Browse
Abstract
The likelihood of threatening events is often simplified for members of the public and presented as risk categories such as the “watches” and “warnings” currently issued by National Weather Service in the United States. However, research (e.g., Joslyn and LeClerc) suggests that explicit numeric uncertainty information—for example, 30%—improves people’s understanding as well as their decisions. Whether this benefit extends to dynamic situations in which users must process multiple forecast updates is as yet unknown. It may be that other likelihood expressions, such as color coding, are required under those circumstances. The experimental study reported here compared the effect of the categorical expressions “watches” and “warnings” with both color-coded and numeric percent chance expressions of the likelihood of a tornado in a situation with multiple updates. Participants decided whether and when to take shelter to protect themselves from a tornado on each of 40 trials, each with seven updated tornado forecasts. Understanding, decision quality, and trust were highest in conditions that provided percent chance information. Color-coded likelihood information inspired the least trust and led to the greatest overestimation of likelihood and confusion with severity information of all expressions.
Abstract
The likelihood of threatening events is often simplified for members of the public and presented as risk categories such as the “watches” and “warnings” currently issued by National Weather Service in the United States. However, research (e.g., Joslyn and LeClerc) suggests that explicit numeric uncertainty information—for example, 30%—improves people’s understanding as well as their decisions. Whether this benefit extends to dynamic situations in which users must process multiple forecast updates is as yet unknown. It may be that other likelihood expressions, such as color coding, are required under those circumstances. The experimental study reported here compared the effect of the categorical expressions “watches” and “warnings” with both color-coded and numeric percent chance expressions of the likelihood of a tornado in a situation with multiple updates. Participants decided whether and when to take shelter to protect themselves from a tornado on each of 40 trials, each with seven updated tornado forecasts. Understanding, decision quality, and trust were highest in conditions that provided percent chance information. Color-coded likelihood information inspired the least trust and led to the greatest overestimation of likelihood and confusion with severity information of all expressions.
Abstract
This study examines risk perceptions and evacuation planning for those residents affected by Hurricane Laura—the first major hurricane evacuation during the COVID-19 pandemic—and Hurricane Sally, prior to the widespread availability of vaccines. Research on hurricane evacuation behavior and risk perceptions during a pandemic is critical for quantifying the intersect of these compounding threats. Analyses captured how people perceive public shelters and whether evacuation choices changed in light of the pandemic. Many study participants considered themselves vulnerable to COVID-19 (39.4%), and two-thirds believed it would be “very serious” if they or their loved ones contracted COVID-19, but this had no impact on their actual evacuation decision-making. Approximately 75% of the sample stayed at home during Hurricanes Laura or Sally, and, of these, just over 80% indicated that COVID-19 was a somewhat important deciding factor. This reflects the partial role that COVID-19 played in balancing individual and household protective action decision-making during complex disasters, whereas 15.5% wanted to evacuate but waited until it was too late. For those who evacuated to a hotel, many found that staff and guests wore masks and socially distanced in common spaces. Of particular interest is that individuals have a continued negative perception of public shelters’ ability to safeguard against COVID-19 that was coupled with a significant decrease in the number of respondents that would potentially use shelters in 2020 as compared with before the COVID-19 pandemic. These results have informed and will inform future hazard mitigation planning during the current pandemic or future pandemics or infectious disease outbreaks.
