Evaluating the SIRIUS wheat crop simulation model in Canterbury

Abstract

An on-farm trial was conducted in an irrigated commercial wheat crop on a property near Dorie in the 2001/2002 and 2002/2003 seasons to validate the SIRIUS crop simulation model. Three nitrogen treatments were applied, a no nitrogen (N) and farmer or model recommended timing and rates of nitrogen. Crop biomass was sampled at four dates throughout the growing season, anthesis date recorded and the crop factors total biological yield (TBY), total grain yield (TGY), thousand seed weight (TSW) and grain protein (GP) recorded at harvest.

The original SIRIUS cultivar settings were used to calculate predictions for the six crops and these were compared with the field observations using a root mean squared deviation (RMSD) calculation. With the exception of TSW in the first season, all crop factors achieved an RMSD of less than 10%, however anthesis date was predicted as much as 9 days early. SIRIUS predicted TGY excess of 13 t/ha, GP in excess of 11% and TSW in excess of 46 mg. This was as productive as that achieved by the farmer.

The cultivar factors, phyllochron, minimum leaf number (MLN) and day length response (DLR) were identified to influence anthesis date. An exact prediction of anthesis date (RMSD = 0) was achieved for the phyllochron when it was extended from 110°Cd to 123°Cd or, minimum leaf number when extended from 8.00 to 9.85 or daylength response extended from 0.40 to 0.64.

The adjusted phyllochron of 123°Cd was selected for further validation retaining the original settings for MLN and DLR. When validated against crops grown with and without nitrogen in two different seasons, the 123°Cd phyllochron improved the original settings RMSD% of the mean crop factor for anthesis date from 2.38 to 0.53% and maintained the overall RMSD% around 10%.

The SIRIUS prediction of TGY of the no N crop in the 2001/2002 season was almost half that of the observed value. This was attributed to poor soil N input information because the soil was only sampled to 0.3 m and it was likely that the wheat roots reached soil nitrogen supplies unaccounted for by the sampling. Assuming that nitrogen was the main cause of this inaccurate TGY prediction, SIRIUS was successfully used to predict that up to an extra 80 kg N/ha was required in the soil to move the predicted grain yield to the observed 10.5 t/ha yield, without affecting grain protein.

The current study showed that SIRIUS could reliably predict crop yield and yield components of on-farm trial crops. The model therefore has potential for use in the future by farmers and consultants to aid crop management decisions.