Temporal and Spatial Weighting of Head and Concentration Observations for a Large-Scale Transient Inverse Model
A regional-scale, three-dimensional groundwater flow and transport modeling effort is ongoing to quantify the environmental consequences of past waste disposal activities and support environmental management activities at the U.S. Department of Energy’s 560-square-mile Hanford Site in southeastern Washington. On the order of one thousand wells in the deep surficial aquifer have been monitored over several decades of site operations (beginning in the 1940’s), and tens of thousands of observations of water table elevation (head) and contaminant concentrations (primarily tritium) have been made over that same period. These data are currently being used as the basis for a site-wide inverse modeling effort to identify model parameters and quantify model uncertainty. Several issues complicate the assignment of appropriate weights to the observations used in the inverse modeling process. The precision of available monitoring techniques has changed significantly over the modeled time period, and the associated error weighting should reflect the methods used (which were in some cases not well documented). In some cases, the detection limits are poorly defined, and some analytical techniques can give rise to non-physical results (such as negative measured concentrations). In addition, the data are strongly clustered both in space and time. This presents the possibility of the inverse solution being too strongly influenced by a cluster of similar values. However, the elimination of some data by declustering techniques, or alternatively, the adjustment of observation weights used in the objective function, raises problems with interpretation and regulatory acceptance of model predictions and uncertainty estimates. This paper presents the methods we have utilized to assign appropriate weights to head and concentration observations and discusses potential issues associated with the weighting scheme employed.
- Research Organization:
- Pacific Northwest National Lab. (PNNL), Richland, WA (United States). Environmental Molecular Sciences Lab. (EMSL)
- Sponsoring Organization:
- USDOE
- DOE Contract Number:
- AC05-76RL01830
- OSTI ID:
- 961706
- Report Number(s):
- PNNL-SA-39048; 1941; EY3542301; TRN: US200923%%242
- Resource Relation:
- Conference: Proceedings of Modflow and More 2003: Understanding through Modeling, 610-614
- Country of Publication:
- United States
- Language:
- English
Similar Records
Declustering in the spatial interpolation of air quality data
Distribution Development for Residual Inventory at the New York West Valley Site - 20400