一种联合历史和实时资料的定量降水短时预报技术研究

Objective calibration method on short-term quantitative precipitation forecast combining historical and real-time data

本文提出一种基于百分位映射,使用实时和历史资料联合订正模式定量降水短时订正预报技术,并与仅使用历史资料或实时资料的订正效果进行对比。结果表明:三种方案均能有效订正模式原始降水预报偏差,提高0~12 h降水预报准确率。对于晴雨预报,采用联合订正方案,预报效果最优。在0~7 h预报时效内,仅采用实时资料的订正方案准确率明显优于仅采用历史资料的订正方案,在8~12 h预报时效内,后者准确率稍高。所有预报时效内,仅采用实时资料的订正方案降水预报范围较仅采用历史资料的方案略偏大。对于较强降水预报,采用联合订正方案准确率为三种方案最优,仅采用实时资料的方案预报准确率虽优于仅采用历史资料的方案,但预报范围及量级较实况明显偏大。

By jointly using the historical and observation data,an objective correction method on short-term quantitative precipitation forecast was proposed based on percentile mapping.Moreover,the correction effect was compared with the results of the correction schemes that only use a historical data or observation data.Results show that the three schemes can effectively correct the deviation of the original model precipitation forecast and improve the accuracy of the 0-12 h precipitation forecast.For the sunny/rainy forecast,the joint correction scheme by using both data performs the best.For the 0-7 h precipitation forecast,the accuracy of the scheme by using only observation data is significantly better than that of using only historical data.For the 8-12 h precipitation forecast,the latter has a slightly higher accuracy.For all the forecast leading time,the precipitation range forecasted by the scheme using only observation data is slightly larger than that of using only historical data.For the heavy rainfall forecast,the accuracy of the joint correction scheme is the best among the three schemes,and the forecasted precipitation range is closer to the observation.Although the forecast accuracy of the scheme using only observation data is better than that using only historical data,the forecasted precipitation range and magnitude are obviously larger than observations.