Oxygen isotope values of charred tree bark as an indicator of forest fire severity

The objective of this study was to determine if oxygen isotope values of charred tree bark could be used to reconstruct fire severity. The study was completed north of River Valley, Ontario, Canada, where a wildfire burned approximately 2500 hectares of white pine (Pinus strobus L.) forest in 2018. We established a network of field plots, collected charred bark samples from standing white pine stems, and estimated burn severity based on a standard field assessment protocol known as the Composite Burn Index (CBI). We also analyzed pre- and post-fire Sentinel-2 imagery of the burn area to compute various Normalized Burn Ratio (NBR)-based change detection algorithms, which are known to produce reliable predictions of CBI. We developed simple linear regression models to predict CBI using either the δ¹⁸O values of charred bark or versions of the NBR. Models developed from the δ18O values of charred bark revealed a significant negative relationship between CBI and plot-level δ¹⁸O, with the strongest relationship being with maximum δ¹⁸O (r² = 0.179, RMSE = 0.565). There were significant positive relationships between all NBR indices and CBI, with better fit statistics than the δ¹⁸O models. The results demonstrate that δ¹⁸O can be used as a predictor of fire severity; however, the scale of measurement of fire severity is finer (tree-level) than the plot-level CBI and NBR indices. The advantage of using the δ¹⁸O method is that it can be used to reconstruct fire severity when satellite or field data are unavailable.