PHOTON TRANSPORT IN 3-DIMENSIONAL STRUCTURES TREATED BY RANDOM-WALK TECHNIQUES - MONTE-CARLO BENCHMARK OF OCEAN COLOR SIMULATIONS

The analysis of light-rays penetrating transparent media like air and water constitutes a pertinent problem in climatic research and in the development of algorithms for the retrieval of bio-geo-chemical parame ters of suspended matter and dissolved pollutants from remotely sensed ocean colour data. On the basis of the neutron transport code TIMOC a visible and near-infrared photon transport code, called PHO-TRAN, has been developed and tested both against a theoretical benchmark of mod els and on real data. The photon transport allows for reflection, refr action, absorption and all relevant light-scattering models, such as e lastic scattering by molecules and strongly forward scattering process es by particulates, in quite general three-dimensional geometries. The scores are flux, radiance, irradiance, averaged over regions or surfa ces as well as point detectors. In some cases it is also possible to s olve the adjoint problem, as for example in the case of strongly colli mated light sensors with extended sources. Adjoint problems - as encou ntered when complex geometry structures are to be considered in the vi cinity of a detector - are solved to evaluate the effects of sensor sh adowing and self-shadowing, in order to provide vicarious calibration information for operational satellite remote sensors. While the featur es listed above, are not all unique, PHO-TRAN offers also the possibil ity to sample first-order derivatives of almost all responses with res pect to certain input parameters, like material densities and characte ristic constants used in scattering models. The knowledge of derivativ es, which can be determined with very little supplementary computing e ffort, provides valuable additional information which provides sensiti vity profiles and allows for multivariate perturbation estimates, unce rtainty analysis and data adjustment. This paper describes mainly the benchmarking of the developed Monte Carlo procedures. (C) 1998 IMACS/E lsevier Science B.V.