Knee Point-Based Multiobjective Optimization for the Numerical Weather Prediction Model in the Greater Beijing Area

Abstract

Determination of the optimal parameter values in numerical weather prediction (NWP) models has a significant impact on predictions. Here, we propose a knee point-based multiobjective optimization (KMO) method to find an optimal solution of the NWP model parameters. We apply it to optimize the Weather Research and Forecasting (WRF) model's summer precipitation and temperature simulations for the Greater Beijing area. The results showed that it required fewer than 125 samples (i.e., 25 times the number of dimensions of the parameter space) to obtain the WRF model's optimal parameter values. The optimal parameters determined by KMO outperform the default parameters in WRF simulations for summer precipitation and temperature prediction in the Greater Beijing area, across all periods (calibration, validation, and testing). Additionally, clear physical interpretations are provided to explain why the optimal parameters lead to improved precipitation and temperature forecasting. Overall, the proposed method is effective and efficient to improve NWP.