The ability to characterize fine-scale fuels in southeastern U.S. longleaf pine (Pinus palustris Mill.) forests across a range of fuel types and moistures is critical to understanding fire effects that affect species diversity and pine regeneration. To begin addressing these questions requires new and spatially robust accounting of fuels that drive fire intensity over the landscape. We utilize a combination of terrestrial laser scanning (TLS) and a new novel approach for simulating fuelbeds using 3D mesh objects to represent fuels at multiple scales, assessing fuel load and type present in frequently burned sites in the pine sandhills of Eglin Air Force Base, NW Florida, USA. We utilize voxel analysis of both TLS and fuelbed simulations compared to dry weight biomass collected pre-burn in 2012 for nineteen plots (0.5m2). Biomass predictions are estimated from occupied voxel volume using Leave-one-out cross-validated linear models for the TLS and fuelbed simulations in relationship to observed total dry weight biomass. Weibull distribution analysis is used to compare and contrast shape and slope parameters between the TLS-based and simulated fuelbeds to determine if there is under sampling of the fuelbeds using TLS data due to occlusion of the laser pulse due to interception from near distance objects.
Results/Conclusions
We found that modeled surface fuelbed mass from the TLS-based voxel volumes predicted biomass well (R2 = 0.84, p<0.001, RMSE = 20.3%). Characterization of litter fuels, the primary carrier of fire and most difficult to estimate using remote sensing, demonstrated a strong relationship to simulated fuelbeds (R2 = 0.74, p<0.001, RMSE = 22.3%). Mass loss estimates from pre- and post-burn TLS data suggest that more variable fuel consumption occurred in grass and herbaceous dominated sites compared to sites dominated by longleaf pine needle litter and saw palmetto (Serenoa repens), where consumption was near complete. From the Weibull analysis, 3D simulated fuelbeds had a more consistent representation of fuel volume and mass, particularly in dense vegetation, compared to TLS measurements. Although TLS data has been linked to spatial fire intensity measurements using infrared thermography, the next step is relating the simulated fuelbeds to fire and fire effects. Simulating calibrated fuelbeds, and even overstory structure, across large areas provides a consistent representation of landscape fuel characteristics. These can be used to scale up fine-scale fire effects (and associated plant community models) that are linked to overstory derived fuels, all of which are the overarching premise of this five year project.