An inverse method for soil permeability estimation from gas pump tests

This paper applies an inverse method to analyzing the sensitivities of gas flow for the estimation of horizontal and vertical soil permeabilities. Because a Galerkin finite-element model is used to simulate gas flow, the inverse method provides flexibility in the treatment of complex system conditions (for example, a partially confined vadose zone). The analysis of sensitivities to vertical and horizontal soil permeabilities provides important information for optimal design of the gas pump test. Theoretical examples demonstrate that the magnitudes of sensitivity coefficients and their spatial correlation are largely dependent on the locations of data points, the radial extent of surface seal, and soil anisotropy. A surface seal can increase the magnitude of the sensitivities and reduce the correlation between the sensitivities. Therefore, the quality of the estimation of soil permeability is improved. Results of computational experiments show that when the correlation between the sensitivities is larger, computation of the parameter values may require more iterations to converge from a narrower range of initial estimates and the calculated soil parameter values are generally less reliable. Measurement error could result in high uncertainty in parameter estimation.