Remote Sensing in Land Science Studies

Mapping Land-Use Land-Cover Change (LULCC) through satellite remote sensing offers valuable insights into understanding the trajectories, patterns, drivers, and consequences of land-cover alterations, being at the core of Land System Science. Throughout this course, we will delve into the evolution of classification and change detection techniques designed to map LULCC and gauge land-use intensity.

The course curriculum encompasses non-parametric machine learning classification methods, including SVM, Random Forests and hybrid classifications. We will explore techniques to enhance classifications and boost classification accuracies, state-of-the-art accuracy assessment methods, multisource image modality techniques, and the analysis of large data composites, including cloud computation environments like Google Earth Engine. Our primary focus will be on the utilization of freely accessible datasets, including optical imagery such as Landsat, Sentinel-2, and MODIS, as well as radar data from Sentinel-1 and microsatellites like PlanetScope. Additionally, we will cover applications in socio-economic and environmental domains, including examples from contemporary research e.g., mapping of woody resources, characterization of wetland types and dynamics, land-cover changes and socio-economic footprints.

This course complements the course “Remote Sensing of the Biogeosphere” and serves as a stepping stone into the more advanced course “Satellite Image Processing and Analysis in the Big Data Era”. The course is particularly tailored for students interested in the human dimensions of land system changes.