Title: Combining GLUE and BMA to account for conceptual model and scenario uncertainties in groundwater modelling – Application to the Walenbos area, Belgium
Authors: Rojas Mujica, Rodrigo
Feyen, L
Dassargues, Alain
Kahunde, Samalie
Peeters, Luk
Batelaan, Okke #
Issue Date: 2008
Conference: HydroPredict 2008 location:Prague, Czech Republic date:15-18 Sept 2008
Abstract: Groundwater models are often used to predict the behaviour of groundwater systems under future
stress conditions. These models may vary in complexity from simplified groundwater system
representations to more elaborated versions. It has been suggested that uncertainties in groundwater
model predictions are largely dominated by uncertainties arising from the definition of alternative
conceptual models. Different external factors such as climatic conditions or groundwater abstraction
policies, on the other hand, may also play an important role. Rojas et al., (2008) proposed a
multimodel approach to account for predictive uncertainty arising from inputs, parameters, and
alternative conceptualizations. We extend upon this methodology to include uncertainties arising from
the definition of alternative conceptual models and optional scenarios. The improved methodology is
applied to a real aquifer system underlying the Walenbos Natural Reserve area in Belgium. Three
alternative conceptual models comprising different levels of geological knowledge are postulated.
Additionally, three recharge estimates to evaluate scenario uncertainties are proposed. A joint
estimation of the total uncertainty including input, parameter, conceptual model, and scenario
uncertainties is determined for groundwater budget terms in the Walenbos area. Finally, results
obtained using the proposed multimodel approach are compared with the results obtained from
multimodel methodologies that include a calibration step and use a model selection criterion to
discriminate between models. Results show that scenario uncertainties have a significant impact on
total predictive uncertainties. Besides, conceptual model uncertainties play a significant role even for
the case when a model is preferred over the others. Predictive distributions showed to be considerably
different in shape, central moment, and spread among alternative conceptualizations and scenarios
analyzed. The latter reaffirms the idea that relying on a single conceptual model driven by a particular
scenario, will likely produce bias and under-dispersive estimations of the total predictive uncertainty.
Multimodel methodologies based on model selection criteria produced ambiguous results. In the frame
of a multimodel approach, these inconsistencies are critical and can not be neglected. The results of
this study strongly advocate the idea to address conceptual model uncertainty. Additionally, including
alternative scenarios permits to obtain more realistic, and possibly, more reliable estimations of the
total predictive uncertainty estimations.
Publication status: published
KU Leuven publication type: IMa
Appears in Collections:Department of Earth and Environmental Sciences
Division of Geology
# (joint) last author

Files in This Item:

There are no files associated with this item.


All items in Lirias are protected by copyright, with all rights reserved.