| 摘要: |
| 利用2019年4月1日～7月31日的0～240h欧洲中期天气预报中心2m温度预报和2019年4月1日00时～2019年7月31日00时贵州省境内364个自动气象观测骨干站点的温度观测资料,分别基于时间持续偏差订正(滑动平均)和类卡尔曼滤波的递减平均统计降尺度方法构建单模式的温度订正方案和客观算法模型,进而对欧洲中心2019年7月1日～2019年7月15日的2m温度预报进行客观订正。结果表明：时间持续偏差法构建的方案对订正效果不好,平均绝对误差值(MAE)均比原始数值预报大。相比之下,采用类卡尔曼滤波的递减平均统计降尺度方法构建订正方案对部分站点的预报有着非常显著的订正效果,平均绝对误差值在大多数时效上比数值预报本身小,且经过参数优化方案之后,又进一步提升了订正效果。订正模型的训练方案仍然有很大的改进空间,有望提高有效订正站点的数量。 |
| 关键词: 时间持续偏差;类卡尔曼滤波;递减平均统计降尺度;模式订正 |
| DOI： |
| 投稿时间：2020-12-21修订日期：2021-03-02 |
| 基金项目: |
|
| Research on objective correction algorithm of localized temperature in Guizhou Province |
| Kong Dexuan,Yang Chunyan,Zhu Wenda,Tang Haopeng |
| (Qianxinan Prefecture Meteorological Bureau;Guizhou meteorological station) |
| Abstract: |
| Used by the 0-240h ECMWF 2m temperature forecast from April 1 to July 31, 2019, and the temperature observation data of 364 automatic meteorological observation backbone stations in Guizhou Province from 00:00 on April 1 to July 31, 2019, a single model temperature correction scheme and an objective algorithm model are constructed based on the time continuous deviation correction (moving average) and Kalman filter like decreasing average statistical downscaling method. And the 2m temperature forecast of the European center from July 1, 2019 to July 15, 2019 is objectively revised. The results show that the scheme constructed by time persistent deviation method is not good for correction, and the mean absolute error (MAE) is larger than the original numerical prediction. Compared with the method of Kalman filtering, the effect of the modified scheme is much smaller than that of the other two methods. There is still much room for improvement in the training program of the revised model, which is expected to increase the number of effective correction sites. |
| Key words: time duration deviation; Kalman like filtering; decreasing average statistical downscaling; model correction |