Abstract
This study examines risk perceptions and evacuation planning for those residents affected by Hurricane Laura—the first major hurricane evacuation during the COVID-19 pandemic—and Hurricane Sally, prior to the widespread availability of vaccines. Research on hurricane evacuation behavior and risk perceptions during a pandemic is critical for quantifying the intersect of these compounding threats. Analyses captured how people perceive public shelters and whether evacuation choices changed in light of the pandemic. Many study participants considered themselves vulnerable to COVID-19 (39.4%), and two-thirds believed it would be “very serious” if they or their loved ones contracted COVID-19, but this had no impact on their actual evacuation decision-making. Approximately 75% of the sample stayed at home during Hurricanes Laura or Sally, and, of these, just over 80% indicated that COVID-19 was a somewhat important deciding factor. This reflects the partial role that COVID-19 played in balancing individual and household protective action decision-making during complex disasters, whereas 15.5% wanted to evacuate but waited until it was too late. For those who evacuated to a hotel, many found that staff and guests wore masks and socially distanced in common spaces. Of particular interest is that individuals have a continued negative perception of public shelters’ ability to safeguard against COVID-19 that was coupled with a significant decrease in the number of respondents that would potentially use shelters in 2020 as compared with before the COVID-19 pandemic. These results have informed and will inform future hazard mitigation planning during the current pandemic or future pandemics or infectious disease outbreaks.
Abstract
Although there is clear evidence that proximity to a tornado or forecast tornado increases an individual’s risk perception, the specific relationships between risk personalization and spatial variables are unclear. It has also been established that one’s own evaluation of distance does not always match objective measurement. This study sought to explain the differences in the distance at which an individual would personalize the risk from a tornado across personally relevant geospatial factors such as the distance between places frequented (e.g., home and work), urban/rural classification of the area, and the length of residence in the county. A survey of 1023 respondents across eight states (Alabama, Arkansas, Georgia, Kentucky, Louisiana, Mississippi, Missouri, and Tennessee) was used to obtain risk personalization distances, which were distinguished as “worry distances” (the distances at which one would worry about their house or loved ones, or take protective action) and “confirmation distances” (the distances at which one would expect to see, hear, or feel the effects of a tornado). We found that individuals who traveled greater distances and traveled more frequently to the grocery store and another location, those who self-defined their area as urban, and those with advanced degrees had increased risk personalization distances. Lengthier residency in the county influenced these distances as well. Future research is required to better comprehend the relationship of place, risk perception, and geographic mobility on protective action when a tornado occurs.
Significance Statement
Greater tornado risk personalization distances were associated with self-defining as urban, having an advanced degree, and driving farther and more frequently to the grocery store and to another location. Longer length of residence was associated with shorter risk personalization distances. With rural participants expressing shorter tornado risk personalization distances, warning communicators with the ability to tailor messages to multiple communities may wish to adjust messages no the basis of whether they are targeted to rural communities or to urban communities.
Abstract
Although there is clear evidence that proximity to a tornado or forecast tornado increases an individual’s risk perception, the specific relationships between risk personalization and spatial variables are unclear. It has also been established that one’s own evaluation of distance does not always match objective measurement. This study sought to explain the differences in the distance at which an individual would personalize the risk from a tornado across personally relevant geospatial factors such as the distance between places frequented (e.g., home and work), urban/rural classification of the area, and the length of residence in the county. A survey of 1023 respondents across eight states (Alabama, Arkansas, Georgia, Kentucky, Louisiana, Mississippi, Missouri, and Tennessee) was used to obtain risk personalization distances, which were distinguished as “worry distances” (the distances at which one would worry about their house or loved ones, or take protective action) and “confirmation distances” (the distances at which one would expect to see, hear, or feel the effects of a tornado). We found that individuals who traveled greater distances and traveled more frequently to the grocery store and another location, those who self-defined their area as urban, and those with advanced degrees had increased risk personalization distances. Lengthier residency in the county influenced these distances as well. Future research is required to better comprehend the relationship of place, risk perception, and geographic mobility on protective action when a tornado occurs.
Significance Statement
Greater tornado risk personalization distances were associated with self-defining as urban, having an advanced degree, and driving farther and more frequently to the grocery store and to another location. Longer length of residence was associated with shorter risk personalization distances. With rural participants expressing shorter tornado risk personalization distances, warning communicators with the ability to tailor messages to multiple communities may wish to adjust messages no the basis of whether they are targeted to rural communities or to urban communities.
