Study of Fire Impacts on Siberian Larch Succession using UAV-based LiDAR and Multispectral Imaging: Yakutia, Russia

Wildfire is natural to the Siberian boreal forest ecosystem and shapes forest succession and vegetation structure. Due to climate warming, recent fire intensity and frequency increased and redefines the prevalent fire regime. Local field studies at plot scale and satellite-based regional-scale provides temporal and spatial information on forest structure, biodiversity and biomass but at inadequate gradient or in coarse resolution. Here, the study of vegetation using high-resolution imaging and LiDAR provided by unoccupied aerial vehicles (UAVs) raises the opportunity to complement remote sensing and field-based research.

Our study region is situated between the Verkhoyansk Mountains and the lowlands of the city Yakutsk (Republic of Sakha, Russia). This region has experienced increasing forest fire within the last decades (Talucci et al., 2022) with a severe 2021 fire season. For this study we acquired a forest succession data set of UAV-based LiDAR and multispectral images in August 2021. We employed a DJI M300 UAV which hosted a 6-band multispectral camera and a 2-return LiDAR sensor. LiDAR data (point density 500-1500 pts/m3) with GNSS post-processing was used to co-register multispectral data for consistency. Multispectral data were processed using photogrammetric workflows and converted to reflectance.

To describe the succession stages, we assessed forest composition and structural parameters, i.e. tree and canopy height distributions and crown area to detect change patterns. We modeled stages based on an apparent time-series from selected forest plots that represent five defined succession stages (Furyaev et al., 2001). We employed area-based analysis and single tree metrics to complement the inventory. Biomass measures were calibrated with terrestrial inventory, literature and subsequent laboratory analysis. Spectral properties of vegetation indicate vitality differences and vegetation response in burned areas, characterizing regime differences (i.e. crown vs. ground fire), and species occurrence. Our results highlight the feasibility of UAV-based measurements in inaccessible forests. Future campaigns should revisit the test sites to increase the validation sample size, collect field loggers and re-investigate the state of the plots.