Spatiotemporal characterization of forest disturbance in the Hunsrück-Hochwald National Park

The characterization of forest disturbance is a critical challenge for remote sensing forest monitoring, which has made promising progress in the last decade, helped by the ever-growing volume of freely available data. There are many viable approaches for mapping forest disturbance causality using spectral and temporal metrics derived from annual Landsat time series nevertheless, in most cases, those implementations depend on visual interpretation of the spectral signal or very high-resolution imagery with fewer cases using field data for training or validating the attribution models. To tackle the missing ground truth problem, we use the ICP-Forest Visual assessment of crown condition and damaging agents survey to address the lack of historical forest disturbance field data. The ICP-Forest Network collects forest health information annually in approximately 6,200 plots distributed across Europe since 1985. The crown condition survey provides the percentage of defoliation of individual trees and additionally the cause of the defoliation. The main goal of this project is to determine the suitability of the ICP Forest data to be used as training in a machine learning framework (ML)