This paper is in the following e-collection/theme issue:

Infodemiology and Infoveillance (1558) Medicine 2.0: Social Media, Open, Participatory, Collaborative Medicine (1911) Environmental Health (123) Infoveillance and Social Listening (313) Environmentally Sustainable Health Care and Climate Change Initiatives (28) Climate Change and Health (36) Quantified Self and Wellness (113) Health Risk and Perception in Cognitive Reasoning (30) Health and Risk Communication (75)

Published on in Vol 27 (2025)

Preprints (earlier versions) of this paper are available at https://preprints.jmir.org/preprint/70336, first published .
Decoding the Public’s Real-Time Emotional and Cognitive Responses to the Changing Climate on Social Media: Computational Analysis Using Weibo and Meteorological Data

Decoding the Public’s Real-Time Emotional and Cognitive Responses to the Changing Climate on Social Media: Computational Analysis Using Weibo and Meteorological Data

Decoding the Public’s Real-Time Emotional and Cognitive Responses to the Changing Climate on Social Media: Computational Analysis Using Weibo and Meteorological Data

Authors of this article:

Yucan Xu1 Author Orcid Image ;   Jiehu Yuan1 Author Orcid Image ;   Sijia Li2 Author Orcid Image ;   Qiuyan Liao1 Author Orcid Image
Article Authors Cited by (1) Tweetations Metrics

    Original Paper

    1School of Public Health, LKS Faculty of Medicine, University of Hong Kong, Hong Kong, China (Hong Kong)

    2Department of Social Work and Social Administration, Faculty of Social Sciences, University of Hong Kong, Hong Kong, China (Hong Kong)

    Corresponding Author:

    Qiuyan Liao, PhD

    School of Public Health

    LKS Faculty of Medicine

    University of Hong Kong

    Patrick Manson Building, 7 Sassoon Rd, Pok Fu Lam

    Hong Kong

    China (Hong Kong)

    Phone: 852 39179289

    Email: qyliao11@hku.hk


    Background: Climate change poses a significant threat to mental health and well-being worldwide. Existing research on the associations between climate change–related events and mental well-being primarily focuses on clinical outcomes and often measures associations at single time points. The long-term effects and variability of the changing climate on more subtle nonclinical but widespread mental well-being remain relatively unexplored. Additionally, the underlying mechanisms that link changing climate events to real-time emotional well-being and pro-environmental actions have rarely been studied. Revealing real-time nonclinical mental well-being and its underlying mechanism is crucial for the early detection of at-risk individuals. This knowledge can also inform future interventions aimed at improving the public’s risk perception and empowering communities to manage related challenges effectively.

    Objective: This study aimed to understand the association between the changing climate and expressed emotional well-being by integrating multiple data sources, including social media posts about climate change on Weibo (N=76,514), 20 years of regional meteorological data (N=216,476 records), and regional vulnerability data in China.

    Methods: This study proposed and tested a new mechanism that connects meteorological factors with expressed emotional well-being through three cognitive responses identified from social media posts: thinking styles, social affiliations, and somatosensory experiences. Psycholinguistic analysis, structural equation modeling (SEM), and multiple regression models were used to examine the mediation of these three conceptual factors, as well as the moderating effects of regional vulnerability and seasonal changes on the influence of climate change on the public’s expressed emotional well-being and downstream pro-environmental tendencies.

    Results: The SEM results revealed that extreme hot days are associated with decreased emotional well-being when talking about climate change (total effect=–0.712, 95% CI –0.894 to –0.531, P<.001), and these effects were mediated by three proposed mediators: social affiliations (indirect effect=–0.445, 95% CI –0.537 to –0.347, P<.001), analytical-intuitive thinking style (indirect effect=–0.100, 95% CI –0.126 to –0.073, P<.001), and somatosensory experiences (indirect effect=0.022, 95% CI 0.005-0.041, P=.02). Additionally, regression analysis indicated that the association between increased temperatures and expressed emotional well-being is moderated by seasonal changes (β=–.091, 95% CI –0.159 to –0.023, P=.009) and regional population density (β=–.068, 95% CI –0.118 to –0.018, P=.007). In the crude model examining associations between weather indices and expressed pro-environmental tendencies, the results showed that extreme hot days are associated with reduced pro-environmental tendencies (odds ratio [OR]=0.802, 95% CI 0.747-0.861, P<.001). However, after controlling for expressed emotional well-being and cognitive responses, such associations were less pronounced.

