Predicting the spatiotemporal chlorophyll-a distribution in the Sea of Japan based on SeaWiFS ocean color satellite data

We developed a new statistical spatiotemporal model for chlorophyll-a (chl-a) distribution over the Sea of Japan, derived from the satellite-based Sea-viewing Wide Field-of-view Sensor (SeaWiFS). Because preliminary analysis showed that the SeaWiFS data exhibit anisotropy in space and autocorrelation in time, we propose a new spatiotemporal model for chl-a distribution and its predictor. Numerical prediction experiments applying the SeaWiFS data showed that the predictor could forecast chl-a distributions in summer and early fall well, although further changes in the model structure will be necessary to predict aspects of the spring and late fall blooms.