A hierarchical bayesian model averaging approach to cope with sources of uncertainty in conceptual ground water models
Document Type
Conference Proceeding
Publication Date
7-21-2011
Abstract
By conducting a facies analysis for the East Baton Rouge aquifers system through the use of indicator variogram functions, the interpolation method seems to be much more sensitive to the sand-clay line cutoff and sand-clay cutoff probability in comparison to the selection of different variogram models. Thus, by changing these two parameters, one gets considerably different conditional stratigraphical realizations. This study introduces a hierarchical Bayesian model averaging (HBMA) to best utilize all possible realizations to estimate the sand-clay distribution under Bayesian statistical framework. The HBMA is applied to twelve stratigraphical models for subsurface elevations from 1460 to 1650 feet below mean sea level (msl) in the Baton Rouge area, Louisiana. The model structure uncertainty considered arises from the sand-clay line cutoff and sand-clay cutoff probability. Although only two sources of uncertainty are considered, the method can be readily extended to account for other sources of structural uncertainty such as fault morphology, dip angle, or borehole elevation. © 2011 ASCE.
Publication Source (Journal or Book title)
World Environmental and Water Resources Congress 2011: Bearing Knowledge for Sustainability - Proceedings of the 2011 World Environmental and Water Resources Congress
First Page
1089
Last Page
1098
Recommended Citation
Tsai, F., & Elshall, A. (2011). A hierarchical bayesian model averaging approach to cope with sources of uncertainty in conceptual ground water models. World Environmental and Water Resources Congress 2011: Bearing Knowledge for Sustainability - Proceedings of the 2011 World Environmental and Water Resources Congress, 1089-1098. https://doi.org/10.1061/41173(414)112