Land use classification with high-resolution satellite radar for estimating the impacts of land use change on the quality of ecosystem services

Abstract

Ecosystems provide several services for human well-being. The quality of these ecosystem services is among others affected by agriculture, the main land user in Europe. Thus, it is essential to provide information about land use in agricultural areas. In this research, land use classification of agricultural areas is carried out based on high-resolution Spotlight TerraSAR-X images (TSX-images) of two different polarisations (HH and VV). A stack of several radar images taken during the vegetation season is used for multi-temporal classification of land cover. The typical phenology of agricultural vegetation types and their individual phenological development during the year cause differences in the backscatter of the radar signal over time. Two different study areas are investigated, one in the North East of the city of Hannover, Germany called "Fuhrberger Feld" and one in the "Gorajec area" in the very South East of Poland. These two areas represent extremely diverse European regions with regard to agro-technological level, population density, cultivation form as well as geological and geomorphological conditions. Thereby, the radar signal backscatter for different regions is tested. Preliminary results show significant differences in the backscatter of crop types in SAR data of about 3 m, especially for grasslands, grain and broad-leaved crops. Furthermore the VV polarised radar signal has clearly lower backscattering for grains during summertime and for grasslands in general than for broad-leaved crops.