Abstract
This paper explores the internal migration of fishers from coastal communities of Bangladesh in response to extreme weather events. It also assesses the vulnerabilities to extreme weather events of these coastal areas, in general, and of targeted fishing communities, in particular. This qualitative study employs a combination of methods, semistructured interviews and observations, in two villages located in the eastern part of Kalapara Upazila, Patuakhali district of Bangladesh. The results indicate that the participants of the study are susceptible to the vulnerability of extreme weather events due to their households’ socioeconomic and geographical location. This study shows that most people from the fishing communities do not migrate to other places to escape from the vulnerabilities as they have high dependency on fish-related activities. Also, there are various socioeconomic and cultural factors that hinder their migration, including the Mohajon–Dadon system, migration costs, lack of skills and resources, and fear of income insecurity. Instead of migrating, they develop their own traditional adaptation mechanisms to ensure their survival. These people remain underrepresented and are not adequately recorded in national or regional migration data.
Abstract
This paper explores the internal migration of fishers from coastal communities of Bangladesh in response to extreme weather events. It also assesses the vulnerabilities to extreme weather events of these coastal areas, in general, and of targeted fishing communities, in particular. This qualitative study employs a combination of methods, semistructured interviews and observations, in two villages located in the eastern part of Kalapara Upazila, Patuakhali district of Bangladesh. The results indicate that the participants of the study are susceptible to the vulnerability of extreme weather events due to their households’ socioeconomic and geographical location. This study shows that most people from the fishing communities do not migrate to other places to escape from the vulnerabilities as they have high dependency on fish-related activities. Also, there are various socioeconomic and cultural factors that hinder their migration, including the Mohajon–Dadon system, migration costs, lack of skills and resources, and fear of income insecurity. Instead of migrating, they develop their own traditional adaptation mechanisms to ensure their survival. These people remain underrepresented and are not adequately recorded in national or regional migration data.
Abstract
Integration of the social sciences into climate assessments enhances report content and actionable science. The literature has identified the benefits and challenges in achieving coequal intellectual partnerships between the social and biogeophysical sciences in climate research. Less has been written on how to rectify the issue in the particular institutional context of a climate assessment. This article uses qualitative research methods to analyze social science integration in the United States’ Fourth National Climate Assessment. It presents findings from focus groups held with social science– and nonsocial science–trained report authors. It finds that knowledge governance, or the formal and informal mechanisms shaping how information is produced and used, and cultural worldviews about the role of social sciences in assessments and assessments in society, affected social science integration. Report authors’ principal orientation toward the social sciences was as a means of achieving what they saw as the assessment’s public function, namely, to support education, decision-making, and action. Author expertise, report framing, and knowledge systems were other key themes that emerged. Based on this analysis, we propose potential pathways toward coequal intellectual partnerships in assessments by expanding the diversity of chapter teams’ expertise, enhancing connections between authors and society, reconsidering report framing, and broadening inclusion of knowledge systems. We also discuss the potential role of applying social science theories and methods throughout the report life cycle from framing and engagement to evaluation.
Significance Statement
We wanted to understand why the Fourth U.S. National Climate Assessment used the social sciences as it did in order to identify opportunities and obstacles for enhancing social science inclusion. To do so, we held focus groups with assessment authors on their experiences with writing the report. This approach lends insight into the evolving integration of social sciences in climate assessments. Its implications for how to better integrate the social and biogeophysical sciences may be of particular interest to authors and managers of global change assessments and to other readers working on interdisciplinary climate research projects. Future studies could investigate similarities and differences in incorporating the social sciences into global, national, and state-level assessments.