    Conclusions: The findings highlight the need for interventions that promote mental well-being in response to climate change and the importance of cognitive responses in developing positive coping strategies and enhancing emotional resilience. This approach could empower individuals to create a positive self-reinforcing cycle that encourages pro-environmental behaviors.

    J Med Internet Res 2025;27:e70336

    doi:10.2196/70336

    Keywords

    climate change; emotional well-being; pro-environmental behaviors; regional vulnerability; cognitive responses; psycholinguistic analysis; Weibo; infodemiology 


    Background

    Climate change is a significant threat to mental health and well-being worldwide [1]. Subacute or long-term climate change–related events, such as heatwaves and higher temperatures, have been found to be associated with increased mental health issues and even suicide rates [2]. For example, a large sampled study examining the relationship of historical meteorological data and information of two million US residents found that shifting of monthly temperatures between 25 °C and 30 °C or above increased the probability of mental health difficulties by 0.5% points, and 1°C of 5-year warming was associated with a 2% point increase in the prevalence of mental health issues [3]. Measuring and monitoring the impacts of climate change on mental health requires a long-term and pragmatic approach by incorporating multiple data sources. Unfortunately, most studies to date investigating the impact of climate change on mental well-being have been conducted with cross-sectional surveys, which may suffer from recall biases or fail to evaluate the underlying mechanisms [4]. In addition, they have focused solely on clinical mental health outcomes, such as hospital attendance or admission, and suicide. Little is known about the real-time emotional impacts, such as negative moods, and the more subtle but widespread impact on expressed mental well-being related to climate change. The more subtle mental well-being impact can serve as an early screening signal to identify people who are at risk for clinical mental health issues and inform future interventions, which informs communication to promote public awareness of risk and strategies to empower the public to cope with related risks. There is also insufficient understanding of how climate change–induced emotional well-being influences downstream pro-environmental behaviors. To answer those inquiries and address these gaps, this research aimed to link social media data and meteorological data to provide a more comprehensive understanding of the impact of climate change on the public’s emotional well-being, the potential mechanisms through cognitive responses, and their impact on pro-environmental tendencies.

    Climate Change and Emotional Well-Being

    There has been extensive evidence that climate change–related events are associated with adverse mental health outcomes, including various mental disorders, hospitalizations due to mental health, and even suicide [4,5]. Both more chronic meteorological changes and acute extreme weather pose significant threats to individuals’ psychological well-being. Concerning the impacts of long-term meteorological changes, a systematic review suggested most meteorological factors, such as temperature increase, aridity, and heatwaves, are associated with mental health outcomes, including hospital admissions, exacerbation of mental health conditions, sleeping difficulty, and suicide [6]. It has been consistently reported that acute extreme weather events, such as flooding and drought, are associated with diverse mental health issues, including posttraumatic stress disorder, anxiety, and depression [7]. Previous studies have examined the links between weather indicators and mental health outcomes, but many of them have focused primarily on posttraumatic scenarios, such as heat waves [8]. This has resulted in a lack of attention to nonacute meteorological changes, such as gradual temperature shifts. It is posited that the impact of different meteorological events, acute and nonacute, may vary and is worth further investigation. As for the outcomes, existing studies on the impact of climate change on mental health outcomes mainly focus on clinical outcomes, including suicides, hospitalizations for mental illness, and severity of symptoms [4]. Except for clinical outcomes, various nonclinical outcomes, such as mood, positive and negative affect, and overall well-being, are also used to assess the impact of climate change on mental health [4]. Although those measures may not directly indicate the clinical profile of individuals, they allow for a broader understanding of the effects of climate change on mental health in wider populations. This broader generalizability can be beneficial for the early detection of individuals who are at risk.

