This project will provide data, maps, and modeling capabilities to aid managers in using prescribed fire to improve forest conditions on Department of Defense (DoD) military lands. The research team aims to provide a state-of-the-art modeling tool to better predict smoke plumes and forecast emissions when planning fuels treatments. Building on the success of two prior Strategic Environmental Research and Development Program (SERDP) projects, the team will take a novel approach to map fuel structures, predict fuel combustion, and model smoke plumes and fire emissions. Employing a new technique using an object-oriented mapping approach, researchers intend to develop a model with seamless transitions across local and regional scales that will be relevant to military land use.

Multi-scale field and remotely sensed measurement datasets will be used as the primary data sources to develop the approach. The research team will aggregate data for fine-scale fuels, including consumption and estimates of energy release (FRP/FRE) associated with prescribed surface fires.  These datasets will be upscaled to the level of trees, forest stands, and management units. Using an object-oriented approach, the research team will map fuel bed components at fine (sub-meter) scales as image objects, and link them to existing libraries of fuel characteristics from which to predict consumption. Next they will develop a complementary library of spatial and temporally high-resolution measurements of Fire Radiative Power (FRP). The measurements will be collected at appropriate scales and linked to the fuel models and emission factors (EF). Estimates of Fire Radiative Energy (FRE) integrated from the near-complete FRP observations will be used to test the model and calibrate and validate the estimates of fuel consumption and emissions. These highly accurate, coupled-reference measurements of consumption, predicted emissions, and FRE attributed to objects at fine scales will be assigned to similar fine-scale image objects mapped from high-resolution imagery based on deep learning. Fuel objects will be aggregated to coarser-scale objects—trees, stands, management units, and entire DoD installations—based on a nested hierarchy of image segmentation that the team will develop.

This project will provide DoD land managers and others with improved tools for managing forests and vegetation using prescribed fire. The research will improve source characterization at the scale of large operational burns (>100 hectare [ha]). This proposal leverages knowledge gained in two other SERDP projects which were initiated in FY19. Building on the success of these projects, this proposal will further advance the state of fuels and fire science using physics-based approaches. The benefits to managers will require translation to operational scales that are much larger than the spatial domain of existing physics-based models. As articulated in this proposal, we will translate fine-scale measurements to operational scales using established empirical data interpolation, integration, and aggregation strategies. The results will provide managers with maps of fuels and consumption by combustion phase, which can be used to improve fuels, fire, and smoke management across large DoD military land bases.