With the exception of the probability of precipitation, information about uncertainty is rarely communicated in public weather forecasts, even though it could improve weather-related decisions with important economic and safety consequences. A recent National Research Council report found that assessing, communicating and disseminating forecast uncertainty information to the end-user is the largest as yet unaddressed task before the weather forecasting community. The purpose of this project is to bridge that gap with the efforts of a multi-disciplinary team of psychologists, atmospheric scientists, statisticians, and applied physicists.
This project will define the psychological processes involved in understanding and using weather forecast uncertainty in realistic contexts among non-expert users. Research will be conducted to determine the impact of alternative methods of presenting information, including verbal explanations as well as visual images and graphs.
The results of this research will be applied to producing an unprecedented prototype weather forecasting website that presents forecasts to the public that are entirely probabilistic. It will give uncertainty-bracketed forecasts for temperature, precipitation and wind speed. It will also provide warning information for extreme events, such as high winds, extreme temperatures and extreme amounts of rainfall. Although such warnings have important safety implications, there has been very little research devoted to the methods for producing or communicating this information. This will require the development of innovative methods of analyzing and displaying probabilistic information for extreme events.
The project will also develop and provide graphics that compare past probabilistic forecasts to observed values, to demonstrate to users that although observed values may not always match single value forecasts, they are often within the boundaries of the probability range. This feature will enhance understanding of the probabilistic nature of the forecast as well as improving user trust.
