Printer Friendly

Field study of pesticide leaching in an allophanic soil in New Zealand. 2: comparison of simulations from four leaching models.


Computer simulation models are useful for understanding the behaviour of pesticides in the soil and groundwater. They are increasingly being used to predict the transport and fate of agrochemicals and to perform risk assessments of the environmental impacts of agrochemical use. Most of the commonly used models have been assessed against field data (e.g. Leonard et al. 1990; League and Green 1991), although the number of studies where several models have been compared with one set of high quality field data is small (Jarvis et al. 1995). Models are generally developed for a range of purposes and include different levels of complexity for the processes of water movement, adsorption, and persistence of chemicals. Three models of differing complexity were selected and inverse modelling was used to estimate leaching parameters from the observed field data. The 3 models were: GLEAMS (Leonard et al. 1987), which is designed to assess management effects on pesticide movement and uses a tipping bucket approach for water movement; LEACHM (Hutson and Wagenet 1995), which is a mechanistically based model and simulates water movement using the Richards' equation; and HYDRUS-2D (Simunek et al. 1996), which also simulates water movement using the Richards' equation and is a 2-dimensional model. A fourth model, SPASMO (Soil Plant Atmosphere System Model), which is currently under development in New Zealand by Green (2001) and simulates water movement using a water capacity approach that considers the soil to have both mobile and immobile pathways for water and solute movement, was used as an independent test of whether the optimised parameters from one model can be used for another model.

This study is part of a nationwide approach to assess pesticide behaviour in key New Zealand soils under different climatic conditions. It follows 2 trials in the relatively dry Hawkes Bay region on sandy/gravelly soils, which ran from 1993 to 1997 (Close et al. 1999). An experimental site was established in the Waikato region (Close et al. 2003) on an allophanic soil. These soils, with higher topsoil carbon levels, are representative of a large part of the regional soil resource. Allophane is a clay mineral with variable charge, which means that, depending on soil conditions, it can adsorb compounds with negative as well as positive charge. This may influence pesticide adsoq3tion, which is normally considered to be controlled mainly by soil organic matter. A large variability is reported in the literature for the mobility and persistence characteristics of pesticides, although there are some 'best available' values recommended for modelling purposes (Wauchope et al. 1992). In some cases, values derived from other field studies in New Zealand have been found to be quite different to those values, particularly in the subsoil (Close etal. 1999).

The objectives of this study were to (i) monitor the movement of selected pesticides and tracers through a free-draining allophanic soil into underlying groundwater, (ii) determine their mobility and persistence characteristics, and (iii) evaluate the performance of several leaching models of different complexity. A companion paper (Close et al. 2003, this issue) provides relevant background, describes the field study, and presents results of tracer and pesticide monitoring in the soil profile and groundwater. This paper describes the simulation results using 4 different pesticide leaching models and compares the estimates of mobility and degradation with other literature values.


Study description

The experimental site was located about 10 km south o f Hamilton, iii the North Island on a Horotiu silt loam classified as a Typic Orthic Allophanic Soil following the New Zealand Soil Classification (Hewitt 1992). The topsoil contains a significant amount of allophane (up to 12%), which exhibits variable charge characteristics. Two trials were set up at the site. Atrazine, hexazinone, procymidone, and terbuthylazine were applied to Site A (area 15 by 15 m), and 2,4-D, picloram, and triclopyr were applied to Site B (5 by 15 m) iii November 1997. Two tracer compounds, bromide and deuterated water, were applied to both trials. The application rates are given by Close et al. (2003) and were 1-7 times the recommended label rates. The high rates were used to assist in collection of sufficient data for estimation of leaching parameters and evaluation of the performance of the solute transport models. The movement and persistence of the chemicals were monitored for about 2 years, using soil sampling down to a maximum of 1 m and 9 suction cups located between 0.2 and 2.5 m down the soil profile. At Site B, the monitoring was restricted to soil sampling only. The underlying groundwater was also monitored using an array of wells screened from 4 to 6 m below ground level (bgl). The suction cups were sampled by applying a vacuum of 65 kPa for 10-20 h at intervals ranging from 1 to 4 weeks. Soil samples were taken every 3-4 months to a maximum depth of 1 m. Groundwater samples were collected infrequently at the start of the study, then with increasing frequency. A more detailed description of the study site, sampling, and analysis methods is contained in Close etal. (2003).

Model simulations

Simulations of water, bromide, and pesticide movement were carried out using GLEAMS, LEACHM, HYDRUS-2D, and SPASMO. The simulated soil moisture profiles were compared with the observed profiles as a check on model performance. As the soil profile contains a significant amount of allophane, to which bromide can adsorb, the levels of allophane in the profile were used instead of organic carbon for the bromide simulations. Input parameters used for each model are given in Table 1. In all cases, the SPASMO simulations used the median values of the pesticide transport parameters.

GLEAMS is a 'tipping bucket' model that uses the relative difference between field capacity and wilting point to simulate water movement through the soil profile. It allows soil, climate, and farm management practices to be varied. GLEAMS uses one value of [K.sub.oc] for the whole profile with varying organic matter content for each soil layer. Organic carbon values were converted to organic matter content for use in GLEAM S, and the degradation half-life was converted to a first-order degradation rate for comparison with the other models. A more detailed description of GLEAMS can be found in Leonard et al. (1987).

LEACHM (LEACHP is the pesticide module of the LEACHM model) is a mechanistically based model of water and solute movement and pesticide chemistry (Hutson and Wagenet 1995). It uses a finite difference approximation to solve the Richards' equation fur water movement and the convection-dispersion equation for solute transport. The model must be supplied with a function to describe water retention and hydraulic conductivity. LEACHM uses a modification of Campbell's function (Hutson and Cass 1987), where the exponential function is replaced by a parabolic function at high potentials to give a better representation of water retention in real soils. A variety of pesticide transformations and adsorption processes can be simulated. The equilibrium linear isotherm was used for the simulations, because the variability in the observed concentration data did not justify using more complex descriptions of adsorption. The temperature was kept constant at the average soil temperature, and no soil moisture or temperature related adjustments were made to the various rate constants. A more detailed description of LEACHM can be found in Hutson and Wagenet (1995).

HYDRUS-2D is a Windows-based modelling environment for analysis of water flow and solute transport in variably saturated porous media (Simunek et al. 1996). The program solves the Richards' equation for saturated unsaturated water flow, uses a Fickian-based advection-dispersion equation for solute transport, and includes provisions for linear equilibrium adsorption, zero-order production, and first-order degradation. The governing equations are solved using a Galerkin type linear finite clement scheme. As HYDRUS-2D is a 2-dimensional model, it was used to simulate movement of bromide and hexazinone downwards through the soil profile, into and through the groundwater system. For these simulations, the degradation rate for hexazinone in the aquifer was taken as zero, based on the results of a groundwater tracing experiment at the site (Pang and Close 2001).

