Detecting Bark Beetle-Induced Changes in Coniferous Alpine Forests Using Sentinel-2 Time Series and In-Situ Felling Data

Abstract

Abstract. Mapping forest areas affected by bark beetle infestation using remote sensing imagery is crucial for effective hazard management and risk assessment. This study evaluates the potential of Sentinel-2 satellite image time series (SITS) in combination with in-situ felling data to detect bark beetle infestation in coniferous forests in Pokljuka, Slovenia. The analysis uses the CuSum method, all Sentinel-2 spectral bands and key spectral indices such as NDVI and NBSI to identify changes and areas of forest loss in the period 2017–2021. The resulting geospatial dataset, which integrates these remote sensing results with field data, serves as a basis for further analyses using advanced machine and deep learning methods and various remote sensing data such as hyperspectral datasets. In addition, we found that the most useful bands for detecting the loss of alpine coniferous forests are SWIR (B11, B12), Red (B04) and Red-Edge (B05) as well as the two spectral indices used, NDVI and NBSI.