Estimation of Leaf Area Index from Sentinel-2 Satellite Imagery: a case study of silver fir-beech forests from the Slovenian Karst region

Leaf Area Index (LAI) is a critical parameter for understanding the exchange of gases and water between the atmosphere, soil, and vegetation. Advances in remote sensing have facilitated automated LAI estimation; however, applying these techniques to forested karst landscapes presents challenges due to their highly heterogeneous forest cover, specific karst topographic features (dolines, interdoline areas) and dynamic aquifer systems. This study investigates the potential of remote sensing for monitoring LAI seasonality in a Slovenian karst region dominated by silver fir-beech forests affected by ice storms and bark beetle infestations. Sentinel-2 imagery was utilized to estimate LAI using six vegetation indices (VIs) and a neural network-based approach implemented in the Sentinel Application Platform (SNAP). LAI field data from eight sites and six timeframes in 2021, using the LAI2200 Plant Canopy Analyzer, provided in situ validation. Results revealed that Sentinel-2 imagery retrieval accuracy, with R² values between 0.80 and 0.85, varied depending on the vegetation index applied. The Random Forrest Regression shows that Enhanced Vegetation Index (EVI) has the highest performance, achieving a Mean Absolute Error (MAE) of 0.69 and a Mean Squared Error (MSE) of 0.91. These findings highlight the efficacy of remote sensing for LAI estimation in complex karst environments and underscore its potential for forest health assessment and UAV-based vegetation monitoring in similarly challenging terrains.