SPASMO (Green 2001) is a solute transport model currently under development by HortResearch, New Zealand. It was decided to use the SPASMO simulations only with the median optimised leaching parameters derived thorn the other 3 established models, rather than use SPASMO to estimate the leaching parameters by optimisation. The outputs from the SPASMO simulations were compared with the observed data and with the other model simulations as an independent test of the optimised parameters. Water transport in SPASMO is modelled using a water capacity approach (Hutson and Wagenet 1993) that considers the soil to have both mobile and immobile pathways for water and solute transport. The division between mobile and immobile water is set to the soil water content at a pressure potential of -50 kPa. The calculations ale run on a daily time step. SPASMO assumes a linear, equilibrium partitioning of the solute between the soil, water and gas phases, and first order degradation of solutes with standard functions to account for the effects of soil moisture and temperature on the degradation rates. A more detailed and complete description of SPASMO is contained in Green (2001).

Parameter estimation and goodness of fit statistics

The pesticide attenuation parameters ([K.sub.oc] and soil degradation rate or soil half life) were optimised to fit the observed data using s non-linear parameter optimisation package called PEST, Parameter ESTimation (Doherty 1994). PEST is a model-independent non-linear parameter estimator, characterised by a powerful optimisation algorithm and an ability to interface with any model through that model's own input and output files, thus requiring no alterations to the model. PEST uses a residual sum of squares procedure to optimise the fit to the observed data. One degradation rate was used for the topsoil horizon (0-0.2 m) and a second degradation rate was used for the remainder of the profile.

There were 114 and 106 soil water observations, and 51 and 43 bulk soil samples for bromide and pesticides at Site A, respectively. At Site B, there were 51 and 43 bulk soil samples for bromide and pesticides, respectively. The differences in the numbers of soil water samples arise because the pesticides had an additional soil sampling to a depth of 0.2 m early in the study and bromide had an extra soil sampling to a depth of 1.0 m at the end of the study. In addition, some suction cups sometimes collected low volumes and could not be analysed for everything on all occasions. The parameter optimisations for Site A were carried out using 3 combinations of field observations: just the soil water data; just the soil sample data; and the combined water/soil data. The simulated solute concentrations using each set of these optimised parameters were compared with both the observed soil sample and soil water data. HYDRUS-2D could only be used to optimise parameters for the soil water data because the output solute concentrations are not expressed in mg/kg. These units have to be subsequently calculated and depend on the [K.sub.oc] value, thus it was not possible to optimise the leaching parameters using the soil sample data and the PEST program.

The following 4 goodness of fit statistics were selected for evaluation of the parameter estimates and the leaching models:

Residual sum of squares (SSres) = [summation of][([S.sub.i] - [O.sub.i]).sup.2]

Coefficient of residual mass (CRM) = [summation of]([O.sub.i] - [S.sub.i])/[summation of]([O.sub.i])

Modelling efficiency (EF) = [summation of][([O.sub.i] - [O.sub.m]).sup.2] - [summation of][([S.sub.i] - Om).sup.2]/ [summation of][([O.sub.i] - [O.sub.m]).sup.2]

Coefficient of determination ([r.sup.2]) = [[summation of]([O.sub.i] - [O.sub.m])([S.sub.i] - [S.sub.m])]].sup.2]/ [summation of][([O.sub.i] - [O.sub.m]).sup.2][summation of]([S.sub.i] - [S.sub.m]).sup.2]

where [S.sub.i] and [O.sub.i] are the simulated and observed values for sample i; [O.sub.m] and [S.sub.m] are the means of the observed and simulated data. The lower limit for SSres and [r.sup.2] is zero and the upper limit for EF and [r.sup.2] is one. Both EF and CRM can become negative. If EF is negative then the model-predicted values are worse than simply using the observed mean (Loague and Green 1991). The ideal values for SSres, CRM, EF, and [r.sup.2] would be 0.0, 0.0, 1.0, and 1.0, respectively.

Results and discussion

Simulation of soil moisture

A comparison of observed and simulated soil moisture levels is shown in Fig. 1. The observed data were a combination of gravimetric samples (taken at the same time as the soil samples) and time domain reflectometry (TDR) measurements. There were fewer observations below 0.5 m depth because only the shallower depths were sampled at the start of the study and because the TDR was only installed down to 0.5 m. There was reasonable agreement between the observed and simulated soil moistures for LEACHM and HYDRUS-2D down to 0.4 m. GLEAMS had a consistently lower value for the first sampling prior to the application date, but thereafter produced a reasonable average agreement with the observed data. Below 0.2 m, there was little variation in the soil moisture levels simulated by GLEAMS. This reflects the capacity type model of water movement combined with the fact that there was no vegetation on the plot after the application of the pesticides.


There is some variability between the observed and simulated soil moisture for depths between 0.4 and 0.6 m (Fig. 1). LEACHM and GLEAMS simulated values are consistently lower than the HYDRUS-2D values and the observed soil moisture oscillates between the 2 levels of simulated data. One reason for a difference in the simulated values could be the different way each model represents the soil depths and horizons. HYDRUS-2D uses the exact depth of each horizon, LEACHM uses fixed depth intervals so that some interpolation is required, and GLEAMS has a maximum of 5 soil layers so some averaging across horizons is needed. The oscillation of the observed soil moisture data (gravimetric samples al these depths) probably reflects the random spatial nature of the sample collection combined with the variation observed in the depth of some of the lower soil horizons over the site (Close et al. 2003). Although the depth of the A horizon was fairly consistent (0.18-0.20 m), and there was always at least 0.12 m of Bwl horizon, there was significant variation in the depth at which the BC horizon started (0.31-0.74 m) with some profiles having no Bw2 and/or Bw3 horizon. Thus, there would be spatial variation in the physical and hydraulic properties of the soil at these depths (0.3-0.6 m), which could be captured by the random spatial gravimetric samples as illustrated in Fig. 1.

Estimation of persistence for atrazine, terbuthylazine, and 2, 4-D

Atrazine, terbuthylazine, and 2,4-D disappeared much more quickly from the trial sites than the other pesticides. Unfortunately, the soil sampling frequency did not provide us with sufficient data to adequately characterise their leaching parameters using inverse modelling procedures with the solute transport models. There were only a few detections of these pesticides in the suction cup samplers. Atrazine was only detected on one occasion in the 0.2 m suction cups and was not detected in either of the 0.4 m suction cups. Terbuthylazine was not detected in the 0.2 m cups and was only detected once in the 0.4 m suction cups. One of the atrazine degradation products, desethyl atrazine, was detected in the suction cups down to 2.5 m but the analysis of the degradation products is beyond the scope of this paper.

Degradation rates for these pesticides can be estimated using mass recoveries from the soil sample data and assuming a simple first-order degradation. The soil on Day 43 was only sampled to a depth of 0.15 in, and so mass recoveries for several of the pesticides were actually greater on the next sampling occasion (Day 149) when the soil was sampled to a greater depth (Close et al. 2003). Data for Day 149 were used to estimate the degradation rates for atrazine, terbuthylazine, and 2,4-D. The mobility of these pesticides could not be estimated.

