Applying Time-Series Analysis on Multi-Sensor Imagery to Map Forest Change
Publication from Digital
The launch of the Sentinel satellites has significantly increased the temporal density of available high resolution imagery. Therefore, the inclusion of time-series analysis is one of the most challenging topics for remote sensing based forest monitoring. We designed two forest change detection workflows, one for high resolution optical data and one for Synthetic Aperture Radar (SAR). The workflows were tested at a tropical forest site in the Republic of Congo (Fig.1) to map forest degradation and forest disturbance during the last 15 years. The optical approach includes multi-sensor imagery from Landsat-4/5/7, SPOT-4 and RapidEye. The presented SAR approach uses Sentinel-1 GRD imagery. The results from the SAR approach are compared with the existing Landsat based Forest Alert layer (GLAD-FA) from University of Maryland, Global Land Analysis and Discovery laboratory.