Using citizen reports for urban flood prediction with machine learning approach and historical rainfall data (Case Study: Qom, Iran)

Articles in Press

Document Type : Research Paper

Authors

1 Kish international Campus, University of Tehran, Kish, Iran

2 Department of Environmental Engineering, Faculty of Environment, University of Tehran, Tehran, Iran

3 Department of Water Engineering, Faculty of Agricultural Technology, University College of Agriculture & Natural Resources, University of Tehran, Tehran, Iran

10.22059/jwim.2025.396294.1235

Abstract

This research aims to develop a scientific and practical method for urban flood prediction using citizen reports and satellite rainfall data. The city area is divided into a grid with 3,000-meter intervals, and the intersection points of the grid lines are selected as rainfall measurement stations. Days are classified as wet, dry, and normal based on citizen reports and observed rainfall, and machine learning models including Support Vector Machine, Logistic Regression, and Random Forest are trained on these data. The results show that the Random Forest model has the highest performance, with 98percent precision and an F1 score of 66 percent in correctly identifying flood events. Considering Qom’s semi-arid climate and the non-lethal damages caused by floods, emphasis was placed on the precision of models in correctly detecting flood cases to enhance confidence and prevent community anxiety due to false alarms. Stations 1 and 23 are identified as critical locations for rainfall equipment installation based on feature importance analysis. Subsequently, a critical rainfall threshold curve at a 90 percent probability level was plotted for flood prediction, and validated with real data. Rainfall above the threshold generally results in floods, while values below are considered safe. This approach is an effective tool for rapid warning systems and can aid urban disaster management to prevent financial and human losses.

Keywords

Main Subjects

Water and Irrigation Management

Articles in Press, Accepted Manuscript
Available Online from 15 August 2025

Files

Share

How to cite

Statistics