- Library Home /
- Search Collections /
- Open Collections /
- Browse Collections /
- UBC Theses and Dissertations /
- On the quantification of the effect of model error...
Open Collections
UBC Theses and Dissertations
UBC Theses and Dissertations
On the quantification of the effect of model error on groundwater model predictions and risk assessments Gaganis, Petros
Abstract
Errors arising from an imperfect model structure (model error) may significantly degrade the usefulness of model calibration in predictive modeling and result in misleading uncertainty and risk analyses. Model error is not random but systematic. Its effect on model predictions varies in space and time and differs for the flow and solute transport components of a groundwater model. Model error does not necessarily have any probabilistic properties that can be easily exploited in the construction of a single-objective model performance criterion. The effect of model error on the solution of the inverse problem is evaluated in the parameter space using a per-datum approach to model calibration where a model is calibrated at each data point separately. For each dependent variable, the location of each per-datum parameter estimate in the parameter space is a function of the magnitude of model error at the given sampling location and time. These parameter estimates are translated into a probabilistic description of model output that represents the level of confidence in model performance evaluated in terms of each model prediction. This approach provides useful information regarding the strengths and limitations of a model as well as the performance of classical calibration procedures. The quantification of model error in the presence of parameter uncertainty is also evaluated within the Bayesian framework. Insight gained in updating the prior information on the parameter values is used to assess the correctness of the model structure, which is defined relative to the required accuracy by model predictions. Model error is evaluated in terms of each measurement of the dependent variable through an examination of the correctness of the model structure for different accuracy levels. The spatial and temporal variability of estimated model error can be used in identifying its possible causes, as well as in discriminating among models in terms of model structure correctness. Application of perdatum calibration and the Bayesian model error quantification to a groundwater contamination problem at the Chernobyl site in the Ukraine indicates that evaluating the effect of model error on estimated risks in hydrogeologic decision analysis offers an attractive alternative to adopting a bias towards conservative values.
Item Metadata
Title |
On the quantification of the effect of model error on groundwater model predictions and risk assessments
|
Creator | |
Publisher |
University of British Columbia
|
Date Issued |
2000
|
Description |
Errors arising from an imperfect model structure (model error) may significantly
degrade the usefulness of model calibration in predictive modeling and result in misleading
uncertainty and risk analyses. Model error is not random but systematic. Its effect on model
predictions varies in space and time and differs for the flow and solute transport components
of a groundwater model. Model error does not necessarily have any probabilistic properties
that can be easily exploited in the construction of a single-objective model performance
criterion. The effect of model error on the solution of the inverse problem is evaluated in the
parameter space using a per-datum approach to model calibration where a model is
calibrated at each data point separately. For each dependent variable, the location of each
per-datum parameter estimate in the parameter space is a function of the magnitude of model
error at the given sampling location and time. These parameter estimates are translated into a
probabilistic description of model output that represents the level of confidence in model
performance evaluated in terms of each model prediction. This approach provides useful
information regarding the strengths and limitations of a model as well as the performance of
classical calibration procedures.
The quantification of model error in the presence of parameter uncertainty is also evaluated
within the Bayesian framework. Insight gained in updating the prior information on the
parameter values is used to assess the correctness of the model structure, which is defined
relative to the required accuracy by model predictions. Model error is evaluated in terms of
each measurement of the dependent variable through an examination of the correctness of
the model structure for different accuracy levels. The spatial and temporal variability of
estimated model error can be used in identifying its possible causes, as well as in
discriminating among models in terms of model structure correctness. Application of perdatum
calibration and the Bayesian model error quantification to a groundwater
contamination problem at the Chernobyl site in the Ukraine indicates that evaluating the
effect of model error on estimated risks in hydrogeologic decision analysis offers an
attractive alternative to adopting a bias towards conservative values.
|
Extent |
13531951 bytes
|
Genre | |
Type | |
File Format |
application/pdf
|
Language |
eng
|
Date Available |
2009-07-23
|
Provider |
Vancouver : University of British Columbia Library
|
Rights |
For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.
|
DOI |
10.14288/1.0089690
|
URI | |
Degree | |
Program | |
Affiliation | |
Degree Grantor |
University of British Columbia
|
Graduation Date |
2000-11
|
Campus | |
Scholarly Level |
Graduate
|
Aggregated Source Repository |
DSpace
|
Item Media
Item Citations and Data
Rights
For non-commercial purposes only, such as research, private study and education. Additional conditions apply, see Terms of Use https://open.library.ubc.ca/terms_of_use.