Climate sensitive decisions- fast graphs for slow thinking

Take home messages

• Many of the decisions faced by growers have outcomes which are sensitive to the climatic conditions received after the decision is made.
• As with many decisions, slowing down and thinking through the potential outcomes (or ‘imagining the future with rigour’) can support the decision-making process.
• Analysing the expected range of outcomes (rather than focussing on a specific point) can provide an increased level of robustness.
• ‘Seasonal outlook’ forecasts can be useful in providing additional information to be used in the decision-making process, but their application needs to be carefully considered.

Background

Climate variability is a major source of risk to grain grower profitability in the southern Australian grain growing region. GRDC has recognised this with a significant investment aimed at improving the way the grains industry manages climate risk and specifically improving the understanding of how and when imperfect but improving seasonal outlook forecasts (SCF) can be used to improve grain grower profitability in the southern region.

The project has three main components:

1. Extending the "The Break" e-newsletter and video communication to cover the whole GRDC southern region. This will involve a South Australian, Victorian and Tasmanian version.
2. Working closely with approximately 20 advisers from South Australia, Victoria and Tasmania through two workshops to explore if and how seasonal climate forecasts can be better incorporated into the management of grain farms.
3. Creating a summary publication with case studies and worked examples to educate GRDC Southern region growers and advisers on the use of seasonal forecast information to better target crop inputs, manage risk and increase profitability.

The project is led by Graeme Anderson and Dale Grey from Agriculture Victoria. In addition to overall project management, they lead the extension of the successful “The Break” suite of communication products across the Southern region.

This paper reports primarily on the early thinking behind Component 2;.the development and running of workshops with 20 advisors, which is led by Dr. Peter Hayman (SARDI) working with Barry Mudge (low rainfall grower, Port Germein, SA) and Mark Stanley (Regional Connections).

A definition of decision making

Decision-making can be defined as ‘The thought process of selecting a logical choice from the available options. When trying to make a good decision, a person must weigh the positives and negatives of each option and consider all the alternatives. For effective decision making, a person must be able to forecast the outcome of each option as well, and based on all these items, determine which option is the best for that particular situation.’

Growers are clearly faced with an array of decisions, the outcome of which can be heavily dependent on the weather/climate which occurs after the decision is made. As growers, we take responsibility for these decisions and being comfortable with the decision-making process is an important psychological hurdle to overcome. Decisions can range from simple to complex; usually the complexity is increased as the time scale increases. For example:

• Short term operational decision such as whether or not to apply urea is a relatively simple decision and could be heavily influenced by short term weather forecasts.
• Medium term decision of how much nitrogen (N) to apply is a complicated decision and the result could be heavily dependent on rainfall for the balance of the season but often tempered by existing knowledge of soil water status, N supply, etc.
• Longer term changes in farm program, for example; from high return/ high risk farming system to lower return but lower risk. This is a complex decision due to the potential longer-term rotational influences and consequences of changing crop types.

Given the influence that the future climate can have on the decision outcomes, it is understandable that SCF are widely regarded as the ‘holy grail’ of our climatically exposed farming systems. However, it is important to unpack that premise to see how it stands up under closer scrutiny. This involves looking at both the areas where seasonal forecasts can be effectively harnessed for better outcomes and the concept of how good the forecast needs to be before it is considered reasonable information to act on.

But before we can attempt to pull seasonal outlook forecasts into some sort of decision making process, we need to understand the implications of climate on the potential outcomes of the decisions.

‘Imagining the future with rigour’

‘Imagining the future with rigour’ is a statement attributed to Professor Bill Malcolm from The University of Melbourne which describes his interpretation of the process of decision making.

Rational decision-making consists of a sequence of steps designed to rationally develop a desired solution. Intuitive decision making is almost the opposite, being more instinctive, subjective and subconscious in nature. This is the slow and fast thinking referred to by Nobel Prize winner Daniel Kahneman in his book ‘Thinking, Fast and Slow’. Both are important mechanisms and both are used extensively in our daily lives.

Effective decision making (or ‘imagining the future with rigour’) could be seen as having an understanding and knowledge of three areas:

1. Recognition of the current state (the ‘known knowns’). While this may seem obvious, there can be an enormous amount of information contained within our knowledge of current circumstances. In an agronomy sense, aspects such as current soil moisture levels, crop stage, weed and disease levels, etc. are the obvious examples but knowledge of aspects such as historical varietal performance, landscape influences and client’s attitudes can also have important influences.
2. The likelihood and consequence of the various states which may occur after the decision is made (the ‘unknown knowns’)
3. A methodology to combine the information contained in point 1 and 2 with other potential externalities (the ‘unknown unknowns’) into a process to arrive at a ‘decision’.

Fast graphs for slow thinking

One way of providing a level of rigour to this process is to describe the decision problems in terms of a graph which can allow consideration of all three components. Our experience is that people with a good understanding of farming systems (e.g. advisers and growers) can create these graphs quite quickly (i.e. 20 minutes or less). It usually just involves taking information out of the head and putting it down on paper.

Consider two examples of a climate sensitive decision:

1. A comparatively simple decision involving the application of urea to a moderately responsive wheat crop. Table 1 shows the components of the decision.

Table 1. The net result across subsequent seasonal rainfall deciles of applying 46kg N compared with no application of N to a moderately N responsive crop.

The result can be shown as a graph (Figure 1).

Figure 1. Graph of the comparison of the net result across subsequent seasonal rainfall deciles of applying 46 kg N with no application of N to a moderately N responsive crop.

