skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Hanford groundwater modeling: a numerical comparison of Bayesian and Fisher parameter estimation techniques

Abstract

This report is concerned with the performance of two promising techniques for estimating the parameters of groundwater models developed and applied by Rockwell Hanford Operations at the US Department of Energy Hanford site. The report begins with a fairly general formulation of the parameter estimation problem which helps to explain the capabilities and differences of the two estimation techniques (weighted least-squares and maximum likelihood). Particular emphasis is placed on the role of model input errors due to initial condition, boundary condition and source term uncertainties. It is shown that the weighted least-squares estimation algorithm (a Fisher estimator) essentially ignores input errors while the maximum likelihood estimation algorithm (a Bayesian estimator) accounts for these errors in a systematic way. Performance differences between these estimators are negligible when input errors are minor. When input errors are large, however, the maximum likelihood estimator appears to give significantly better estimates. This is illustrated with a numerical example based on an application of the RHAFE groundwater flow model to a hypothetical twelve element, three dimensional aquifer. Recommendations indicating the most promising directions for future research and application are provided at the end of the report.

Authors:
Publication Date:
Research Org.:
Resource Management Associates, Lafayette, CA (USA)
OSTI Identifier:
6636751
Report Number(s):
RHO-C-24
TRN: 81-005035
DOE Contract Number:  
AC06-77RL01030
Resource Type:
Technical Report
Country of Publication:
United States
Language:
English
Subject:
58 GEOSCIENCES; 12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE WASTES FROM NUCLEAR FACILITIES; 11 NUCLEAR FUEL CYCLE AND FUEL MATERIALS; AQUIFERS; FLOW MODELS; MATHEMATICAL MODELS; GROUND WATER; COMPARATIVE EVALUATIONS; HAPO; LEAST SQUARE FIT; MAXIMUM-LIKELIHOOD FIT; RADIOACTIVE WASTE DISPOSAL; UNDERGROUND DISPOSAL; HYDROGEN COMPOUNDS; MANAGEMENT; NATIONAL ORGANIZATIONS; NUMERICAL SOLUTION; OXYGEN COMPOUNDS; US AEC; US DOE; US ERDA; US ORGANIZATIONS; WASTE DISPOSAL; WASTE MANAGEMENT; WATER; 580100* - Geology & Hydrology- (-1989); 052002 - Nuclear Fuels- Waste Disposal & Storage; 053000 - Nuclear Fuels- Environmental Aspects

Citation Formats

McLaughlin, D B. Hanford groundwater modeling: a numerical comparison of Bayesian and Fisher parameter estimation techniques. United States: N. p., 1980. Web.
McLaughlin, D B. Hanford groundwater modeling: a numerical comparison of Bayesian and Fisher parameter estimation techniques. United States.
McLaughlin, D B. Sat . "Hanford groundwater modeling: a numerical comparison of Bayesian and Fisher parameter estimation techniques". United States.
@article{osti_6636751,
title = {Hanford groundwater modeling: a numerical comparison of Bayesian and Fisher parameter estimation techniques},
author = {McLaughlin, D B},
abstractNote = {This report is concerned with the performance of two promising techniques for estimating the parameters of groundwater models developed and applied by Rockwell Hanford Operations at the US Department of Energy Hanford site. The report begins with a fairly general formulation of the parameter estimation problem which helps to explain the capabilities and differences of the two estimation techniques (weighted least-squares and maximum likelihood). Particular emphasis is placed on the role of model input errors due to initial condition, boundary condition and source term uncertainties. It is shown that the weighted least-squares estimation algorithm (a Fisher estimator) essentially ignores input errors while the maximum likelihood estimation algorithm (a Bayesian estimator) accounts for these errors in a systematic way. Performance differences between these estimators are negligible when input errors are minor. When input errors are large, however, the maximum likelihood estimator appears to give significantly better estimates. This is illustrated with a numerical example based on an application of the RHAFE groundwater flow model to a hypothetical twelve element, three dimensional aquifer. Recommendations indicating the most promising directions for future research and application are provided at the end of the report.},
doi = {},
url = {https://www.osti.gov/biblio/6636751}, journal = {},
number = ,
volume = ,
place = {United States},
year = {1980},
month = {11}
}

Technical Report:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that may hold this item. Keep in mind that many technical reports are not cataloged in WorldCat.

Save / Share: