Suchergebnisse

In recent years, methodologies for deforestation detection that use satellite data have been developed, primarily using optical data, which cannot detect deforestation in the presence of clouds. In this paper, we discuss a methodology developed to detect deforestation using Sentinel-1 data and that aims to complement typical early warning system based on optical satellite images such as one the Peruvian Government employs. The methodology was applied in three pilot areas in the tropical humid forest of Peru. Sentinel-1 data were acquired in Interferometric Wide Swath (IW) mode and VH polarization. We use a Gamma-Map filter to reduce the speckle noise, and the average of 3 chrono-sequentially continuous images to reduce the multi-temporal variation of the forest backscattering. This produced 6 time series for each pilot area. For the detection of deforestation, we used an algorithm based on the difference and ratio between the images before and after deforestation. The accuracy assessment revealed a user's accuracy greater than 95%. We also made a multitemporal comparison between our results and the early warning tropical forest loss alerts that use only Landsat data, which showed that until the end of the study period 33.26% of the deforestation we detected was not detected by the early warning alerts that use Landsat data.