Understanding defoliation of Pinus plantations in the Mediterranean mountains using tree segmentation and ALS time...

Forest decline is observed across several spatial and temporal scales. Remote sensing studies have traditionally focused on severe, stand-replacing disturbances such as widespread tree mortality, whereas more gradual declines driven by drought or pest outbreaks have received less attention, largely due to their high spatial and temporal variability and the difficulty of detecting subtle changes in forest canopy and structure. In this study, we explored climatic-driven pine defoliation using time-series Airborne Laser Scanning (ALS) data (2008, 2013, 2018) in Pinus sylvestris and P. nigra plantations in south Spain. We derived three key forest attributes: tree segmentation, height, and defoliation from ALS point clouds using non-parametric methods (e.g., K-nearest neighbors k-NN, Random Forest RF and neural networks NN) and aggregated these at both tree and pixel scales. Then Kernel Density Estimation (SKDE) was used to characterize defoliation patterns. Based on these analyses, we identified temporal defoliation trends and their relationships with environmental and forest drivers. We obtained high accuracy values for tree segmentation, height estimation, and defoliation assessment. Considering the proportion of defoliated categories at tree- and stand-scale assessments, non-defoliated trees and defoliated tree were significantly related to forest attributes SKDE and climate variables. These findings demonstrate that ALS-based approaches can capture gradual forest decline, providing insights into ecological processes and supporting adaptive forest management strategies to adapt forests over space and time.