Remote Sensing of Grasslands: Performance Comparison of Radar and Optical Data in Machine Learning Classification
Journal
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences - ISPRS Archives
Date Issued
July 28, 2025
DOI
10.5194/isprs-archives-XLVIII-G-2025-295-2025
Abstract
Classification of grasslands has an important role in environmental monitoring, and management. This study compares and evaluates the performance of various machine learning and deep learning algorithms in grassland classification using remote sensing data from Sentinel-1 and Sentinel-2 satellites. Sentinel-1 satellite provide Synthetic Aperture Radar data, which captures structural and moisture-related information. Sentinel-2 captures high-resolution optical images with rich spectral details. Both datasets from Sentinel-1 and Sentinel-2 satellites were used to train and evaluate a variety of machine learning models including Random Forest, Support Vector Machines, Logistic Regression, XGBoost and Deep Neural Networks. The results of this study show that Random Forest performs best on Sentinel-1 data and Neural Networks perform best when it comes to grassland classification using Sentinel-2 data. These results show how important it is to select a model based on the characteristics and the nature of the dataset.
File(s)![Thumbnail Image]()
Name
remote_sensing.pdf
Size
1.17 MB
Format
Adobe PDF
Checksum (MD5)
2b7486de8b6807a714cbaab90cbaeb6a