Improving wheat simulation capabilities in Australia from a cropping systems perspective.

A methodology to objectively compare model components within a cropping systems model is introduced. It allows effective and efficient comparisons of modelling approaches with the help of a versatile cropping systems shell. This highly modular simulation environment allows inclusion of desired modules at the click of a button. The methodology is applied to some key wheat models currently in use for systems analysis and decision support in Australia. Thus, comprehensive data sets for model testing were required. One such data set, comprising various levels of applied nitrogen and water, is analysed using a crop physiological framework that provides all necessary parameter values for inclusion into a predictive wheat model of intermediate complexity. Further, detailed measurements of light interception during early growth showed that leaf sheaths and stems intercept a substantial amount of light during this phase. If this effect is not accounted for in a model, it can lead to a significant underestimate of anthesis dry matter when a maximum leaf area index of 2 is not exceeded. Data sets from Northern and Southern Australia, New Zealand and the USA were then used to evaluate performance of four wheat and one barley model currently used in Australia. In particular, resource utilization (water and nitrogen) was tested since the condition of the soil at the end of one cropping cycle determines the starting conditions of the next. Based on the strong and weak points highlighted during testing, the Integrated Wheat Model (I-WHEAT) was developed. Its main objective is to provide better predictive wheat modelling capabilities for inclusion in a cropping systems model. I-WHEAT combines well performing approaches from the tested models with some newly developed components. The number of input parameters needed is kept to a minimum and all coefficients can be easily derived from experimental data. It avoids the necessity of having to simulate green leaf dry matter as a means to predict leaf area. This avoids sensitive feedbacks that can generate significant error. I-WHEAT performed better than any of the tested models for resource utilization, leaf area and grain nitrogen content. Amongst others, it will be applied in Australia to investigate options for manipulating either the crop or the cropping system as an aid to pursuing improved sustainable farming practices