Any changes to the Earth’s surface should be monitored in order to maintain and update the spatial reference system. To establish a global model of ground movements for a large area, it is important to have consistent and reliable measurements. However, in dealing with mass data, outliers may occur and robust analysis of data is indispensable. In particular, this paper will analyse Synthetic Aperture Radar (SAR) data for detecting the regional ground movements (RGM) in the area of Hanover, Germany. The relevant data sets have been provided by the Federal Institute for Geo-sciences and Natural Resources (BGR) for the period of 2014 to 2018. In this paper, we propose a data adoptive outlier detection algorithm to preprocess the observations. The algorithm is tested with different reference data sets and as a binary classifier performs with 0.99 accuracy and obtains a 0.95 F1-score in detecting the outliers. The RGMs that are observed as height velocities are mathematically modeled as a surface based on a hierarchical B-splines (HB-splines) method. For the approximated surface, a 95% confidence interval is estimated based on a bootstrapping approach. In the end, the user is enabled to predict RGM at any point and is provided with a measure of quality for the prediction.