|
|
|
|
| NCAR > SIP > Projects | ||||
 |
|
|||
|
|
Projects
Communicating Uncertainty in Weather Forecasts: A Survey of the U.S. Public Weather forecasts are inherently uncertain, and meteorologists have substantial situation-specific information about weather forecast uncertainty that is not readily available to most forecast users. This uncertainty information has potential to benefit users by helping them make more informed forecast-related decisions. Yet effectively communicating weather forecast uncertainty to non-meteorologists remains challenging. Improving uncertainty communication requires research-based knowledge that can inform decisions on what weather forecast uncertainty information to communicate and when and how to do so. To help build such knowledge, this research examines non-meteorologists’ perspectives on weather forecast uncertainty and uncertainty information through a nationwide survey. Based on a national sample survey of over 1500 households, results show that most people infer uncertainty into deterministic weather forecasts. Although individuals’ confidence in forecasts varied, respondents tended to have more confidence in shorter lead time forecasts and more confidence in temperature than precipitation forecasts. Examining people’s interpretations of probability of precipitation forecasts, the article finds that most do not know the technical definition of these uncertainty forecasts, despite their familiarity in the U.S., but argues that a more important question is whether the forecast information benefits people’s decisions. A significant majority of respondents were willing to receive weather forecasts that included uncertainty information, and at least in the contexts tested, many respondents preferred such forecasts to deterministic forecasts. Respondents also had preferences about how forecast uncertainty information is conveyed. Further research is needed to investigate these findings in other contexts and to explore key aspects of interpretation, communication, and use of weather forecast uncertainty information in greater detail. Mental Modeling: Expert Influence Diagrams and Communication of Forecast Uncertainty Mental modeling is the art and science of understanding people's concepts and thinking in a particular domain. Different members of the weather forecasting community and its stakeholders consider the weather forecast system using different mental models, leading to gaps in the production, communication, and use of weather forecast information. To bridge these gaps, SIP researchers are conducting two projects using mental modeling methods. The first project uses interviews with research meteorologists to develop a "consensus mental model" of the weather forecasting system, called an expert influence diagram, then illustrates how the diagram provides a framework for discussing several issues of interest to the weather forecasting community. The second project uses interviews with forecasters, intermediaries, and end-users in specific areas of weather and climate forecasting to explore how understanding the groups' different mental models can help improve generation of probabilistic forecasts and communication of forecast uncertainty. Sensitivity of U.S. Economic Productivity to Weather Although it is known that many economic sectors are sensitive to weather and weather information, this sensitivity has not yet been rigorously evaluated. To fill this gap, the SIP initiated a project to quantify the impact of weather and weather information on twelve major sectors of the U.S. economy, including agriculture, construction, utilities, and manufacturing. Sectoral economic productivity is modeled against standard economic production inputs of capital, labor, and energy, and as a function of temperature, precipitation, and major storm events. The sector and state-specific model estimates are then aggregated nationally to quantify the sensitivity of the entire U.S. economy to weather variation. The results represent an objective evaluation of the potential upper-bound value for atmospheric products and services. Societal Value of Improved Quantitative Precipitation Forecasts: A Synthesis The U. S. Weather Research Program (USWRP) and U. S. National Weather Service (NWS) have identified Quantitative Precipitation Forecasts (QPFs) as an important area for future high-impact forecast improvements. In support of the USWRP's and NWS's QPF-related goals, SIP researchers are conducting a project with three primary objectives: 1) to synthesize existing knowledge about the societal value of current and improved QPFs; 2) to summarize methodologies that have been or could be used to elicit the value of QPFs and other weather forecasts; and 3) to investigate which types of QPF improvements — such as spatial resolution, timing, and forecast lead time — are likely to be "highest-impact," i.e., to most benefit society. A report summarizing the results will be available in summer 2005. THORPEX: A Global Atmospheric Research Programme THORPEX: A Global Atmospheric Research Programme is an international program (under the auspices of the World Meteorological Organization) to improve the skill of high-impact weather forecasts for the benefit of society, the economy, and the environment. The NCAR SIP team is playing a major role in developing the Societal and Economic Applications (SEA) component of the International, North