Utilizing Both Radarsat-2 And Terrasar-X Polarimetric Data For Crop Growth Stages Estimation

This paper uses RADARSAT-2 quad Polarimetric Synthetic Aperture Radar and TerraSAR-X dual polarimetric SAR data to estimate agriculture crop growth stages. Ten RADARSAT-2 Fine Quad Wide beam modes data and 13 sets of Stripmap TerraSAR-X data with duel HH and VV polarizations were used in this study. Polarimetric features such as differential reflectivity bands ratio, entropy, anisotropy, alpha angle, lambda, scattering diversity and polarization index were extracted and evaluated for wheat and canola crop types. The results from both RADARSAT-2 and TerraSAR-X data were compared; they demonstrated clear correlations between crop growth stages obtained from the ground truth data and polarimetric parameters. A stepwise regression feature selection technique was employed for determine the most suitable parameters for different crop types by utilizing both RADARSAT-2 and TerraSAR-X data. By integrating of the selected polarimetric parameters from both Radarsat-2 and TerraSAR-X data, the more accurate estimate of the crop growth stages was achieved and validated using crop growth stage ground truth data. The estimation of crop growth stages was performed on the classified crop image subsets acquired at different dates. The results indicates that the mean square errors of the estimated crop growth stages are reduced using integrated/combined polarimetric parameters from both RADARSAT-2 and TerraSAR-X data, as opposed to using a single sensor.