Peterson, A. R. Institute for Computational Earth System Science, University of California Santa Barbara, email@example.com
Siegel, D. R. Institute for Computational Earth System Science, University of California Santa Barbara, firstname.lastname@example.org
Garver, S. A. Department of Geography and Anthropology, California State Polytechnic University,
OPTIMAL TUNING OF SEMI-ANALYTICAL OCEAN COLOR ALGORITHMS FOR THE GLOBAL OCEAN
Semi-analytical ocean color models potentially will provide a great deal more information concerning the state of ocean ecosystem than their simpler empirical counterparts. However, due to their complexity and reliance on literature estimates, it is difficult to rigorously tune a semi-analytical inversion model. Here, we apply a global parameter optimization procedure, simulated annealing, to tune these model parameters. We employ this procedure to a synthetic data set of remote sensed reflectance and three data products: chlorophyll concentration, and coefficients for non-algal absorption and particulate backscatter. The optimization procedure retrieves the exact model parameter values under perfect conditions (no added noise). Using the synthetic data set with 1, 2 and 5% added noise, reasonable parameter retrievals are still found (typically within 10% of the correct value). When the tuning is performed using only one product from the synthetic data set (the chlorophyll concentration), accuracy in the parameter retrievals may be comprised. We apply this model to field data--a large (N = 919), global data set assembled as part of the recent SeaWiFS BioOptical Algorithm Meeting (SeaBAM)--and compare these results to others in the literature.
Day: Tuesday, Feb. 2
Time: 11:45 - 12:00pm
Location: Sweeney Center