Explore the vast range of titles available.

Individuals’ past experiences with a hazard can encompass many different aspects, which can influence how they judge and respond to a future hurricane risk. This study, which utilizes survey data from coastal residents who are at risk from hurricanes, adds to understanding of past hazard experience in two ways. First, it examines six different aspects of people’s past hurricane experiences and the relationships among them. Then, it draws on risk theories of behavioral responses to explore how these different experiences influence people’s evacuation intentions for a hypothetical hurricane as mediated through multiple dimensions of risk perception (cognitive, negative affective) and efficacy beliefs (self efficacy, response efficacy). The results suggest that people can experience emotional or otherwise severe impacts from a hurricane even if they do not have experiences with evacuation, property damage, or financial loss. The results also reveal that different past hurricane experiences operated through different combinations of mediating variables to influence evacuation intentions. Some of these processes enhanced intentions; for instance, experience with evacuation, financial loss, or emotional impacts heightened negative affective risk perceptions, which increased evacuation intentions. Other processes dampened evacuation intentions; for instance, people with past hurricane-related emotional impacts had lower self efficacy, which decreased evacuation intentions. In some cases, these enhancing and dampening processes competed. Exploring people’s different past weather experiences and the mechanisms by which they can influence future behaviors is important for more deeply understanding populations at risk and how they respond to weather threats.

We introduce the National Science Foundation (NSF) AI Institute for Research on Trustworthy AI in Weather, Climate, and Coastal Oceanography (AI2ES). This AI institute was funded in 2020 as part of a new initiative from the NSF to advance foundational AI research across a wide variety of domains. To date AI2ES is the only NSF AI institute focusing on environmental science applications. Our institute focuses on developing trustworthy AI methods for weather, climate, and coastal hazards. The AI methods will revolutionize our understanding and prediction of high-impact atmospheric and ocean science phenomena and will be utilized by diverse, professional user groups to reduce risks to society. In addition, we are creating novel educational paths, including a new degree program at a community college serving underrepresented minorities, to improve workforce diversity for both AI and environmental science.

U.S. National Weather Service (NWS) forecasters assess and communicate hazardous weather risks, including the likelihood of a threat and its impacts. Convection-allowing model (CAM) ensembles offer potential to aid forecasting by depicting atmospheric outcomes, including associated uncertainties, at the refined space and time scales at which hazardous weather often occurs. Little is known, however, about what CAM ensemble information is needed to inform forecasting decisions. To address this knowledge gap, participant observations and semistructured interviews were conducted with NWS forecasters from national centers and local weather forecast offices. Data were collected about forecasters’ roles and their forecasting processes, uses of model guidance and verification information, interpretations of prototype CAM ensemble products, and needs for information from CAM ensembles. Results revealed forecasters’ needs for specific types of CAM ensemble guidance, including a product that combines deterministic and probabilistic output from the ensemble as well as a product that provides map-based guidance about timing of hazardous weather threats. Forecasters also expressed a general need for guidance to help them provide impact-based decision support services. Finally, forecasters conveyed needs for objective model verification information to augment their subjective assessments and for training about using CAM ensemble guidance for operational forecasting. The research was conducted as part of an interdisciplinary research effort that integrated elicitation of forecasters’ CAM ensemble needs with model development efforts, with the aim of illustrating a robust approach for creating information for forecasters that is truly useful and usable.

Reducing loss of life and harm when a hurricane threatens depends on people receiving hurricane risk information that they can interpret and use in protective decisions. To understand and improve hurricane risk communication, this article examines how National Weather Service (NWS) forecasters at the National Hurricane Center and local weather forecast offices, local emergency managers, and local television and radio media create and convey hurricane risk information. Data from in-depth interviews and observational sessions with members of these groups from Greater Miami were analyzed to examine their roles, goals, and interactions, and to identify strengths and challenges in how they communicate with each other and with the public. Together, these groups succeed in partnering with each other to make information about approaching hurricane threats widely available. Yet NWS forecasters sometimes find that the information they provide is not used as they intended; media personnel want streamlined information from NWS and emergency managers that emphasizes the timing of hazards and the recommended response and protective actions; and emergency managers need forecast uncertainty information that can help them plan for different scenarios. Thus, we recommend that warning system partners 1) build understanding of each other's needs and constraints; 2) ensure formalized, yet flexible mechanisms exist for exchanging critical information; 3) improve hurricane risk communication by integrating social science knowledge to design and test messages with intended audiences; and 4) evaluate, test, and improve the NWS hurricane-related product suite in collaboration with social scientists.

