A dynamic early warning model for flash floods based on rainfall pattern identification

In this study, a dynamic early warning model for flash floods was proposed based on the real-time correction of rainfall patterns and the HEC-HMS model, which improved the accuracy and forecast lead time of flood warnings. A dynamic early warning model was developed for the Xinxian, Tengzhou, and Luofubao small watersheds.

The study came to three main conclusions:

1. Characteristic rainfall patterns obtained by cluster analysis reasonably represent regional rainfall patterns. The characteristic rainfall pattern database included various cases of rainfall peak in the front, middle, and later stages of a rainfall process as well as overall uniform distribution.
2. The HEC-HMS model demonstrated reliable simulation performance for the flooding process. After constructing the HEC-HMS model for the Xinxian watershed and using 10 historical flood events to calibrate and validate the model, the NSE values of the model were higher than 0.8, and the time difference of the peak flow occurrence was within 1 hour. 
3. The dynamic early warning model for flash floods based on rainfall pattern identification can effectively improve the accuracy of flash flood warnings and prolong forecast lead time. The accuracy of the warning is gradually improved time by time compared to the traditional warning model that obtains the complete rainfall process and then warns about it.