Keyword: Artificial Inteligence, Spatial Database, Spatial Modeling, Web Developement, Data Analysis
2022
Research
Forecasting Wildfire Risk in California explores how data analytics and spatial modeling can help predict and prevent wildfires across the state. Using logistic regression and environmental datasets, the project generates wildfire probability maps that reveal when and where fires are most likely to occur, offering new tools for proactive emergency planning and response.
As climate change intensifies droughts and high temperatures, wildfires in California have grown more frequent and destructive. In 2021 alone, over 8,600 fires burned 2.6 million acres. This project aims to help agencies and communities anticipate high-risk periods through data-driven forecasting, reducing damage and supporting long-term climate resilience.
The model combines geospatial datasets including elevation, land cover, precipitation, temperature, and historical wildfire records. Data were processed into two-mile grids and analyzed for spatial and seasonal patterns linked to fire occurrence. A logistic regression model was trained and optimized to prioritize early detection, achieving high accuracy after cross-validation. Twelve monthly models were also developed to capture shifting fire risks throughout the year, revealing peak activity between June and September.
Developed in 2022 as part of the Master of Urban Spatial Analytics (MUSA) program at the University of Pennsylvania, this project was completed by myself and Shengao Yi under the guidance of Professors Elizabeth Delmelle, Michael Fichman, and Matt Harris. All analysis was conducted in R, with results presented through a reproducible markdown report and interactive web visualizations.
The forecasting model achieved strong predictive performance, identifying clear seasonal wildfire patterns and producing detailed monthly risk maps. The resulting insights informed the creation of Favigator, a web application that visualizes fire probability and suggests patrol routes for emergency responders. Together, the model and application demonstrate how spatial data science can strengthen wildfire preparedness and support more resilient, data-informed decision-making in California.