The first-order degradation rates (per day), with halt-lives in parentheses, were 0.028 (25 days), 0.021 (34 days), and 0.022 (32 days) for atrazine, terbuthylazine, and 2,4-D, respectively. The mass recoveries for these pesticides at 149 days after application was quite low ranging from 1.6 to 4.7%; so there is quite a bit of uncertainty regarding these estimates. It is also possible that some of the 2,4-D was lost from the soil surface due to volatilisation, and this could introduce further uncertainty regarding the initial mass inputs to the soil. The same procedure was used to calculate the degradation rate for procymidone and this was compared with the optimised degradation rate from LEACHM and GLEAMS simulations using the complete soil dataset. Procymidone was selected because it was one of the least mobile and would be least affected by the shallower soil sampling depths at the start of the study. The degradation rate (per day) for procymidone, as estimated from mass recovery, was 0.0029 (235 days) compared with the 'optimised' GLEAMS estimate of 0.0027 (260 days) and the corresponding LEACHM estimate of 0.0038 (184 days). This indicates good agreement between the parameter estimates.

Unless otherwise stated all literature values for pesticide mobility and degradation have been taken from the United States ARS Pesticide Properties database ( as at February 2002). For the purposes of modelling, the 'best available' degradation half-lives are 60 days for atrazine and terbuthylazine, and 6 days for 2,4-D, although there is a wide range in values for atrazine (15 330 days) and for 2,4-D (1-9 days). Our results indicate that atrazine and terbuthylazine are less persistent and 2,4-D is more persistent on a Horotiu soil than would be expected based on 'best available' values of degradation half-life.

Simulation results for chemicals

A comparison of the model simulated solute concentrations in soil and soil water with the observed data is shown in Figs 2-8. Comparisons from both the suction cups and soil samples are given for bromide (Figs 2 and 3) and for hexazinone (Figs 4 and 5). In the case of procymidone, only the soil sample results are shown (Fig. 6) as procymidone was detected mainly in the top 0.1-0.3 m and there were only a few low-level detections from the suction cups lower in the profile. For pesticides at Site B, picloram (Fig. 7) and triclopyr (Fig. 8), there were only soil sampling data and they are presented as log-concentration plots because it is easier to see both the high and low values. However, caution needs to be exercised in interpreting the log-plots because any discrepancies at the lower values appear to be magnified on a log scale. The estimates of the mobility and degradation parameters, together with the goodness of fit statistics for bromide and all the pesticides are given in Table 2.



The median [K.sub.oc] value for bromide was 5.2 mL/g indicating slight adsorption to the soil medium. As stated in the Methods section, the [K.sub.oc] values for bromide were normalised with respect to allophane instead of organic carbon. The only model to indicate no adsorption was LEACHM, based on the soil data only. However, the [K.sub.oc] value from this optimisation resulted in poor agreement with the soil water observations (Table 2). GLEAMS was the best model for bromide and all the goodness of fit statistics showed a similar pattern regardless of which set of optimisation data or observation data was used. Similar [K.sub.oc] values were obtained for both Site A and Site B. The slight adsorption of bromide is consistent with the retardation observed with respect to the deuterated water tracer (Close et al. 2003) and would result from the high allophane content (up to 12%) in the top 0.5 m of the sod profile. Figure 2 shows that all simulations of bromide soil water concentrations were generally good with the exception of LEACHM, where the bromide did not leach as rapidly and the arrival of the bromide at depths below 1 m was delayed compared with the observed data. In the case of the soil bromide concentrations (Fig. 3), there was good agreement up until 300 days, but thereafter the LEACHM simulations predicted more bromide in the profile than was observed. The simulations for the bromide soil data from SPASMO were also over-estimated, although the log-scale for Fig. 3 does tend to emphasise any discrepancies at low concentrations.


The median [K.sub.oc] value for hexazinone was 23.3 mL/g with all estimates being between 12 and 30 mL/g. These are lower than the literature value of 40 mL/g (range 34-74). The median degradation half-life was 185 days in the topsoil and 259 days in the subsoil. These values are higher than the literature value of 88 days (range 27-218). This means that it is both more mobile and more persistent than would be expected from the literature values. One of the main uses for hexazinone is in forestry and there are large areas of forest located on the allophanie soils in the central North Island. In light of our results, there may need to be an assessment of the likely impacts of the usage of hexazinone. Fortunately, there is often little usage of groundwater in these areas. The goodness of fit statistics indicated that all the models and combinations of optimisation data gave good fits to the observed data with the exception of the parameters optimised with just the soil data compared with the soil water observations (Table 2).


The median [K.sub.oc] value for procymidone was 352 mL/g with a range in estimates from 170 to 668 mL/g (Table 2). These are much lower than the best available value from the literature of 1500 mL/g. The median estimate of degradation half-life was 184 days in the topsoil, with estimates ranging from 28 to 436 days. This degradation rate is much slower than expected from the best available literature value of 15 days (range 7-120). Most models indicated little degradation in the subsoil but there was a lot of variability in the estimates and there was probably insufficient data at depth for these estimates to be reliable. As procymidone was more strongly adsorbed than hexazinone and bromide, the soil water concentrations were much lower and often close to the analytical detection limit. The water optimisation data resulted in both the maximum and minimum estimates for both [K.sub.oc] and half-life and the estimates from the soil data were more consistent. The goodness of fit statistics indicated reasonable results for the soil observation data with the soil or combined optimisation data. The best fit obtained was for the LEACHM model using the soil optimisation data. In all cases, there was good agreement between the observed and simulated soil concentrations for procymidone (Fig. 6). The persistent nature and relatively low mobility of procymidone is shown by the higher levels of pesticide in the top 0.2 m towards the end of the study (Fig. 6), compared with hexazinone (Fig. 5).

The discrepancy between our optimised parameters and the literature values is quite significant. However, when the literature references for procymidone were further investigated, both [K.sub.oc] and half-life values were found to be derived from regression equations using compound solubility and a correction for crystal energy. In the introduction to the 1996 version of the ARS database, Hornsby et al. (1996) indicated that these sort of regression equations should predict the majority of [K.sub.oc] values to within a factor of 3 and nearly all [K.sub.oc] values within a factor of 10. The median optimised [K.sub.oc] value differs from the 'best available' value by a factor of 4.3. The degradation data in the literature is fairly sparse with one reference (Cabras et al. 1987) giving a half-life of 120 days based on 4 soils. The other reference (The Pesticide Manual 1994) indicated persistence of procymidone in soil for about 28-84 days. This translates to a degradation half life of between 7 and 21 days, assuming persistence equals 4 half-lives (Hornsby et al. 1996). It is not clear why a best available value of 14 days was chosen for the ARS database but the upper range of literature values are closer to the values found in this study. The much greater mobility and persistence of proeymidone indicate a much greater likelihood for leaching. This is consistent with detections of procymidone of up to 3 [micro]g/L in groundwater from around this region (Hadfield and Smith 1999). The mobility and degradation of procymidone in a range of New Zealand soils, not including an allophanic soil, was measured in the laboratory by McNaughton et al. (1999). They found a [K.sub.oc] value of 580 mL/g mad a degradation half-life of 34 days (s.d. 23 days). These values are much closer to the field-derived values for procymidone.