    Cognitive Responses as Potential Mediators

    Although climate change–related events can directly affect individuals’ mental well-being [5], the literature has documented several important mechanisms that could explain how climate change can indirectly impair human mental health. First, the physiological mechanism argues that exposure to heat waves may induce biological changes in neural mechanisms that may further affect and exacerbate symptoms in individuals with mental health conditions [9]. Second, cognitive perspectives have found that sleep deprivation caused by environmental changes can affect human cognitive capacity and contribute to mental health problems [10]. Third, the environmental and social process states that climate change–related events can erode social and community well-being by reducing economic and agricultural outputs, ultimately leading to mental health problems [11]. However, there remains limited knowledge regarding how cognitive responses, such as individuals’ thinking styles, affiliative strategies, and personal somatosensory experiences, interact with climate change–related events and mental well-being. Investigating these cognitive strategies can offer important insights into mitigating the impacts of climate change on mental well-being.

    Thinking Style

    In accordance with the Stimulus-Organism-Response framework, meteorological changes in the environment can induce cognitive and emotional responses in human beings and ultimately impact human health directly or indirectly [12]. According to Daniel Kahneman, the cognitive responses to a stimulus or information involve two styles: intuitive thinking (system 1) and analytical thinking (system 2) [13]. Intuitive thinking is a process that is usually fast, automatic, and sometimes unconscious, while analytical thinking is a process that is more logical and deliberative. People are usually predisposed to a specific cognitive style, the more intuitive or analytic one [14]. Individuals who adopt an analytical thinking style tend to perceive more control in a specific situation compared to those who mainly adopt the intuitive thinking style [15,16]. Conversely, intuitive thinking is associated with greater skepticism toward climate change [17]. Nevertheless, these studies have not examined thinking styles as a potential mediator of the impacts of the changing climate on emotional well-being.

    Social Connectedness and Affiliations

    Based on social safety theory, social connectedness and affiliations (perceived as being connected and affiliated with others) serve as an important factor that buffers the emotional impacts of major environmental hazards [18]. Extensive research has demonstrated that individuals with high affiliative motives are more likely to experience positive emotions and exhibit better mental health outcomes compared to those with self-focused motives [19,20]. Engaging in affiliate motives and altruistic acts could alleviate and help prevent mental health difficulties, leading to better emotional outcomes [21]. In turn, those positive emotions encourage individuals to help others who are in need and form an ongoing cooperative relationship [22]. When faced with climate change events with low-dread risks, such as heat waves, individuals tend to prioritize individual adaptation strategies, such as seeking shade, decreasing clothing, or using air-conditioning, rather than engaging in the more costly pro-environmental behaviors [23]. A recent review on drivers for climate action revealed that social groups and the individual attachment to these groups can be influential on individual climate actions [24]. However, there is limited knowledge regarding the triangle relationship between affiliate motives, changing temperatures and heat waves, and the public’s emotional well-being.

    Somatosensory Experiences

    The existing literature elucidates the mechanism linking somatosensory, brain, and affect systems, revealing the pathway of how somatosensory systems mediate the impact of climate change on mental health [10,25]. The process begins with the body’s sensory system receiving signals from climate change–related stimuli and subsequently transmitting those signals to the brain systems that control affective function. The somatosensory system is responsible for the conscious perception of various bodily somatic sensations, including touch, pressure, pain, temperature, body position, movement, and vibration [26]. Previous research on the association between different body sensations and emotions has demonstrated that somatosensory experiences could trigger conscious emotional responses and that different bodily sensations are associated with different emotions [27]. Given the connection between somatosensory experiences and emotional responses, it is reasonable to speculate that individuals may perceive and interpret different bodily signals when exposed to increased temperatures and extreme weather events, potentially leading to the activation of negative emotional states. However, empirical testing is required to confirm this hypothesis.

