A comprehensive research framework using computational wildfire simulations, geospatial data science (remote sensing, GIS, network science), and decision-making for complex systems (wildland urban interface). The project examines how wildfire risk emerges, evolves, and can be mitigated - from global pyrogeographic shifts down to the neighborhood and individual parcel scale.
Wildfires represent one of the most dynamic and destructive natural hazards on Earth. Fire behavior is shaped by the complex nexus of climate, landscapes, and humans -- all of which are dynamic factors interacting simultaneously. This research program addresses wildfire risk from a multi-scale perspective: understanding how global fire regimes are evolving, how synchronous fire weather is enhancing the likelihood of extreme fires, how extreme fires emerge in unexpected geographies, how simulations can improve operational response, and how risk can be shared and mitigated at the community scale through shared governance. The projects presented here span from global pyrogeographic analysis using machine learning down to parcel-level responsibility networks for neighborhood fire risk mitigation.
Using unsupervised ML on 24 years of global fire data to cluster into pyromes and track how they transition and intensify over time.
Quantifying concurrent high-danger fire weaether conditions using synchronous fire weather days (SFW) and assessing the conditional probability of fire occurrence under extreme weather.
Benchmarking the record-breaking March 2025 South Korea wildfire against global megafire databases and analyzing the anomalous fire weather conditions.
Coupling Cell2Fire with WindNinja to produce spatially-explicit, downscaled wind fields and to simulate on the 2025 Los Angeles Palisades Fire for first 32-hour spread.
Automating fire potential polygon generation via hydrology-inspired modeling and building suppression strategy networks - together with the Catalan Fire Service.
Rethinking defensible spaces as overlapping buffers and developing spatial responsibility networks that quantify shared and owed risk between neighboring parcels.