Towards a detection of grassland cutting practices with dual polarimetric TerraSAR-X data

Abstract

In this study, polarimetric synthetic aperture radar (SAR) parameters are analysed and compared with in situ measurements in order to develop a methodology for detecting cutting practices within grassland areas. The grasslands were monitored with TerraSAR-X radar imaging in dual polarization HH/VV mode and are located near the banks of the Kasari River, close to the Baltic Sea coast of Estonia. The parameters analysed include HH, VV, HH + VV, and HH – VV backscatter, HH/VV polarimetric coherence magnitude and phase, T₁₂ polarimetric coherence magnitude and phase, and also dual polarimetric entropy, alpha, and alpha dominant parameters. Using these parameters derived from the dual polarimetric TerraSAR-X data set, it was virtually impossible to distinguish tall grass (height >30 cm) from short grass (height <30 cm). On the other hand, it proved feasible to detect areas where grass had been cut and left on the ground. Several parameters showed specific behaviour for the state of grassland and the most notable change was found in the dual polarimetric dominant scattering alpha angle. This angle changed from 10° to 25° after tall grass had been cut and left on the ground. This behaviour of the dominant scattering alpha angle can effectively be described using a particle scattering model for vegetation backscattering.

Acknowledgements

This study was carried out under the DoRa scholarship programme of the Archimedes Foundation during a research stay at the Microwaves and Radar Institute of the German Aerospace Center (DLR). The authors would like to thank DLR for the TerraSAR-X imagery used in the study, distributed under project LAN1123. We would also like to thank the Estonian Agricultural Registers and Information Board for the geospatial data and information about the grasslands used in the study.