DMS model calibration using Genetic Algorithm

Abstract

Recent researchers suggested Dimethyl sulphide (DMS) flux emission in Arctic Ocean plays an important role for the global warming. A Genetic Algorithm (GA) method was developed and used in calibrating the DMS model parameters in Barents Sea in Arctic Ocean (70-80N, 30-35E). Two-step GA calibrations were performed. First step was to calibrate the most sensitive parameters based on Chlorophyll_a (CHL) satellite SeaWIFS 8-day data. DMS model was then calibrated for another 5 most sensitive parameters. The best fitness was as good as -0.76 for CHL calibration in 1998-2002. The GA proved an efficient tool in the multiple-parameter calibration task. Model simulations indicate significant inter-annual variation in the CHL amount leading to significant inter-annual variability in the observed and modeled production of DMS and DMS flux in the study region in Arctic Ocean.