245soil pH, erosion and a full range of management controls. APSIM has been tested in
a diverse range of systems and environments including many small householder
systems in Africa (Robertson et al. 2005 ; Whitbread et al. 2010 ; Dimes et al. 2015 ).
APSIM has been used in a broad range of applications including support for on-
farm decision making, farming systems design for production or resource manage-
ment objectives, assessment of the value of seasonal climate forecasting, analysis of
supply chain issues in agribusiness activities, development of waste management
guidelines, risk assessment for government policy-making and as a guide to research
and education activities.
2.4 Risk Management Using APSIM
One of the most well-known applications of modelling in dryland agricultural sys-
tems is to inform risk and uncertainty. Here we define risk as imperfect knowledge
where the probabilities of possible outcomes are known – ‘known unknowns’.
Uncertainty exists when these probabilities are not known – ‘unknown unknown’.
The distinction is important, as the major source of risk in dryland agricultural sys-
tems is usually climate variability, which can be informed using long-term climate
records and seasonal climate forecasts – when available. The nature, time frame and
intensity of more uncertain events can be difficult to foresee, e.g. price fluctuations,
climate change, pest and diseases. In dryland systems, climate change is usually
studied using climate projections derived from global circulation models (Rodriguez
et al. 2014 ).
Although discovered by Walker in the 1920s, it was not until 75 years later that
the Southern Oscillation Index (SOI) became the major tool to inform climate vari-
ability and the state of the El Nino in Australia (Stone et al. 1996 ). Seasonal climate
forecasts based on the SOI have been used, in combination with simulation model-
ling, probability theory, profit function, and finance techniques, to inform practice
change (Meinke and Hochman 2000 ). A relevant example is the use of SOI and SST-
based forecasts in June/July to predict spring rainfall or wheat yields in southeast
Australia (Anwar et al. 2008 ) (Fig. 3 ). The authors found that the overall predictive
skill for spring rainfall and simulated wheat yields was 60–83 % consistent. Levels
of predictive skill that should allow farmers use the information to make tactical in
crop management interventions, such as top dressing of nitrogen fertilisers.
2.5 Modelling G×E×M Interactions in Dryland Agriculture
A more recent application of systems modelling for dryland agricultural systems is
in the area of crop design. Crop design uses dynamic models to match crop physi-
ological traits (G) and managements (M) to specific environments (E) in G×E×M
studies. For example, APSIM wheat (Chenu et al. 2011 ), sorghum (Chapman 2008 )
Modelling Dryland Agricultural Systems