Representing weather-year variation in whole-farm optimisation models: Four-stage single-sequence vs eight-stage multi-sequence

The trade-off between accuracy and complexity is a common issue faced in farm systems analysis. To provide insights into the importance of representing weather-year sequence in farm modelling, two whole-farm optimisation models are constructed and applied to a mixed enterprise farming system in a subregion of Western Australia. The frameworks are (i) four-stage single-sequence stochastic programming with recourse (4-SPR) to capture weather-year variation and management tactics tailored to each weather-year and (ii) eight-stage multi-sequence stochastic programming with recourse (8-SPR) to outline weather-year sequences and management tactics tailored to particular weather-year sequences. Results show that single-year stochastic programming generates similar expected profit and strategic management as multi-year stochastic programming. However, optimal tactical farm management is affected by the outcome of the previous year. Tactical decision-making in response to the outcome of the preceding weather-year increases profitability by 14%. Technology changes over the last decade, particularly the increase in computer speed and computational power, increase the ease of construction and application of the 4-SPR and 8-SPR frameworks. Nonetheless, choosing which framework is best to apply to a particular issue or opportunity remains a challenge.

Automated summary

The trade-off between accuracy and complexity is a common issue in farm systems analysis. This study compares the performance of two optimization models in representing weather-year sequence in farm modeling. The results show that single-year stochastic programming generates similar expected profit and strategic management as multi-year stochastic programming. However, optimal tactical farm management is affected by the outcome of the previous year. Technology changes over the last decade have made it easier to construct and apply the optimization models, but choosing the best framework for a specific issue or opportunity remains a challenge.