Predicting sulphate leaching in yellow-brown pumice soils

Abstract

Sulphur (S) is an important plant nutrient, and is often deficient for plant growth on the yellow-brown pumice soils of the Central North Island, New Zealand. In order to remove this limitation to plant growth in the most efficient and economic way it is necessary to understand the fate of fertiliser S applied to these soils. The experiments reported in this thesis aimed to determine the fate of sulphate applied to yellow-brown pumice soils, particularly as affected by leaching, and to use this information to make predictions of fertiliser availability. Experiments with intact soil cores showed that leaching of sulphate in the yellow-brown pumice soils was rapid and efficient. The breakthrough curves produced indicated that solute transport was dominated by miscible displacement, combined with adsorption. This contrasts with the preferential flow patterns seen in many other intact soils with stronger physical structure.

Sulphate breakthrough curves for six soil types were described using the Gompertz equation. This allowed comparisons between soil types to be made. The leaching results indicated that there were significant differences in leaching rate between soil types related to sulphate retention and field capacity moisture content. Calculations of drainage from a water balance indicated that sulphate applied to all sites in autumn would be mostly leached by the following spring, but that there was a large range in October-April leaching of spring applied sulphate. These results confirm that leaching of sulphate, leading to low plant uptake of applied fertiliser S, is likely to be a problem in many areas of yellow-brown pumice soils.

Modelling of solute transport was a central part of this study. Initially chloride leaching was modelled to isolate physical transport processes from the chemical and biological processes which affect sulphate concentration and leaching rate. The Rose model (a partially analytical solution to the convective-dispersive equation) simulated the chloride breakthrough curves with reasonable accuracy, and this was improved when a retardation factor (R) was included to allow for anion exclusion. Inclusion of R made the Rose model a two-parameter model, referred to as Rose2. The convective log-normal transfer function model (CLT) gave excellent simulation of the chloride breakthrough curves. Neither the Burns equation nor the Barraclough model were able to reproduce the chloride breakthrough curves.

The sulphate breakthrough curves were simulated equally well by either the Rose2 or CLT models when all parameters were optimised. When an attempt was made to simulate the sulphate breakthrough curves using parameters from the chloride breakthrough curves, and retardation factors from sulphate adsorption isotherms, neither model was particularly successful.

Examination of the results indicated that the mean solute velocity was overestimated by this method, possibly due to biological immobilisation or cycling. When the optimised mean travel time parameter for each model was used, along with the dispersion parameter from the chloride breakthrough curves, then the CLT gave a good simulation of the sulphate breakthrough curves, while the Rose2 model underestimated the amount of dispersion.

These results indicated that the CLT was better able to simulate solute transport in the yellow-brown pumice soils because its underlying assumption, that dispersion increases with the square of depth, was more accurate than the assumption in the Rose2 (and all CDE models) that dispersion increases linearly with depth. Using two tracers (one adsorbing, one non-adsorbing) together seems to be a useful test of the depth-dispersion relationship, which is the key difference between CDE and CLT models.

When the CLT model was applied to a field site it gave reasonable simulation of plant S availability, although underestimating the residual effect of fertiliser sulphate. It seems most likely that organic S cycling increased the degree of sulphate retardation in the field site above that measured in the laboratory experiments.

The role of organic S cycling was studied in an incubation experiment. This study indicated that the addition of sulphate stimulated both mineralisation of native organic Sand immobilisation of added S. The net result of this was immobilisation of added S. Increasing the rate of S addition increased the susceptibility of immobilised S to remineralisation. This is consistent with the observation from the field site that addition of sulphate led to a residual effect greater than that expected from calculated leaching losses.

The field test of the model indicated that leaching has an important effect on the availability of sulphate fertilisers, but that organic S cycling is also important in extending the time of availability. More work to understand and quantify this residual effect is required. The relative importance of these processes in determining the availability of elemental S fertilisers also requires study.