Macropore flow in soils and pesticide risk assessment

Summary

Agricultural use of pesticides may result in contamination of groundwater being used as a drinking water source. Additionally, the emission of these compounds via drainage into surface waters can adversely affect aquatic ecosystems. Pesticide risk assessment is applied to evaluate the potential for health and ecological effects of a pesticide and is commonly performed using mechanistic models. Pesticides currently used in agriculture are soluble in water. Hence, both pesticide transport and water flow through the soil must be studied together through an environmental risk assessment of pesticides.

Water flow and pesticide transport can be described for field conditions as uniform or preferential. While uniform flow leads to stable wetting fronts parallel to the soil surface, preferential flow generates unstable wetting fronts, differences in water pressure and solute concentrations, and rapid flow through parts of the soil matrix. This Ph.D. research project focuses on one source of preferential flow; macropore flow. Macropore flow produces fast vertical water flow and pesticide transport in a small soil volume, bypassing the reactive soil matrix. One effect of this is that some of the applied pesticides cannot degrade in the soil and instead arrives in groundwater or surface waters, negatively affecting water quality. Therefore, mechanistic models must incorporate macropore flow in pesticide risk assessment to ensure accurate simulations.

Macropores are originated mainly by biological activity, drying and wetting cycles, and shrinking clays, which results in natural variation of the number of macropores in time and space. Some macropores directly end within the soil matrix, referred to as ‘dead-end macropores.’ We have designated the spatial variation of macropores over depth as ‘heterogeneous macropore geometry.’

Dual-permeability models such as HYDRUS and SWAP are conventional mechanistic models utilized in pesticide risk assessment studies. The parametrization of these models for heterogeneous macropore geometries as produced by dead-end macropores is still a challenge. Two parameters commonly utilized for water flow and pesticide transport related to the description of heterogeneous macropore geometries in both models are the relative macroporosity and the effective aggregate width. We saw that the determination of both parameters under field conditions is poorly understood through an extensive literature review. Increasing the understanding of the parametrization of the relative macroporosity and the effective aggregate width for field conditions would be essential for HYDRUS and SWAP.

To achieve this, the following objectives were proposed: (1) Understand the effect of dead-end macropores on water redistribution and outflow; (2) Simulate water flow with HYDRUS and SWAP in heterogeneous macropore geometries as produced by dead-end macropores; (3) Generate a new methodology for the determination of the relative macroporosity and the effective aggregate width under field conditions; (4) Compute the importance of the relative macroporosity and the effective aggregate width on different pesticide outputs of the SWAP-PEARL models.

Chapter 2 reports the first two objectives, which were investigated utilizing two soil columns under controlled laboratory conditions with artificial macropores ending at different depths. We observed that dead-end macropores do not increase outflow proportionally to the increase in the relative macroporosity from the experimental results. The simulation results indicated that both HYDRUS-1D and SWAP performed well in the system with dead-end macropores. This is relevant for performing pesticide risk assessments at different scales because we can trust the accuracy of both model simulations.

In Chapter 3, we present the results of Objective 3. Our methodology is based on Watson & Luxmoore (1986) but expands it beyond cylindrical macropore shapes by pore-scale modeling for rings, hexagons, bricks, and rectangular slabs macropore-matrix shapes using a transformation factor. We found that the actual relative macroporosity is underestimated, and the effective aggregate width is overestimated if always presuming cylindrical shapes for field conditions. The transformation factor obtained was constant for all the non-cylindrical shapes and equaled 1.5. We also introduce a mathematical relation between the relative macroporosity and the effective aggregate width for all the shapes analyzed. The methodology can be applied to former disk infiltrometer databases following only three conditions.

We found that the physical assumptions performed in Chapter 3 to obtain the transformation factor tended to overestimate the water flow. Therefore, we improved the methodology by calibration with HYDRUS 2D/3D and presented those results in Chapter 4. The representation of heterogeneous macropore geometries in HYDRUS 2D/3D implies a greater number of parameters than for homogeneous macropore systems (Chapter 2). We tackled this issue by introducing a general meta-model which reduces the number of parameters required to represent heterogeneous macropore geometries. The meta-model parameters obtained, and macropore parameters from previous research were utilized to obtain a complete initial parametrization of HYDRUS 2D/3D, which then required some calibration to obtain a good match between observations and simulations. Subsequently, a dimensional reduction of HYDRUS 2D/3D parameters was utilized to parametrize HYDRUS 1D and SWAP. Therefore, HYDRUS 2D/3D can now be utilized as a transitional model for obtaining macropore parameters for HYDRUS 1D and SWAP utilizing disk infiltrometer measurements.

Chapter 5 answers the last objective. A Morris screening and Sobol-Jansen sensitivity analysis were performed using the SWAP-PEARL model for the Andelst site, including complex hydrology. The seven most essential parameters selected through the Morris elementary effect were analyzed using the Sobol-Jansen method to quantify the parameter importance. The results indicated that parameters describing degradation and sorption of the pesticide are essential parameters for all output types and that for leaching to a depth of 1 m and below, the depth of the static macropores and the depth of the internal catchment were critical parameters. Therefore, we also found that the relative macroporosity and the effective aggregate width were not relevant parameters for pesticide outputs. This finding implies that a higher uncertainty in obtaining those parameters is acceptable in SWAP-PEARL models.

The new equations and procedures generated are relevant for more realistic HYDRUS and SWAP parameterization in heterogenous macroporous systems. Our findings make an important contribution to accurate forecasting of potential contamination of groundwater and surface water by pesticide compounds or other agrochemicals, thus ensuring the availability of good quality water for current and future inhabitants of this planet.