As suggested, there is a lot of knowledge captured in the information presented (Table 1, Figure 1). The yield estimates in this case have been derived from Yield Prophet®. If this site has been correctly parameterised, then information about the current state (soil water, current N levels, etc.) is caught in the model’s estimates of yield. Likelihood and consequence of different climate states occurring after the decision point is shown in the deciles versus yield estimates. And if the numbers are considered robust and accepted by the decision maker, then a methodology for decision making can be adopted from the information. For example; using the decile break-even point as a guide or comparing the upside and downside (the potential losses compared with the gains).

2. A more complex decision involving changing the crop type at the start of the season. In this example, we are comparing a low rainfall scenario of switching from a higher risk/higher return crop such as lentils to a lower risk/lower return pasture option such as vetch.

Table 2. Comparison of the net result across seasonal rainfall deciles of sowing lentils for grain or vetch for pasture in a low rainfall environment.

Figure 2. Comparison of the net result across seasonal rainfall deciles of sowing lentils for grain versus vetch for pasture in a low rainfall environment.

The numbers in this example have been derived from a gross margins program developed by Rural Solutions (Rural Solutions Farm Gross Margins 2018 Excel Spreadsheet).

The greater level of complexity in this second decision example includes the difficulty of robustly identifying inter-seasonal influences such as the rotational advantages and/or compromises of the different scenarios.

Examples 1 and 2 have used rainfall as the climate variable and have adopted deciles as the measure of risk. Clearly, there will be many other instances where other climate factors (e.g. temperature extremes (frost, heat), excessive rainfall, etc.) will be the drivers of variability with the consequences being dependent on different states of each.

The real value in developing the ‘fast graphs’ lies in our ability to interrogate and ask questions of the output (i.e. the slow thinking). For example;

• How confident are we that we have reasonably captured the essence of the decision problem?
• How does the upside and downside risk compare?
• What externalities would/could change the graphs?

When attempting to use a rational approach to analysing climate sensitive decisions, a number of areas (and pitfalls) can be identified that growers/advisers need to be aware of:

• The value of simplicity (or ‘It is important to keep things simple, but no simpler’). There will always be outriders which cannot be effectively captured in objective analysis. These do not necessarily diminish the value of the analysis, rather we need to identify externalities that could have an influence and temper the outcomes with a healthy dose of perspective.
• There may be biases which are difficult to ignore when analysing a situation. These could include recency bias (greater emphasis on recent experiences), anchoring and availability (information is based on what is readily available).

Where do seasonal climate forecasts fit in?

Seasonal climate forecasts potentially provide some additional information that can be brought into the decision-making process. Once a climate sensitive decision has been described in terms of a graph, it is a relatively simple process to apply the additional information contained in a SCF.

Consider the lentils versus vetch example: A grower has various ways that SCFs can be used (or choose not to use) when confronted with this crop choice decision:

• Prepare, don’t predict:In this case, the grower accepts there is a range of possible outcomes, and prepares for all eventualities. There is no attempt to alter historic climatology by applying any changed probabilities from any seasonal forecast. In the lentil/vetch example, the grower would most likely plant lentils due to the higher upside benefit compared with downside risk.
• Predict, then prepare: The grower may obtain a deterministic forecast which indicates that a particular outcome is likely e.g. a decile 3 season. The grower then uses this as a base for their decision- in the decile 3 case; the grower would lean towards planting vetch as pasture. For this selection to occur, a grower needs to have a good level of confidence (either valid or not) in the SCF.
• Prepare and predict: Similar to prepare, don’t predict the grower accepts there is a range of possibilities but tempers their expectations by accepting the shift in probabilities as provided in a SCF. Again, the grower needs to have a good level of confidence in the skill being demonstrated in the forecast.

An Excel based program has been developed which allows users to examine the effect of changes in probabilities of different climate outlooks on expected outcomes. While this is not meant as a decision support tool, it does allow for discussion on how much the patterns on the chocolate wheel would need to change before a decision might realistically be altered. In the lentil/vetch example, increasing the chance of the driest tercile from 33% (historic climatology) to 50% increases the size of the downside ‘wedge’ but it is still outweighed by the potential gains if the (less likely) better seasons occur (Figure 3). A decision maker in this situation could still be well satisfied with the selection to plant lentils, due to an unwillingness to forego the potential for substantial gains.

Figure 3. Comparison of the effect of changes in seasonal outlooks on the range of outcomes from planting either lentils for grain or vetch for pasture.

Conclusion

• Growers are regularly faced with the need to make important decisions, the outcomes of which can be highly variable depending on the climatic conditions which occur after the decision is made. While not appropriate in all cases, some ‘slow thinking’ objectivity about the range of possible outcomes can be beneficial to the decision-making process.
• There are times when the additional information contained in SCFs can have an important role in making decisions more robust. Identifying these times will be aided by due recognition of both the level of skill in the forecast and the amount of change from historic climatology.
• Equally important will be the need to identify those times when either the skill of the forecast is poor, or the projected change is insufficient to affect the decision process.
• In the long run there are likely to be economic benefits of ‘slow thinking’ but also psychological benefits. Clarifying the difference between good decisions that used the best information at the time and decisions that are lucky or unlucky depending on the season is likely to reduce stress.

Useful resources

The Break subscription or email: the.break@ecodev.vic.gov.au

Acknowledgements

Graeme Anderson and Dale Grey of Agriculture Victoria and Mark Stanley of Regional Communications.

Contact details

Barry Mudge
Barry Mudge Consulting
0417826790
theoaks5@bigpond.com
@MudgeBarry

Peter Hayman
SARDI Climate Applications
0401 996 448
Peter.Hayman@sa.gov.au