Short course training: Robust decision making and uncertainty

Submitted by Sukaina Bharwani | published 25th Mar 2011 | last updated 15th Jun 2018

Introduction

Having talked a lot during the week about the uncertainties associated with both climatic and socio-economic projections of the future, this session started to focus on how to make adaptation decisions given these uncertainties.

We discussed whether climate prediction and the need for better models is a 'limit to adaptation' - if you had $100bn to spend on adaptation research, how would you spend it? The feeling in the group seemed to be that while we should be trying to improve climate modeling there should be a big focus on pilot testing activities to find out what works and what doesn't on the ground. While some of the areas of uncertainty in climate models are likely to be able be reduced through further research, some of the uncertainty is likely to be irreducible, due to the complex and partly chaotic nature of the climate system, and certain limits to our ability to model complex systems (Stainforth and Harrison 2009). In order to start to deal with this uncertainty we need to get a sense of what the range of projections of future climate looks like; this is what the Climate Change Explorer was designed to help with.

Importantly, we need to accept these uncertainties and work with them, rather than assuming that we can predict the future and then adapt to this future state. Over-confidence in our projections of future climatic conditions, and reliance on one single projection of the many possible futures, could well lead to Maladaptation.

Not just climatic uncertainties. . .

Have you seen any flying cars recently?

 

Have you seen any flying cars recently?

 

Wilby and Dessai 2009 Cascade of Uncertainty
Wilby and Dessai (2009) Cascade of Uncertainty

 

When we are talking about adaptation and vulnerability it is not just changes in climate that will have an effect but also future socio-economic, political, cultural and technological developments (for example population growth, market prices, communication technologies etc), which in many cases will play have a greater effect on vulnerability than climatic factors. The difficulty of trying to forecast these changes is at least as great as trying to predict future climate, as illustrated by the top image, which is taken from a 1960s publication about how we would be living in 2000. . .

The diagram above, taken from Wilby and Dessai (2009), illustrates the way that uncertainty is amplified throughout the system.

All this is to say that we cannot see into the future and we should not assume that we can and try to take a 'predict and adapt' approach. This does not mean we don't know anything; we have models which can give us valuable information about possible future conditions, and help us to take more informed decisions. The approaches outlined below can be used as ways of taking informed adaptation decisions that take into account uncertainty but allow us to act.

Robust Decisions

Framework for decisions under uncertainty: Wilby and Dessai 2009
Framework for decisions under uncertainty: Wilby and Dessai 2009

A way round uncertainty is to look for adaptation strategies that are not sensitive to specific combinations of future conditions, but will be of benefit under a wide range of possible future states. For example rather than planning for a damn that will have a 100 year lifespan and fulfill its purpose only under the set of conditions that temperature will increase by no ore than 3°C, precipitation has to increase by 20% and population of the nearby city won't exceed 2 million people, we are looking for options that perform well under a range of states. A formalised way of doing this is to use a Robust Decision-Making framework developed by the RAND corporation.

Another way of addressing the problem is by using the framework illustrated in the diagram above, also from Wilby and Dessai 2009. The basic steps are:

[Given your problem/what you are looking at]

  • Create a plausible list of adaptation strategies that you think would improve the situation.
  • Of these options, which are a set of preferred options (i.e are socially desirable, affordable, technically feasible etc)
  • Create a list of plausible scenarios of future conditions (both climatic and socio-economic) based on available information and understanding.
  • Test your preferred options against the different scenarios; how well would each option perform under different scenarios of the future? Here we are looking for options that are likely perform well across the range of scenarios (this is not necessarily the BEST option for any given scenario).
  • Choose an option that you find sufficiently robust given your decision-context. There will always be some risk involved, but given the situation, what is acceptable to you?

The session also touched on the idea of adaptive management; making the first decision on an adaptation pathway, assessing how this action is performing, assessing any new information and then deciding whether to carry on with that course of action or to reassess and take a different course of action. This links to ideas of Adaptation Landscapes discussed in stream 1 on Adaptation Principles and Planning.

Resources

Download the presentation here

Papers