京津冀夏季雷达定量降水估测的误差统计及定量气候校准

Characteristics of summer QPE error and a climatological correction method over Beijing-Tianjin-Hebei region

  • 摘要: 雷达定量降水估测(QPE)是短时临近预报的关键部分,在定量降水预报(QPF)、强降水预警、城市积水内涝、地质山洪灾害、精细化天气服务等方面具有重要作用。利用京津冀地区雷达定量降水估测资料和逐时自动气象站降水观测数据,分析了2011—2016年夏季京津冀地区雷达定量降水估测的误差空间分布特征,并重点提出了一种新的雷达本地化定量气候校准算法。结果表明,京津冀地区雷达定量降水估测较好地反映了总降水量东北—西南带状分布特征,但西北部山区、东北部山区及西南部山区估计偏弱,东北部山前地带估计偏强,西北部存在虚假降水估计,而北京市城区估计最为准确。利用雷达本地化定量气候校准算法对1 h雷达定量降水估测进行气候尺度上的约束订正,检验结果表明,经过校准后的雷达定量降水估测偏差(BIAS)、平均绝对误差(MAE)、均方根误差(RMSE)和均方根相对误差(RRMSE)均减小。绝大部分站点偏差减小幅度超过50%,京津冀东部及南部平原地带平均绝对误差、均方根误差和均方根相对误差减小幅度在20%左右,而北部及西南部山区误差减小幅度相对较小。降水个例检验结果表明,经过雷达定量气候校准后的雷达定量降水估测强度更接近自动气象站观测的降水量级,且降水结构细致,偏差、平均绝对误差和均方根误差均减小,与自动气象站观测降水的相关系数增大,因此该算法有助于改进雷达定量降水估测的准确度。

     

    Abstract: Radar Quantitative Precipitation Estimation (QPE) is a key component of short-term and nowcasting. The radar QPE plays an important role in many fields such as Quantitative Precipitation Forecast (QPF), heavy precipitation warning, urban waterlog, geological mountain torrent disaster, and fine weather service. In this paper, the spatial distribution characteristics of radar QPE over Beijing-Tianjin-Hebei region are analyzed using the radar QPE data and automatic weather station hourly precipitation data in the summers of 2011-2016. Specifically, a new type of climatological correction algorithm is proposed. Results show that the radar QPE can well reflect the northeast to southwest distribution of observed total rainfall. However, the radar QPE is underestimated over mountain areas in the northwest, northeast and southwest, and overestimated over the piedmont zone of the northeastern mountain area. Besides, false precipitation estimation is found over the northwest of Beijing-Tianjin-Hebei region. Precipitation over the urban area of Beijing estimated by radar is the closest to observations. The newly proposed climatological correction algorithm is then applied for hourly radar QPE to calibrate previous radar estimates. Results of tests show that the BIAS, the mean absolute error (MAE), the root mean square error (RMSE) and the relative RMSE (RRMSE) all are reduced after the correction. In particular, the reduction percentage of BIAS reaches more than 50% at most stations. The reduction percentages of MAE, RMSE and RRMSE are about 20% in the eastern and southern plain, but they are relatively small in the northern and southwestern parts. The test of precipitation cases shows that the strength of radar QPE is closer to station observations and the high-resolution structure of precipitation is well captured by radar after the correction. Furthermore, the BIAS, MAE and RMSE of precipitation cases all are reduced, and the correlation coefficient between the radar QPE and observations is increased. Hence, the climatological correction algorithm can improve the accuracy of the radar QPE. The correction algorithm is useful for operational weather forecasting and can be widely used in other areas.

     

/

返回文章
返回