Abstract
Integration of the social sciences into climate assessments enhances report content and actionable science. The literature has identified the benefits and challenges in achieving coequal intellectual partnerships between the social and biogeophysical sciences in climate research. Less has been written on how to rectify the issue in the particular institutional context of a climate assessment. This article uses qualitative research methods to analyze social science integration in the United States’ Fourth National Climate Assessment. It presents findings from focus groups held with social science– and nonsocial science–trained report authors. It finds that knowledge governance, or the formal and informal mechanisms shaping how information is produced and used, and cultural worldviews about the role of social sciences in assessments and assessments in society, affected social science integration. Report authors’ principal orientation toward the social sciences was as a means of achieving what they saw as the assessment’s public function, namely, to support education, decision-making, and action. Author expertise, report framing, and knowledge systems were other key themes that emerged. Based on this analysis, we propose potential pathways toward coequal intellectual partnerships in assessments by expanding the diversity of chapter teams’ expertise, enhancing connections between authors and society, reconsidering report framing, and broadening inclusion of knowledge systems. We also discuss the potential role of applying social science theories and methods throughout the report life cycle from framing and engagement to evaluation.
Significance Statement
We wanted to understand why the Fourth U.S. National Climate Assessment used the social sciences as it did in order to identify opportunities and obstacles for enhancing social science inclusion. To do so, we held focus groups with assessment authors on their experiences with writing the report. This approach lends insight into the evolving integration of social sciences in climate assessments. Its implications for how to better integrate the social and biogeophysical sciences may be of particular interest to authors and managers of global change assessments and to other readers working on interdisciplinary climate research projects. Future studies could investigate similarities and differences in incorporating the social sciences into global, national, and state-level assessments.
Abstract
Tornadoes are responsible for considerable property damage and loss of life across Oklahoma. While several studies have explored drivers of tornado adjustment behaviors, their results are not consistent in terms of their significance and direction. To address this shortcoming in the literature, we surveyed households using a disproportionate stratified sampling procedure from counties in Oklahoma that frequently experience tornado threats to explore drivers of adjustments. We used structural equation modeling (SEM) to explore relationships among variables highlighted in the protection motivation theory (PMT) and related literature that affect adjustment intentions and risk perceptions. Overall, we found that the factors highlighted in the PMT are effective at explaining households’ intentions of adopting adjustment behaviors associated with tornado hazards. Threat appraisals, however, were less important than coping appraisals in explaining tornado hazard adjustment intentions. In further analysis, we grouped adjustments as 1) basic (e.g., flashlight, food supply, and water supply) and 2) complex (e.g., insurance and storm shelter), and we found that while coping appraisals are significant drivers of both adjustment categories, the effect of threat appraisals is only significant for complex adjustment intentions. We also found that emotional responses to hazards are major drivers of threat appraisals, stronger than perceived knowledge and hazard salience. Moreover, we found that demographic characteristics affect both adjustment intentions and threat appraisals. The additions to the PMT and categorization of adjustment activities improve our understanding of the PMT in different contexts. Such insights provide scholars and emergency managers with strategies for risk communication efforts.
Significance Statement
Tornadoes have caused considerable property damage and loss of life across the state of Oklahoma. Here, we utilize the protection motivation theory (PMT) to explore drivers of tornado hazard adjustment intentions by surveying households from counties in Oklahoma that frequently experience tornadoes. Overall, we found that threat appraisals and coping appraisals produce differential effects depending on the type of hazard adjustment in question. Our findings show that risk perceptions are not a significant explanatory variable of basic adjustments (e.g., flashlight, food supply, and water supply) but are a significant explanatory variable of complex adjustments (e.g., insurance and storm shelter). Future work should provide broader perspectives on how to advance the PMT to better explain adjustment intentions for various hazards.
