Flood risk analysis using random forest machine learning method (Case study: Mashhad city)

Document Type : Research Article

Authors

1 Environment

2 Department of Envi. Sci., Gorgan University of Agri. Sci. & Natural Resources, Gorgan, Iran

3 Dept. of Environmental Planning & Design - Gorgan Univ. of Agricultural Sciences & Natural Resources (GUASNR) Gorgan, Iran

4 Associate Professor, Department of Geography, University of Rennes, Rennes, France, CNRS.

10.22059/ije.2022.346677.1667

Abstract

Flood risk analysis using random forest machine learning method

(Case study: Mashhad city)







Abstract



Flood is one of the most common natural disasters and causes a lot of financial and human losses. Although the amount of rainfall is low in many regions of Iran, in many regions, the highest amount of annual rainfall occurs in only one day or a short period of time, which leads to floods. Due to the geological structure and the destruction of the ecosystem, the flowing water during the flood can be very polluted and often carries a lot of mud, which adds to the damage of the flood. To reduce potential flood damage, planners and decision makers must be aware of when and where floods occur. This requires the use of new flood forecasting methods and prevention of its damages. In this study, the Random Forest (RF) machine learning method was used to predict the location of floods in Mashhad City and its performance was investigated. Also, the effect of each of the factors of Elevation, slope, aspect, TWI, SPI, distance from waterways, geology, land use, density of waterways, and rainfall in this forecast is examined. Validation results using training data for RF models show the area under the curve (AUC) (95%). In general, the results showed that the RF model has a high accuracy in determining flood sensitive areas in the Mashhad basin.



Keywords: Food risk assessment, random forest, machine learning, Mashhad city.

Keywords

Main Subjects



Articles in Press, Accepted Manuscript
Available Online from 19 December 2022
  • Receive Date: 02 August 2022
  • Revise Date: 27 October 2022
  • Accept Date: 19 December 2022
  • First Publish Date: 19 December 2022