Using drones to derive fuel load estimates in semi-arid rangelands

Fires are a key process in landscapes globally. However, the frequency, size, and intensity of wildfires are shifting. In the Great Plains of North America, wildfires are larger, more frequent, and burn at higher severities. These changes have resulted in megafire outbreaks, where multiple large-scale (>10,000-ha) fires occur simultaneously and destroy infrastructure and livestock. Understanding the fuel loads that drive these megafires in rangelands at a landscape scale is crucial for mitigation and planning. The objective of this study was to estimate herbaceous fuel loads in a rangeland system using high-resolution drone imagery and field sampling and to determine how vegetation height and status (invasive/native grasses and forbs) affect fuel load. We used the 428,352-ha Smokehouse Creek Fire that burned across the south-central United States, as a case study. We collected ~800 ha of RGB imagery with a Trinity Pro fixed-wing drone and collected 225 fuel load plots from four different sites within the burn scar. To calculate fine fuel load (kg/ha) from the ground, 0.5 by 0.5 m herbaceous vegetation plots were clipped, dried, and weighed. Fine fuel load (cm³) estimates from the drone were calculated by generating a digital elevation model (DEM), taking the mean DEM value from each plot, and multiplying it by pixel size. We ran a two-way ANOVA to analyze the impacts on vegetation height, status, and vegetation height-status on fuel loads (cm³ or kg/ha). In both methods, vegetation height had a significant impact on fuel load (drone: p = 2 x 10^-16; ground: p = 4.85 x 10^-9). Status and vegetation height-status did not significantly affect fuel load in either method. This indicates that managers looking to reduce fine fuel load should focus on reducing overall vegetation height, as opposed to exclusively targeting tall invasive grass/forb patches.