CFSv2模式对淮河流域夏季气温降水预测能力的评估

Evaluation on summer temperature and precipitation predictions in Huai River Basin by CFSv2 model

基于美国环境预测中心提供的CFSv2模式回报数据以及淮河流域172个气象台站的气温降水观测数据,评估了该模式对淮河流域夏季气温降水的预测能力。结果表明:模式对气温气候平均态的模拟与实况较为接近,降水虽然能够反映出南多北少的分布特征,但气候平均值明显偏小。模式对于气温和降水均方差的模拟均偏小。从频率分布来看,回报气温的频率分布与实况较为接近,回报降水不仅存在很大的负偏差,出现异常降水的频率也比实况明显偏低。对ROC曲线的分析进一步表明模式预测高温事件的能力明显好于低温事件,预测降水异常事件的能力在不同起报月有差异。从主要模态的时空结构来看,模式能够反映出其空间结构,对于增暖的趋势模拟也相当不错,但增暖的速率高于实况,虽然模式没有能够反映出降水主要模态的年代际变化特征,但对于年际变化有较高的预测技巧。这些评估为短期气候预测和模式解释应用提供了参考。

Based on CFSv2 model reforested data supplied by the National Climate Center and observed temperature and precipitation data at 172 stations in the Huai River Basin, the skills of model prediction for summer temperature and precipitation are evaluated. Results show that simulated climatological mean temperature is similar to observation, while simulated precipitation is clearly less than observation mainly due to the underestimation in July and August, despite of the similar northern dry and southern wet spatial distribution. The negative bias is also found in root mean square deviation. From the view of frequency distribution, reforested temperature is similar to observation, but there are great differences for precipitation both in a negative bias and a low frequency of extreme events. Analysis of ROC further indicates that the skill of predicting high-temperature events is better than that of predicting low-temperature events, while the skill of predicting abnormal precipitation depends on leading months. The main spatial structures are well displayed in the main modes. The warming trend is well simulated but with a positive bias in warming speed. There is some skill in predicting annual variability of the main precipitation EOF mode, though the decadal change cannot be embodied. The results from this article can serve as a reference to climate prediction and model interpretation and application.