Gilberto Camara
Sessions
This work integrates optical and radar data cubes to detect forest disturbances in tropical regions.
Our method identifies initial degradation and selective logging, often precursors to deforestation, demonstrating its utility in early-warning systems. These results emphasizes the crucial role of integrating optical and radar data to improve the precision and dependability of monitoring systems, essential for sustainable forest management. These findings highlight the value of integrating multi-source data cubes to enhance precision in monitoring forest disturbances, thereby supporting more responsive and reliable environmental management.
The SITS package (Satellite Image Time Series) is designed for the analysis and classification of satellite image time series using machine learning. It provides a comprehensive framework for managing, modelling, and classifying time series data derived from remote sensing imagery. In version 1.5.3, SITS supports both R and Python APIs and has included support for CDSE and OGH cloud providers. SITS supports large-scale operational analysis on data cubes, and has state-of-the-art functions for deep learning, post-processing, uncertainty estimation, and texture measures. It allows the merging of Sentinel-1 and Sentinel-2 data, as well as Landsat data with Sentinel-2, and enables the inclusion of DEM and climate data as additional bands.
This study explores transfer learning in the Brazilian Amazon over the period from 2015 to 2022.