Model uncertainty – parameter uncertainty versus conceptual models

Abstract

Uncertainties in model structures have been recognised often to be the main source of uncertainty in predictive model simulations. Despite this knowledge, uncertainty studies are traditionally limited to a single deterministic model and the uncertainty addressed by a parameter uncertainty study. The extent to which a parameter uncertainty study may encompass model structure errors in a groundwater model is studied in a case study. Three groundwater models were constructed on the basis of three different hydrogeological interpretations. Each of the models was calibrated inversely against groundwater heads and streamflows. A parameter uncertainty analysis was carried out for each of the three conceptual models by Monte Carlo simulations. A comparison of the predictive uncertainties for the three conceptual models showed large differences between the uncertainty intervals. Most discrepancies were observed for data types not used in the model calibration. Thus uncertainties in the conceptual models become of increasing importance when predictive simulations consider data types that are extrapolates from the data types used for calibration.

Uncertainty in geological and hydrogeological data

“…Harrar et al (2003) and Højberg and Refsgaard (2005) present two different examples, both using three different conceptual models, based on three alternative geological interpretations for multi-aquifer system representative of eastern part of Denmark with glacial till plains (Højberg and Refsgaard, 2005) and in sandy outwash plains in the western part of Denmark (Harrar et al, 2003). Each of the models was calibrated against piezometrical head data using inverse optimisation.…”

“…However, when the models were used to extrapolate beyond the calibration data for predictions of solute transport and travel times the three models differed dramatically. When assessing the uncertainty contributed by the model parameter values using Monte Carlo simulations, the overlap of uncertainty ranges between the three models by Højberg and Refsgaard (2005) significantly decreased when moving from groundwater heads to capture zones and travel times. The larger the degree of extrapolation, the more the underlying conceptual model dominates over the parameter uncertainty and the effect of calibration.…”

“…Structural uncertainty has long been recognized to be a dominating factor (Carrera and Neuman, 1986;Harrar et al, 2003;Troldborg, 2004;Højberg and Refsgaard, 2005;Poeter and Anderson, 2005;Eaton, 2006). This is especially important in groundwater modelling, where the geological structure is dominant for the groundwater flow but where specific knowledge of the geology at the same time is very limited.…”

A Bayesian belief network approach for assessing uncertainty in conceptual site models at contaminated sites

“…Conceptual uncertainty is a well described phenomenon [Beven, 2009;Konikow and Bredehoeft, 1992;Refsgaard et al, 2006;Walker et al, 2003], which is often found to be a major source of uncertainty and must therefore be considered [Bredehoeft, 2005;Harrar et al, 2007;Højberg and Refsgaard, 2005;Troldborg et al, 2007]. The challenge of quantifying conceptual uncertainty has been discussed in the literature and many methods have been proposed.…”

Challenges in conditioning a stochastic geological model of a heterogeneous glacial aquifer to a comprehensive soft data set

“…This can lead to alleged correct results, for instance when addressing the water balance on catchment scale, but will often prove to be inadequate for simulations beyond general flows and heads, e.g., contaminant transport modeling. Therefore, it is proposed by numerous studies that the uncertainty on the geological conceptualization is crucial when assessing uncertainties on flow paths (Neuman, 2003;Bredehoeft, 2005;Hojberg and Refsgaard, 2005;Troldborg et al, 2007;Seifert et al, 2008). One of the strategies often recommended for characterizing geological uncertainty and assessing its impact on hydrological predictive uncertainty is the use of multiple geological models (Renard, 2007;Refsgaard et al, 2012).…”