基于强度尺度法的雷达和数值模式定量降水预报检验对比

VERIFICATION OF QUANTITATIVE PRECIPITATION FORECAST BETWEEN RADAR AND NUMERICAL MODEL BASED ON INTENSITY-SCALE METHOD

为提高雷达外推和数值模式对短临预报能力的认识，采用强度尺度方法对比检验了2012年浙江省梅汛期内的雷达外推和WRF、T639、ECMWF4类定量降水预报（QPF）产品。检验结果表明：短临预报的大部分误差来自较小的时空尺度，在强降水方面表现尤为明显。对于雷达和模式的预报能力交叉点（3h前后）的QPF检验，雷达的误差比率随着空间尺度增大先减后增，而模式随尺度增大误差比率总体递减。在中小尺度（2°及以下）雷达总体较准确，但0．25°尺度是个特例，在较大尺度（4°及以上1的弱降水上雷达不如模式。雷达在0．5°、1°尺度上相对误差贡献最少；在2°、4°尺度下，ECMWF表现最优。同时，雷达对于中-β和较小的中-α尺度系统的预报检验效果最好，对尺度达几百公里的弱系统不如数值模式，对于10-30km的中-γ尺度系统雷达预报带来近74％的空报，今后需提高其对中-γ尺度系统生消的预报能力。

In order to improve the understanding of the short-term quantitative precipitation forecast, four quantitative precipitation forecast products （QPF） including radar extrapolation and three numerical models （WRF, T639, ECMWF） during a 2012 Meiyu rainstorm in Zhejiang province were verified. The results showed that most errors of short-term QPF come from small spatial and temporal scale, especially for intense rain. For 3 hour QPF verification, the error ratio of radar first drops and then increases as the spatial scale increases while model forecast generally decreases. It is found that radar was more accurate for spatial scale lower than 2 o except for 0.25°, yet less accurate for weak rainfall for spatial scale higher than 4°. The relative error was the least for the 0.5° and 1° spatial scale, and in the meantime ECMWF was the best for the 2°and 4°spatial scale. In other words, the test results of the radar extrapolation was the best for M-β and small M-α spatial scale system, opposite to weak rainfall in a system less than 1 000 kilometers. Meantime, about 74% of the error comes from M-γ, spatial scale in radar QPF. Therefore, further study is needed to improve radar extrapolation forecast on the M-γ, spatial scale.