MODELLING CHLOROPHYLL A OF BOHAI BAY BASED ON CA-SVM USING REMOTE SENSING DATA

In this research, an integrated cellular automata (CA) and support vector machine (SVM) model was developed to model the concentration of chlorophyll a in Bohai Bay using remote sensing data. In order to take into account the spatial heterogeneity and local interaction of the blooms, cellular automata paradigm was implemented. The SVM was used here to constructing nonlinear transition rules for CA, for its better capacity of dealing with nonlinear complex relationships. The CA-SVM model was established with three different cases, which reflect the effect of 8-neighbouring cell of chlorophyll a concentration, the spatial position of the cell, and external factors such as SDD, SSC, SST. The model was calibrated and verified by remote sensing data of Bohai Bay. Through this study, it is shown that the spatial position, and external factors such as SDD, SSC, SST have a great effect on the distribution of chlorophyll a, and the proposed CA-SVM model could be preferably to simulate and predict the concentration of chlorophyll a in Bohai Bay, and capture the non-linear information in ecological processes.