Sentinel-2 Satellite Image Processing using Machine Learning Algorithms of the Manombo Nature Reserve

Tsaramanana, Valerien Eugene

JGISE-Journal of Geospatial Information Science and Engineering
Vol 7, No 2 (2024): December

This paper is based on the fields of satellite image processing and analysis using Sentinel-2 satellite images with machine learning algorithms under Google Earth Engine for the study of land cover evolution in the Manombo Madagascar, nature reserve. The objectives of the study are to identify the elements that occupy the land in the reserve. During our experiments, we compared the best machine learning algorithm using CART, Random Forest, Naive Bayes, SVM to determine the best machine learning algorithm for our Sentinel-2 data. So, we have proposed a methodology to do the treatment and in the end we have treatment results. From our treatments, we can conclude that the use of Random Forest classifier gave the most accuracy on the correct classification.