Detection of Rapid Ground Surface Deformation using Satellite Interferometry Data and Deep Learning

Scientifically developed methods to analyse ground deformation have found broader applications and have become more effective. Satellite-based Interferometric Synthetic Aperture Radar (InSAR) is widely used for mapping and monitoring deformation when it comes to large-scale deformation studies. The development of innovative algorithms such as the Small Baselines Subsets (SBAS) has increased the accuracy of the data and has improved the reliability of the data. Data scientists and AI professionals have exploited the capability of machine learning integrally with operational data analysis. The significance of Phase Unwrapping Network (PUNet) has decreased the rate of errors in interference due to loss of accuracy during the phase unwrapping and subsidence detection. This artificial neuronal system has obtained very good results, being able to detect deformation effects that the traditional methods, mainly in cases of more difficult environments such as mining, lack. This research integrates the use of InSAR data and deep learning to detect rapid ground surface deformation in the Legnica-Głogów Copper Belt (LGCB) mining area in Poland. The data spanned from January 2016 to January 2022. The SBAS InSAR dataset was utilized, and it consists of 329 obtained datasets. The data were used to calculate displacement and identify areas that experienced rapid ground surface deformation due to the underground mining activities in the LGCB region.