Flooding is the natural hazard that imposes the most profound economic and social impacts within the United States, and these impacts are progressively intensifying over time. Major freshwater flooding occurrences from 2004 to 2014 incurred an average of $9 billion in direct damage and claimed 71 lives per year. In the United States, FEMA offers nationally recognized flood hazard maps. However, these maps are outdated and fail to incorporate the increasing intensity of precipitation. This limitation leaves a lot of homeowners at risk of flooding. This study uses a historical extreme precipitation event in the study area as a proxy for projected precipitation to model flood susceptibility. Seven flood drivers including DEM, slope, distance to streams, soil type, impervious surface area, land use/land cover and precipitation to create a flood model for Greater Boston by employing GIS and Machine Learning techniques. The Random Forest and Light Gradient Boosting Machine (LightGBM) algorithms were used to predict flood susceptibility. Both models achieved the same accuracy and precision with LightGBM slightly performing better at recall than the Random Forest model. The trained LightGBM was used to generate the final flood susceptibility map for the study area. The spatial pattern of flood map indicated that very high to high flood risk zones exist outside of FEMA’s Special Flood Hazard Areas, leaving a lot of people at risk of flooding. The findings also revealed the potential economic losses as well as population at risk in these flood-prone zones not mapped by FEMA. The result is expected to inform more accurate flood mapping and improve risk mitigation efforts.

Author

• Owusu-Ansah, Richard

Language

• en

Keyword

• FEMA Flood maps
• Greater Boston
• Flood Susceptibility
• Geospatila analysis
• Climate Change
• Machine Learning

Date created

• 2025

Type of Work

• Master's Thesis

License

• All rights reserved

Rights statement

• In Copyright

GW Unit

• Geography

Degree

• M.S.

Advisor

• Rain, David

Committee Member(s)

• Mann, Michael

Persistent URL

•  https://scholarspace.library.gwu.edu/etd/8910jv65r