At Site B, only soil samples were collected so there was only one set of optimisation data for parameters. The mean [K.sub.oc] value for picloram was 138 mL/g with a range fur the GLEAMS and LEACHM models of 103-174 mL/g. This is much higher than the 'best available' value from the literature of 29 mL/g and well above the maximum value given in the ARS database (range 7-48). The mean degradation half life in the topsoil was 218 days, which is much higher than the literature value of 90 days, although still within the literature range of 20-277 days. There was wide variation in the degradation half-lives in the subsoil for both picloram and triclopyr, and the estimates are probably not reliable. The good agreement between the observed and simulated data for all models can be seen in Fig. 7. Picloram had the highest average persistence in the topsoil of all the pesticides in the study and relatively high levels remained in the soil profile at the end of the study (Fig. 7).


The mean [K.sub.oc] value fur triclopyr was 297 mL/g with a range for the GLEAMS and LEACHM models between 108 and 485 mL/g. This is much higher than the 'best available' value from the literature of 68 mL/g and above the literature range of 12-160. The mean degradation half-life in the topsoil was 183 days, which is much higher than the literature value of 32 days (range: 8-69). The levels of triclopyr decreased rapidly with depth (Fig. 8), with no triclopyr being detected below 0.3 m throughout the study. The simulations, particularly from LEACHM and SPASMO, tended to show a more gradual reduction in concentration with depth.

Comparison with parameter estimates from the earlier Hawkes Bay studies

There were only 2 pesticides, atrazine and picloram, that were also applied in the earlier studies carried out on 2 soils in Hawkes Bay (Close et al. 1999). Only the degradation half-life for atrazine of 25 days could be estimated based on the mass recovery in the soil samples in this present study on the Horotiu silt loam. This compares with degradation half lives of 20-28 days for the Twyford soil and 60-126 days for the Te Awa soil (Close et al. 1999; Pang et al. 2000), indicating similar rates of degradation for atrazine in the Twyford soil but much greater persistence in the "re Awa soil. Picloram was much more mobile in the Twyford soil ([K.sub.oc] = 45-47 mL/g) and the Te Awa soil ([K.sub.oc] = 19-30 mL/g) than in the Horotiu soil ([K.sub.oc] = 103-174 mL/g). This is consistent with the greater adsorption of picloram to the allophane in the Horotiu soil. Picloram was more persistent in the Twyford soil ([T.sub.1/2] = 364-480 days) and the Te Awa soil ([T.sub.1/2] = 200-1268 days) compared with a [T.sub.1/2] = 204-235 days estimated for the Horotiu soil. Values of degradation half-life for all 3 soils were much higher than the 'best available' literature value of 90 days.

SPASMO simulations

The goodness of fit parameters for the SPASMO simulations, together with the best and worst values for the other models from Table 2, are given in Table 3. The results in this table, together with Figs 2-8, indicate that SPASMO simulated the observed data fairly well and was comparable to established solute transport models. In these comparisons SPASMO tended to give better results for the soil solution simulations, whereas it tended to overestimate soil concentrations compared with the observed data. This can be seen clearly for the hexazinone data (Table 3, Figs 4 and 5), where there is very good agreement for the soil solution data, but the soil sample data is significantly overestimated for nearly every data point. It should be noted that the leaching parameters were not optimised for SPASMO, but were the median optimised values from Table 2. There is a reasonable range in the optimised parameter values in Table 2 for the 3 models. The comparison results for SPASMO would have improved if the parameters had been optimised for this model, but that was not the intended use. Rather, SPASMO has been used as an independent check of the fitted parameters. SPASMO used only one value for degradation (the median optimised topsoil value) and allowed it to vary down the profile in response to changes in soil moisture and temperature, whereas the other 3 models used 2 constant degradation rates, one for the topsoil and one for the subsoil.

Comparison of models

GLEAMS was the best performed for the bromide data, both soil samples and soil water data, from Site A (Table 2), although LEACHM performed slightly better than GLEAMS for the bromide data for Site B (soil samples only). HYDRUS-2D only had goodness of fit statistics for the soil water data and performed slightly worse than GLEAMS and slightly better than LEACHM for bromide simulations at Site A. All the models had similar very good performances for hexazinone, while LEACHM gave the best results for procymidone. Picloram and triclopyr were only simulated by GLEAMS and LEACHM and both models provided very good results. SPASMO simulations were carried out on a different basis so direct comparisons are not valid and the SPASMO simulations are discussed separately.

In terms of data requirements, GLEAMS has the lowest demand for input data, whereas both LEACHM and HYDRUS-2D are more complex and require more data, particularly for charaeterisation of the water retention curve. The computational demands are low for GLEAMS, moderate for LEACHM, and high for HYDRUS-2D. All 3 models require a degree of expertise with modelling and unsaturated zone processes, with HYDRUS-2D having the greatest requirements. However, HYDRUS-2D does have the advantage of being able to simulate the linkage between the unsaturated zone and the groundwater system.

General discussion

The study provides a good opportunity to assess the relative benefits and limitations of soil sampling compared with suction cup samples of soil water. In addition to our previous comments (Close et al. 2003), the simulations carried out for the soil water illustrate some of the issues related to suction cup sampling, at a fixed location over time, and soil samples which are taken at random spatial locations within the study area. The suction cups give a consistent trine series at the location of each suction cup, but there can still be unexpected variations between suction cups. For example, Fig. 4 shows the peak observed concentrations of hexazinone being higher in the suction cups at 0.6 m compared with those at 0.4 m, which are also higher than those at 0.2 m. This would mainly be a result of spatial variability in the soil properties, and resulting leaching patterns, probably combined with the timing of sampling, where peak leachate concentrations may not have been sampled by some of the suction cups. Variability in soil sampling can be seen for both bromide (Fig. 3) and triclopyr (Fig. 8), where the total mass in the profile is greater on Day 503 compared with Day 388 (see also Table 3 in Close el al. 2003). The variation that can occur with the random spatial pattern of soil sampling is also illustrated for the soil moisture data in Fig. 1 for depths between 0.4 0.6 m.

Comparisons between the optimised leaching parameters obtained from either suction cup or soil sample data can only be made for Br, hexazinone, and procymidone from Site A. Bromide was the most mobile and both the suction cup and soil sample data gave similar results, especially for the GLEAMS model which gave the best results. Hexazinone was moderately mobile and persistent, and there was reasonable agreement between the models for estimated values of [K.sub.oc], with the soil sample data giving slightly lower estimates of [K.sub.oc] (Table 2). Most of the optimised degradation rates for hexazinone were similar, except for one low rate estimated by GLEAMS based on the soil data alone. Procymidone was much less mobile and therefore the soil sample data were much better than the suction cup data for the purpose of parameter estimation, because the top of the soil profile, where procymidone was in higher concentrations, was better represented. The estimates for the procymidone leaching parameters were quite variable. The goodness of fit statistics indicated that the best estimates were derived from the soil sample data. The soil sample data were best at predicting the observed soil sample data, and likewise for the suction cup data, as would be expected. The difference in cross-predictions became greater as the mobility of the pesticide decreased.

