Transport modeling of trace elements with an extend Freundlich isotherm – Consideration of spatial variability and data uncertainty in leaching prediction
Country / Region: Germany
Begin of project: October 1, 2001
End of project: March 30, 2005
Status of project: March 30, 2005
The prediction of future contaminant concentration trends in the transition zone between the unsaturated and saturated zone is central to the groundwater pollution risk assessment strategy outlined in the German soil protection ordinance (BBodSchV, 1999). Within the framework of legislation the risk assessment aims mainly at environmental situations at the field scale where materials with low contamination or certain recycling products are disposed, or where soils are contaminated unintentionally. The outlined prediction procedure is aimed at risk assessment of trace element leaching in these situations when information is scarce.
In general, when carrying out a model-based groundwater risk assessment on the scale of interest (frequently the field scale), there is often a mismatch between the limited amount of data available for an investigated site and the need of a transport model for detailed parameter information. This mismatch will cause uncertainty in the risk assessment as the limited availability of data requires simplified models which are less demanding with regard to their input but also less precise.
For strongly sorbing trace elements as investigated here, the application of a transport model is often prohibited because site specific sorption isotherms are not available or their determination is too time consuming. To overcome this situation, a pedotransfer function relating parameters relevant for the description of sorption or transport processes to easily measurable soil properties can be applied. The derivation of a generally applicable extended Freundlich isotherms for Cd, Cr, Co, Cu, Mo, Ni, Pb, Sb, Tl and Zn on a nation-wide scale is described in Utermann et al. (2005). An advantage of these functions is that the spatial variability of the sorption relevant properties on the field scale can easily be incorporated in the transport risk assessment.
The above mentioned strategy, where a simplified model is preferred over a complicated one, can only be utilized at the expense of forecast accuracy. In this study, one important source of uncertainty is the limited accuracy of the applied extended Freundlich equation when predicting solute concentrations in the solid or liquid phase. In addition element sorption to soil varies in space because of the spatial variability of soil properties. Spatial variability of these properties is an intrinsic feature of soils independent of their origin. The characterisation of this variability is an additional source of uncertainty because quality, availability and amount of site-specific soil information are often very heterogeneous. The detailed knowledge of the dynamic soil water regime is only of minor importance for the trace elements mentioned above. Because these elements are generally strongly sorbing and nondegradable in soils, the assumption of a stationary water regime for solute flux modeling is possible without introducing a significant error compared to the other sources of uncertainty.
Our transport modeling approach is based on the parallel column model, where the investigated site is divided into individual soil columns. Solute transport in each column is modeled with the convection-dispersion-equation taking into account sorption. Sorption coefficients are calculated with the extended Freundlich equation mentioned above. Uncertainty of the extended Freundlich equation as well as spatial variability and uncertainty are explicitly considered by using a 2-dimensional (2-D) Monte-Carlo technique which allows the discrimination between uncertainty’s and variability’s share in the variance of the modeled output distribution. Figure 1 illustrates the principle steps of such a simulation.
Figure 1: Scheme of the 2-dimensional (2-D) Monte-Carlo technique illustrating the n x n local transport simulations and the averaging in dimension variability to obtain n field-averaged concentration-depth profiles for further statistical evaluation in the context of risk assessment.
Source: BGR
For the risk assessment according to the BBodSchG (1998) results of the 2 D-Monte Carlo simulation are spatially averaged to yield a distribution of site or field averaged concentrations that reflects only uncertainty of the forecast. When evaluating this distribution with regard to threshold values, it is necessary to choose between the mean, median or a certain percentile depending on the directives of the risk assessment.
To test the models capability for risk assessment, we applied it to a case study dealing with Cd transport at the field scale. A comparison of the model results from a hindcast simulation at a site having an 85 year history of Cd immision from a metal smelter with measured data allows an evaluation of the approach.
Figure 2 shows that the results of the hindcast approach and the measured data agree quite well. Figure 3 shows a prediction of the future development of Cd concentration at the site of assessment (groundwater level) according to the BBodSchV (1999) up to the year 2175.
Figure 2: Measured depth distribution of Cd at the Nordenham site after 85 years of immission vs. the Cd depth distribution (including the 95 % confidence interval) predicted with the extended Cd-Freundlich equation derived by Utermann et al. (2005).
Source: BGR
Figure 3: Leaching prediction up to the year 2175 at the Nordenham site. The site of assessment according to the BBodSchV (1999) is at 60 cm below soil surface.
Source: BGR
The results illustrate the models capability for a risk assessment when available data is limited like in a preliminary investigation.
Downloads:
Literature:
- BBodSchV, Bundesgesetzblatt (1999): Bundesbodenschutzverordnung: BGBl I, S. 1554
- Utermann, J., Heidkamp, A., Meyenburg, G., Gäbler, H. E., Altfelder, S., Böttcher, J., (2005): Pedotransfer-functions for Sorption of Trace Elements in Agricultural & Forest Soils, Consoil 2005
Partner:
- Bundesministerium für Bildung und Forschung
- Prof. Dr. T. Streck, Fg. Biogeophysik, Universität Hohenheim, Stuttgart
- Prof. Dr. J. Böttcher, Institut für Bodenkunde, Universität Hannover