PNNL

Abstract

This report describes the use of a multipoint geostatistics (MPS) framework with multiple data types for developing alternative conceptual-mathematical models of the subsurface at the Hanford Site. Developing multiple conceptual-mathematical models of the site is important for assessing potential uncertainties related to the features, events, and processes that control the subsurface flow of groundwater and the transport of contaminants of concern (COCs). Evaluating the effects of these uncertainties can result in better site characterization, monitoring, and remedial decisions. Contaminant plume maps suggest that paleochannels have a strong influence on subsurface flow and transport behavior at the Hanford Site. Recent surface-based geophysical data provide additional evidence for paleochannel features in some areas. The primary focus of this work is on developing an approach for evaluating the potential effects of paleochannel features on subsurface flow and contaminant transport. The multiple data types that were used in this study include grain-size distribution, electrical resistivity tomography (ERT), and Light Detection and Ranging (LiDAR)-based digital elevation data. The LiDAR data show outlines of relic paleochannel features that were formed by the ancestral Columbia River and flooding events in the past. These relic paleochannel features were assumed to be representative of paleochannel features that exist in the subsurface. The use of this combination of data in an MPS framework for generating realizations of paleochannel features at the Hanford Site is unique. A training image used by the MPS algorithm was developed from the LiDAR-based digital elevation data. Binned grain-size distribution metrics from selected borehole samples, and ERT data from selected measurement transects, were also used, as hard (known) and soft (uncertain) categorical data, respectively, for representing paleochannel features in the MPS algorithm. The methodology reproduces the general shapes of the relic paleochannel features visible in the LiDAR-based topography data and honors the hard data at their measurement locations. Multiple realizations of paleochannel features were generated and superimposed on a base or reference model. The alternative models were implemented numerically using the subsurface flow and transport simulator eSTOMP (https://www.pnnl.gov/estomp ). Simulations were performed of the transport of three COCs – tritium, carbon tetrachloride, and iodine-129 – using the alternative conceptual models, and a base or reference case that was developed from the Plateau-to-River (P2R) model of the Hanford Site. The alternative models containing paleochannel features generated by the MPS algorithm produced slightly faster transport results relative to the base model, but the general trajectories of the centers of mass of the simulated plumes were very similar for all of the models. It is anticipated that larger variations in transport results would occur with more alternative model realizations, and if different methods were used for parameterization of the material properties assigned to the paleochannel features. The described methodology provides a systematic approach for using multiple types of site characterization data to develop alternative conceptual-mathematical models of the subsurface that include paleochannels. The models developed using this approach are consistent with all available characterization data used in their development. Using the alternative models for subsurface flow and transport modeling illustrates some of the potential effects of uncertainties in features such as paleochannels on the predicted transport and fate of selected COCs at Hanford. Assessment of these uncertainties can help to inform site characterization, monitoring, and remediation decisions.