Communication of uncertainty in seasonal prediction and climate projection
Convener: Haleh Kootval  | Co-Conveners: Anca Brookshaw , Tanja Cegnar 
 / Thu, 09 Oct, 16:30–18:30  / Room Tycho
 / Attendance Thu, 09 Oct, 10:30–11:30  / Display Wed, 08 Oct, 15:00–Fri, 10 Oct, 14:00  / Meridian Right Front

The aim of this session is to improve communication of uncertainty in seasonal forecasts and climate projections by sharing experience on how to communicate uncertainty, which is inherent in the seasonal forecasts and climate projections, in an understandable way for stakeholders.

Seasonal forecasts and climate projections are new products compared to other climate services. It is not intuitive for all the potential users what benefits could be obtained using information that is conveyed in a probabilistic form. Many times people have difficulty interpreting the meaning of probabilistic information. The terms »probability« and »uncertainty« are themselves often interpreted in various ways. There is potential for misunderstanding probability information, therefore there is a need to determine exactly what problems stakeholders have with probability information, and how they may be rectified.

We are aiming to investigate use of graphical presentations of uncertainty and nongraphical ways, but also to present probability interpretation tools, which could be used at stakeholder workshops and as part of routine communication with stakeholders. While difficulties with probabilities are sometimes attributed to lack of experience or education, others argue that the problem lies in how information is presented, rather than in some internal cognitive problem. This session seeks to understand some of the ways in which probabilistic information can be communicated in a way that maximizes its comprehension by a wide range of stakeholders.

Several issues that influence how people interpret numerical probabilities will be also addressed, including:
• low probabilities tend to be treated as if they are equal to zero;
• low probabilities in relative risk formats tend to lead to risk aversion compared to
numerical formats;
• specifying the reference class on the probability in question is extremely important to facilitate understanding;
• there is a large variation in how interpret probability words (i.e., the same word can be taken to mean a wide range of numerical values);
• self-serving bias can affect the way people interpret probability words; and
• natural frequency presentation greatly enhances the understanding of probabilities by both field experts and lay people