Improved Road Flood Predictability and Disruption Response Through the Synergistic Integration of Geospatial Databases, Process-Based Modeling, and Machine Learning

Major flood events can have devastating impacts on communities, ecosystems, and infrastructure. Heavy rainfall in urban areas often overwhelms existing infrastructure, resulting in localized street or section flooding. Flooded roads hinder access to essential services and pose significant challenges for emergency management. Predicting these floods in near-real-time and with high resolution is difficult due to limited data and the computational cost of detailed models. The research team has already developed and tested a framework (Bhattarai et al., 2024). This project will test the modeling framework around the Jackson, Mississippi, downtown and surroundings. For instance, events like floodwater beneath the railroad bridge on Monument Street near Mill Street in Jackson (reported on Wednesday, January 24, 2024, and similar events). The project will compile information on flooded road and railway networks from local and regional news portals and X (formerly Twitter). Using location keywords (Jackson’, ’Jackson downtown’, ’Jackson MS’) and flood-related terms (’flood’, ’flooding’, ’road flood’, ’urban flood’, ’flash flood’, ’road closure’, ’rainfall’), the research team will identify flooding dates and affected road locations for the recent time and geolocate flooded locations using QGIS, that will serve as training-testing data for the machine learning model. Then the project will develop and test machine learning models (base learner models, such as random forest, support vector machines, and ensemble of these base learners). The research team will use datasets of covariates from other available hydrodynamic models, satellite rainfall estimates, traffic cameras (if available), flood-control infrastructure databases, and basin characteristics to predict flood inundation at street-level resolution. The research team believes these machine learning-based models offer significant improvements in computational efficiency while maintaining accuracy and consistency. In a nutshell, the research team will identify the most susceptible road and rail networks to critical urban facilities.

Language

• English

Project

• Status: Active
• Funding: $77,500.00
• Contract Numbers:

  69A3552348331

• Sponsor Organizations:

  Office of the Assistant Secretary for Research and Technology

  University Transportation Centers Program
  Department of Transportation
  Washington, DC  United States  20590
• Managing Organizations:

  Maritime Transportation Research and Education Center (MarTREC)

  University of Arkansas
  4190 Bell Engineering Center
  Fayetteville, AR  United States  72701
• Performing Organizations:

  Jackson State University, Jackson

  Department of Civil and Environmental Engineering
  Jackson, MS  United States  39217-0168
• Principal Investigators:

  Talchabhadel, Rocky

• Start Date: 20250801
• Expected Completion Date: 20260531
• Actual Completion Date: 0
• USDOT Program: University Transportation Centers

Subject/Index Terms

• TRT Terms: Emergency management; Flood protection; Floods; Geographic information systems; Highways; Machine learning; Predictive models; Railroads
• Geographic Terms: Jackson (Mississippi)
• Subject Areas: Data and Information Technology; Highways; Planning and Forecasting; Railroads; Security and Emergencies; Transportation (General);

Filing Info

• Accession Number: 01951358
• Record Type: Research project
• Source Agency: Maritime Transportation Research and Education Center (MarTREC)
• Contract Numbers: 69A3552348331
• Files: UTC, RIP
• Created Date: Apr 10 2025 2:38PM