A Local-Global Pattern Matching Approach for Dynamic Data Integration

Abstract:

Inverse modeling is essential for generating reliable subsurface flow and transport models that can inform groundwater resource management and aquifer remediation efforts. Multiple point statistics (MPS) based models, beyond the traditional two-point statistics based methods, offer an alternative to simulate complex geological features and pattern, conditioned to the measured conductivity data. Parameter estimation of MPS conductivity field, using measured dynamic data such as piezometric head data, remains one of most challenging tasks in inverse modeling. We propose a new local-global pattern matching method to integrate dynamic data into geological models. The local pattern is composed of hydraulic conductivity and piezometric head, sampled from joint training images that are composed both of spatial variations in conductivity as well as the corresponding simulated piezometric head. Subsequently, a global constraint is enforced on the simulated geologic models in order to honor the entire temporal profile of measured head data. Meanwhile, the training image models are refined on the basis of new accepted models at each time step in order to reduce the computational cost of pattern matching. As a consequence, the final suite of models preserves both the geologic patterns of variability as well as honor the dynamic data in a computationally efficient fashion. This local-global pattern matching method is demonstrated for modeling a two dimensional bimodal distributed heterogeneous conductivity field. The results indicate that the characterization of conductivity and predictions of flow and transport are improved when the piezometric head data are integrated into the geological models.