A Novel Flood Classification Method Based on Machine Learning to Improve the Accuracy of Flood Simulation: A Case Study of Xunhe Watershed DOI Open Access

Xi Cai,

Xiaoxiang Zhang, Changjun Liu

и другие.

Water, Год журнала: 2025, Номер 17(4), С. 489 - 489

Опубликована: Фев. 9, 2025

Flood disasters pose one of the greatest threats to humanity. Effectively addressing this challenge requires improving accuracy flood simulation. Taking Xunhe watershed in Shandong Province as study area, Random Forest model was utilized classify historical events within based on rainfall conditions, such varying durations, intensities, and total precipitations. Multiple sets hydrological parameters were established conduct classification simulation, reducing error caused by using a single parameter set for entire watershed. The results indicate that can be applied simulation Compared unclassified simulations, method proposed leads an improvement Nash coefficient 0.06 0.14, reduction relative peak discharge 3% 11.24% volume 1.46% 9.44%. has certain applicability errors under different scenarios watershed, providing new insights control disaster efforts.

Язык: Английский

A Novel Flood Classification Method Based on Machine Learning to Improve the Accuracy of Flood Simulation: A Case Study of Xunhe Watershed DOI Open Access

Xi Cai,

Xiaoxiang Zhang, Changjun Liu

и другие.

Water, Год журнала: 2025, Номер 17(4), С. 489 - 489

Опубликована: Фев. 9, 2025

Flood disasters pose one of the greatest threats to humanity. Effectively addressing this challenge requires improving accuracy flood simulation. Taking Xunhe watershed in Shandong Province as study area, Random Forest model was utilized classify historical events within based on rainfall conditions, such varying durations, intensities, and total precipitations. Multiple sets hydrological parameters were established conduct classification simulation, reducing error caused by using a single parameter set for entire watershed. The results indicate that can be applied simulation Compared unclassified simulations, method proposed leads an improvement Nash coefficient 0.06 0.14, reduction relative peak discharge 3% 11.24% volume 1.46% 9.44%. has certain applicability errors under different scenarios watershed, providing new insights control disaster efforts.

Язык: Английский

Процитировано

0