A. M. L. Lindahl

Influence of input uncertainty on prediction of within-field pesticide leaching risks

Previous research has suggested that pesticide losses at the field scale can be dominated by a small proportion of the field area. The objective of this study was to investigate whether site-specific applications (i.e. avoiding high-risk areas) at the field scale can contribute to a reduction of pesticide leaching despite uncertainty in the underlying model-based leaching risk map. Using a meta-model of the dual-permeability model MACRO, the annual average pesticide leaching concentrations were estimated for 162 sample sites on a 47 ha field. The procedure was repeated for different scenarios describing different patterns of spatial variation of degradation half-lives and the partition coefficient to soil organic carbon. To account for interpolation uncertainty, maps of predicted pesticide leaching risk were produced by the method of sequential Gaussian simulation. The results of the case study show that larger reductions of predicted leaching were achieved by site-specific application than by that of a comparable uniform dose reduction. Hence, site-specific-applications may be a feasible method to reduce pesticide leaching at the field-scale providing that the model approach gives reasonable estimates of the spatial pattern of pesticide leaching.

A Study of K Variability and Its Effect on Solute Transport in Subsurface-Flow Sand Filters by Measurement and Modelling

Abstract Hydraulics of subsurface flow filters (SSF) was studied by measurement of soil hydraulic conductivity (K) variation and performing tracer tests in two SSF filters consisting of 1–4 mm Ca rich sand (shell sand). Soil samples were carefully taken at several locations in Filter I. A tracer experiment was conducted in the undisturbed Filter II using KI. The measured K variability in Filer I was used to analyze the variations in tracer breakthrough. The spatially distribution of K was obtained by fitting a variogram to observed data and interpolation using Kriging. The tracer residence probability density function (PDF) was determined by modelling the tracer movement with a 3-D groundwater model. The observed and simulated tracer arrival was compared for cases with constant K, constant K and dispersion (D), and for spatially variable K and dispersion. The results show that groundwater models were well suited to simulate solute movement in the SSF system studied. An almost perfect fit to observed tracer PDF was obtained when variable K and dispersion was included in the model. This indicates that information on K variability and dispersion is important for studying solute movement in SSF constructed wetlands.