POSTER Session 2

BING: THE BAYESIAN INFERENCES WITH GORDON COEFFICIENTS (BING) PACKAGE

We will present results derived from the BING package, a new sandbox algorithm for the retrieval of inherent optical properties (IOPs). The primary reason we developed BING was to facilitate forecasts of IOP retrieval accuracy and precision, in past, current and future observational datasets. To demonstrate its value, we re-assess the predictive power of SeaWiFS and MODIS ocean color observations and demonstrate that these have the statistical strength to recover three parameters total describing IOPs absorption and backscattering. We specifically address the challenge posed by the unknown shape parameter used to describe exponential absorption by CDOM and detritus, thereby revealing limitations in retrieving phytoplankton properties from multispectral observations We then forecast the statistical power of PACE, finding that one can retrieve 4 total parameters and possibly a fifth. Our results emphasize the value of Bayesian inference techniques in ocean color analysis, especially the complete handling of error and information content analysis and the explicit definition of priors that constrain the degenerate and difficult problem of IOP retrievals.

J. Prochaska, University of California, Santa Cruz, USA, https://orcid.org/0000-0002-7738-6875

Robert Frouin, Scripps Institution of Oceanography, USA

Code of Conduct
Privacy Policy