Abstract
Tornadoes are responsible for considerable property damage and loss of life across Oklahoma. While several studies have explored drivers of tornado adjustment behaviors, their results are not consistent in terms of their significance and direction. To address this shortcoming in the literature, we surveyed households using a disproportionate stratified sampling procedure from counties in Oklahoma that frequently experience tornado threats to explore drivers of adjustments. We used structural equation modeling (SEM) to explore relationships among variables highlighted in the protection motivation theory (PMT) and related literature that affect adjustment intentions and risk perceptions. Overall, we found that the factors highlighted in the PMT are effective at explaining households’ intentions of adopting adjustment behaviors associated with tornado hazards. Threat appraisals, however, were less important than coping appraisals in explaining tornado hazard adjustment intentions. In further analysis, we grouped adjustments as 1) basic (e.g., flashlight, food supply, and water supply) and 2) complex (e.g., insurance and storm shelter), and we found that while coping appraisals are significant drivers of both adjustment categories, the effect of threat appraisals is only significant for complex adjustment intentions. We also found that emotional responses to hazards are major drivers of threat appraisals, stronger than perceived knowledge and hazard salience. Moreover, we found that demographic characteristics affect both adjustment intentions and threat appraisals. The additions to the PMT and categorization of adjustment activities improve our understanding of the PMT in different contexts. Such insights provide scholars and emergency managers with strategies for risk communication efforts.
Significance Statement
Tornadoes have caused considerable property damage and loss of life across the state of Oklahoma. Here, we utilize the protection motivation theory (PMT) to explore drivers of tornado hazard adjustment intentions by surveying households from counties in Oklahoma that frequently experience tornadoes. Overall, we found that threat appraisals and coping appraisals produce differential effects depending on the type of hazard adjustment in question. Our findings show that risk perceptions are not a significant explanatory variable of basic adjustments (e.g., flashlight, food supply, and water supply) but are a significant explanatory variable of complex adjustments (e.g., insurance and storm shelter). Future work should provide broader perspectives on how to advance the PMT to better explain adjustment intentions for various hazards.
Abstract
Hazardous weather conditions can pose a threat to the functioning of transportation systems. While the impacts of extreme weather events (e.g., hurricanes/tornadoes and flooding) on transportation disruptions have received significant attention, minor transient disturbances in traffic and transport systems due to rainfall events have remained understudied. Given that a road network experiences rainfall events on a regular basis, which in turn likely reduces its efficiency through short-term disruptions, it is imperative to assess the influence of variations in rainfall intensity on the traffic speed. By synergistically using crowdsourced probe vehicle speed data and spatially explicit meteorological data, this study quantifies the sensitivity of traffic speed to rainfall events of different intensities over 1151 road sections within Alabama. It is observed that instead of variations in the rainfall intensity, traffic speed sensitivity is primarily influenced by a road section’s free-flow speed (uninterrupted speed during dry pavement conditions) and antecedent traffic volume. Relative sensitivity of road sections exhibits high consistency over different rainfall intensities across all road sections, thus underscoring the possibility of assessing sensitivities based only on speed data collected during rainfall intensities that are much more frequent. These results may be used to identify road sections and time periods with high sensitivity to rainfall, thus helping in prioritization of mitigation measures.
Significance Statement
To safeguard against hazardous driving conditions during rainfall events, from either compromised visibility or reduced friction between tires and pavement, drivers often reduce vehicle speed. However, the influence of rainfall intensity on traffic speed reduction remains unclear. This study analyzes the sensitivity of traffic speed to rainfall intensity. Our results indicate that, while rainfall indeed leads to traffic speed reductions, the extent of reduction is predominantly influenced by free-flow speed (uninterrupted vehicle speed) of the road section and the traffic volume on it instead of the rainfall intensity. These results may be used to identify high-sensitivity time periods and locations and guide prioritization of mitigation measures.
