A novel approach for the categorization of cropland and grassland based on multi-seasonal high and medium resolution satellite imagery

Kurzfassung

The implementation of effective and sustainable cultivation procedures is a key element in the framework of the European Community agricultural production. However, political, economic and environmental factors impact the cultivation strategies directly and indirectly, and therewith strongly determine the condition and transformation of the cultivated and natural landscape. Accordingly, a frequent and area-wide monitoring of cropland and grassland is required to assess the actual status, identify basic trends, and mitigate major threats with respect to the agricultural production and its impact on the cultural and natural landscape. Earth Observation (EO) data already proved to be effective for the area-wide and spatially detailed provision of up-to-date geo-information on the agricultural land use and the properties of the cultivated landscape. In particular, one of the major advantages is the possibility of multi-seasonal analyses facilitating the study of the phenological behaviour of the main crop and grassland types and the impact of land use intensity on the environment. Nevertheless, currently available EO-based land-use/land-cover (LULC) datasets (e.g., national topographic data, CORINE Land Cover, etc.) generally exhibit poor spatial and semantic resolution both for cropland and grassland. To overcome this limitation, we present an operational and application-oriented approach aimed at improving the level of thematic/geometric detail for given LULC datasets. A novel system is then proposed for the categorization of agricultural cropland and grassland combining multi-resolution EO data with ancillary geo-information available from currently existing databases. Multi-seasonal high (HR) and medium resolution (MR) satellite imagery is used for determining crop types as well as differentiating between cropland and grassland, respectively. In our experimental analysis, we investigated a test site located in Mecklenburg (Germany). Two images acquired by the IRS-P6 LISS-3 sensor (20m) are first employed to delineate the field parcels in potential agricultural and grassland areas. Next, a stack of seasonality indices is generated based on 5 image acquisitions (i.e., the two LISS scenes and three additional scenes from the IRS-P6 AWiFS sensor (60m)). Finally, a C5.0 tree classifier is applied to identify main crop types and grassland based on the input imagery and the derived seasonality indices. The classifier is trained using sample points provided by the European Land Use/Cover Area Frame Survey (LUCAS). Final results assess the effectiveness of the proposed approach and demonstrate that a multi-scale and multi-temporal analysis of satellite EO data can provide spatially detailed and thematically accurate geo-information on crop types and the cropland-grassland distribution, respectively.