Cloud-Based Remote Sensing for Conservation

This project aims to develop web-based landcover classification tools for Virunga National Park in Democratic Republic of the Congo, leveraging the rich information provided by Sentinel-2 multi-spectral imagery. The tool will enable researchers, park managers, and other stakeholders to analyze land cover changes, identify potential threats, and develop targeted conservation strategies.

However, working with multi-spectral imagery in tropical regions like Central Africa poses significant challenges due to persistent cloud cover. Hence, developing effective cloud detection systems is a prerequisite for obtaining reliable analysis-ready imagery. These detection systems must be able to distinguish between clouds and other similar features like bare soil and bright urban features, while also accounting for the spatial and temporal variability of cloud cover in the region.

The tools that were developed integrate cloud detection algorithms and image processing techniques to deliver accurate, high-quality imagery. Additionally, the tool employs machine learning and deep learning techniques to perform automatic land cover classification and provide users with an intuitive map-driven interface.

This web-based remote landcover classification tool provides park managers and researchers in Virunga, as well as other Congolese national parks, with a powerful platform for analyzing land cover changes, helping to support conservation efforts and promote sustainable land use practices.