    Emotional Well-Being and Pro-Environmental Tendencies

    The cognitive and emotional impacts of climate change can go beyond and influence people’s behaviors. Research indicates that negative emotions triggered by climate change can have a bidirectional impact on pro-environmental behaviors [28]. On the one hand, coping mechanisms suggest that negative emotions triggered by climate change–related events can motivate individuals to take action to mitigate the negative impacts of climate change [29]. On the other hand, some argue that chronic negative mental health problems may lead to eco-paralysis, a state in which individuals passively respond to climatic and ecological challenges, feeling hopeless and powerless [30]. A recent study also reported an association between higher climate anxiety and fewer pro-environmental behaviors, highlighting that climate anxiety leads to lower efficacy and dampens individuals’ ability to take climate actions [31]. Despite some evidence, existing studies have yet to examine whether the more real-time negative impact of climate change on emotional well-being would harm individuals’ pro-environmental intentions and tendencies.

    Social Media Data and Climate Change

    Numerous studies have demonstrated social media’s potential to understand the public’s perception and emotional responses to climate change–related events [32-34]. Through the use of natural language processing and network analysis, researchers have been able to study public reactions and sentiments to climate policy actions on social media platforms, such as X (formerly known as Twitter) [35]. Previous studies have demonstrated that linguistic expressions, including content features (eg, words and context), style features (eg, punctuations and emoji), and word categories (eg, discrete emotion dictionaries), have shown strong predicting power for various mental well-being outcomes, including depression and suicide [36,37]. Sentiment and emotional distress identified from such digital content can be used for early detection of individuals who are in need and provide in-time interventions [38]. Among those psycholinguistic methods, the Linguistic Inquiry and Word Count (LIWC) dictionary has demonstrated great potential and validity in capturing social media users’ emotional and cognitive responses to science communication agendas, such as climate change [39,40] and the COVID-19 pandemic [41,42].

    Although social media data has been widely used to comprehend public reactions to climate change, only a limited number of studies have used a combination of social media data and offline meteorological data to monitor the influence of offline temperature changes and extreme weather events on the public’s emotional responses to climate change [43,44]. This approach may provide a more comprehensive picture of real-time interactions between climate change and the public’s affective responses in a large-scale population.

    Study Aims

    Building on the knowledge gaps discussed earlier, this study aimed to investigate the real-time associations between acute extreme meteorological events, including extreme daily temperatures and precipitations; more long-term meteorological changes, including deviations in daily temperatures and precipitations; and the public’s expressed emotional well-being and cognitive responses when discussing related concepts using one popular online social media platform: Weibo data and offline regional meteorological and vulnerability data. Guided by the climate response–related theories as our theoretical foundation, we proposed and tested one novel mechanism connecting meteorological indicators and emotional responses through three cognitive pathways. Specifically, our hypotheses are:

    • Hypothesis (H)1: Extreme meteorological events and meteorological changes are associated with the public’s emotional well-being, indicated by climate change discourse. Specifically, extreme hot days and increased temperatures are associated with more negative emotional well-being.
    • H2: The association of H1 varies by region and time. People who live in regions with higher vulnerability and in seasons with hotter days (eg, summer) are more prone to be influenced.
    • H3: The association of H1 is mediated by individuals’ intuitive-analytical thinking style, social affiliations, and somatosensory experiences. Specifically, extreme hot days and hotter temperature changes are associated with an intuitive-leaning thinking style, a less affiliative attitude, and a higher level of somatosensory experience, which are further associated with more negative emotional well-being.
    • H4: Extreme hot days have a negative impact on people’s pro-environmental tendencies. This effect is channeled through their emotional well-being, where extreme hot days are associated with poorer emotional well-being and hence a lower willingness to be pro-environmental.

    Dataset

    Three parts of data were collected to answer the research questions and were available publicly: First, meteorological data were obtained to identify and measure extreme meteorological events and changes. Second, social media data from Weibo were gathered to gauge the public’s cognitive responses and emotional well-being, as well as their inclinations to be pro-environmental. Finally, regional vulnerability data were collected to ascertain regional variations. Details of the three data collection processes are reported in the Data Collection section in Multimedia Appendix 1.

