Abstract

Estimation of spatial random fields (SRFs) such as transmissivity or porosity is required for predicting groundwater flow and subsurface contaminant movement. Similarly, distributed parameter fields such as terrain roughness and evapotranspiration coefficients are required by other areas of environmental and earth sciences modeling. This dissertation presents an inverse modeling framework for characterizing SRFs called MAD#, which is an end-user software implementation of the Bayesian inverse modeling technique Method of Anchored Distributions (MAD). The MAD# framework allows modelers to “wrap” existing simulation modeling tools using an extensible driver architecture that exposes model parameters to the inversion engine. A compelling aspect of this model wrapping approach is that it does not require end-users to modify model configuration files; rather the model driver manages dynamic changes to model input and configuration files at run time. The MAD# framework is implemented in an open source software package with the goal of significantly lowering the barrier to using inverse modeling in education, research, and resource management. Toward this end, we introduce and test an intentionally simple user interface for simulation configuration, model driver integration, spatial domain and model output visualization, and evaluation of model convergence.

Degree

PhD

College and Department

Ira A. Fulton College of Engineering and Technology; Civil and Environmental Engineering

Rights

http://lib.byu.edu/about/copyright/

Date Submitted

2014-12-01

Document Type

Dissertation

Handle

http://hdl.lib.byu.edu/1877/etd7409

Keywords

Inverse modeling, Method of Anchored Distribution, Groundwater

Share

COinS