ISSN 1000-0526
CN 11-2282/P
CN 11-2282/P
Archive > Volume 45 Issue 4 > 2019,45(4):469-482. DOI:10.7519/j.issn.1000-0526.2019.04.003 Prev Next
Verification of China Extreme Temperature Forecasts in 2016 Based on T639 Ensemble Forecast
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
Based on the realtime forecast and analysis data of 2 m temperature from the T639 ensemble forecast system in China, the climatic percentile distribution is estimated for both realtime forecast and analysis data, and the forecasts of extreme high temperature events and extreme low temperature at typical stations over China in 2016 are verified and evaluated. The analysis results show that the forecasted climate percentile distribution at different lead time is basically the same as the percentile distribution of the analysis for a given station site, but they are different to some degree between different regions and different seasons. The results of the Talagrand distribution show that there is a warm bias in the forecasts of Harbin and Changsha Stations by the T639 ensemble system and a cold bias in the predictions of Beijing and Lhasa Stations. The other typical stations show a low ensemble spread. Based on the definition of historical climate percentile, the 2016 extreme temperature identified, forecasted and verified. The results of TS score show that the T639 ensemble system has a certain forecasting performance for extreme temperatures in China, but the prediction skill has apparent difference in different regions. The extreme high temperature has higher prediction skills in the south to the Yangtze River and the Northeast China, and the extreme low temperature has higher prediction skills in the northern and southern parts of China. The comparisons of different forecasting methods show that the ensemble mean method has a smoothing effect on the extreme signals and the overall prediction skill is low. The ensemble model method improves obviously the extreme low temperature skill, while the method of using maximum or minimum in all ensemble members can amplify the extreme signals and increase significantly the prediction skill of extreme high temperature, but not clear for low temperature. Thus, we can see that proper extraction of extreme information in ensemble forecasting is of crucial importance.
Keywords:
Project Supported:
Clc Number:
Mobile website
