ECMWF模式的延伸期日最高气温预报偏差估计及订正分析

Error Estimation and Correction of the Extended-Range Daily Maximum Temperature in the ECMWF Model

使用欧洲中期天气预报中心(ECMWF)20 a的集合预报回算数据,检验分析了延伸期第16~30天预报时效其对我国日最高气温的预报性能。结果表明,西部地区预报误差明显大于中东部地区。全国平均而言,模式预报较实况偏低1.1~1.39℃,均方根误差为4.6~4.9℃。进一步分析指出,第16天均方根误差最小,且随着时效的延长其略有增大。夏季模式预报效果最好,春季和秋季的部分时段预报效果较差。基于历史偏差订正方法,对2018年6月—2019年6月的日最高气温预报进行了误差订正试验。结果显示,订正后的预报准确率提升了15.2%~19.2%。聚焦2018年7月的一次中东部地区大范围高温过程,模式原始预报明显低估了高温强度,订正预报更接近实况,显示其具有一定的订正效果。

Based on the 20-year ensemble forecast data of European Medium-Range Weather Forecast(ECMWF),the verification of daily maximum temperature forecast in the extended-range(days 16~30) in China was used to estimate the forecast skill. The results showed that:forecast errors of ECMWF model in western China are significantly great than that in the mid-eastern China.For the whole country,the predicted daily maximum temperature is 1.1 ~1.39 ℃ lower than the observation and the root mean square error(RMSE) is 4.6 ~4.9 ℃. Further analysis demonstrates,RMSE of the day 16 forecast is the smallest among all valid times,and it increases slightly as the forecast time goes on. The performance of model is the best in summer,while in some periods of spring and autumn,it is poor. The historical deviation correction method is used to calibrate the daily maximum temperature forecast from June 2018 to June 2019 in this study. The verification demonstrates that the forecast accuracy of corrected daily maximum temperature forecast is improved by 15.2% ~ 19.2%. The extent of a high temperature process in the mid-eastern China in July 2018 is obviously underestimated,while the calibrated forecast provides closer information to the observation.It shows that this bias correction method is effectual.