Inverse method to estimate mineralisation rate constants for nitrogen simulation models: interaction between sampling strategy and quality of parameter estimates

Abstract

Sustainable agricultural practices and land application of wastes require that the accession of nitrate to groundwater be within acceptable limits. Simulation modelling is a valuable aid to the development and testing of management practices that achieve this goal, but requires unbiased and precise parameter estimates. Here we consider the role of simple lysimeter-based techniques, which may yield only a single integral observation in the form of total solute leached or total drainage, in supplementing infrequent concentration data for the purposes of parameter optimisation.

The utility of such techniques was evaluated using a simulation model to create a ‘no-error data set’ of nitrate concentration values and summary observations of the total mass of nitrate leached and total drainage over a 182-day period. From that no-error data set a more realistic data set incorporating random error was created. By using those concentrations, the value of the mass of nitrate leached or total drainage was evaluated by their effect on the unbias and precision of optimised mineralisation or evaporation parameters. The effects of observation weight, error in the observations, and 2 other experimental strategies involving a higher intensity of solute sampling were also tested.

It was found that the summary observations, such as those obtainable from simple lysimeter-based techniques, had the potential to reduce bias and improve the precision of the optimised parameters. The consequence of this effectiveness was that error in the summary observations led to considerable error in the optimised parameters.