All the goodness of fit statistics generally provided consistent rankings regarding the model and optimisation data that gave the best fit to the observed data (Table 2). The [r.sup.2] statistic, while easy to interpret, provided less discrimination between the various models and datasets. This is probably because it is bounded between 0 and 1, whereas the other statistics are only bounded at either the maximum or minimum level. The consistency between the statistics gave a greater degree of confidence in selecting the model with the best fit.

The pesticides in the study ranged from weak bases (atrazine, terbuthylazine, and hexazinone) to weak acids (2,4-D, triclopyr, and picloram), with one non-ionic pesticide being procymidone (Weber 1994). Allophane is expected to adsorb weak acids slightly more than would be expected from the organic matter content alone. Triclopyr and picloram both had greater adsorption than would be expected on the basis of organic carbon content, with [K.sub.oc] values being 1.6 7.1 times higher than 'best available' literature values. Three of the 4 optimised [K.sub.oc] values were above the maximum values reported in the literature (Table 2 of Close et al. 2003). This is consistent with additional adsorption of pesticide to the allophane. It is probable that the allophane also resulted in repellancy towards hexazinone, a weak base, as the optimised [K.sub.oc] values were all lower than the lowest literature values. This would result in much greater mobility than would be predicted for hexazinone, and by implication, other weakly basic pesticides, such as atrazine and terbuthylazine, in allophanic soils.

The linkage between the soil profile and the groundwater could be simulated using the 2-dimensional model HYDRUS-2D, which is also able to simulate transport through variably saturated porous media. Observed and simulated groundwater concentrations for bromide and hecazinone are shown for 3 monitoring wells (Fig. 9). The horizontal distances are with respect to the centre of the plot, so the well at 15 m is about 7.5 m from the edge of the plot. The large peak at around 670 days corresponds to an intensive irrigation period near the end of the study when about 156 mm was applied over 5 days. The peaks around 360-400 days correspond to winter recharge events (Close et al. 2003). HYDRUS-2D simulates the position of the peak around 670 days very well. Two different simulations are shown for hexazinone. The first simulation uses the degradation rate derived for the subsoil frown the suction cup data (0.2-2.5 m) for the remainder of the unsaturated zone as well and the second simulation assumes zero degradation from 1.6 to 4.5 m. The groundwater hexazinone concentrations are well simulated by assuming zero degradation for the lower part of the unsaturated zone, indicating that this is the more likely situation. The groundwater concentrations are well simulated for bromide around 670 days. The fit of the simulations from a 2-dimensional model to a 3-dimensional groundwater system would have been assisted by 2 factors that would have reduced effective mixing in the groundwater. The aquifer layer was very thin at the time of the intensive recharge experiment (around 1 m) and the size of the source area (15 by 15 m) was large relative to the distance to the down-gradient wells (maximum distance = 12.5 m from edge of plot). HYDRUS-2D did predict increases in groundwater concentrations for bromide and hexazinone at 300 and 400 days, but the simulated increases were much lower than the observed concentrations, and cannot be seen in Fig. 9. The excellent simulation results of bromide and hexazinone groundwater concentrations at around 700 days give some confidence in predicting groundwater concentrations of tracer and pesticides, particularly following a large pulse of recharge. However, the failure to simulate the peak values around 300-400 days (factor of around 40) with lower amounts of recharge indicates that there are still important processes occurring (possibly macropore flow) that the models are not able to identify.



In several instances there were marked differences in the degradation and mobility parameters recorded for the pesticides used in this study compared with those reported in the literature. This was particularly the case for procymidone where the field-derived values from the Horotiu silt loam indicated that procymidone was much more mobile and persistent than the 'best available' literature values. Some differences in mobility in this study could be associated with the allophanic nature of the soil. The weak acids, picloram and triclopyr, were less mobile and the weak base, hexazinone, was more mobile, which is consistent with the effect of allophane in the soil. Differences such as those discussed highlight the issues associated with assigning a single persistence or mobility value to a pesticide irrespective of the type of soil involved and provides an inbuilt error into any model simulation involving different soils. The simulations carried out using either the suction cup or soil sample data for Site A indicated that both the suction cup and soil sample data gave similar results for the mobile, persistent compounds, bromide, and hexazinone. Procymidone was much less mobile and the soil sample data were much better than the suction cup data for the purpose of parameter estimation. This is because the suction cups were all located below 0.4 m in the profile, and there were little detections in the suction cups, whereas the soil sampling covered the top of the soil profile, where procymidone was in higher concentrations, much better. HYDRUS-2D was used to simulate the linkage between the unsaturated zone and the groundwater and simulation of bromide and hexazinone levels in 3 monitoring gave a high goodness of fit to observed data following a large recharge pulse. Observed concentrations following normal winter recharge events were significantly under-predicted by HYDRUS-2D.
Table 1. Soil and chemical properties for model simulations for the
Horotiu silt loam

GLS, gravelly loamy sand; properties for horizon C were used for the
unsaturated zone from 1.56 to 4.5 m. BD, bulk density; OC, organic
carbon; a, b: Campbell's coefficients for water retention curve;
[K.sub.sat], saturated hydraulic conductivity; RETC-estimated hydraulic
parameters: [[theta].sub.r], residual water content; [[theta].sub.s],
saturated water content; [alpha], n, coefficients in water retention

 OC Allophane
Horizon Depth(m) Lithology (%) (%)

Ap 0-0.20 Loamy silt 8.00 10
Bw1 0.20-0.40 Loamy silt 1.88 12
Bw2 0.40-0.53 Loamy silt 0.93 12
Bw3 0.53-0.67 Loamy silt 0.43 5
BC1 0.67-1.05 Loamy sand 0.13 2
BC2 1.05-1.28 Slightly cemented 0.10 2
BC3 1.28-1.56 GLS 0.07 1
C >1.56 Gravelly sand 0.08
Aquifer (C) 4.5-10 Coarse sand 0.11 0.8

 BD (g/ Porosi- Field Wilting
Horizon [cm.sup.3]) ty (A) capacity (A) point (A)

Ap 0.88 0.63 0.54 0.22
Bw1 0.85 0.67 0.36 0.22
Bw2 1.03 0.61 0.29 0.18
Bw3 1.11 0.60 0.21 0.14
BC1 1.11 0.60 0.21 0.14
BC2 1.03 0.60 0.21 0.14
BC3 1.03 0.60 0.21 0.14
C 1.46 0.60 0.21 0.14
Aquifer (C) 1.02 0.30

 a (B) b (B) sub.r] (C) [[theta].sub.s] (C)

