Operative Automatic Detection of Tropical Rainforest Disturbances in X-Band SAR Time-Series Based on Convolutional Neural Network

The tropical rainforest, with its biodiversity, is a priceless natural heritage on which the entire existence of our planet and its climate balance depend. Unfortunately, it is threatened by human activities such as deforestation. The detection of forest cut areas was mainly studied on optical imagery.Motivated by these scenarios, this work shows the development, training and validation of an artificial intelligence model based on convolutional neural network (DRIAS) that puts together the power of deep learning with the availability of high-resolution X-Band Synthetic Aperture Radar (SAR) time-series data provided by the COSMO-SkyMed first-and second-generation constellation to better support tropical rainforest safeguard. Main outcomes of its integration into operational information service provisioning chain are detailed as well.