A Model Management System for Numerical Simulations of Subsurface Processes

Numerical models related to subsurface fluid flow, contaminant transport, and chemical and biological reactions play key roles in several DOE program areas. Decision makers and stakeholders dealing with groundwater issues need to determine, from the large number of models available, which models have the potential to produce useful information. At present, decision makers have no systematic method for identifying simulation models that meet a set of specified requirements for use on a given project. This project will create a Model Management System for numerical simulations of subsurface science phenomena. Phase I developed a preliminary ontology for subsurface science simulations; constructed a model metadata vector with a structure that is rich enough to enable a detailed description of each model and provide quantitative comparisons of different models; created a prototype relational database of mathematical models that simulate subsurface flow, heat transport, and chemical and biological reactions; developed an initial web site that enables users to use a ¿model choice wizard¿ to assist in identifying models appropriate for use on a particular problem; and initiated beta testing of the prototype database and web site. Phase II will augment the current cosine based model comparison metric with a Latent Semantic Analysis metric; add a wiki (a set of web pages for community editing) to the model management system to provide a forum for model users to discuss models and share data; develop a generic model management framework; and continue the development of tools for creating and managing metadata. Commercial Applications and Other Benefits as described by the awardee: In addition to the application to subsurface science, a web-based model management framework should find use in numerous physical sciences areas.