Simulation of picloram, atrazine, and simazine leaching through two New Zealand soils and into groundwater using HYDRUS-2D

Two 15 m x 15 m field plots, a Te Awa silt loam and a Twyford fine sandy lo am., located in Hawkes Bay, New Zealand, were applied with bromide, piclora m, atrazine, and simazine. The Te Awa subsoil was a heterogeneous coarse sa nd and sandy gravel, and the Twyford subsoil was a more homogenous fine san dy loam. The underlying aquifers were composed of alluvial gravels at both sites with the water tables generally between 4-5 m below ground level. The sites were monitored for 2.2-3.5 years at approximately monthly intervals using suction cups in the unsaturated zone and monitoring wells in groundwa ter. HYDRUS-2D was used to simulate water movement and solute transport in soil and groundwater in a domain with a depth of 10 m and length of 68 m, i ncluding a 4.5-m unsaturated zone. The model simulated well the general tre nd of field observations for soil water content (theta) and potential (psi( s)), and the values matched better for the soils with less heterogeneity. F or the soils with significant surface cracks, the simulated theta values we re overestimated. On the other hand, for the soil layer perching on top of a less permeable layer, the simulated theta values were underestimated. Simulated pesticide concentrations using the "best available literature val ues" (BALVs) of organic carbon distribution coefficient (K-oc) and half-lif e (T-1/2) were generally lower than those observed. At early times in the t rails, most simulations using BALVs were still within the same order of mag nitude as observed concentrations for the shallow depths. However, at great er depths and later rimes, there were major differences between observed an d simulated concentrations. The model was then calibrated for K-oc and T-1/ 2 values using observed data with an aid of the PEST optimisation package. Despite higher organic contents found in the topsoil, optimised K-oc values for pesticides were consistently lower for the topsoil than for the subsoi l, and were also lower than the BALVs except for picloram, possibly as a re sult of preferential flow in the topsoil. Compared to the BALVs. picloram w as much more persistent but slightly more sorptive in both soil types, and atrazine and simazine were more persistent in the Te Awa soils but less per sistent in the Twyford soils. The optimised K-oc values for all three pesti cides were generally greater, or close to, the BALVs in the subsoil of both sites. HYDRUS-2D provided a reasonably good link for pesticide transport in bath t he unsaturated zone and groundwater. Simulated pesticide concentrations in groundwater using optimised values were generally similar to the observed v alues. Both observed and simulated bromide and pesticide concentrations ind icated that solutes leached more quickly through the soils that were coarse r and more heterogeneous, but were more diluted in the groundwater system t hat was more heterogeneous, conductive, and dispersive. Significant levels of picloram were found in groundwater 22 and 53 m down-gradient of the Twyf ord and Te Awe plot, respectively. Both simulated and observed atrazine and simazine concentrations in groundwater were less than detection limits. (C ) 2000 Elsevier Science B.V. All rights reserved.