Assessment of Post-Disturbance Changes in Spruce Mountain Forests Using Sentinel-2 Data and Machine Learning

Climate change is increasingly altering ecosystems, especially in mountainous areas. The study area located in Poland includes spruce mountain forests of the Tatra Mountains affected by numerous disturbances caused, among others, by the bark beetle infestation. The aim of the study is to analyses changes in spruce stands and identify the rate of stand decline based on Sentinel-2 optical satellite data for the years 2015-2023 combining them with information on climatic conditions based on the Standardised Precipitation Evapotranspiration Index. For this purpose, Sentinel-2 multi-temporal compositions for individual years were classified with the Random Forest machine learning algorithm achieving high classification accuracy values (overall accuracy: 0.91-0.99). The values of the proportion of dead trees in the stands varied from 13.8% in 2015 reaching a peak in 2019 of 20.9% to 1.8% in 2023. Which is combined with SPEI values which reaches a minimum (average -1.2) in 2019 and gradually increases which is associated with forest stand regeneration.