HEPEX-SIP Topic: Communication and Decision Making (3/4)

HEPEX-SIP Topic: Communication and Decision Making (3/4)

Contributed by Florian Pappenberger, Liz Stephens, Jutta Thielen, Schalk-Jan van Andel and Maria-Helena Ramos

So, what are the needs of decision-makers on communication and use of forecasts? What can we find in the literature?

Decision-maker requirements

dialogueIn order to communicate uncertain forecast, there is the need to harmonize the goals with particular needs of users. It is beneficial to develop ways of presentation of EPS results that allow for the user to connect to the results his or her expert knowledge. Expert-based knowledge interpretation involves methods of interpreting model outputs, specifically when it is related to extreme conditions that are not regularly experienced by forecasters and have not been part of data records of calibration periods (Joslyn and Savelli, 2010). It should be noted that often, even without EPS, forecaster’s judgments are already uncertain in nature. By taking into account secondary information, such as pre-event conditions, earlier performance of the forecast system, and regional weather conditions, forecasters add their own confidence bound to a particular deterministic forecast.

Training and workshops performed (Ramos et al. 2007 and 2013; Demeritt et al, 2007; Pappenberger et al., 2013) show that there is an increase in understanding of how to use uncertainty data in making decisions. Training, e.g.through decision games/simulations,  has an effect on the use of probabilistic forecasts, and it serves to collect feedback from users to develop communication products of EPS. Effective communication of the final forecast products depends on assessment of how users perceive uncertainty and tend to react to information on uncertainty.

Literature that addresses communication and decision-making of ensemble hydrological forecasts

As part of communication, the effectiveness of different visualisation approaches is also a topic for research (Ramos et al., 2010, Zappa et al., 2010, Bruen et al., 2010, Cloke et al., 2009). Ramos et al (2010) and Demeritt et al (2010) have documented a wide range of opinions among operational forecasters across Europe about what is the most important information to extract from HEPS. There are a number of different ways in which such information could be displayed, but there is, as yet, no systematic assessment of their communicative effectiveness. Different users are likely to require different kinds of information from HEPS and thus different visualisations; expert users may require access to multiple parameters, such as previous observed values, skill scores and the full suite of ensemble members, all combined in a single, information-rich display, whereas other users may prefer simpler displays showing simply the probability of threshold exceedance.

Ramos et al.(2007) describe how products were developed in collaboration with users for a concise and useful visualization of probabilistic results in the European Flood Alert System (EFAS). Demeritt et al. (2007) discuss with users of EFAS ensemble predictions what are their perceptions of risk, uncertainty, and error in flood forecasting.

information-gap-decision-theory-decisions-under-severe-uncertaintyBen-Haim (2001) shows how Information-gap decision theory can be used in decisions under severe uncertainty, and Kowalski-Trakofler et al. (2003) analyse judgement and decision making under stress.

Hydrological forecasters can benefit from literature on decision making and communication from other fields (see for example Spiegelhalter et al., 2011). In health risk communication, for example, it is well recognized that drug treatments are regarded more favourably when described in terms of relative rather than absolute risk reduction or the number of patients needed to be treated in order for one actually to benefit (Covey, 2007). Kurz-Milcke et al. (2008) have assessed the effectiveness of visual displays in communicating relative as against absolute risks. Gigerenzer (2002) has shown that medical professionals and the lay public alike understand risk better when it is communicated in terms of natural frequencies rather than probabilities, although Joslyn and Nichols (2009) found the opposite to be true in the case of weather forecasts. There is also an extensive body of research on how probability of precipitation forecasts are understood by the general public (Murphy et al. 1980; Gigerenzer et al. 2005; Morss et al. 2008) as well as other work on public responses to the idea of the 100-year flood (Bell and Tobin 2007) and to the cone of uncertainty visualisations used by the US Hurricane Centre to communicate the likelihood of different storm track paths (Broad et al. 2007).

flood-comms-diagram
Designing effective flood warning systems in Scotland (source: Knowledgescotland)

Proceeding from an implicit ‘deficit model’ much of this work has focused on how the design of risk messages can improve public understandings of scientific uncertainty (Demeritt et al. 2011). However, it is increasingly clear that hydro-meteorological forecasters (Doswell 2004; Elia and Laprise 2005; Demeritt et al. 2007), like other professional groups (Kostopoulou et al. 2009), are not immune to cognitive and other biases that may distort their understanding of uncertainty and risk, and so there are calls for more attention to be paid to the communication of uncertainty between the different professional groups (i.e. from meteorologists to hydrologists, and from hydrologists to civil protection authorities) involved in flood incident management (Faulkner et al. 2007).

________________

Part 1: What is communication and decision making (https://hepex.irstea.fr/hepex-sip-topic-communication-and-decision-making-14)

Part 2: Communication and decision making within HEPEX and related activities (https://hepex.irstea.fr/hepex-sip-topic-communication-and-decision-making-24)

Part 4: Challenges and research needs (https://hepex.irstea.fr/hepex-sip-topic-communication-and-decision-making-44)

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.