Investigation of the Feasibility of Data Assimilation Applied to the Management of Mining-Induced Ground Movements for Underground Gas Storage

Abstract Subsurface cavities created by mining operations lead to subsidence at the Earth’s surface. To minimise risks and damage to the environment, infrastructure, and people, accurate and reliable predictions of mining-induced ground movements are required. Models are used for these predictions, which depend on unknown parameters that must be determined from known observations using so-called inverse parameter estimation. One frequently employed method for inverse parameter estimation is the adjustment of observations. When using this method, however, it is not straightforward to consider non-linear relationships, data from different sources, or dynamic changes in the parameters over time. Possible alternatives are the Kalman filter, which was developed for sequential data assimilation in linear dynamical systems, and the ensemble Kalman filter, an extension of the Kalman filter to non-linear problems. This study compares these methods for inverse parameter estimation to show that the Kalman filter and the ensemble Kalman filter are suitable substitutes for the adjustment of observations, using three artificial test studies simulating annual levelling over salt domes and one test with real-world data. This lays the foundation and builds confidence for fully exploiting the alternative methods in future work.