ISSN 1000-0526
CN 11-2282/P
CN 11-2282/P
ECMWF Precipitation Calibration Based on the Kalman Dynamic Frequency Matching Method
- Article
- Figures
- Metrics
- Preview PDF
- Reference
- Related
- Cited by
- Materials
Abstract:
Based on ECMWF high resolution model grid precipitation forecast data from January 2019 to February 2020 and 388 meteorological stations precipitation observation data, the forecast performance of the model precipitation frequency is objectively verified, and the ECMWF grid precipitation forecast is corrected by using the Kalman dynamic frequency. The main conclusions are as follows. The ECMWF model has significantly higher forecast frequency for small-scale precipitation than the observation, but lower frequency for torrential rain. The frequency of model forecast and observed precipitation are significantly different in different seasons. Matching the frequency of forecast precipitation with that of observed precipitation does not get the highest precipitation forecast score. Based on the Kalman filtering method, the forecast and observed precipitation frequency can be dynamically matched, the model forecast frequency can be revised to be basically consistent with the observation, the standard deviation of the model forecast precipitation can be improved, and the phenomenon that the model forecast errors are more for light precipitation but less for severe precipitation can be significantly adjusted. Due to the position or time deviation of precipitation forecasted by the model, appropriate coefficients are selected so that the frequency of torrential rain forecast is slightly higher than the observed frequency, and the light rain frequency is slightly lower than the observed frequency, thus, a better precipitation forecast score can be obtained. According to the precipitation characteristics of different regions, the Kalman precipitation dynamic frequency is calculated separately to correct the precipitation, which can effectively improve the TS forecast score of torrential rain, but the accuracy of sunny and rain forecast is not significantly improved.
Keywords:
Project Supported:
Clc Number:
Mobile website
