Interpreting climate data visualisations to inform adaptation decisions

摘要

Highlights ? An online survey examined subjective interpretations of climate data visualisations. ? Altering visualisation style and/or information content impacts on interpretation. ? Respondents who interpret climate changes as more likely express higher confidence. ? African focused respondents interpret a higher likelihood of drying in the future. ? Variations in individual interpretations are larger than variations between groups. Abstract The appropriate development of graphical visualisations to communicate climate data is fundamental to the provision of climate services to guide climate change adaptation decisions. However, at present there is a lack of empirical evidence, particularly in Africa, to help climate information providers determine how best to communicate and display climate data. To help address this issue, an online survey, primarily targeted at the African vulnerability, impacts and adaptation community, was designed and disseminated widely. The survey examines the interpretation of climate data as a function of the style and information content of graphical visualisations. It is shown that choices made when constructing the visualisations, such as presenting percentile information versus showing the range, significantly impact on interpretation. Results also show that respondents who interpret a higher likelihood of future changes to climate, based on the visualisation of climate model projections, express greater confidence in their interpretations. The findings have relevance to the climate risk community in Africa and elsewhere across the world, and imply that a na?ve approach to visualising climate data risks misinterpretation and unjustified levels of trust, with the potential to misinform adaptation and policy decisions. prs.rt("abs_end"); Keywords Climate change ; Communication ; Uncertainty ; Confidence ; Africa Introduction Visualising climate data is a central component of communicating climate science research findings and climate model results. With a growing demand for climate information to guide climate change adaptation decisions ( Hewitt et al., 2012 ) there is pressure on scientists to ensure that data visualisations are aesthetically attractive and tailored for specific user communities. However, this creates a tension as some visualisation approaches and techniques risk distorting the interpretation of the data; for example, Stauffer et al. (2014) discuss the use of different colour palettes in meteorological visualisations and highlight their potential to mislead. It is therefore crucial to have robust evidence to inform the appropriate design and dissemination of data visualisations to avoid misinterpretation. Yet in the field of climate science and climate change adaptation the current evidence base remains weak. Issues in visualising data to inform decision making are not unique to climate science. In his seminal work on the graphical display of information, Edward Tufte provides a set of principles to guide the development of visualisations across disciplines. Tufte (1983) states that graphical displays should “avoid distorting what the data have to say” and “serve a reasonably clear purpose: description, exploration, tabulation or decoration”. The effectiveness of different visualisation approaches has since been well explored in the health ( Hawley et al., 2008 , Garcia-Retamero and Galesic, 2010 , Galesic, 2011 and Garcia-Retamero and Cokely, 2013 ), environmental hazards ( Gahegan, 1999 , Appleton et al., 2002 , Bostrom et al., 2008 , Martin et al., 2008 and Pang, 2008 ) and computer science literatures ( Robertson, 1990 , Keller et al., 2006 , Aigner et al., 2007 and Vande Moere et al., 2012 ). Other studies routed in the psychological and cognitive science literatures describe common issues in communicating scientific data using visualisations. Carpenter and Shah (1998) discuss the multiple influencing factors that affect a user’s ability to interpret visual displays of information, noting that interpretations are heavily influenced by the viewer’s expectations about, or familiarity with, the graph’s content. In a review of the recent literature, Glazer (2011) similarly reflects on the issue of familiarity but also stresses the importance of scientific literacy, highlighting the need for better education to help people interpret graphical information. There is an increasing literature on the implications of different visualisations to communicate weather data and hazardous weather events ( Haase et al., 2000 , Broad et al., 2007 , Demuth et al., 2012 , Stephens et al., 2012 , Cox et al., 2013 , Radford et al., 2013 and Ash et al., 2014 ), and in the last decade scientists have begun to investigate the use of visualisations to communicate climate change information ( Nicholson-Cole, 2005 and Johansson et al., 2010 ). Kaye et al. (2012) present different approaches to mapping climate data, paying particular attention to the complex issue of communicating