Testing a data assimilation approach to reduce geomechanical uncertainties in modelling land subsidence

The compaction of a gas/oil bearing reservoir or an aquifer system due to subsurface fluid production may result in land subsidence as has been observed worldwide during the 20th century. Uncertainties on geomechanical parameters typically affect model prediction of anthropogenic land settlement. Usually, soil compressibility, Young’s modulus, and the Poisson ratio, that is, the most important parameters characterising the rock geomechanical properties, are derived from laboratory tests and/or in situ measurements, whose reliability may be limited in some cases. In the present work, the authors test the capability to reduce the uncertainty on geomechanical parameters by assimilating a given number of surface displacements. A data-assimilation algorithm, known as ensemble smoother (ES), is used along with a radial-symmetric finite element (FE) code in a realistic orthotropic geological setting, where a 1200-m deep diskshaped reservoir is assumed to be developed. The results show that the ES constitutes a quite promising tool to reduce geomechanical uncertainties in modelling land subsidence.