Ground-Based Synthetic Aperture Radar (GB-SAR) has been used to show the difference in polarimetric behavior of a metal sphere object located in different azimuth positions. However, this study has a problem with long data collection in experiment and the probability of measurement error due to antenna vibration during the movement in antenna rail. In this paper, Compressive Sensing (CS) is used in simulation by doing the sampling process to the measurement data from experiment. Full GB-SAR images are produced by solving l1 minimization problem using the sampled data. The result shows that CS can reduce measurement to around 1%, while giving better quality images and give the same conclusions to the conventional system.