Satellite-Based Forest Disturbance Dataset from Sentinel-1 SAR (2016–2021)

This dataset provides an independent, satellite-derived record of forest disturbances across a study region, generated from Copernicus Sentinel-1 Synthetic Aperture Radar (SAR) data spanning 2016–2021. A total of 17 Sentinel-1 tiles in the southeast of the USA were processed, rescaled to a 20 m × 20 m resolution, and mapped to the EQUI7 grid system, including terrain corrections to account for topographic variation. Forest disturbances were detected using a pixel-wise change detection method based on time series analysis. For each tile, yearly land-cover transitions were identified within a moving two-year observation window (July of the previous year through June of the following year). Detected changes were assigned to the central calendar year, with a ±0.5-year temporal uncertainty due to the buffer around the attribution window. Recurrence Quantification Analysis (RQA) was applied to quantify structural changes in forest cover. The method measured time-step similarity, and RQA-Trend values were calculated from the distance of recurring patterns relative to the main diagonal. Pixels with an RQA-Trend above –1.28 were classified as stable (no significant change), while values below the threshold indicated disturbance. This resulted in annual binary raster layers indicating whether structural forest changes occurred. Paper describing the process: F. Cremer, M. Urbazaev, J. Cortés, J. Truckenbrodt, C. Schmullius and C. Thiel, "Potential of Recurrence Metrics from Sentinel-1 Time Series for Deforestation Mapping," in IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 13, pp. 5233-5240, 2020, doi: 10.1109/JSTARS.2020.3019333. Data sources: Copernicus Sentinel-1 SAR data (2016–2021) License:Creative Commons Attribution (CC-BY 4.0)