Citation

Lin K, Zhou J, Liang R, Hu X, Lan T, Liu M, Gao X, Yan D. Nat. Hazards 2021; 108(2): 1427-1448.

Copyright

(Copyright © 2021, Holtzbrinck Springer Nature Publishing Group)

DOI

10.1007/s11069-021-04739-0

PMID

unavailable

Abstract

Flash flood disaster, with strong suddenness and tremendous destructiveness, is one of the most severe natural disasters in China that seriously threaten the lives and property safety of people and social development. Owing to the complex terrain and limited rainfall and runoff monitoring gauges, it is arduous to effectively prevent and control flash flood disasters in small-sized and medium-sized hilly watersheds. Identifying rainfall threshold, critical discharge and warning periods for flash flood, is critical in disaster prevention in such regions. This study adopted two approaches, the entropy-based decision approach and the hydrological model approach in calculating rainfall thresholds under different antecedent moisture condition (AMC). In particular, the entropy-based decision approach was improved, by using the Frank copula to calculate the multivariate joint distribution of the cumulative rainfall and the corresponding peak discharge. These two approaches were validated in a typical basin, located in the Pearl River Delta in South China that is characterized by frequently heavy rainfall and floods in the monsoon.

RESULTS showed that both approaches have the ability to quantify rainfall thresholds. Relatively, the Bayesian method exhibited higher rainfall thresholds, comparing to the other methods. In particular, for AMC I, the utility-entropy risk function method (with λ = 1) exhibited the best-applied practicable forecast lead time of 2.0 h; as for AMC II, the two methods showed the same forecast lead time of 1.5 h, and for AMC III, the hydrological model-based approach showed an optimal forecast lead time of 6.0 h. Meanwhile, the entropy-based approach exhibited the same performance as hydrological modeling approaches, indicating that this method is practicable in flood forecasting. The results would be helpful for floods prevention and mitigation in such small-sized and medium-sized hilly watersheds.

Language: en