Atmospheric corrections and interpretation of marine radiances in czcs imagery, revisited

In a previous study (Bricaud and Morel, Oceanologica Acta, 1987), a reflectance model developed for oceanic Case 1 waters was introduced into CZCS data processing, mainly with the aim of improving the atmospheric correction scheme and permitting discrimination between Case 1 and turbid Case 2 waters. The present paper, not limited to CZCS, examines the possible extension of the use of this reflectance model. In particular, it is proposed that the relationships required in iterative computations when estimating marine radiances in the four (visible) channels be those which derive from the model rather than the empirical expressions. Also, in order to ensure an internal coherency, it is proposed to operate the modelled relationships between the reflectance ratios and the chlorophyllous pigment concentration as algorithms for pigment retrieval. The performances of a pixel-by-pixel procedure (PPP), which iteratively produces the best estimates of both the atmospheric and the marine signals, are assessed through sensitivity analyses. To this end, the application of PPP (or any other procedure) is simulated by using a modelled ocean-atmosphere system, in which the aerosol properties, the pigment content, the sun and viewing geometrical conditions can be changed. We also examine the consequences: of using a mean Angstrom exponent instead of those which result at the pixel level from the application of the PPP; of considering the "full" pigment concentration range when applying the PPP, i.e. a range which considerably exceeds that of "clear water"; of the inevitable deviation of actual waters with regard to the modelled Case 1 waters (which are underlying as reference waters in such a processing). It is believed that improvements can be expected from the present procedure, with the proviso that well calibrated, remotely sensed data should be available.