Co-Funding Deed AgScore project under the Managing Climate Variability Program

Summary

This project made use of an innovative software tool providing robust comparison of seasonal climate models using agricultural relevant metrics to help primary producers assess seasonal forecasts for more profitable decision making and climate risk management.

AgScore is a cloud-based tool which executes the Agricultural Production Systems sIMulator (APSIM) with uploaded climate model data and analyses the results against identical APSIM simulations using baseline climatology for the same period. This is done to identify biases and weaknesses in climate models for predicting agriculturally relevant metrics for different industries, such as crop yield in grains. This approach was applied to multiple global climate models and different Australian agricultural industries with the results communicated at different levels to growers, advisors, extension specialists in the climate risk and broader climate science community.

Real-world case study farms were used to demonstrate the additional value a forecast might offer over a baseline management scenario that assumed average conditions. The case studies included grains enterprises across a range of climates and geographies, as well as cotton, sugar and rice. For the grains case studies, were focussed largely on sowing related decisions including crop species, variety, sowing density and nitrogen fertilisation.

Objectives

To model the economic value of using seasonal climate forecasts information to aid strategic on-farm decisions such as crop choice prior to sowing.

Key findings

This project gauged the performance of various seasonal forecasting systems on an even playing field. The regional analysis involved verifying 12 different forecast models based on past performance matched with their observations. It presents measures of forecast quality via an interactive dashboard, allowing users to explore their regions and seasons of interest.
Overall no single model stood out as superior to the group of forecast systems evaluated. The skill, or ability of the forecasts to improve upon the baseline climatology, was relatively low, with increased skill observed during spring.
The results of this type of analysis are heavily influenced by the availability of hindcast data (forecasts produced in the past). This limits the statistical power and hence the confidence placed in which models performed better or worse.

Benefits to industry

Forecasts translated into yield or productivity-based predictions have obvious benefit to users in that they incorporate multiple climate drivers i.e. rainfall and temperature, and integrate seasonal trajectories of plant growth. However, the results indicate that the overall signal and the corresponding accuracy of the yield forecast may be similar to the rainfall forecast for the same time of year and location.

Future research

Investigate the potential for incorporating other sources of local farm information that might strengthen predictions of yield or productivity when generating a forecast. For example, soil water estimates, particularly early in the season, might bolster forecasts of crop yield.

Investigate the concept of a ‘perfect knowledge’ forecast as a means to gauge the extent to which management decisions might be optimised to the potential options available within a given enterprise.

Explore the role of multi-week forecasts that might address decisions that have received less attention in the past. This may include spray planning, irrigation scheduling in response to temperature fluctuations and harvest logistics.