Evaluation of multi-models forecast of precipitation in Xin’anjiang River Basin during the ultra-long Meiyu in 2020
Article
Figures
Metrics
Preview PDF
Reference
Related
Cited by
Materials
Abstract:
The continuous heavy rainfall during the Meiyu season in 2020 (May 29 to July 17) led to event since the construction of the Xinanjiang Reservoir and the first time that all the sluices were fully opened. Refined precipitation forecasts are of great support to flood control efforts in the basin. Based on observed precipitation data from various stations, this study examines the forecast performance of four global models and five regional models regarding both overall precipitation patterns and area-specific rainfall within Xin"an River Basin. Additionally, it focuses on evaluating the predictive capabilities of these models regarding extreme and cumulative precipitation effects in the basin to determine whether they can meet the demand of reservoir flood discharge forecasting service. Furthermore, an analysis is conducted to assess how topographic height influences each model"s precipitation forecasts. The results show that: (1) The global model consistently yielded lower precipitation forecast results compared to the regional model. The regional model exhibited high accuracy but had relatively large variability among its predictions. The regional multi-model ensemble average demonstrated superior forecast performance than single model results. (2) The regional model performed well in forecasting rainfall ranging from rainstorm to heavy rainstorm; however, there were some discrepancies in predicting the magnitude and timing of heavy rainstorms. (3) Compare to forecast evaluation of single-day precipitation in models it is more instructive to comprehensively consider the cumulative effects and extreme of precipitation predicted by the models. (4) Topographic height significantly influences extreme rainfall prediction for heavy rainfall events and above. As topographic height increases, the advantage of using a regional model becomes evident while the predictive ability of a global model for heavy rainfall events decreases. Especially for ZJWARMS and ZJWARRS, the TS score has increased from below 0.1 to approximately 0.15. Additionally, moderate or lighter intensity rains do not exhibit significant prediction effects.
Keywords:
Project Supported:
Science and Technology Plan Projects of Zhejiang Province (2020ZD06、2022ZD14、2023ZD14), The Project Supported by Zhejiang Provincial Natural Science Foundation of China joint meteorological found (ZJMA23D50007)