Estimating Biological Parameters of a Coupled Physical-Biological Model of the Indian Ocean Using Polynomial Chaos

A statistical emulator technique, namely polynomial chaos, has been used to estimate two time-dependent biological parameters of a coupled physical-biological model of the Indian Ocean. This has been achieved by minimizing a distance function representing misfit between model simulated and satellite-derived surface chlorophyll. First, the parameters have been assumed to be constant in time and optimized values have been found by minimizing a time-averaged distance function. Since no significant improvement in model simulation has been found using a fixed set of optimum parameters, minimization has been carried out daily, assuming the parameters to be time-dependent. Emulation with this set of parameters has led to a significant improvement in the simulated surface chlorophyll. Smoothing of the parameters with singular spectrum analysis has caused less noisy simulations, at the cost of increasing the model data misfit. Time-varying parameters have been found to be more suitable for the hindcast of daily averaged chlorophyll both in the Arabian Sea and the Bay of Bengal.