APSIM's water and nitrogen modules and simulation of the dynamics of water and nitrogen in fallow systems

APSIM (Agricultural Production Systems Simulator) is a software system which provides a flexible structure for the simulation of climatic and soil management effects on growth of crops in farming systems and changes in the soil resource. The focus of this paper is the predictive performance of APSIM for simulation of soil water and nitrate nitrogen in contrasting soils (vertisols and alfisols) and environments. The three APSIM modules that determine the dynamics of water, carbon, and nitrogen in the soil system (viz. SOILWAT, SOILN and RESIDUE v.1) are described in terms of the processes represented with particular emphasis on aspects of their coding that differ from their precursors in CERES and PERFECT. The most fundamental change is in SOILN, which now provides a formal balance of both carbon and nitrogen in the soil and includes a labile soil organic matter pool that decomposes more rapidly than the bulk of the soil organic matter. Model performance, in terms of prediction of soil water and nitrate, is evaluated during fallows, thereby avoiding complications arising from water use and nitrogen uptake by a crop. One data set is from a long-term experiment on a vertisol in southeast Queensland which studied two tillage treatments (conventional and zero tillage) in combination with fertiliser nitrogen inputs for the growth of wheat; soil water and nitrate were measured twice each year (pre-planting and post-harvest). The second comes from experiments at Katherine, Northern Territory, where legume leys growing on alfisols were chemically killed and ensuing changes in soil water and nitrate were measured during a single season. For both datasets, the predictive ability of the model was satisfactory for water and nitrate, in terms of both the total amounts in the whole profile and their distribution with depth. Since neither of these datasets included measurements of the runoff component of the water balance, this aspect of model performance was evaluated, and shown to be generally goad, using data from a third source where runoff had been measured from contour bay catchments. (C) 1997 Elsevier Science Ltd.