Feasibility study and mapping of grassland habitat types with remote sensing with emphasis on separation of intensive and extensive grasslands

In a study on the applicability of Sentinel-1 and Sentinel-2 satellite data for mapping grassland habitat types, we investigated the potential for distinguishing between intensive and extensive grasslands. Both Sentinel-1 and Sentinel-2 imagery is freely available through the European Copernicus programme. Each of the two systems consists of two satellites, allowing high temporal resolution. The two Sentinel-1 satellites use Synthetic Aperture Radar (SAR) sensor in four different modes; they typically image the land with a spatial resolution of up to 10 m and a temporal resolution of 12 days and 6 days from April 2014 and April 2016, respectively. The two Sentinel-2 optical satellites acquire with a multispectral sensor in thirteen spectral bands, with a spatial resolution of up to 10 m. Sentinel-2 data has been available every ten days since December 2015 and every five days since July 2017.

We studied characteristics of vegetation development between March and November, for all grasslands in Slovenia, for three test years (2017, 2018 and 2019). We computed the results using machine learning on time series of satellite imagery and ancillary data. For the latter, we used the available habitat type mapping of the client IRSNC; for the definition of the study area (read: grassland polygons), we used the state data of the Register of Actual Use of Agricultural and Forest Land ("RABA") and the Graphical Unit of Agricultural Land Use ("GERK").