This paper is based on the study of the ensemble spread–error relationship to obtain more valuable information than historical data, and applicate the information to the ensemble-based quantitative precipitation forecasts of Jiangsu province. The TIGGE and the CMORPH data were used. The conclusions can be drawn as the following.
1, The ensemble spread is highly correlated with the ensemble mean, so confidence index is constructed to measure the spread which removed the effect of the ensemble mean.
2, At some Leadtime, when the spread is greater than a certain threshold, the error is usually positive, and the value of the error and the ensemble spread Showing a linear correlation. Correction can improve TS score.
3, When the confidence index is greater than a certain threshold, using the median instead of the ensemble mean improve the TS score without the increase of the average absolute error.
4, FMM method is slightly better than the spread correction method, but increase the average absolute error, and the spread correction method can be superimposed with the effect of the FMM method.
In summary, these applications of ensemble spread–error relationship in this paper has obvious improvement on ensemble-based quantitative precipitation forecasts of Jiangsu province.Besides, it can be superimposed with other methods and rather easy to compute; therefore, it can be easily implemented on the operational prediction system.
- Indicates paper has been withdrawn from meeting
- Indicates an Award Winner