Improved prediction model for flood-season rainfall based on a nonlinear dynamics-statistic combined method

Autor: Gui-Quan Sun, Kaiguo Xiong, Yong-Ping Wu, Rong Zhi, ZhiHai Zheng, Jie Yang, Junhu Zhao, Zhiqiang Gong, Guolin Feng, Shaobo Qiao, Ziheng Yan
Rok vydání: 2020
Předmět:
Zdroj: Chaos, Solitons & Fractals. 140:110160
ISSN: 0960-0779
DOI: 10.1016/j.chaos.2020.110160
Popis: Precipitation predictions during the flood season are critical and imperative on continents, especially in monsoon-impacted areas. However, majority of current dynamical models failed to predict the flood-season rainfall very well, although their simulations are high correct. In this study, based on the EOF decomposition of multi-factors field, we used a similar-error correction method to improve model prediction effect, which we call dynamic–statistic combined prediction method. Chinese Global atmosphere-ocean Coupled Model/Climate System Model was combined with dynamic–statistic combined prediction method as a case and the real-time prediction during 2009-2019 were implemented. The spatial anomaly correlation coefficient between predicted and observed values was used to assess the effectiveness of the improvement. The results show that the average anomaly correlation coefficient scores of dynamic–statistic combined prediction method (0.16) is 0.12 higher than that of Chinese Global atmosphere-ocean Coupled Model/Climate System Model (0.04), implying that dynamic–statistic combined prediction method has a broad application prospects in precipitation prediction. We suggest that dynamic–statistic combined prediction method should be promoted to other models for testing.
Databáze: OpenAIRE
Externí odkaz: