Post-fire regeneration dynamics of heterogeneous Mediterranean ecosystems using Landsat and ALS data

Wildfires are major ecological disturbances that profoundly alter forest structure and function. However, the long-term structural recovery of Mediterranean ecosystems remains insufficiently understood, particularly across heterogeneous vegetation types. This study addresses this gap by developing and validating an integrated remote sensing methodology that combines multi-temporal Landsat imagery and low-density Airborne Laser Scanning (ALS) data to assess post-fire vegetation recovery over a 20-30-year period in Mediterranean forests dominated by Quercus ilex, Pinus halepensis, and shrubland. Random Forest models were used to relate ALS-derived structural attributes-vegetation height, canopy cover, and height diversity-to Landsat vegetation indexes and topographic predictors and applied Bayesian Additive Regression Trees (BART) to identify key drivers of regeneration. The ALS-Landsat models showed high predictive performance (R² = 0.68-0.94), with height diversity index (LHDI) achieving the highest accuracy across vegetation types. BART models explained up to 72 % of the variability in post fire structural metrics, identifying temperature, plant community, and succession stage as primary determinants of recovery. Results reveal distinct regeneration trajectories: shrublands exhibited rapid recovery in cover and height, while Pinus forests showed the slowest structural rebound, with many areas failing to regain pre-fire height after 30 years. These findings underscore the value of integrating ALS and Landsat data for monitoring postfire recovery of forest structural variables offering practical insights to guide post-fire restoration efforts.