The problem

In order for decision makers to manage the risks posed by climate change, climate model projections need to be communicated clearly. Placing climate risk in the context of other risk factors is essential to understand the relative importance of climate change and how climate information ultimately affects a decision. However, there is often a mismatch between the information contained within climate model projections and that which is needed by decision makers.

There is a considerable demand for more accurate and more detailed climate projections. Given both the practical and scientific challenges of providing such information to decision makers, it is essential to communicate climate model output in a sensible and consistent manner. Given that probabilistic information is now available within the UK, e.g. UK Climate Projections ’09 (UKCP09), and mindful of the need to communicate such information appropriately, tools must be developed which are tailored to specific climate change adaptation decisions.

The approach

A Bayesian Network (BN) is a graphical model with an underlying probabilistic framework, which characterises and quantifies an outcome of interest, and the variables and their interactions associated with this outcome. They allow for the explicit quantification of risk and uncertainty combining evidence from diverse sources incorporating subjective beliefs and objective data.

In order to determine how they might aid climate change adaptation decisions, we developed a case study applying the tool to a specific decision problem. We decided to focus on the UK Health Infrastructure and more specifically on the cooling options available to hospitals and health facilities to cope with potential future temperature increases. The case study centred on consultations with two UK hospitals; Guy’s and St Thomas’ Foundation Trust in central London and the Royal Devon and Exeter Foundation Trust in the southwest.

Using the software Netica, we built a BN for a typical capital project at both of the hospitals and investigated the way that probabilistic information from UKCP09 can be combined with observed climate data. In a simple illustrative example, we showed that current evidence indicates an increased risk of exceeding a crucial temperature threshold (28 °C) within hospital buildings towards the end of the century.