    Study Measures

    Extreme Weather Indicators and Temperature Changes

    There are two common approaches to defining extreme weather. The first approach uses a constant value to set the threshold. Various temperature thresholds, ranging from 25 °C to higher than 40 °C, have been used to define extreme heat indices [45]. For example, one study used 25°C to indicate hot days [46], while the National Weather Service in the United States adopts a threshold of 40.6°C in the daytime and greater than or equal to 26.7°C in the nighttime for its excessive heat watch and warning system [47]. Additionally, another study considered a heat wave as the maximum temperature at a station reaching at least 40 °C for the plains and at least 30 °C for hilly regions [48]. The second approach is to choose a relative threshold that is varied by the empirical situation of the location [49]. For instance, using the 90th percentile as the cutoff for extreme weather indices ensures that 10% of events are classified as “extreme.” This method is beneficial and more robust because it considers the geographical context and relative heat adaptations. We selected the 10th and 90th percentiles as cutoff points for defining extreme weather indices in line with the recommendations by Stephenson et al [49]. We recognized that using the 5th and 95th percentiles might yield different results. However, choosing more extreme values as cutoffs could result in fewer data points, which would limit the interpretability of the data. Based on this approach, we created four indicators: extreme hot days, extreme cold days, temperature changes, and precipitation changes. The detailed method is reported in the Weather Indicators section in Multimedia Appendix 1.

    Outcomes
    Expressed Emotional Well-Being

    Emotional well-being refers to an individual’s overall emotional state, encompassing how positively one feels about their life in general [50]. Consequently, emotional well-being expressed regarding climate change captures the feelings and emotional states that arise when facing climate change. Expressed emotional well-being was the primary outcome identified through the analysis of psycholinguistic cues in individuals’ Weibo posts. These cues were evaluated using the affect categories of the Simplified Chinese version of the Linguistic Inquiry and Word Count (SC-LIWC) dictionary, which is a validated adaptation of the original LIWC dictionary [51]. The affect categories encompass positive and negative sentiments, as well as three discrete negative emotions: anger, anxiety, and sadness. Following Monzani et al’s [41] method, we created an emotional well-being index by subtracting the negative sentiment score from the positive sentiment score to indicate the overall emotional well-being expressed for a post. A higher emotional well-being index indicates a more positive emotional well-being loading. Instead of measuring each user’s long-term subjective well-being [52], we assessed expressed emotional well-being at the post level. The goal is to understand the relationships between extreme weather indices and expressed emotional well-being in the context of relevant weather events. By limiting our analysis to posts specifically related to climate change, we have ensured that the discussions are directly tied to climate change issues. In contrast, focusing on all posts from each user may introduce confounding factors, as their emotional well-being could be influenced by unrelated topics (eg, COVID-19) and daily life events (eg, relationship problems) rather than the weather indices.

    Pro-Environmental Tendency

    The pro-environmental tendency of Weibo users was a secondary outcome evaluated using a binary classification of their posts. The keywords and phrases used for the classification were primarily based on the Pro-Environmental Behavior questionnaire (PEB) curated and validated by Kalamas et al [53], covering 50 PEB items in 7 categories: saving energy, consumption, activism, recycling, transportation, diets, and waste management. The full keyword list can be found in Table S1 in Multimedia Appendix 1.

    Cognitive Responses

    To assess the cognitive responses of the public, posts from individual accounts were analyzed using the LIWC dictionary. The analysis focused on three proposed cognitive responses: analytical-intuitive thinking style, social affiliations, and somatosensory experiences. The detailed information for measuring three cognitive responses is reported in the Measuring Cognitive Responses section in Multimedia Appendix 1.

    Regional Vulnerability Indicators

    Followed by a national study of regional health vulnerability to extreme heat in China [54] and the framework of vulnerability [55], three indicators were used to measure regional vulnerability: exposure, sensitivity, and adaptability. A detailed description of the three vulnerability indicators is reported in the Vulnerability Indicators section in Multimedia Appendix 1.