Abstract
Hazardous weather conditions can pose a threat to the functioning of transportation systems. While the impacts of extreme weather events (e.g., hurricanes/tornadoes and flooding) on transportation disruptions have received significant attention, minor transient disturbances in traffic and transport systems due to rainfall events have remained understudied. Given that a road network experiences rainfall events on a regular basis, which in turn likely reduces its efficiency through short-term disruptions, it is imperative to assess the influence of variations in rainfall intensity on the traffic speed. By synergistically using crowdsourced probe vehicle speed data and spatially explicit meteorological data, this study quantifies the sensitivity of traffic speed to rainfall events of different intensities over 1151 road sections within Alabama. It is observed that instead of variations in the rainfall intensity, traffic speed sensitivity is primarily influenced by a road section’s free-flow speed (uninterrupted speed during dry pavement conditions) and antecedent traffic volume. Relative sensitivity of road sections exhibits high consistency over different rainfall intensities across all road sections, thus underscoring the possibility of assessing sensitivities based only on speed data collected during rainfall intensities that are much more frequent. These results may be used to identify road sections and time periods with high sensitivity to rainfall, thus helping in prioritization of mitigation measures.
Significance Statement
To safeguard against hazardous driving conditions during rainfall events, from either compromised visibility or reduced friction between tires and pavement, drivers often reduce vehicle speed. However, the influence of rainfall intensity on traffic speed reduction remains unclear. This study analyzes the sensitivity of traffic speed to rainfall intensity. Our results indicate that, while rainfall indeed leads to traffic speed reductions, the extent of reduction is predominantly influenced by free-flow speed (uninterrupted vehicle speed) of the road section and the traffic volume on it instead of the rainfall intensity. These results may be used to identify high-sensitivity time periods and locations and guide prioritization of mitigation measures.
Abstract
This research examines tornadoes and their fatalities by light condition (i.e., daytime and nighttime) for the United States. The study has two primary objectives: 1) to catalog and reassess differences in daytime and nighttime, or nocturnal, tornadoes and their fatalities from spatial and temporal perspectives and 2) to employ a spatially explicit Monte Carlo simulation technique to calculate differences in daytime and nocturnal tornado–population impact potential by combining climatological tornado risk data with fine-scale, gridded estimates of day and night population density. Results reveal that nocturnal tornadoes remain a substantial impediment to mitigating tornado casualties despite long-term improvements in detection and warning of these events. Nocturnal tornadoes are nearly 2 times as deadly as daytime events, with fatalities stemming from overnight (i.e., from local midnight to sunrise) tornadoes increasing fourfold since the late nineteenth century. The proportion of all tornado fatalities that occurred during daytime hours has decreased 20% over the last 140 years while the nocturnal fatality proportion has increased 20%. The stall, or even slight growth, in U.S. tornado mortality rates over the last 30 years has, at least in part, been driven by increasing nocturnal tornado fatalities. Overall, nocturnal tornadoes affect 13% more people on average than daytime tornadoes, revealing the importance of time of day in mitigating tornado–population impacts and disasters. Emergency managers, forecasters, first responders, policy makers, and researchers should continue to focus efforts on understanding nocturnal tornadoes, especially with regard to how populations receive warnings and respond to these nocturnal threats.
Abstract
This research examines tornadoes and their fatalities by light condition (i.e., daytime and nighttime) for the United States. The study has two primary objectives: 1) to catalog and reassess differences in daytime and nighttime, or nocturnal, tornadoes and their fatalities from spatial and temporal perspectives and 2) to employ a spatially explicit Monte Carlo simulation technique to calculate differences in daytime and nocturnal tornado–population impact potential by combining climatological tornado risk data with fine-scale, gridded estimates of day and night population density. Results reveal that nocturnal tornadoes remain a substantial impediment to mitigating tornado casualties despite long-term improvements in detection and warning of these events. Nocturnal tornadoes are nearly 2 times as deadly as daytime events, with fatalities stemming from overnight (i.e., from local midnight to sunrise) tornadoes increasing fourfold since the late nineteenth century. The proportion of all tornado fatalities that occurred during daytime hours has decreased 20% over the last 140 years while the nocturnal fatality proportion has increased 20%. The stall, or even slight growth, in U.S. tornado mortality rates over the last 30 years has, at least in part, been driven by increasing nocturnal tornado fatalities. Overall, nocturnal tornadoes affect 13% more people on average than daytime tornadoes, revealing the importance of time of day in mitigating tornado–population impacts and disasters. Emergency managers, forecasters, first responders, policy makers, and researchers should continue to focus efforts on understanding nocturnal tornadoes, especially with regard to how populations receive warnings and respond to these nocturnal threats.