American, and U.S. THORPEX efforts. International THORPEX SEA research goals include i) evaluating the net economic benefits of THORPEX improvements in weather forecasting; ii) assessing and improving weather forecast information; and iii) assisting with product development and the transfer of tools and knowledge, especially to developing countries. Value of current and improved weather forecasts to U.S. households SIP researchers are collaborating with Stratus Consulting to conduct the first comprehensive national random sample survey of U.S. households to explore their uses, perceptions, sources, and preferences and values for current and improved short to mid term (up to 14 days) weather forecast information. This work builds on a research program initiated in 1999 and previously funded by the National Oceanographic and Atmospheric Administration (NOAA) to explore households' values for weather forecast information. Leveraging existing survey instruments, and building on guidance from an external expert review panel conducted on behalf of NOAA, the current project advances the prior work to elicit information from households on their (1) sources of daily weather forecasts, (2) perceptions of the quality of different types of weather forecast information, (3) uses of daily weather forecasts in decision making, (4) values for current weather forecast services, (5) values for improved weather forecast services in the short to near term for different attributes of weather forecast products (e.g., more accurate precipitation or temperature forecasts), and (6) general preferences for improved information on other weather forecast products. Digital Library for Societal Impacts The SIP is developing the Digital Library for Societal Impacts (DLSI), a Web-based resource for collecting and disseminating research findings related to the use and value of weather forecasts. The library will provide the research and policy-making communities with easy, organized access to a wide variety of resources, including research results, case studies, Web sites, and decision support tools. The current DLSI collection focuses on literature related to tropical cyclones. You can access a prototype version of the digital library here. THORPEX International Outreach In collaboration with UCAR's Cooperative Program for Operational Meteorology Education & Training (COMET), SIP visitor John Cahir (The Pennsylvania State University) is developing a blueprint for how the application of new science and technology developed in THORPEX can improve the capability of meteorological services internationally, particularly in developing countries. The planned work will be organized along societal and economic themes that have the potential to be impacted by results from THORPEX. New Methodologies for Evaluating Benefits of Improved Forecasts The SIP will host a two-part workshop in 2005 to develop a new cadre of researchers exploring the benefits of weather information. The workshop will be developed and organized by SIP visitor Eve Gruntfest (University of Colorado at Colorado Springs). Building on tools from social sciences (including economics, geography, planning, communication, and public administration), workshop participants will develop and test methodologies for effective evaluation. The workshop will offer new methodologies to the societal impacts arena, as well as providing the groundwork for future case studies and facilitating applied studies. After the second session, a methods workbook including the lessons from participants' case studies will be available for broad dissemination. Tropical Cyclone Social Science Research Agenda Tropical cyclones have significant social and economic impacts that may be mitigated in part by forecasting and warning systems. Greater incorporation of economic and social dimensions into tropical cyclone forecasting and warning promises large dividends in terms of relevance and user response. To promote such incorporation, the SIP organized a workshop in the spring of 2005 to identify social science research capabilities, needs, and priorities with respect to the tropical cyclone forecasting and warning system. The workshop report (Draft Workshop Report) details an initial compilation of concepts for tropical cyclone social science research, including study of the communication, perception and understanding, behavioral responses, and costs and benefits of hurricane forecast information products. Applied Probabilistic Meteorology Leonard Smith and collaborators at the London School of Economics are examining the problem of constructing probabilistic weather and climate forecasts from an ensemble of model simulations. Issues being addressed include: (1) the underlying theoretical justifications for methods used to generate estimates of the probability distribution conditioned on ensemble forecasts, (2) the relative performance of parametric and non-parametric methods for generating the distributions used for "dressing" individual ensemble members, (3) how the value of forecast error archives varies with their size, (4) the relative performance of different classes of ensemble and ensemble formation schemes, and (5) resource allocations between improving resolution and increasing ensemble size. The researchers are looking at two case studies: the efficiency of combined-cycle gas turbine generators and long-line transmission of electricity. |
|||
|
||||
