Abstract
China is one of the countries in the world that seriously affected by flash floods disasters. The flash flood caused by extreme rainfall occurred at mountainous small-sized watersheds in China often leads to serious economic damages and obstructs the social development. Setting up an efficient forecasting system for flash flood has been widely accepted as one of the key non-structural measures to improve the control and prevention capability of China. However, due to the data limitation, establishing forecast models in those flash flood areas is challenged by the lack of parameter references. This paper proposed a new machine learning approach based on the Random Forest (RF) algorithm for model parameter regionalization. Integrated with distributed deterministic hydrological models of 20 small-sized watersheds in Henan province, the RF algorithm has been applied for defining the watersheds’ similarity and further transferring the parameters from sample watersheds to the objective watershed. Validated through leave-one-out approach, the RF model is able to effectively improve the simulation accuracy of flash floods in Henan province. The presented approach showed high-levelled applicability to be extended in other flash flood areas in China for providing effective reference for parameter regionalization.
Highlights
Flash floods in small-sized mountainous watersheds cause serious damage to life and property in China
The results show that the descriptors of similarity for small mountain watersheds in Henan Province are underlying surface conditions and climate
The parameter regionalization approach has been widely accepted as one of the main measures. With both measured and simulated data from 20 flash flood watersheds in Henan province of China, a new machine learning approach for parameter regionalization has been proposed with Random Forest (RF) algorithm
Summary
Flash floods in small-sized mountainous watersheds cause serious damage to life and property in China. From 1991 to 2015, the number of deaths due to flash flood disasters was around 27,000, with an average annual death rate nearly 1093 per year. The average annual economic loss caused by flash flood disasters exceeded RMB 40 billion [1]. Improving the accuracy of flood forecasting is the key point to solve the problem of flood disaster reduction. It is difficult to determine the parameters of the hydrological model for the ungauged watersheds. The study the parameter regionalization of flash flood modelling based on machine learning has scientific significance and practical value
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