Abstract
Evidence exists that exposure to weather hazards, particularly in cities subject to heat island and climate change impacts, strongly affects individuals’ physical and mental health. Personal exposure to and sentiments about warm conditions can currently be expressed on social media, and recent research noted that the geotagged, time-stamped, and accessible social media databases can potentially be indicative of the public mood and health for a region. This study attempts to understand the relationships between weather and social media sentiments via Twitter and weather data from 2012 to 2019 for two cities in hot climates: Singapore and Phoenix, Arizona. We first detected weather-related tweets, and subsequently extracted keywords describing weather sensations. Furthermore, we analyzed frequencies of most used words describing weather sensations and created graphs of commonly occurring bigrams to understand connections between them. We further explored the annual trends between keywords describing heat and heat-related thermal discomfort and temperature profiles for two cities. Results showed significant relationships between frequency of heat-related tweets and temperature. For Twitter users exposed to no strong temperature seasonality, we noticed an overall negative cluster around hot sensations. Seasonal variability was more apparent in Phoenix, with more positive weather-related sentiments during the cooler months. This demonstrates the viability of Twitter data as a rapid indicator for periods of higher heat experienced by public and greater negative sentiment toward the weather, and its potential for effective tracking of real-time urban heat stress.
Significance Statement
Social media such as Twitter allow individuals to broadcast their opinions in real time, including perceptions and sensations related to weather events. Evidence from two cities exposed to hot weather—one equatorial and one desert subtropical—indicates that tweets were sensitive to seasonal temperature differences even within a small range. For Twitter users exposed to no strong temperature seasonality, generally negative sentiments to hot weather were seen year-round. In Phoenix with more pronounced seasonality, tweets were more positive in sentiment during the cooler months. This result shows promise for the medium as a rapid real-time indicator—or a snapshot—for societal sentiment to weather events.
Abstract
Evidence exists that exposure to weather hazards, particularly in cities subject to heat island and climate change impacts, strongly affects individuals’ physical and mental health. Personal exposure to and sentiments about warm conditions can currently be expressed on social media, and recent research noted that the geotagged, time-stamped, and accessible social media databases can potentially be indicative of the public mood and health for a region. This study attempts to understand the relationships between weather and social media sentiments via Twitter and weather data from 2012 to 2019 for two cities in hot climates: Singapore and Phoenix, Arizona. We first detected weather-related tweets, and subsequently extracted keywords describing weather sensations. Furthermore, we analyzed frequencies of most used words describing weather sensations and created graphs of commonly occurring bigrams to understand connections between them. We further explored the annual trends between keywords describing heat and heat-related thermal discomfort and temperature profiles for two cities. Results showed significant relationships between frequency of heat-related tweets and temperature. For Twitter users exposed to no strong temperature seasonality, we noticed an overall negative cluster around hot sensations. Seasonal variability was more apparent in Phoenix, with more positive weather-related sentiments during the cooler months. This demonstrates the viability of Twitter data as a rapid indicator for periods of higher heat experienced by public and greater negative sentiment toward the weather, and its potential for effective tracking of real-time urban heat stress.
Significance Statement
Social media such as Twitter allow individuals to broadcast their opinions in real time, including perceptions and sensations related to weather events. Evidence from two cities exposed to hot weather—one equatorial and one desert subtropical—indicates that tweets were sensitive to seasonal temperature differences even within a small range. For Twitter users exposed to no strong temperature seasonality, generally negative sentiments to hot weather were seen year-round. In Phoenix with more pronounced seasonality, tweets were more positive in sentiment during the cooler months. This result shows promise for the medium as a rapid real-time indicator—or a snapshot—for societal sentiment to weather events.
