Usages of AI in Predicting Accidental Risks of Soil and Water Management in Urban Areas
1 Department of Soil, Water & Environment, University of Dhaka, Dhaka, Bangladesh.
2 Department of Soil Science, University of Chittagong, Chattogram, Bangladesh.
3 General Banking Division, Islami Bank Bangladesh PLC, Gazipur, Bangladesh.
4 Independent Researcher, Dhaka, Bangladesh.
5 Department of Mechanical Engineering, Bangladesh University of Engineering and Technology, Dhaka, Bangladesh.
Research Article
International Journal of Science and Technology Research Archive, 2025, 09(02), 036-043.
Article DOI: 10.53771/ijstra.2025.9.2.0065
Publication history:
Received on 12 November 2025; revised on 24 December 2025; accepted on 27 December 2025
Abstract:
The main goal of this research is to develop a reliable and effective AI-based framework for predicting the risk of soil and water hazards in urban areas amid rapidly accelerating urbanization and climate change. A GIS-based risk prediction model has been developed by integrating multiple data sources, including remote sensing imagery, IoT sensor readings, and historical accident records. Comparative analysis demonstrated that the Random Forest model achieved the highest performance, with an accuracy of 0.87. Using this model, the northern part of the city (Zone C) was identified as the primary hotspot, with the highest risk score (0.83). Importantly, feature importance analysis confirmed that soil moisture (31%) and poor drainage (27%) were the main determinants of increased risk. These results lay the foundation for an early warning system and provide a proactive decision-support tool for policymakers. Ultimately, this model makes a significant contribution to increasing urban resilience by prioritizing resource allocation and infrastructure interventions.
Keywords:
Urban Risk Forecasting; AI in Water Management; Random Forest Classification; Soil Moisture Sensing; Drainage Deficit Modeling; Decision Support Systems
Full text article in PDF:
Copyright information:
Copyright © 2025 Author(s) retain the copyright of this article. This article is published under the terms of the Creative Commons Attribution Liscense 4.0