Terrestrial lidar scanning provides efficient measurements of fire-caused crown scorch in longleaf pine

Background: Fire is a dominant ecological process that influences the structure, function, and biodiversity of many ecosystems across the globe. Crown scorch is a primary predictor of tree mortality and is crucial for understanding the mechanisms of tree mortality. Yet, conventional scorch measurements use ocular estimates which are subjective and variable and typically only completed in small areas. Terrestrial lidar systems (TLS) can measure proxies for physiological processes such as leaf water and chlorophyll content and therefore may detect crown injury. Objective measurements of crown scorch that can be rapidly conducted over large areas can advance the mechanistic understanding of tree injury and improve predictions of fire effects. Results: We predicted crown scorch based on patterns of return intensity (range-corrected amplitude) from terrestrial lidar scanning in 253 trees with scorch severity ranging from 0 to 100%. We found that crown scorch consistently altered the empirical distribution of lidar return intensity, particularly in the range of −7.5 to −5 dB, shifting from bimodal to nearly unimodal. Predictions of crown scorch from the best model trained on manually segmented trees were highly correlated with ocular measurements (R² = 0.683, RMSE = 0.187). Predictions from automatically segmented trees were less accurate (R² = 0.460, RMSE = 0.227) than the higher-quality manually segmented dataset primarily due to segmentation errors among small trees (e.g., <10 cm). To demonstrate the utility of the methodology, we predicted scorch over a 15-ha study area and found the data collection rate was ca. 400 trees h⁻¹ of active time—20 times faster than the traditional ocular method. We created an R package CrownScorchTLS to facilitate accessibility of techniques for assessing lidar scorch predictions using the methods herein. Conclusions: Measuring crown scorch with TLS allows accurate and rapid measurements of detailed crown scorch for individual trees at relatively large scales (10s of ha) compared to what is practical with ocular measurements. Moreover, because TLS provides return intensity with high spatial resolution, within-tree measurements of crown scorch may be possible, allowing mapping injury to specific tissues. TLS methodologies and the accompanying R package allow estimates of crown scorch at operational scales, providing much-needed objectivity to improve our mechanistic understanding of the impact of fire on forest ecosystems worldwide.