ISSN 1000-0526
CN 11-2282/P
CN 11-2282/P
Archive > Volume 47 Issue 5 > 2021,47(5):586-600. DOI:10.7519/j.issn.1000-0526.2021.05.007 Prev Next
Analysis on Prediction Skills of S2S Models for Extreme Precipitation During Flood Season in Sichuan Province
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
Based on the precipitation data of Sichuan meteorological stations and the hindcast data of eight models in the World Meteorological Organization (WMO) sub-seasonal to seasonal (S2S) forecasting project from 1995 to 2010, forecasting skills of the models for extreme precipitation events during the flood season in Sichuan Province are evaluated and analyzed with five methods. These methods are probability of detection (POD), false alarm ratio (FAR), Heidke’s skill score (HSS), mean error and absolute error. As the results show, the prediction skills of S2S models for extreme precipitation in Sichuan Province are overall low, showing a feature of “low hit rate, high false alarm rate, and the forecast value far less than the observation with large deviation”. The forecast capability of all models improves with the lead-time shortening and their prediction performances on the synoptic scale are better than that on the sub-seasonal scale. Spatially, the qualitative prediction skill of the models is the highest in the south of western Sichuan Plateau, and the lowest in the east of the basin or Panxi Region. Meanwhile, the prediction absolute error present the distribution of “the largest in the basin, the secondly largest in the Panxi Region, the smallest in the western Sichuan Plateau”, and the maximum absolute error appears along the mountains in the western part of the basin. The prediction skills are different in each month during the flood season. The qualitative prediction skills are higher in the main flood season, especially in midsummer than in other months. However the quantitative prediction skill is the lowest in midsummer. Under consideration of quantitate and qualitative forecast ability, among the eight models, models from the United Kingdom Met Office (UKMO) and the Institute of Atmospheric Sciences and Climate of the National Research Council (CNR-ISAC) perform better in forecasting the extreme precipitation in Sichuan Province on the synoptic and the sub-seasonal scale, respectively. The best prediction models for the basin and Panxi Region are consistent with that of the whole province. However, for western Sichuan Plateau, model from the Korea Meteorological Administration (KMA) is optimal on the synoptic scale while models from the Australian Bureau of Meteorology (BoM) and CNR-ISAC are the best forecasting models on the sub-seasonal scale.
Keywords:
Project Supported:
Clc Number:
Mobile website
