MODES系统对贵州月气温、降水预测初步评估

Assessment of Multi-model Downscaling Ensemble Prediction System for Monthly Temperature and Precipitation Prediction in Guizhou

通过对2013年1月—2015年6月(MODES)发布的最优月预测产品在贵州省月平均气温距平和降水距平百分率的预测检验评估,发现MODES对全省平均气温有较好的预报,分析时段内预测与实况的相关系数为0.24,距平同号率为65.5%,且对气温偏高预测的可参考性高于其对气温偏低的预测。相比于气温,MODES对降水预测能力较弱,参考性也相对较低,其中对贵州全省平均降水偏多趋势的预测技巧要优于对全省平均偏少趋势的预报技巧。逐站分析显示,MODES对贵州气温预测效果较好的地区在西部、北部和东部,对降水偏多的预测效果较好的地区位于除西北部和北部边缘地区外的其余大部地区。通过对MODES与预报员综合预报的结果评估发现,MODES月预测总体效果较预报员好,且稳定性高于预报员,可为预报员提供参考信息。

Based on the outputs of different climate models, National Climate Center of ChinaMeteorological Administration (CMA)established a multi-model downscaling ensemble predictionsystem (MODES)in 2011. Since then the system has been used widely in the domestic seasonalprediction operation. By assessing and verifying the prediction of MODES for monthly temperatureand precipitation in Guizhou province during January 2013 to June 2015, this paper indicates thatthe system has a good temperature prediction skill in the province. The correlation coefficient andthe total ration of same anomaly symbol between MODES and observation are 0.24 and 65.5%.Prediction skill of positive temperature anomaly is higher than the negative anomaly. Compared totemperature prediction, the prediction skill for precipitation in MODES is lower, but the predictionskill of positive precipitation anomaly is higher than the negative anomaly. Spatial distribution ofverification results show that the high skill scores appear in the western, northern and eastern partof Guizhou province for temperature, and in most regions for precipitation except in part ofnorthwestern and northern regions. During the research period, the prediction skill scores are stablyhigher than the scores of the subjective forecasts made by the forecaster, especially for negativetemperature anomaly and for positive precipitation anomaly.