Citation

Yang M, Zhong PA, Li J, Liu W, Li Y, Yan K, Yuan Y, Gao Y. Environ. Monit. Assess. 2020; 192(6): e387.

Copyright

(Copyright © 2020, Holtzbrinck Springer Nature Publishing Group)

DOI

10.1007/s10661-020-08351-w

PMID

unavailable

Abstract

Global climate change and human activities aggravate the frequency of flood disasters. Flood risk includes natural flood risk and risk of economic and social disasters, which is displayed intuitively by flood risk zonation maps. In this paper, we take the disaster-causing factors, the disaster environment, the disaster-bearing body, and the disaster prevention and mitigation capability into consideration comprehensively. Eleven influencing indexes including annual maximum 3-day rainfall and rainfall in flood season are selected, and the virtual sown area of crops is innovated. Taking the Huaihe River Basin (HRB) as the research area, the flood risk prediction of the basin is explored by using the long short-term memory (LSTM). The results show that LSTM can be successfully applied to flood risk prediction. The short-term prediction results of the model are good, and the area where the risk is seriously underestimated (the high and very high risk are identified as the very low risk) accounts for only 0.98% of the total basin on average. The prediction results can be used as a reference for watershed management organizations, so as to guide future flood disaster prevention.

Language: en

Keywords

GIS; Crops; Flood risk; Huaihe River Basin; Intelligent prediction; Neural network