Use of the APSIM wheat model to predict yield, drainage, and NO3- leaching for a deep sand

High rates of drainage and leaching of nitrates in deep sands in Western Australia are contributing to groundwater recharge and soil acidification in this region. Strategies are being soughtto increase water and nitrogen (N) use in the legume-based cropping systems. Choice of appropriate management strategies is complicated by the diversity of soil types, the range of crops, and the inherent season to season variability. Simulation models provide the means to extrapolate beyond the bounds of experimental data if accurate predictions of key processes can be demonstrated. This paper evaluates the accuracy of predictions of soil water content, evapotranspiration, drainage, inorganic N content insoil, nitrate (NO-3) leaching, wheat growth, N uptake, and grain yields obtained from the Agricultural Production Systems Simulator (APSIM) model when this was initialised with appropriate information on soil properties and wheat varieties commonly grown on deep sands in the 500 mm rainfall zone west of Moora in Western Australia. The model was found to give good predictions of soil water content,evapotranspiration, deep drainage, and overall NO-3 leaching. Temporal changes in inorganic N insoil were simulated, although the small concentrations in soil inorganic N precluded close matching of paired observed and predicted values. Crop growth and N uptake were closely predicted up to anthesis, but a poor fit between observed and predicted crop growth and N uptake was noted postanthesis. Reasons for the discrepancies between modelled and observed values are outlined. The model was run with historical weather data (81 years) and different initial soil water and inorganic soil N profiles to assess the probability of drainage and NO-3 leaching, and the grain yield potentials for wheat grown on deep sands in the region west of Moora. Simulation showed that thesoil water and the soil inorganic N content at the beginning of each season had no effect on grain yield, implying that pre-seed soil NO-3 was largely lost from the soil by leaching. There was a 50% probability that 141 mm of winter rainfall could drain below 1·5 m and a 50% probability that 53 kgN/ha could be leached under wheat following a lupin crop, where initial soil water contents andsoil NO-3 contents used in the model were those measured in a deep sand after late March rainfall. Simulated application of N fertiliser at sowing increased both grain yield and NO-3 leaching. Splitting the N application between the time of sowing and 40 days after sowing decreased NO-3 leaching,increased N uptake by wheat, and increased grain yield, findings which are consistent with agronomic practice. The high drainage and leaching potential of these soils were identified as the main reasons why predicted yields did not approach the French and Schultz potential yield estimates based on 20 kg grain yield per mm of rainfall. When the available water was reduced by simulated drainage, simulated grain yields for the fertilised treatments approached the potential yield line.