United States (1985–2024): Urban Wildfire Trends, Severity & Recurrence (Ongoing)

Summary

  • Question. Where and how are wildfires intersecting urban land changing over time in the U.S., and what factors drive likelihood, severity, and recurrence in cities and the WUI?

  • What we did. Combined the full MTBS wildfire time series (1985–2024) with a yearly urban extent (LCMS land cover + ≥ 150 people/km²) to isolate urban-intersecting fires. Analyzed national trends, burned-area growth, per-pixel burn frequency (30 m), and severity using the MTBS burn-intensity raster. Added Sentinel-2 detail for 2017–2024 large events.

  • Outcome. Counts of urban fires are flat, but total urban area burned is rising significantly, indicating larger/more intense urban fire footprints. Recurrence is rare but not negligible: 1× 96.3% (262,783 pixels); 2× 3.4% (9,317); 3× 0.23% (617); 4× 0.03% (88); 5× 0.02% (54). Results will guide risk planning, building codes, and vegetation management in urbanized landscapes.

Research Challenge

Planners need more than statewide wildfire metrics—they need to know what’s happening inside and at the edge of urban fabric. The challenge is to:

  1. Dynamically define “urban” each year so we don’t over/under-count;

  2. Separate changes in fire counts from changes in area burned and severity; and

  3. Quantify recurrence and drivers at scales that matter to on-the-ground decisions.

Approach

Data

  • MTBS (fire perimeters & burn intensity/severity rasters), 1985–2024.

  • LCMS annual land-cover classifications to delineate developed/urban.

  • Population density (≥ 150 people/km²) to refine urban extent year-by-year.

  • Landsat surface reflectance stacks (30 m) for drivers at the MTBS scale.

  • Sentinel-2 (10 m) for 2017–2024 fine-scale subsets of large urban fires.

  • Optional covariates where available: slope/elevation, canopy/impervious, distance to structures/roads, recent burn history, drought/weather indices.

Methods

  • Built an annual urban mask (LCMS + population density threshold) and intersected it with MTBS perimeters to extract urban-intersecting fire pixels.

  • Computed national time-series of (a) number of urban fires, (b) total urban area burned, and (c) burn severity distributions.

  • Calculated burn recurrence per 30-m pixel (1×…5×) and mapped hotspots.

  • Modeled likelihood (urban pixel burned vs. not), severity, and stability/recurrence against candidate drivers (land cover, canopy/impervious, proximity variables, antecedent vegetation indices, drought) at Landsat scale, with Sentinel-2 repeats for 2017–2024 large events.

  • Performed trend tests and cross-scale consistency checks (Landsat ↔ Sentinel).

Tools

  • Google Earth Engine (masking, time series, zonal stats), ArcGIS Pro/ArcPy (spatial ops, intersections), Python (modeling, QA/QC, figure generation).

Map showing multiple urban wildfire burns along the coast from 1990 to 2024, with a legend indicating the number of repeat burns in different colors: gray for 1 burn, yellow for 2, green for 3, blue for 4, and red for 5 burns.

Outputs

  • National workbook: trends in counts, area burned, severity mix; maps of recurrence (1×–5×) and high-risk corridors.

  • City/WUI briefs: metro-level summaries with priority zones and recommended mitigations (vegetation management, setbacks, material/defensible-space flags).

  • Method notes: yearly urban mask recipe, assumptions, and sensitivity to thresholds.

  • Data & code: GeoTIFF/GeoPackage exports; reproducible GEE/Python scripts.

  • Status: Ongoing; Sentinel-2 refinement for 2017–2024 large fires in progress.

Impact / How it’s used

  • Risk planning & zoning: focus mitigations where burned area is growing despite flat counts; use recurrence maps to prioritize inspections, fuel breaks, and evacuations planning.

  • Urban forestry & landscaping: align canopy placement, species choice, and maintenance with severity/recurrence patterns (e.g., setbacks, breaks).

  • Building codes & outreach: translate severity and recurrence signals into defensible space and materials guidance for at-risk neighborhoods.