Ap -4.19 6.94 0.000 0.590
Bw1 -0.163 6.87 0.225 0.634
Bw2 -0.0496 7.383 0.236 0.663
Bw3 -0.0028 8.245 0.243 0.642
BC1 -0.0028 8.245 0.195 0.438
BC2 -0.316 9.443 0.140 0.441
BC3 -0.316 9.443 0.140 0.441
C -0.018 5.125 0.140 0.441
Aquifer (C) 0.000 0.300

 [alpha] (C) n (C) [K.sub.sat] (C)
Horizon (1/m) (m/day)

Ap 0.982 1.178 0.199
Bw1 7.821 1.330 5.14
Bw2 11.443 1.300 8.42
Bw3 12.226 1.279 14.69
BC1 36.621 3.013 40.78
BC2 46.605 3.215 68.64
BC3 46.605 3.215 68.64
C 46.605 3.215 68.64
Aquifer (C) 62.86 3.485 131.0

(A) Used for GLEAMS simulations.

(B) Used for LEACHM simulations.

(C) Used for HYDRUS-2D simulations.

Table 2. Summary of [K.sub.oc] and soil degradation parameters

CRM, coefficient of residual mass; EF, modelling efficiency; Dr1,
degradation rate for topsoil (0-20 cm); Dr2, degradation rate for
subsoil (<20 cm); [T.sub.1/2], corresponding degradation half-life;
W & S, using combined water and soil data for optimisation

Chemical Model Data [K.sub.oc] Dr1
 (mL/g) ([day.sup.-1])

Site A

Bromide LEACHM W & S 5.3
Bromide GLEAMS W & S 5.1
Bromide LEACHM Water 7.2
Bromide GLEAMS Water 5.2
Bromide HYDRUS Water 3.6
Bromide LEACHM Soil 0.0
Bromide GLEAMS Soil 5.3
 Mean 4.5
 Median 5.2
Hexazinone LEACHM W & S 26.0 0.00550
Hexazinone GLEAMS W & S 23.3 0.00354
Hexazinone LEACHM Water 21.0 0.00760
Hexazinone GLEAMS Water 30.2 0.00352
Hexazinone HYDRUS Water 25.6 0.00374
Hexazinone LEACHM Soil 14.5 0.00410
Hexazinone GLEAMS Soil 12.0 0.00085
 Mean 21.8 0.00412
 Median 23.3 0.00374
Procymidone LEACHM W & S 352 0.00805
Procymidone GLEAMS W & S 200 0.00281
Procymidone LEACHM Water 668 0.01050
Procymidone GLEAMS Water 170 0.02500
Procymidone HYDRUS Water 667 0.00159
Procymidone LEACHM Soil 360 0.00376
Procymidone GLEAMS Soil 178 0.00267
 Mean 371 0.00777
 Median 352 0.00376

Site B

Bromide LEACHM Soil 1.00E-10
Bromide GLEAMS Soil 5.17
 Mean 2.59
Picloram LEACHM Soil 174 0.00340
Picloram GLEAMS Soil 103 0.00295
 Mean 138 0.00318
Triclopyr LEACHM Soil 485 0.00440
Triclopyr GLEAMS Soil 108 0.00316
 Mean 297 0.00378

Chemical Model Data [T.sub.1/2] Dr2
 (days) ([day.sup.-1])

Site A

Bromide LEACHM W & S
Bromide GLEAMS W & S
Bromide LEACHM Water
Bromide GLEAMS Water
Bromide HYDRUS Water
Bromide LEACHM Soil
Bromide GLEAMS Soil
Hexazinone LEACHM W & S 126 1.00E-10
Hexazinone GLEAMS W & S 196 4.12E-03
Hexazinone LEACHM Water 91 2.31E-05
Hexazinone GLEAMS Water 197 3.44E-03
Hexazinone HYDRUS Water 185 4.03E-03
Hexazinone LEACHM Soil 169 2.31E-05
Hexazinone GLEAMS Soil 815 2.68E-03
 Mean 168 0.00205
 Median 185 0.00268
Procymidone LEACHM W & S 86 1.00E-10
Procymidone GLEAMS W & S 247 3.80E-02
Procymidone LEACHM Water 66 1.00E-10
Procymidone GLEAMS Water 28 3.62E-05
Procymidone HYDRUS Water 436 1.00E-10
Procymidone LEACHM Soil 184 1.00E-10
Procymidone GLEAMS Soil 260 5.70E-03
 Mean 89 0.00625
 Median 184 1.00E-10

Site B

Bromide LEACHM Soil
Bromide GLEAMS Soil
Picloram LEACHM Soil 204 1.00E-10
Picloram GLEAMS Soil 235 2.48E-03
 Mean 218 0.00124
Triclopyr LEACHM Soil 158 9.00E-06
Triclopyr GLEAMS Soil 219 1.93E-01
 Mean 183 0.09650

Chemical Model Data [T.sub.1/2]

Site A

Bromide LEACHM W & S
Bromide GLEAMS W & S
Bromide LEACHM Water
Bromide GLEAMS Water
Bromide HYDRUS Water
Bromide LEACHM Soil
Bromide GLEAMS Soil
Hexazinone LEACHM W & S 6.9E+09
Hexazinone GLEAMS W & S 1.7E+02
Hexazinone LEACHM Water 3.0E+04
Hexazinone GLEAMS Water 2.0E+02
Hexazinone HYDRUS Water 1.7E+02
Hexazinone LEACHM Soil 3.0E+04
Hexazinone GLEAMS Soil 2.6E+02
 Mean 339
 Median 259
Procymidone LEACHM W & S 6.9E+09
Procymidone GLEAMS W & S 1.8E+01
Procymidone LEACHM Water 6.9E+09
Procymidone GLEAMS Water 1.9E+04
Procymidone HYDRUS Water 6.9E+09
Procymidone LEACHM Soil 6.9E+09
Procymidone GLEAMS Soil 1.2E+02
 Mean 111
 Median 6.9E+09

Site B

Bromide LEACHM Soil
Bromide GLEAMS Soil
Picloram LEACHM Soil 6.9E+09
Picloram GLEAMS Soil 2.8E+03
 Mean 559
Triclopyr LEACHM Soil 7.7E+04
Triclopyr GLEAMS Soil 3.6E+00
 Mean 7


Chemical Model Data Water Soil

Site A

Bromide LEACHM W & S 55200 23900
Bromide GLEAMS W & S 30900 3172
Bromide LEACHM Water 50700 32800
Bromide GLEAMS Water 30900 3729
Bromide HYDRUS Water 40400
Bromide LEACHM Soil 239200 5680
Bromide GLEAMS Soil 30900 3745
Hexazinone LEACHM W & S 3.8 4.4
Hexazinone GLEAMS W & S 3.3 6.5
Hexazinone LEACHM Water 3.5 6.7
Hexazinone GLEAMS Water 2.9 7.7
Hexazinone HYDRUS Water 3.3
Hexazinone LEACHM Soil 15.0 3.5
Hexazinone GLEAMS Soil 26.7 4.0
Procymidone LEACHM W & S 0.00147 21.8
Procymidone GLEAMS W & S 0.05600 17.7
Procymidone LEACHM Water 0.00088 31.6
Procymidone GLEAMS Water 0.00089 93.0
Procymidone HYDRUS Water 0.00083
Procymidone LEACHM Soil 0.00611 5.0
Procymidone GLEAMS Soil 0.90600 17.6

