Geographic-dependent Parameter Optimization based on A-4DEnVar: Simulation with an Intermediate Coupled Model

Coupled parameter optimization (CPO) can reduce the systematic error in coupled model and enhance its capability for weather forecast and climate prediction. However, the implementation of CPO involves in dealing with strong nonlinear processes inherent in coupled model. The analytical four-dimensional ensemble variational (A-4DEnVar) retains the nonlinear processing capability of the four-dimensional variational but gets rid of the dependence on the adjoint model. In this study, a novel dynamic independent point (DIP) scheme combined with a sample-space variable replacement algorithm, which enhances the convexity of the cost function, reduces computational dimensionality, and further expands the parameter subspace, is introduced to A-4DEnVar. Based on the improved A-4DEnVar, we conduct a series of geographic-dependent CPO experiments within an intermediate atmosphere-ocean-land coupled model. Results show that despite the strong nonlinear influence from the coupled model, A-4DEnVar can still accurately capture the geographical characteristics of model parameter, and exhibit high-quality performance in geographic-dependent optimization of cross-component parameter in the scope of strongly coupled data assimilation. Additionally, the DIP scheme presents significant advantages compared to the static independent point scheme, especially with fewer independent points. In the case of only 90 independent points, satisfactory geographic-dependent CPO can be achieved.