Engineers at the University of Houston have developed a new artificial intelligence framework that converts decades of paper-based flood maps into digital datasets, providing a clearer view of how flood risk has changed over time. The research, published in the Journal of Hydrology: Regional Studies, focuses on transforming historical Flood Insurance Rate Maps into georeferenced, high-accuracy digital tools.
The AI-driven system extracts and aligns information from legacy flood maps, allowing researchers to analyze long-term floodplain evolution. As a result, stakeholders can better understand how flood exposure has shifted across different areas. The dataset enables researchers, planners, and policymakers to evaluate historical patterns and use that information to inform decisions on infrastructure, development, and disaster preparedness.
Francisco Haces-Garcia, a former doctoral student at the University of Houston and the lead author of the study, said the dataset provides a longitudinal view of flood evolution. He noted that the data can be used to examine how geophysical factors have contributed to flooding over time. In addition, when combined with other variables, such as population growth or critical infrastructure data, the framework may support strategies to reduce vulnerabilities and improve recovery outcomes following flood events.
The research team applied the framework to the Houston metropolitan area, one of the most flood-prone regions in the United States. Using data spanning from the 1970s through 2025, the team conducted three case studies to test how flood risk has changed across different locations. The selected areas included Meyerland along Brays Bayou, Fifth Ward, and Kashmere Gardens near Hunting Bayou, and Cypress near Bridgeland.
The findings show that flood risk is not consistent across the region. Instead, it varies significantly depending on local conditions, development patterns, and environmental factors.
In Meyerland, located along Brays Bayou, the study found that the 100-year floodplain has expanded over time. This expansion occurred despite major investments in flood control infrastructure. Researchers linked the increase to rapid urbanization and the growth of impervious surfaces, which can accelerate water runoff and contribute to flooding.
In Fifth Ward and Kashmere Gardens, the analysis also showed an expansion of flood risk. The changes in this area appear to be associated with a combination of factors, including modifications to drainage systems, updates in mapping methodologies, increased rainfall, and ongoing development.
By contrast, the Cypress area near Bridgeland showed a different pattern. In this location, the floodplain extent has decreased in some areas. The study suggests that newer regulations and improved drainage systems may have contributed to this reduction. However, the research also notes that flood risk in the region remains complex. Additional factors, such as land subsidence, continue to influence conditions.
Overall, the study demonstrates how digitizing historical flood data can provide a more detailed understanding of long-term risk trends. By unlocking previously inaccessible information from paper maps, the AI framework offers a new way to analyze flood exposure across time and space.
Get the latest insurance market updates and discover exclusive program opportunities at ProgramBusiness.com.