Assessing pre- and post-fire biomass variations in boreal peatlands and uplands using multitemporal and multispectral lidar data

Abstract

This study used pre- and post-fire lidar data to estimate aboveground biomass (AGB) before and AGB losses within a boreal upland/peatland environment after the 2016 Horse River Wildfire, Fort McMurray, AB, Canada. The objectives were to compare patterns of AGB in forests, bogs, and fens and assess how lidar intensity-based active normalized burn ratio (aNBR) and Landsat change in NBR (dNBR) compared with structural losses of biomass from bogs, fens, and forests. AGB was determined within peatlands and forests using plot measurements. Based on linear regression models between point-cloud metrics and AGB, the best predictor of AGB was interquartile range (IQR) in forests (R2 = 0.91; p < 0.05) and quadratic average (qav) in peatlands (R2 = 0.84; p < 0.05). Based on these models, the median AGB in pre-fire forests (3.4 kg m-2) exceeded that in bogs (0.6 kg m-2) and fens (0.48 kg m-2). Subtracting post-fire from pre-fire AGB represented AGB loss following the fire. Forests still retained the highest AGB (0.5 kg m-2) after the fire in contrast to other ecosystems (p < 0.05). Post-fire, fens retained slightly higher AGB (median = 0.2 kg m-2) than bogs (0.17 kg m-2). As expected, more AGB was lost in forests (72% relative to pre-fire AGB; p < 0.05 relative to peatlands) in comparison with bogs (66%) and fens (64%), although biomass loss did not differ significantly between the two peatland ecosystems (p > 0.05). I also found a significant relationship between Landsat dNBR and aNBR (R2 = 0.48; p < 0.01). Overall, the findings demonstrate the utility of lidar with plot measurements in estimating tree aboveground biomass. The study contributes towards a greater understanding of the impacts of fire for use in carbon accounting in sub-humid climate-impacted ecosystems.