Jaclyn N. Brown1, Zvi Hochman2 Dean Holzworth3, and Heidi Horan2.
1 CSIRO, Hobart, Tas, 7001, Jaci.Brown@csiro.au
2 CSIRO, Brisbane, QLD.
3 CSIRO, Toowoomba, QLD.
Skilful predictions of upcoming yield potential allow producers to mitigate risk and implement effective management decisions, thereby increasing productivity and profitability. The season-ahead climate forecast is a key parameter in determining upcoming yield potential yet it is difficult to incorporate into crop models. Seasonal climate models are at lower spatial resolution than the paddock scale needed for yield predictions in crop models and often contain many biases particularly in crucial factors such as rainfall. The science of climate modelling is progressing rapidly and we have explored whether the models now have adequate skill to provide direct daily weather input into crop models, after only a simple downscaling and calibration. We find that in some locations, the daily climate input is very effective for predicting yield. In other locations the biases are still large and the forecasts unreliable. The reasons for this poor predictive skill appears to come from numerous sources including a tendency to always under predict yield, errors in the rainfall amount and distribution, radiation and temperature. A new climate model is being implemented in Australia this year and we expect to soon see improvements from this baseline level.