Previous work, in which Advanced Microwave Sounding Unit (AMSU) data from the Atlantic Ocean and east Pacific Ocean basins during 1999–2001 were used to provide objective estimates of 1-min maximum sustained surface winds, minimum sea level pressure, and the radii of 34-, 50-, and 64-kt (1 kt ≡ 0.5144 m s−1) winds in the northeast, southeast, southwest, and northwest quadrants of tropical cyclones, is updated to reflect larger datasets, improved statistical analysis techniques, and improved estimation through dependent variable transforms. A multiple regression approach, which utilizes best-subset predictor selection and cross validation, is employed to develop the estimation models, where the dependent data (i.e., maximum sustained winds, minimum pressure, wind radii) are from the extended best track and the independent data consist of AMSU-derived parameters that give information about retrieved pressure, winds, temperature, moisture, and satellite resolution. The developmental regression models result in mean absolute errors (MAE) of 10.8 kt and 7.8 hPa for estimating maximum winds and minimum pressure, respectively. The MAE for the 34-, 50-, and 64-kt azimuthally averaged wind radii are 16.9, 13.3, and 6.8 n mi (1 n mi ≡ 1852 m), respectively.

Risk theories and empirical research indicate that a variety of factors can influence people’s protective decisions for natural hazards. Using data from an online survey that presented coastal U.S. residents with a hypothetical hurricane scenario, this study investigates the relative importance of cognitive risk perceptions, negative affect, efficacy beliefs, and other factors in explaining people’s anticipated evacuation decisions. The analysis finds that multiple factors, including individual and household characteristics, previous experiences, cognitive and affective risk perceptions, and efficacy beliefs, can help predict hurricane evacuation intentions. However, the largest amount of variance in survey participants’ evacuation intentions is explained by their evacuation-related response efficacy (coping appraisals) and their perceived likelihood of getting hurt if they stay home during the storm. Additional analysis explores how risk perceptions and efficacy beliefs interact to influence people’s responses to risk information. Although further investigation in additional situations is needed, these results suggest that persuading people at high risk that evacuating is likely to reduce harm can serve as an important risk communication lever for motivating hurricane evacuation.

Confidence is a concept important to weather prediction, shaping how risk information is created, shared, understood, and acted upon. For forecasters in the National Weather Service (NWS) and their partners in public safety, confidence is central to their work, appearing frequently during their decision support services. While confidence has been examined in a variety of literatures, it is often addressed simplistically or as one of many variables in a study. It is rarely the object of study in and of itself, even less so in a naturalistic setting like an operational environment. To build a more robust knowledge of confidence and its many dimensions, we conducted a multisited ethnography of three interrelated sites central to tornado prediction and information dissemination, leading up to and during a cool-season tornado event. In partnership with collaborators from the NWS and emergency management, we simultaneously deployed to a National Center, a local Weather Forecast Office, and an emergency management office. This article explicates confidence from multiple social science theories, considering the scientific, data-based roots of confidence, as well as its affective, relational, and procedural origins. Our results show that confidence emerges in varied and complex ways and at different scales. Confidence can indicate one’s assessment of evidence and agreement (or lack) of it, beliefs about partners’ future behavior based on past experiences, and ritual interactions between offices that create patterned expectations. We argue for a more robust interdisciplinary analysis of confidence given how it shapes weather-related policies and practices, technologies, and communication strategies.

