5 Combinations of leaching models and hydrological models

As outlined above, leaching models include the description of attenuation processes within soils, while hydrological models partly describe attenuation processes during horizontal transport. Therefore, combinations of both model genera could improve the predictions of pesticide emissions via runoff, but to date only few studies successfully tested such combinations to calculate pesticide loads to or concentrations in surface waters.

Ramanarayanan et al. (2005) applied a combination of GLEAMS, elements of PRZM, and SWAT to provide an estimate of pesticide concentrations within surface waters on a watershed scale. PRZM’s edge-to-field prediction was transformed to a watershed scale using a convection-dispersion equation. GLEAMS was integrated into SWAT to describe the fate of pesticides in the terrestrial environment. This combination differentiates between the soluble and the sorbed phase, and pesticide leaching is calculated for each soil layer. Transformation processes are described by first-order-relationships. Four factors were found to determine residues of pesticides in surface waters: watershed morphology, magnitude of timing of runoff or drainage events, management practices, and degradation rate within the water body. The validation with monitoring data showed good and significant correlation.

Miao et al. (2003*) tested a combination of RICEWQ (Capri and Miao, 2002) and VADOFT to simulate pesticide fate and transport in rice paddies and underlying soils. RICEWQ is a multiple dimension flow model that describes precipitation, irrigation, evapotranspiration, runoff, and seepage. Miao et al. (2003) found a high sensitivity of soil permeability and management practices. The combination was successfully validated by means of a two-year field study.

A combination of LEACHP with AS, an attenuation factor model was tested by Chatupote and Panapitukkul (2005*). LEACHP was used to simulate downward fluxes of pesticides in soils, and AS working on a GIS-platform served to extrapolate simulation results to a catchment in Thailand. The extrapolation showed that efficient measures to reduce pesticide concentrations in surface waters are the reduction of application rates and the optimization of irrigation measures. However, Chatupote and Panapitukkul (2005) did not verify their results.

Tournebize et al. (2006*) combined PCPF-1 (Watanabe and Takagi, 2000), a lumped model simulating pesticide behaviour in paddy water and soil with SWMS-2D (Wu et al., 1995), a finite element numerical model that solves the Richards and the advection-dispersion equation for solute transport in soil. The coupling involves interactions of water flow and concentrations of the soil interface. Monitoring data were used to parameterize and calibrate soil hydrodynamics. The coupling of both models was conducted by linking percolation flux and pesticide concentration at the soil interface. A sensitivity analyses highlighted the impact of daily fluctuating water levels on pesticide concentrations, but again a validation was missing in the study of Tournebize et al. (2006).

Tiktak et al. (2002*) used a model combination which is based on an analytical expression that describes the mass fraction of pesticide leached. To have the seepage and drainage fluxes correctly described, PEARL was loosely coupled with a regional groundwater model. Comparison with a standard scenario showed that the latter was not applicable to the full range of registered pesticides. The metamodel EuroPEARL (Tiktak et al., 2006*), which is a probabilistic application of a model combination further developed, could explain over 90% of the variation of the original model with only four independent spatial attributes, but until present, validation results of EuroPEARL have not been available.

The combination of leaching models with other models that describe runoff seems to be a promising development that started during the last decade. However in some studies, validation is missing, and some combinations are useful only for special land use. Further combinations have to be tested in order to find more general solutions applicable to a wide range of landscapes.