ISSN 1000-0526
CN 11-2282/P
CN 11-2282/P
Archive > Volume 51 Issue 5 > 2025,51(5):581-594. DOI:10.7519/j.issn.1000-0526.2025.010202 Prev Next
Evaluation of Multi-Model Forecast of Precipitation in Xin’anjiang Basin During the Ultra-Long Meiyu Season in 2020
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
The refined precipitation forecasts from numerical prediction models are of a crucial support to flood control efforts within river basins. The continuous heavy rainfall during the Meiyu season in 2020 led to the heaviest flood event since the construction of the Xin’anjiang Reservoir and it was the first time that all the sluices were fully opened. Based on precipitation observation data from various stations, this paper examines the forecast performance of four global models and four regional models regarding both overall precipitation patterns and areal rainfall within Xin’anjiang Basin. Additionally, it focuses on evaluating the predictive capabilities of these models regarding extreme and cumulative precipitation effects in the basin to understand whether they can meet the demand of reservoir flood discharge forecasting service. Furthermore, an analysis is conducted to assess how terrain height influences models’ precipitation forecasts. The results show that the global models consistently underestimate the precipitation and have weaker prediction ability for extreme precipitation than the regional models. The regional models mostly overestimate the precipitation but have relatively large variations among the predictions. The regional multi-model ensemble average demonstrates a better forecast performance than single-model results. The regional models perform well in forecasting rainfall of rainstorm to heavy rainstorm, but have some discrepancies in predicting the locations and timing of heavy rainstorm. Compare to the model evaluation of single-day precipitation forecast, it is more instructive to comprehensively consider the cumulative effects and extremity of precipitation prediction. Terrain height significantly influences the prediction of rainstorm events and above. As the terrain height increases, the advantage of regional models becomes evident while the predictive ability of global models for rainstorm events decreases. Especially for ZJWARMS and ZJWARRS, the TS scores increase from below 0.10 to approximately 0.15 or so. Additionally, moderate or lighter intensity rains are not affected by terrain so obviously.
Keywords:
Project Supported:
Clc Number:
P456
Mobile website