Site B

Bromide LEACHM Soil 1876
Bromide GLEAMS Soil 4047
Picloram LEACHM Soil 3.5
Picloram GLEAMS Soil 3.9
Triclopyr LEACHM Soil 7.5
Triclopyr GLEAMS Soil 11.6


Chemical Model Data Water Soil

Site A

Bromide LEACHM W & S -0.40 -2.07
Bromide GLEAMS W & S -0.07 0.55
Bromide LEACHM Water -0.13 -2.38
Bromide GLEAMS Water -0.06 0.26
Bromide HYDRUS Water -0.14
Bromide LEACHM Soil -1.52 -0.33
Bromide GLEAMS Soil -0.06 0.24
Hexazinone LEACHM W & S 0.04 0.06
Hexazinone GLEAMS W & S 0.05 0.30
Hexazinone LEACHM Water 0.20 0.29
Hexazinone GLEAMS Water 0.13 0.22
Hexazinone HYDRUS Water 0.19
Hexazinone LEACHM Soil -0.96 -0.13
Hexazinone GLEAMS Soil -1.29 0.34
Procymidone LEACHM W & S 0.35 0.52
Procymidone GLEAMS W & S -4.45 0.15
Procymidone LEACHM Water 0.89 0.63
Procymidone GLEAMS Water 0.95 0.92
Procymidone HYDRUS Water 0.86
Procymidone LEACHM Soil -1.00 0.10
Procymidone GLEAMS Soil -7.27 0.10

Site B

Bromide LEACHM Soil -0.19
Bromide GLEAMS Soil 0.31
Picloram LEACHM Soil 0.60
Picloram GLEAMS Soil 0.10
Triclopyr LEACHM Soil -0.05
Triclopyr GLEAMS Soil -0.06


Chemical Model Data Water Soil

Site A

Bromide LEACHM W & S 0.20 -2.19
Bromide GLEAMS W & S 0.55 0.58
Bromide LEACHM Water 0.26 -3.38
Bromide GLEAMS Water 0.55 0.50
Bromide HYDRUS Water 0.41
Bromide LEACHM Soil -2.48 0.24
Bromide GLEAMS Soil 0.55 0.50
Hexazinone LEACHM W & S -0.04 0.90
Hexazinone GLEAMS W & S 0.12 0.85
Hexazinone LEACHM Water 0.07 0.85
Hexazinone GLEAMS Water 0.21 0.83
Hexazinone HYDRUS Water 0.11
Hexazinone LEACHM Soil -3.06 0.93
Hexazinone GLEAMS Soil -6.22 0.91
Procymidone LEACHM W & S -1.51 0.79
Procymidone GLEAMS W & S -95.09 0.83
Procymidone LEACHM Water -0.50 0.69
Procymidone GLEAMS Water -0.52 0.10
Procymidone HYDRUS Water -0.40
Procymidone LEACHM Soil -9.43 0.95
Procymidone GLEAMS Soil -153.65 0.83

Site B

Bromide LEACHM Soil 0.81
Bromide GLEAMS Soil 0.59
Picloram LEACHM Soil 0.96
Picloram GLEAMS Soil 0.95
Triclopyr LEACHM Soil 0.96
Triclopyr GLEAMS Soil 0.94


Chemical Model Data Water Soil

Site A

Bromide LEACHM W & S 0.40 0.13
Bromide GLEAMS W & S 0.55 0.61
Bromide LEACHM Water 0.34 0.06
Bromide GLEAMS Water 0.55 0.53
Bromide HYDRUS Water 0.42
Bromide LEACHM Soil 0.42 0.56
Bromide GLEAMS Soil 0.55 0.53
Hexazinone LEACHM W & S 0.19 0.91
Hexazinone GLEAMS W & S 0.28 0.89
Hexazinone LEACHM Water 0.22 0.87
Hexazinone GLEAMS Water 0.28 0.88
Hexazinone HYDRUS Water 0.23
Hexazinone LEACHM Soil 0.16 0.93
Hexazinone GLEAMS Soil 0.21 0.94
Procymidone LEACHM W & S 0.014 0.96
Procymidone GLEAMS W & S 0.001 0.87
Procymidone LEACHM Water 0.017 0.92
Procymidone GLEAMS Water 0.034 0.57
Procymidone HYDRUS Water 0.003
Procymidone LEACHM Soil 0.006 0.96
Procymidone GLEAMS Soil 0.004 0.86

Site B

Bromide LEACHM Soil 0.84
Bromide GLEAMS Soil 0.62
Picloram LEACHM Soil 0.97
Picloram GLEAMS Soil 0.95
Triclopyr LEACHM Soil 0.96
Triclopyr GLEAMS Soil 0.95

Table 3. Goodness of fit parameters for SPASMO simulations using
median leaching parameters from Table 2

Best, best or most optimal value for other model simulations from
Table 2; Worst, worst value for other model simulations from Table 2;
CRM, coefficient of residual mass; EF, modelling efficiency

Chemical Data SS-residual

 SPASMO Best Worst

Bromide Water 37214 30900 239200
 Soil 24601 3172 32800
Hexazinone Water 2.99 2.90 26.70
 Soil 39.16 3.50 7.70
Procymidone Water 0.0029 0.0008 0.906
 Soil 30.77 5.00 93.00
Picloram Soil 39.80 3.50 3.90
Triclopyr Soil 64.91 7.50 11.60


 SPASMO Best Worst

Bromide Water 0.45 0.55 -2.48
 Soil -2.29 0.58 -3.38
Hexazinone Water 0.19 0.21 -6.22
 Soil 0.12 0.92 0.83
Procymidone Water -3.93 -0.40 -153.00
 Soil 0.70 0.10 0.95
Picloram Soil 0.49 0.96 0.95
Triclopyr Soil 0.65 0.96 0.94

Chemical Data CRM

 SPASMO Best Worst

Bromide Water -0.03 -0.06 -1.52
 Soil -1.34 0.05 -2.38

Hexazinone Water 0.04 0.04 -1.29
 Soil -0.58 0.06 0.34
Procymidone Water -0.90 0.35 -7.27
 Soil -0.42 0.10 0.92
Picloram Soil -0.48 0.10 0.60
Triclopyr Soil -0.73 -0.05 -0.06


 SPASMO Best Worst

Bromide Water 0.50 0.55 0.34
 Soil 0.18 0.61 0.06
Hexazinone Water 0.24 0.28 0.16
 Soil 0.78 0.94 0.87
Procymidone Water 0.02 0.03 0.001
 Soil 0.93 0.96 0.57
Picloram Soil 0.93 0.97 0.95
Triclopyr Soil 0.91 0.96 0.95


The authors thank Danny Thornburrow and staff from Environment Waikato for assistance with the fieldwork. The research was funded by contracts CO3X001 (ESR), CO9X0017 (Landcare Research), and CO6X004 (HortResearch) from the Foundation for Science, Research and Technology (New Zealand) and by Environment Waikato for the monitoring well installation.


