Fire is the dominant disturbance in boreal forests, but fine-scale variability in fire effects is not well constrained by sparse field measurements or moderate resolution satellite data. Multi-temporal airborne lidar data provides critical insight on pre-fire conditions and post-fire changes in vegetation structure, composition, and soils. In remote regions such as interior Alaska, lidar data can both act as a surrogate to field-measured structural data and bridge the scale gap between field and satellite resolutions. In this study, we analyzed transects of airborne lidar from Alaska’s Kenai Peninsula, a region that is ecologically representative of the larger boreal biome. Repeat lidar acquisitions from 2004, 2008, 2009, and 2014 bracket years with large fires (2005, 2007, 2009, and 2014). Lidar data were combined with Forest Inventory and Analysis plot data to address the following questions:
- How do burn severity, forest structure, and recovery vary within a burn, and what are the spatial scales of within-fire heterogeneity?
- How does pre-fire stand structure influence remotely sensed (e.g., differenced Normalized Burn Ratio) estimates of burn severity and the first decade of post-fire recovery?
- What is the fate of fire-killed trees (e.g., snags), and how do snags alter net carbon losses from fire?
Airborne lidar provide a unique perspective on fine-scale variability in fire effects. Among three fires from 2005, we note that within-burn heterogeneity with respect to pre-fire stand structure, burn severity, loss of biomass, and recovery trajectories frequently occur at spatial scales of less than a single Landsat pixel. In our study region, spectral indicators of burn severity often did not vary closely with the magnitude of change in forest structural metrics measured using lidar. Rather, high dNBR values may be more indicative of removal of the organic soil matter and the exposure of the underlying mineral layer. Severe crown fires generally did not lead to complete loss of above-ground biomass. Instead, the prevalence of standing dead trees represents a significant structural remnant on the landscape and, in a modeling context, an underappreciated delay in carbon release from boreal forest fires.