The study reported here explores how to enhance the public value of hurricane forecast and warning information by examining the entire warning process. A mental models research approach is applied to address three risk management tasks critical to warnings for extreme weather events: 1) understanding the risk decision and action context for hurricane warnings, 2) understanding the commonalities and conflicts in interpretations of that context and associated risks, and 3) exploring the practical implications of these insights for hurricane risk communication and management. To understand the risk decision and action context, the study develops a decision-focused model of the hurricane forecast and warning system on the basis of results from individual mental models interviews with forecasters from the National Hurricane Center (n = 4) and the Miami–South Florida Weather Forecast Office (n = 4), media broadcasters (n = 5), and public officials (n = 6), as well as a group decision-modeling session with a subset of the forecasters. Comparisons across professionals reveal numerous shared perceptions, as well as some critical differences. Implications for improving extreme weather event forecast and warning systems and risk communication are threefold: 1) promote thinking about forecast and warning decisions as a system, with informal as well as formal elements; 2) evaluate, coordinate, and consider controlling the proliferation of forecast and warning information products; and 3) further examine the interpretation and representation of uncertainty within the hurricane forecast and warning system as well as for users.

Artificial intelligence (AI) and machine learning (ML) pose a challenge for achieving science that is both reproducible and replicable. The challenge is compounded in supervised models that depend on manually labeled training data, as they introduce additional decision‐making and processes that require thorough documentation and reporting. We address these limitations by providing an approach to hand labeling training data for supervised ML that integrates quantitative content analysis (QCA)—a method from social science research. The QCA approach provides a rigorous and well‐documented hand labeling procedure to improve the replicability and reproducibility of supervised ML applications in Earth systems science (ESS), as well as the ability to evaluate them. Specifically, the approach requires (a) the articulation and documentation of the exact decision‐making process used for assigning hand labels in a “codebook” and (b) an empirical evaluation of the reliability” of the hand labelers. In this paper, we outline the contributions of QCA to the field, along with an overview of the general approach. We then provide a case study to further demonstrate how this framework has and can be applied when developing supervised ML models for applications in ESS. With this approach, we provide an actionable path forward for addressing ethical considerations and goals outlined by recent AGU work on ML ethics in ESS. Plain Language Summary Artificial intelligence and machine learning can make it hard to do science in a way that can be repeated. This can mean redoing a study in the exact same way to see if you can get the same or similar results (reproducibility) or trying to use the same study design on a new problem to see if the results are the same or similar (replicability). These types of scientific repetitions is important for developing robust knowledge, but is hard to do with certain types of machine learning that rely on data that were categorized by researchers. The researchers have to make decisions and categorize their data, which the machine learning algorithm then uses as a guide to make its own decisions. Generally, there is not enough information shared by the researchers about how these decisions were made to repeat the science or evaluate how good it is. In this paper, we provide a way to address these shortcomings. The approach and example we offer illustrates how to (a) create a rulebook that can be shared for how to make decisions and (b) quantitatively measure how consistent the researchers are at using that rulebook to make their decisions. Key Points We provide a rigorous hand labeling procedure to improve the replicability and reproducibility of supervised machine learning (ML) Our case study and step‐by‐step guide clearly outline how the procedure can be applied The procedure is an actionable path forward for addressing ethical considerations and goals for ML development in Earth systems science

This study uses data from a survey of coastal Miami-Dade County, Florida, residents to explore how different types of forecast and warning messages influence evacuation decisions, in conjunction with other factors. The survey presented different members of the public with different test messages about the same hypothetical hurricane approaching Miami. Participants’ responses to the information were evaluated using questions about their likelihood of evacuating and their perceptions of the information and the information source. Recipients of the test message about storm surge height and the message about extreme impacts from storm surge had higher evacuation intentions, compared to nonrecipients. However, recipients of the extreme-impacts message also rated the information as more overblown and the information source as less reliable. The probabilistic message about landfall location interacted with the other textual messages in unexpected ways, reducing the other messages’ effects on evacuation intentions. These results illustrate the importance of considering trade-offs, unintended effects, and information interactions when deciding how to convey weather information. Recipients of the test message that described the effectiveness of evacuation had lower perceptions that the information was overblown, suggesting the potential value of efficacy messaging. In addition, respondents with stronger individualist worldviews rated the information as significantly more overblown and had significantly lower evacuation intentions. This illustrates the importance of understanding how and why responses to weather messages vary across subpopulations. Overall, the analysis demonstrates the potential value of systematically investigating how different people respond to different types of weather risk messages.