL Helsinki Lighthouse G. Dalen Tower N r 2 RMS BIAS SLOPE OFFSET Page 18 Chesapeake Bay Rhode Island 520 0. 01 1. 54 1. 09 0. 05 1. 74 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST N 1497 r 2 0. 07 RMS 1. 30 BIAS 1. 07 SLOPE 0. 46 OFFSET 0. 66

Performance of the Glob. COLOUR products in case 2 waters: CHL – July 2002 Depth < 200 m All pixels Turbid flag <200 m Turbid flag Page 20 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Performance of the Glob. COLOUR products in case 2 waters: CDM – July 2002 Depth < 200 m All pixels Turbid flag <200 m Turbid flag Page 21 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Performance of the Glob. COLOUR products in case 2 waters: bbp – July 2002 Depth < 200 m All pixels Turbid flag <200 m Turbid flag Page 22 Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST

Conclusions • • Validation of merged case I products in coastal case II waters presents the same limitations than with non-merged products Validation of daily-averaged products over variable coastal waters is a challenging task – Spatial discrepancy (4. 6 km vs. point) – Temporal discrepancy (daily average vs. instantaneous) • • Discriminate between algorithm weakness and quality of the in-situ dataset has shown to be difficult A better in-situ dataset for coastal case 2 waters is needed – best results so far from Ferrybox data collection • • • Page 25 Glob. Colour GSM 01 merging algorithm shows to be quite robust over coastal waters (tested so far) Glob. Colour merged products show a potential for investigating seasonal to inter-annual variability in coastal zone: One of the Glob. Colour follow-up? In coastal waters with inherent high optical variability, the use the various OC data sources independently remains an asset for resolving temporal and spatial variability Glob. Colour/Medspiration User consultation – November 20 -22, 2007, Oslo All rights reserved © 2006, ACRI-ST