A pipeline for representing buildings as fuels in wildland urban fire spread and risk modeling

Maria Faye Theodori , Maryam Zamanialaei, Dwi Purnomo, Yiren Qin, Chris Lautenberger, Michael Gollner

Wildfires pose an increasing risk to structures and communities located adjacent to or among vegetative landscapes. Yet most open landscape-scale fire modeling workflows still lack a reproducible way to represent buildings as combustible fuels rather than only as exposed assets or nonburnable developed land. This paper presents FireDX, a geospatial data engine that generates standardized, simulation-ready urban fuel inputs for wildland urban fire spread and risk analyses. FireDX makes four contributions: (1) it introduces a Building Fuel Model (BFM) framework that maps observable structure attributes to ignition and burning parameters; (2) it provides a reproducible pipeline that integrates heterogeneous public datasets into buildings-as-fuels inputs for landscape fire modeling; (3) it implements a tile-based, memory-conscious processing architecture suitable for large geospatial domains; and (4) it produces an interoperable rasterization layer that translates structure-level BFMs and urban geometry into hybrid fuel maps and companion rasters for downstream simulators. We demonstrate the pipeline by generating a California-wide urban fuels dataset that resolves 44.3 million structures into BFM classes and aligned 30 m raster layers. FireDX is not presented as a calibrated structure-loss or urban-spread model. Its contribution is an open data and coupling layer that enables heterogeneous built fuels to be represented explicitly in downstream fire spread, exposure, and mitigation analyses.