Development of generalized mapping tools to improve implementation of data driven computer simulations (04-ERD-083)

The Stochastic Engine (SE) is a data driven computer simulation tool for predicting the characteristics of complex systems. The SE integrates accurate simulators with the Monte Carlo Markov Chain (MCMC) approach (a stochastic inverse technique) to identify alternative models that are consistent with available data and ranks these alternatives according to their probabilities. Implementation of the SE is currently cumbersome owing to the need to customize the pre-processing and processing steps that are required for a specific application. This project widens the applicability of the Stochastic Engine by generalizing some aspects of the method (i.e. model-to-data transformation types, configuration, model representation). We have generalized several of the transformations that are necessary to match the observations to proposed models. These transformations are sufficiently general not to pertain to any single application. This approach provides a framework that increases the efficiency of the SE implementation. The overall goal is to reduce response time and make the approach as ''plug-and-play'' as possible, and will result in the rapid accumulation of new data types for a host of both earth science and non-earth science problems. When adapting the SE approach to a specific application, there are various pre-processing and processing steps that are typicallymore » needed to run a specific problem. Many of these steps are common to a wide variety of specific applications. Here we list and describe several data transformations that are common to a variety of subsurface inverse problems. A subset of these steps have been developed in a generalized form such that they could be used with little or no modifications in a wide variety of specific applications. This work was funded by the LDRD Program (tracking number 04-ERD-083).« less