4.5 Article

GIS-based hybrid machine learning for flood susceptibility prediction in the Nhat Le-Kien Giang watershed, Vietnam

期刊

EARTH SCIENCE INFORMATICS
卷 15, 期 4, 页码 2369-2386

出版社

SPRINGER HEIDELBERG
DOI: 10.1007/s12145-022-00825-4

关键词

Flood susceptibility; Machine learning; Nhat Le-Kien Giang watershed; Vietnam

资金

  1. [QG.22.20]

向作者/读者索取更多资源

This study develops flood susceptibility maps for the Nhat Le-Kien Giang watershed using machine learning algorithms. The results show that the models have high accuracy and can be applied to future development strategies.
Floods is a natural hazard that occurs over a short time with a high transmission speed. Flood risk management in many countries employs flood susceptibility modeling to mitigate against the damage to the economy and loss of life caused by future floods. The objective of this study is the development of a new approach based on the machine learning algorithms namely Multilayer Perceptron (MLP), Archimedes Optimization Algorithm (AOA), Whale Optimization Algorithm (WOA), Water Cycle Algorithm (WCA), Decision Tree (DT), and Adaboost (ADB) to build flood susceptibility maps for the Nhat Le-Kien Giang watershed of Quang Binh province. In total, 1964 flood locations and 14 conditioning factors were split with a ratio of 70:30 for the training model and the validation model respectively. Various statistical indices - namely root-mean-square error (RMSE), mean absolute error (MAE), coefficient of determination (R-2), and area under the ROC curve (AUC-ROC) - were used to evaluate the models. The results show that all the models performed well in building the flood susceptibility map, with an AUC value of more 0.9; the models MLP-WCA (AUC = 0.99) and MLP-AOA (AUC = 0.99) were most successful, followed by MLP-WOA (AUC = 0.98), MLP-ACO (0.98), ADB (0.98), and DT (0.97). The analysis of the flood susceptibility maps shows that the high and very high susceptibility level corresponds to no more than 39% of the study area. The results also show that altitude, slope, rainfall, and land use are most influential on the probability of flood occurrence in the study area. It can be concluded that the machine learning models proposed in this study provide results with high accuracy. The flood susceptibility maps that have been produced will be significant in determining future development strategies, especially in selecting the locations of new urban areas; and the findings of this study may be applied not only to this particular watershed in Vietnam, but also in developing countries around the world.

作者

我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。

评论

主要评分

4.5
评分不足

次要评分

新颖性
-
重要性
-
科学严谨性
-
评价这篇论文

推荐

暂无数据
暂无数据