Modelling water, nitrogen, and crop yield for a long-term fallow management experiment

Abstract

Two models that differ markedly in how they represent the crop-soil system have been used to simulate soil processes and crop production in the long-term experiment at Hermitage Research Station, Warwick, Queensland. The experiment was designed to examine the effects of tillage, stubble management, and nitrogen (N) fertiliser on the productivity of a winter cereal-summer fallow cropping system. it commenced in 1968 and the treatments have been maintained until the present. CENTURY operates on a monthly time step, considers all soil N transformations to occur in a single soil layer, and has a very simple crop growth routine that does not deal with crop phenology. APSIM provides a framework whereby a model of a cropping system is configured from component modules, which operate on a daily time step. For simulating the Hermitage experiment, modules to represent the dynamics of soil-water, N, surface residues, and growth of a wheat crop were used. The water and N modules deal with a multi-layered soil, whilst the wheat module develops leaf area, intercepts light, and accumulates and partitions dry matter in response to weather, soil-water, and N. Both models were specified to simulate the whole experimental period (1969-92) as a continuous run. The ability of these models to simulate grain yields, soil-water and drainage, nitrate-N, and soil organic matter were examined. Both models predict, in agreement with the observed data, that for this continuous cereal cropping system there has been a decline in soil organic matter for all the treatments and a reduction through time in the capacity of the soil to mineralise and accumulate nitrate during the fallows. CENTURY performed better than APSIM in predicting the relative yields of the N treatments but was less satisfactory than APSIM for absolute grain yield, soil-water, and drainage. Yield predictions with APSIM were sensitive to carry-over errors in the water balance from one season to the next, so that in some seasons large errors occurred in the predicted relative yields. Both models reproduced the observations well enough to indicate their suitability for providing useful insights into the behaviour of cropping systems where the focus is on depletion of soil fertility.