PL0538 - Cranfield University
The UK has left the EU, new rules from January 2021
The transition period after Brexit comes to an end this year.
The MACRO model is increasingly used to simulate movement of pesticides to surface waters via drainflow and leaching to depth through soils prone to preferential flow. Recent work funded by Department for Environment, Food and Rural Affairs ( Defra) concluded that MACRO is and should continue to be the preferred preferential flow model for regulatory use. However, the predictive use of MACRO is greatly restricted by difficulties in robust parameter selection, particularly as model predictions are significantly affected by changes in many of the problematic parameters. This study aimed to develop guidance on the estimation of those soil hydraulic parameters which describe preferential flow within the MACRO model (version 4.2). The study investigated whether additional field and laboratory measurements could assist in model parameterisation. Work also explored the potential use of inverse modelling techniques to obtain calibrated values for parameters where direct determination did not prove possible.
Total porosity and water content at the micropore/macropore boundary (TPORV and XMPOR)
Comparisons were made between modelling based on standard laboratory determinations and a novel method based on analysis of field moisture contents. It was found that the use of values derived from field data may better represent field conditions.
Tortuosity of the macropore domain (ZN)
This is a very sensitive parameter in determining the relative influence of preferential flow on pesticide transport. Laboratory work with intact soil cores was undertaken to try to measure representative values. However, it was concluded that the direct measurement of the macropore tortuosity term in MACRO is not practicable and that ZN is probably best considered a parameter that has to be systematically calibrated.
Hydraulic conductivity at the micropore/macropore boundary (KSM)
There are existing field protocols for the measurement of this conductivity value as well as a number of pedotransfer functions. Field measurements were made for soils studied in a previous DEFRA-funded lysimeter experiment. Model runs were compared for various values of KSM. Values selected using expert judgement (original simulation) were always significantly smaller than those derived from field measurements or pedotransfer functions, suggesting that they were not physically realistic. Despite this, simulations of pesticide leaching were very much better when KSM values based on expert judgement were used. Here, direct measurement of the parameter adversely affected the quality of the simulation, suggesting that the parameter could have been compensating for processes of importance not included in the model. Unsaturated hydraulic conductivity is highly variable both spatially and temporally and it is therefore difficult to assign a single measured KSM value, particularly for long-term simulations.
Given the difficulties in independently measuring key parameters, automatic calibration of MACRO (inverse modelling) was undertaken against volumes of leachate and concentrations of a pesticide from a lysimeter experiment to derive values for KSM, ZN and ASCALE (effective diffusive path length in soil). A good match between observed and simulated concentrations of pesticide could be obtained in some circumstances, but the optimised parameter values were often not physically realistic and sharply contrasting sets of parameter values achieved a similar fit to the data. These problems arose because: (i) the dataset used had several limitations; (ii) the MACRO model is highly non-linear and exhibits significant correlation between uncertain parameters; and (iii) advanced techniques in inverse modelling (including the use of alternative packages) were not explored. Automatic calibration is a rapidly developing science, but it is considered that the tools are not yet in place to allow robust derivation of input parameters for MACRO on the basis of inverse modelling in the regulatory framework.
It was concluded that it might not always be possible to improve the fit of MACRO 4.2 simulations to measured data by directly measuring more of the variables or through use of inverse modelling. It may thus be more appropriate to select parameters on the basis of generic guidance. Estimation procedures adopted by Cranfield Centre for Eco Chemistry are set out and ranges and appropriate values are proposed for 14 key parameters that are primarily related to the description of preferential flow. This guidance is based on application of the model to a number of UK datasets for pesticide leaching (both within PL0538 and in previous projects).
Development of guidance on parameter estimation for the preferential flow model MACRO 4.2.