Simulated performance of remote sensing ocean colour algorithms during the1996 PRIME cruise

Coincident pigment and underwater radiometric data were collected during a cruise along the 20 degreesW meridian from 60 degreesN to 37 degreesN in th e north-eastern Atlantic Ocean as part of the Natural Environment Research Council (NERC) thematic programme: plankton reactivity in the marine enviro nment (PRIME). These data were used to simulate the retrieval of two bio-op tical variables from remotelysensed measurements of ocean colour (for examp le by the NASA Sea-viewing wide field-of-view sensor, SeaWiFS), using two-b and semi-empirical algorithms. The variables considered were the diffuse at tenuation coefficient at 490 nm, (K-d(490), units: m(-1)) and the phytoplan kton pigment concentration expressed as optically-weighted chlorophyll-a co ncentration [C-a, units: mg m(-3)]. There was good agreement between the me asured and the retrieved bio-optical values. Algorithms based on the PRIME data were generated to compare the performance of local algorithms (algorit hms which apply to a restricted area and/or season) with global algorithms (algorithms developed on data from a wide variety of water masses). The use of local algorithms improved the average accuracy, but not the precision, of the retrievals: errors were still +/- 36% (K-d) and +/- 117% (C-a) using local algorithms. (C) 2001 Elsevier Science Ltd. All rights reserved.