高级搜索

ECMWF模式对2020年夏季江淮流域降水的预报偏差分析

Analysis of the deviation of precipitation forecast of ECMWF model over the Yangtze-Huaihe River Valley in summer 2020

  • 摘要: 基于CMPAS多源融合降水和ERA5再分析产品,评估ECMWF全球高分辨率确定性预报产品对2020年梅雨期(6月10日—7月20日)极端强降水过程的预报性能。同时,基于面向对象的诊断评估方法(MODE),揭示ECMWF模式对强降水落区的质心经纬度、面积、长度、宽度、轴角等空间特征的预报性能。结果表明,ECMWF模式对于梅雨期的日降水量预报,在雨带的空间位置上,模式预报偏北、偏西的偏差较多;在落区形态上,模式预报的雨带面积偏大,轴角倾斜度更大。观测中江淮流域区域平均降水的日变化主峰值出现在清晨至上午,ECMWF预报能够再现降水日变化特征。针对模式对主雨带南北落区质心位置预报偏差的评估表明,模式预报主雨带位置偏北的频次呈现出双峰分布的日变化特征,峰值出现在夜间和午后。雨带位置预报偏南的频次为单峰分布,峰值在上午。低空急流的日变化特征明显,且峰值时刻超前降水峰值时刻3 h,而ECMWF预报急流峰值时刻则较观测早3 h。ECMWF预报降水落区位置偏差与预报低层南风分量的强弱偏差相关,当对流层低层南风分量偏强时,雨带位置预报易偏北;南风分量较弱时雨带位置预报易偏南。针对ECMWF预报位置偏北和偏差较小的两次典型强降水事件的对比分析,结果表明在小时尺度上急流与降水的日变化一致,ECMWF预报降水落区的偏北与前3 h内强度更强的急流有关。

     

    Abstract: Based on CMPAS precipitation products and ERA5 reanalysis datasets, the performance of the ECMWF global high-resolution products (ECMWF) have been analyzed during the typical Meiyu (from 10th June to 20th July) period in 2020. Using the method of object-based diagnostic evaluation (MODE), the performance of the ECMWF in simulating the spatial characteristics of heavy precipitiation, such as the centroid latitude, the centroid longitude, the rainfall area, the major axis length, width and axis angle of precipitation, have been comprehensively revealed. The results show that the ECMWF tends to have a northward and westward bias in reproducing the Meiyu rain belt, in terms of the daily precipitation. Regarding the spatial morphology characteristics of the rain belt, the ECMWF has a larger rainfall area and has a steeper slope than that in the observation. The diurnal variation of precipitation within the rain belt peaks around early-morning in the observation, and the ECMWF can well forecast the diurnal variation of precipitation. In addition, the results also show that the frequency of the northward bias in forecasting the rain belt peaks in the evening to midnight and afternoon, while the bias of the southward mainly occurs in the morning. The low-level jet exhibits similar diurnal variations but with a 3 hours earlier peak than the observed precipitation. The peak time of low-level jet in the ECMWF forecast is 3 hours earlier than that in observed. Further analysis reveals that the zonal location of the rain belt is highly correlated with the intensity of low-level jet. The rain belt moves further northward when the low-level jet is strengthened, and visa-versa. Finally, two heavy rainfall events were chosen to represent the typical northward bias and relatively accurate forecast of the zonal position of the rain belt. The results show that the northward bias of the rain belt results from the stronger low-level jet 3 hours ahead in the ECMWF forecast.

     

/

返回文章
返回