Use of geostatistics to determine spatial variation in pesticide leaching - Preliminary findings

The risk of pesticides leaching into ground water depends on the characteri stics of the soil that affect their adsorption and degradation, and on wate r and solute fluxes in the soil. The study site is in a groundwater protect ion zone in the Thames Valley (England) where the soil is a sandy loam over lying gravel. Studies have shown that there is a strong linear relation bet ween soil organic carbon content and the adsorption characteristics of atra zine. Using existing data sets for hydrodynamic properties and organic C me asured at the same site we have used geostatistical techniques to interpola te and extrapolate to unsampled locations within a field. Soil organic matt er (SOM) was measured by loss on ignition for the entire field and mean por e water velocity (V) and the solute dispersion coefficient (D) were measure d in a 48 cm by 48 cm area that was adjacent. There were 64 drainage collec tors in the latter area and each had a support of 6 cm by 6 cm. Variograms were computed for the three variables which suggested that the variation in leaching can be both very local, over distances of less than a metre, and can occur over much longer distances of hundreds of metres. The variograms were used with the data to produce kriged estimates of V, D and SOM for map ping. Conditional Gaussian sequential simulation of V and D, using the vari ogram models and the available data, was used to obtain information from a larger area which could be used as an input to the LEACHP model. Assuming a relation between the hydrodynamic properties of the soil and SOM these res ults suggest that there are at least two important scales of spatial variat ion in the pattern of leaching. This has implications for future sampling t o take account of leaching at these very different spatial scales.