The consequences of interpolating or calculating first on the simulation of pesticide leaching at the regional scale

We analysed different approaches to process oil information when simulating pesticide leaching to groundwater at the regional scale. The first approach, calculate alone (CA), consisted of the model application on point data followed by the aggregation of the results to the regional scale. Two further approaches were used to generate spatial output and differed by interpolating after or before the model run on point support (calculate first, interpolate later; CI vs. interpolate first, calculate later; IC).The three approaches were tested with both a linear (modified Attenuation Factor, AF; Rao, P., Hornby, A., Jessup, R., 1985. Indices for taking the potential for pesticide contaminatio of groundwater. Soil and Crop Science Scoiety of Florida Proceedings 44, 1-8) and a non-linear (GeoPEARL; Tiktak, A., de Nie, D., van der Linden, T., Kruijne, R., 2002. Modelling the leaching and drainage of pesticides in the Netherlands: The GeoPEARL model. Agronomie 22, 373-387; Tiktak, A., van der Linden, A., Boesten, J., 2003. The GeoPEARL model. Model description, applications and manual. RIVM report 716601007) leaching model. The non-linearity of GeoPEARL appeared to produce differences between CI and IC that did not occur in the linear model. The results also suggested that the relevance of either CI or IC is dependent on the available input information. Finally, different ways to estimate the prediction precision of the three approaches are discussed.