Adjoint Assimilation Method In A Marine Ecosystem Model In The Bohai Sea And The Yellow Sea

In this study, the adjoint assimilation method is applied to 3-dimensional Nutrient (DIN, DIP)-Phytoplankton-Zooplankton-Detritus (N(2)PZD ) plankton ecosystem model in the Bohai Sea and the Yellow Sea. The undetermined ecological parameters are optimized by using adjoint method. In addition, a series of identical twin experiments and real experiments are carried out.Identical twin experiments, in which uncorrelated parameters are chosen as control variables on the basis of sensitivity analysis, are employed to study the influences of synthetic errors on the recovery of control variables. The results of experiments indicate that the optimal value of control variables are successfully recovered by using adjoint method in following three situations (observed data sets don't include random errors; 20% random errors are added to observed data sets.; the rest parameters which are not chosen as control variables include synthetic errors.) . The success of parameters recovery proves that the adjoint method is valid and reasonable, which provides valuable experiences and grounds on assimilating real color data.SeaWiFS chlorophyll data over period 1998-2006 as observational data are assimilated in reality experiments. The ecological parameters are optimized when they are invariant in space, and a new method is presented: some independent parameter grids are chosen, the value of other grids'parameters are linearly interpolated by these independent grids. A series of experiments are conducted for these two different parameterization assumptions. The results of these experiments show that compared with the invariant biological parameters, the model which runs with spatially variable biological parameters produces higher simulation accuracies. In addition, we find that the simulation result is influenced by the number of independent parameter grids and increasing the number of independent grids can improve the simulation result.