Fire Weather Data Analysis: Modeling Fire Potential Index and High Risk Fire Events. In Data Science & Engineering Master of Advanced Study (DSE MAS) Capstone Projects

Over the course of two quarters, our Data Science and Engineering MAS capstone group teamed up with the San Diego utility company, San Diego Gas and Electric, on improving their method for assessing fire risk potential. The Fire Potential Index (FPI) is a risk assessment mechanism and calculation leveraged in various localized forms across different areas by organizations and fire scientists for preventing and controlling wildfires. The San Diego Gas & Electric company leverages a parametric FPI calculation taking into account vegetation, live and dead fuel moisture, and weather elements for predicting risk of significant fire incident in case of an ignition. This risk information is then assessed and applied to determine the day’s required operating protocols, for preventative measures. Being an electric company, SDG&E have begun to proactively shut down power grids as a form of preventative measures on days of elevated FPI values. The task of this project is to apply data science principles to further optimize the San Diego fire risk calculation and 7 day forecast, specifically through development of a machine learning algorithm, flexible data design for incorporation of new features such as solar radiation from mesowest.utah.edu, and model validation procedure from GeoMAC Wildfire. Our goal is for SDG&E to deploy the final results of our project as a tool for communicating to fire fighters, public officials and San Diego residents on potential wildfire risks, and for mitigating wildfire risks with effective company measures.