Multi-model Ensemble Prediction of Summer Precipitation in China Based on Machine Learning Algorithms

With the development of machine learning (ML), it provides new means and methods for accurate climate analysis and prediction. This study focuses on summer precipitation prediction using ML algorithms. Based on BCC CSM1.1, ECMWF SEAS5, NCEP CFSv2, JMA CPS2 model data, we conducted the multi-model ensemble (MME) prediction experiment using three tree-based ML algorithms,, the decision tree (DT), the random forest (RF), and the adaptive boosting (AB) algorithm. On this basis, we explored the applicability of ML algorithms to ensemble prediction of seasonal precipitation in China, as well as the impact of different hyperparameters on prediction accuracy. Then, the MME predictions based on optimal hyperparameters were constructed for different regions of China. The results show that all three ML algorithms have an optimal maximum depth less than 2, which means that based on the current amount of data, the three algorithms can only predict positive or negative precipitation anomalies, and extreme precipitation is hard to predict. The importance of each model in the ML-based MME is quantitatively evaluated. The result shows that NCEP CFSv2 and JMA CPS2 have a higher importance in MME for eastern part of China. Finally, summer precipitation in China was predicted and tested from 2019 to 2021. According to the results, the method provides a more accurate prediction of the main rainband of summer precipitation in China. ML-based MME has a mean ACC of 0.3, an improvement of 0.09 over the weighted average MME of 0.21 for 2019-2021, which exhibits a significant improvement over other methods. It shows that ML methods have great potential in improving short-term climate prediction. These results provide an important reference for short-term climate prediction in China. ML-based MME has the potential to accurately forecast the main rainbands of summer precipitation in China.