GRAPES_GZ 3 km模式对2019年海南岛暖季非台风降水预报的时空检验

利用基于目标诊断的空间检验方法(MODE)和时空检验方法(MTD)评估了华南3 km高分辨率区域数值模式(GRAPES_GZ3 km)对2019年海南岛暖季非台降水预报性能, 结果显示: (1)模式24 h累积降水预报的空间分布范围偏大、降水强度偏强; (2)模式逐小时降水预报的平均质心总体偏西和偏北, 降水出现时间总体偏早1~3 h, 结束时间总体偏晚2~4 h, 降水持续时间偏长; 预报的降水目标数量偏多, 与实况一致均存在着主峰和次峰形态的昼夜分布特征, 但预报的昼间主峰出现时间比实况偏早2 h; 预报的短时强降水出现频次总体偏多。相对于传统的预报和观测点对点检验评估方法, MODE和MTD方法具有捕捉模式预报偏差特征的优势。 

EVALUATION OF SPATIO-TEMPORAL PARAMETERS OF FORECASTS FROM GRAPES_GZ3KM MODEL: WITH SPECIFIC REFERENCE TO NON-TYPHOON PRECIPITATION DURING THE WARM SEASON IN 2019 IN HAINAN ISLAND

The present study used the method for object-based diagnostic evaluation (MODE) and the time dimension version MODE Time Domain (MTD) to evaluate the performance of non-typhoon precipitation forecast from the GRAPES_GZ3km high-resolution regional numerical model. The non-typhoon precipitation referred to the one that happened during the warm season in 2019 in Hainan Island. The results showed that: (1) The spatial distribution range of the model''s 24 h cumulative precipitation forecast was too large and the precipitation intensity was relatively strong; (2) The average centroid of hourly precipitation forecast from the model was generally westward and northward, and the start time of precipitation was generally 1~3 h earlier and the end time was generally 2~4 h later; the precipitation duration was longer. The forecast rainfall target number was more than the actual number; the day and night distribution features of the main peak and the sub-peak were consistent in the forecast and observation, but the main peak in the day was predicted to appear 2 hours earlier than that the observed. The frequency of short-time heavy rainfall forecast was generally more than the observed. Compared with the point to point evaluation of traditional forecast and observation, MODE and MTD methods have the advantage of capturing the characteristics of model forecast deviations.