Inversion of the Spatially Varying Parameters in a Marine Ecosystem Model

Abstract

Abstract The adjoint variational method is applied to numerical simulation of marine ecosystem dynamics on global scales based on the spatial parameterization in this study. On the basis of previous studies, we make improvements and conduct discussion in detail by assimilating chlorophyll-a data into a simple NPZD model. When the spatially varying Vm (Maximum uptake rate of nutrient by phytoplankton) is estimated alone, new strategies are designed to optimize the step-length that is used to adjust the parameters preferably and the assimilation efficiency is improved. On the condition that the same step is employed, the reduced cost function (RCF), the mean error of phytoplankton in the surface layer (ME), the absolute average error (AAE) and the relative average error (RAE) of Vm between given and simulated values decrease obviously compared with strategy in previous work. How the distribution schemes of spatial parameterization and influence radius affect the results is discussed by utilizing the above strategy, and the optimal influence radius corresponding to each distribution scheme is obtained. On the basis of the above work, when the five key parameters are estimated separately, the two given types of spatial variations could be reproduced and the RAE are less than 3%. It demonstrates that the simulation precision and computing efficiency could be improved by utilizing the improved strategy to modify the step-length and by using the optimized distribution schemes of independent grids.