Monitoring thirty-five years of italian forest disturbance using Landsat time series

Forest ecosystems have a crucial role for biodiversity conservation, providing a large set of ecosystem services. Understanding and assessing forest disturbance regimes on a large spatial and temporal scale is a prerequisite setting up sustainable forest management solutions. In this context, Remote Sensing is an efficient tool frequently used in land-use change detection. The present work is aimed at spatially estimating forest disturbing events occurred in Italy in the period 1985-2019. Using Landsat time series and the 3I3D forest disturbance detection algorithm, we analyzed “extreme” forest disturbance patterns and their evolution in the last 35 years. We found a total of 472 events, with the highest incidence (96) in the period 1990 - 1994. The accuracy of the 3I3D algorithm was estimated using a photo-interpreted dataset of nine random-sampled squared cells of 225 km2 each, distributed in the Italian region. Omission error for the 3I3D map ranged from a minimum of 37.43% to a maximum of 64.62% (mean value of 47.07%) while the commission error between 36.80% and 83.92%, with an average of 49.60%. Results suggest that occurrence of severe disturbance events do not seem to increase over time in the study period.