Fire severity and tree/forest structures derived from pre- and post-fire LiDAR data in a large forest fire in SE Spain

Heterogeneous patterns of trees and openings in forests create a patchy fuel matrix that may burn with different fire severity, which can affect post-fire regeneration. Understanding how forest structures determines fire severity and whether fire severity metrics entails variability in such structures within a given category is important to improve our ability to assess post-fire forest development. Here, we assessed how fire severity changes the vertical and horizontal structure of trees and forest stands, and what are the main post-fire tree/forest structures associated with different fire severities. The study site was a large and mixed severity fire (3,217 ha) occurred in southeast Spain (Yeste, Albacete) in the summer of 2017. Pre-fire forest structures were estimated from LiDAR data (sensor ALS 50 – II) collected in 2016, with a theoretical laser pulse density between 0.5-2 returns m². Post-fire forest structures were estimated from LidarPod data (sensor Velodyne HDL-32e), with a laser pulse density of 312 returns m², collected in April 2018 (8 months after fire) at 3 burned sites plus one unburned control. Fire severity was estimated from the post-fire NBR (Normalized Burn Ratio) and other similar indices derived from Sentinel 2. We found that up to 5 post-fire tree classes and up to 4 forest stand structures were separable, each characterized by different heights, gap fractions, crown properties and fire intensities. There was not a one-to-one relationship between tree/forest structures and standard fire severity levels. The main changes in height, crown and other tree properties were highly correlated with post-fire tree structures and fire severity indices. Accordingly, the trees more severely burned were those with higher losses in height and crown area. Our results indicate that satellite fire severity metrics were highly related to biomass consumption; nonetheless, standard fire severity classifications included several tree/forest structures that without Lidar data it would be impossible to differentiate.