Land Cover detection through Remote Sensing
The goal of this work was to determine the current land cover of a concession in Ethiopia.
The sustainable management of forest concessions requires detailed information about the current land cover and the distribution of the land cover classes like a closed forest, bamboo, grassland, and bare soil. Knowing the properties and spatial distribution will allow precise planning of management measures.
For this land cover classification, we have selected Sentinel 2 satellite imagery. Its great advantage is a global weekly revisiting. This enables us to potentially create cloud-free composites on a monthly base even in regions with a high cloud occurrence. The free accessible data enables a cost-efficient generation of time series, which are a valuable source for a better understanding of the land cover during wet and dry seasons.
Carrying out multi-band analysis combined with ground truth data collected with a mobile app, we have been able to train the classification to distinguish various ground cover types.
Machine learning algorithms help us to get more accurate and scalable results. – Stefan Haas, CEO and Co-Founder, OpenForests
The goal of the explorer.land platform is to support projects with a comprehensive presentation tool that improves their visibility for collaborators, investors, donors, and product buyers by creating transparency and conveying the human dimension inherent in each project.
The result of the classification process: distribution of three different classes of land cover.
