Contribution of climate models and APSIM phenological parameters to uncertainties in spring wheat simulations: Application of SUFI-2 algorithm in northeast Australia

We used SUFI-2 for the first time to calibrate the phenology module of the APSIM-wheat model for 10 spring wheat cultivars cultivated in northeast Australia (south-eastern Queensland). Calibration resulted in an average root mean square error (RMSE) of 5.5 days for developmental stages from stem elongation up to flowering. Projections from 33 climate models under the representative concentration pathway 8.5 were used for simulations at 17 sites. Using adapted sowing times, we simulated significantly shorter crop cycles and grain yield improvements for the period 2036-2065 relative to 1990-2019 for three selected cultivars (Hartog, Scout and Gregory). Photoperiod and vernalisation sensitivities were shown to be the largest and smallest contributors to total uncertainties in the simulated flowering day and grain yield, respectively. Uncertainties in climate models had a relatively minor contribution to the total uncertainties in the simulated values of target traits. This contribution significantly increased when climate change impact on the target traits was quantified.

Automated summary

For the first time, the SUFI-2 method was used to calibrate the phenology module of the APSIM-wheat model for spring wheat cultivars in northeast Australia. Results showed that adjusting sowing times can significantly shorten crop cycles and increase grain yields, with photoperiod and vernalisation sensitivities being major contributors to uncertainties in the simulated values.