Cabras P, Meloni M, Pirisi FM (1987) Pesticide fate from vine to wine. Reviews of Environmental Cantamination and Toxicology 99, 83-117.

Close ME, Magesan GN, Lee R, Stewart MK, Hadfield JC (2003) Field study of pesticide leaching in an allophanic soil in New Zealand. 1: Experimental results. Australian Journal of Sail Research 41, 809-824.

Close ME, Watt JPC, Vincent KW (1999) Simulation of picloram, atrazine, and simazine transport through two New Zealand soils using LEACHM. Australian Journal of Soil Research 37, 53-74.

Doherty J (1994) PEST: a unique computer program for model-independent parameter optimisation. In 'Water Down Under 94'. Adelaide, SA, Australia. pp. 551 554. (The Institution of Engineers, Australia: Barton, ACT)

Green SR (2001) Pesticide and nitrate movement through Waikato and Franklin soils: Interim Progress Report. HortResearch, 2002/007, Palmerston North, NZ.

Hadfield J, Smith D (1999) Pesticide contamination of groundwater in the Waikato region. Environment Waikato, 1999/9.

Hewitt AE (1992) 'New Zealand Soil Classification.' DSIR Land Resources Scientific Report No. 19. (Manaaki-Whenua Press: Lincoln, NZ)

Hornsby AG, Wauchope RD, Herner AE (1996) 'Pesticide properties in the environment.' (Springer: New York)

Hutson JL, Cass A (1987) A retentivity function for use in soil water simulation models. Journal of Soil Science 38, 105 113.

Hutson JL, Wagenet RJ (1993) A pragmatic field-scale approach for modeling pesticides. Journal of Environmental Quality 22,494-499.

Hutson JL, Wagenet RJ (1995) An overview of LEACHM: a process based model of water and solute movement, transformations, plant uptake and chemical reactions in unsaturated zone. Soil Science, 409-422.

Jarvis NJ, Bergstrom LF, Brown CD (1995) Pesticide leaching models and their use for management purposes. In 'Environmental Behaviour of Agrochemicals'. (Eds TR Roberts, PC Kearney) pp. 185-220. (John Wiley & Sons Ltd: New York)

Leonard RA, Knisel WG, Davis FM, Johnson AW (1990) Validating GLEAMS with field data for fenamiphos and its metabolites. Journal of Irrigation and Drainage Engineering 116, 24-35.

Leonard RA, Knisel WG, Still DA (1987) GLEAMS: Groundwater loading effects of agricultural management systems. T ASAE 30, 1403-1418.

Loague K, Green RE (1991) Statistical and graphical methods for evaluating solute transport models: Overview and application. Journal of Contaminant Hydrology 7, 51-73.

McNaughton DE, Holland PT, Clothier B, James T (1999) Parameters for environmental persistence of pesticides in horticultural soils. In 'Environmental aspects of pesticide use'. Abstracts of a workshop, Hamilton. p. 30 (AgResearch: Hamilton, NZ)

Pang L, Close ME (2001) A field tracer study of attenuation of atrazine, hexazinone and procymidone in a pumice sand aquifer. Pest Management Science 57, 1142-1150.

Pang L, Close ME, Watt JPC, Vincent KW (2000) Simulation of picloram, atrazine, and simazine leaching through two New Zealand soils and into groundwater using HYDRUS-2D. Journal of Contaminant Hydrology 44, 1946.

Simunek J, Sejna M, van Genuchten MT (1996) HYDRUS-2D. Simulating water flow and solute transport in two-dimensional variably saturated media. U.S. Salinity Laboratory, USDA/ARS Riverside, California.

The Pesticide Manual (1994) 'Incorporating the agrochemicals handbook.' (The British Crop Protection Council and The Royal Society of Chemistry: UK)

Wauchope RD, Battler TM, Hornsby AG, Augustijn-Beckers PWM, Burt JP (1992) The SCS/ARS/CES pesticide properties database for environmental decision-making. Reviews of Environmental Contamination and Toxicology 123, 1-164.

Weber JB (1994) Properties and behavior of pesticides in soil. In 'Mechanisms of pesticide movement into groundwater'. (Eds RC Honeycutt, DJ Schabacker) Ch. 2. pp. 15-41. (Lewis Publishers: Boca Raton, FL)

Manuscript received 3 July 2002, accepted 16 January 2003

M. E. Close (A,E), L. Pang (A), G. N. Magesan (B,D), R. Lee (B), and S. R. Green (C)

(A) Institute of Environmental Science and Research, PO Box 29-181, Christchurch, New Zealand.

(B) Landcare Research NZ Ltd, Private Bag 3127, Hamilton, New Zealand.

(C) HortResearch, Private Bag 11-030, Palmerston North, New Zealand.

(D) Present address: Forest Research, Private Bag 3020, Rotorua, New Zealand.

(E) Corresponding author; email:
COPYRIGHT 2003 CSIRO Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2003 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Close, M.E.; Pang, L.; Magesan, G.N.; Lee, R.; Green, S.R.
Publication:Australian Journal of Soil Research
Geographic Code:8NEWZ
Date:Sep 1, 2003
Previous Article:Field study of pesticide leaching in an allophanic soil in New Zealand. 1: experimental results.
Next Article:Land use effects on sorption of pesticides and their metabolites in sandy soils. I. Fenamiphos and two metabolites, fenamiphos sulfoxide and...

Related Articles
Predicting the probabilities of groundwater contamination by pesticides under variable recharge.
Measurement and prediction of permethrin persistence in six Malaysian agricultural soils.
Sorption of pesticides used in banana production on soils of Ecuador.
Field study of pesticide leaching in an allophanic soil in New Zealand. 1: experimental results.
Interaction effects of clay swelling and dispersion and CaC[O.sub.3] content on saturated hydraulic conductivity.
Identification of functional horizons to predict physical properties for soils from alluvium in Canterbury, New Zealand.
Fate and behaviour of pesticides in the agroecosystem--a review with a New Zealand perspective.
Leaching of macronutrients and metals from undisturbed soils treated with metal-spiked sewage sludge. 2. Leaching of metals.
Chemical properties of two soils irrigated with thermo-mechanical pulp mill effluent.
Nutrient leaching and changes in soil characteristics of four contrasting soils irrigated with secondary-treated municipal wastewater for four years.

Terms of use | Privacy policy | Copyright © 2019 Farlex, Inc. | Feedback | For webmasters