Abstract

There is a clear need to conduct uncertainty analyses with respect to nitrate leaching at a range of spatial scales, from the field to the catchment scale, dependent on the spatial scale for which an assessment is relevant. At the scale of the Odense catchment in Denmark, an uncertainty assessment most likely leads to an unacceptably high number of simulations as a result of the large number of combinations of distributed data available. In this study we present a framework for an uncertainty assessment for model parameters for the unsaturated component in a linked rootzone and groundwater model and exemplified for the Odense River catchment. The proposed framework consists of (i) a simplification of the model, (ii) identification of uncertain parameters, (iii) generation of model parameter sets by means of Latin hypercube sampling (LHS), (iv) rootzone model simulation of realizations under (iii), (v) uncertainty analysis based on the standardized rank regression coefficient index, (vi) ranking of simulated nitrate leaching, and (vii) simulation of selected parameter sets sampled from the output distribution under (vi). The results show a high sensitivity of the van Genuchten soil water release characteristics parameters for percolation and leaching. The amount of applied slurry has a relatively high sensitivity for leached nitrate from the rootzone. The ranking nitrate leaching derived from the LHS simulations for a 25-yr period is preserved for longer 55-yr simulations. This implies that the proposed framework is applicable to the Odense catchment and likely has general applicability. In this way, uncertainty assessments are feasible at the catchment scale.

You do not currently have access to this article.