PICS Background.

Over half of the U.S. population depends on groundwater for its water supply. Groundwater is an important source of water for public utility consumption, as well as for irrigation and industrial process water. In many regions, available sources of groundwater are the determining factor for development and economic activity. Groundwater supplies are increasingly threatened by the migration of organic, inorganic and radioactive contaminants which have been introduced to the environment by improper disposal or accidental release. Current estimates of remediation costs of contaminated sites at U.S. government facilities alone range into the hundreds of billions of dollars.

Remediation methods remain extremely (and potentially prohibitively) expensive and unpredictable in their success. Mathematically based numerical models are an indispensible tool in the effort to protect uncontaminated regions and to plan remediation strategies for polluted sites. Resolution of current limitations of numerical models for groundwater remediation will enable better, more cost-effective and more reliable engineering approaches to groundwater remediation.

The Partnership In Computational Science was organized to address the limitations of current numerical codes describing groundwater flow and contaminant transport. These limitations are common to numerical models across a broad spectrum of scientific problems. Three dimensional, nonlinear, heterogeneous problems require massive computational resources. It is a goal of basic science (and, in particular, of the Grand Challenge program) to develop parallel computing technology so that computational resources are effectively deployed to solve such problems. Computational limitations have prevented the use of otherwise available methods and data. The first of PICS's basic science objectives is to develop and prove the massively parallel approach by producing a combined state-of-the-art parallel code, GCT, describing Groundwater flow and Contaminant Transport.

Another important generic limitation of numerical methods in groundwater modeling, is a lack of necessary input data in the form of model parameter characterization. The use of techniques including statistical characterization, multiple realizations or ensemble averages, and inverse methods for parameter identification draw upon the collective expertise of consortium personnel in geostatistics from the disciplines of civil and chemical engineering, geology, mathematics, and computer science. Again the problems are computationally intensive, and require innovative, massively parallel approaches. These problems are common to many disciplines, and thus of basic importance to science.