[1] Irrigation is important to many agricultural businesses but also has implications for catchment health. A considerable body of knowledge exists on how irrigation management affects farm business and catchment health. However, this knowledge is fragmentary; is available in many forms such as qualitative and quantitative; is dispersed in scientific literature, technical reports, and the minds of individuals; and is of varying degrees of certainty. Bayesian networks allow the integration of dispersed knowledge into quantitative systems models. This study describes the development, validation, and application of a Bayesian network model of farm irrigation in the Shepparton Irrigation Region of northern Victoria, Australia. In this first paper we describe the process used to integrate a range of sources of knowledge to develop a model of farm irrigation. We describe the principal model components and summarize the reaction to the model and its development process by local stakeholders. Subsequent papers in this series describe model validation and the application of the model to assess the regional impact of historical and future management intervention.

[1] Catchment managers are interested in understanding impacts of the management options they promote at both farm and regional scales. In this third paper of this series, we use Inteca-Farm, a Bayesian network model of farm irrigation in the Shepparton Irrigation Region of northern Victoria, Australia, to assess the current condition of management outcome measures and the impact of historical and future management intervention. To help overcome difficulties in comprehending modeling results that are expressed as probability distributions, to capture uncertainties, we introduce methods to spatially display and compare the output from Bayesian network models and to use these methods to compare model predictions for three management scenarios. Model predictions suggest that management intervention has made a substantial improvement to the condition of management outcome measures and that further improvements are possible. The results highlight that the management impacts are spatially variable, which demonstrates that farm modeling can provide valuable evidence in substantiating the impact of catchment management intervention.

Excess nutrients are challenging the long-term sustainability of grazing-based dairy farming. Whole-farm nutrient-mass balance (NMB) is a well recognised approach to improve on-farm nutrient management decisions. In the present paper, we use a standardised approach for quantifying NMB on grazing-based dairy farms, using a newly developed online tool. Preliminary evaluation, using selected farm data from a previous Australia-wide dairy-farm nutrient study, demonstrated highly comparable estimates of farm area, nutrient fluxes and NMB, with substantial efficiencies in time and sample analysis. Nutrient mass balances were also determined on 16 diverse dairy farms across the five major dairy regions of Victoria, Australia. These results highlighted the importance of purchased feed, fertiliser and milk sales, as major sources of nutrient inputs and outputs, with whole-farm NMB for the 16 dairy farms ranging from 185 to 481 kg/ha for nitrogen, 12–59 kg/ha for phosphorus, 9–244 kg/ha for potassium and –6–55 kg/ha for sulfur. Current industry adoption of the NMB tool has confirmed the benefits of a standardised and efficient collation and processing of readily available farm data to inform nutrient management decisions on commercial dairy farms. We suggest that the standardised assessment of nutrient fluxes, balances and efficiency, as well as feed- and milk-production performance at the whole-farm level, provides dairy farmers, farm advisors and industry and policy analysts with the ability to determine industry-wide goals and improve environmental performance.