    Statistical Analysis

    Overall, this study applied computational social science methods to facilitate the understanding of the emotional and behavioral impact of the changing climate [56]. Psycholinguistic analysis for climate change–related Weibo posts was conducted with the SC-LIWC dictionary. Descriptive analysis was conducted to describe the initial associations between meteorological indicators and the public’s emotional and cognitive responses with mixed correlations. Next, structural equation modeling (SEM) was used to analyze the mediating effect of the intuitive-analytical thinking style, social affiliations, and somatosensory experiences on the associations between meteorological indicators and emotional well-being. SE values were generated through 1000 bootstrapping. The 95% CI was calculated using the adjusted bootstrap percentile (BCa) method, which is the bias-corrected method for CI estimation [57]. Fit indices were used to evaluate the model fit, including the comparative fit index (CFI>0.90), the goodness-of-fit index (GFI>0.90), and the root mean square error of approximation (RMSEA<0.08) [58,59]. Model modification indices were used to indicate the potential correlations between mediators. Multivariate linear regressions were used to detect the moderation and interaction effect of regional vulnerability factors. Multiple logistic regression models were used to estimate the power of environmental indicators in predicting pro-environmental tendencies when controlling different confounders. The variance inflation factor (VIF) was used to detect multicollinearity, where variables with VIF>5 were removed from selected models [60]. SEM was conducted with Mplus (version 8.3), and all other analyses were performed with the Python statsmodel package (version 0.13.1).

    Cross-Dataset Merge

    We initially combined meteorological data from the National Oceanic and Atmospheric Administration (NOAA) with vulnerability data using the IP address at the provincial level. Next, we aligned these combined environmental data with social media data by matching the corresponding date and province (as indicated by the IP address) of each post.

    Ethical Considerations

    Data were used complying with ethical guidelines for internet research [61]. All data are publicly open access data, and all analyses were performed on public and anonymized metadata; no institutional review board approval was required for the use of these datasets or the completion of this study.


    Correlations Between Meteorological Factors, People’s Emotional Responses, and Cognitive Processing

    Figure 1 visualizes the correlations between environmental meteorological factors and interested measures. Results highlighted that extreme hot days are correlated with more negative emotional well-being expressed in climate change–related posts by Weibo individual users, while extreme cold days are correlated with more positive emotional well-being. Additionally, as the temperature increased, more negative emotional well-being was observed in related posts. Relatedly, extreme hot days and hotter temperatures were correlated with higher scores of anxiety, anger, and sadness (H1 supported). Extreme hot days seemed to have an impact on one’s cognitive processing and behavioral tendency. Specifically, extreme hot days were correlated with an intuitive-leaning thinking style, a greater level of somatosensory experience but reduced affiliation, and a lower inclination toward pro-environmental behaviors. As expected, social affiliations, thinking styles, somatosensory experiences, and emotional well-being were significantly intercorrelated. Individuals who exhibited a higher level of affiliation, an analytical-leaning thinking style, and somatosensory experiences tended to display more positive emotional well-being. Similar results were observed in the outcome of pro-environmental tendencies, where more positive emotional well-being, social affiliations, an analytical-leaning style, and somatosensory experiences were correlated with a greater inclination toward pro-environmental behaviors. On the contrary, negative emotions, including anger, anxiety, and sadness, were negatively associated with pro-environmental tendencies.

    ‎
    Figure 1. Mixed correlations between meteorological indicators, emotional responses, and cognitive responses for climate change–related posts (N=24,156). (a) Binary variables and (b) continuous variables. Correlations between two continuous variables were evaluated with Pearson correlation, correlations between one continuous variable and one binary variable were evaluated with point-biserial correlation, and correlations between two binary variables were evaluated with Matthews correlation. The red color indicates positive correlations, while the blue color indicates negative correlations. The darker the color, the stronger the correlation. *P<.05, **P<.01, and ***P<.001. The statistical significance (P values) presented was adjusted using false discovery rate (FDR) corrections.

    Climate Change–Related Expressed Emotional Well-Being by Season

    After controlling for seasons, the presence of extreme hot days remained a robust predictor of emotional well-being (Tables 1 and 2). Specifically, extreme heat or hotter temperature were associated with more negative emotional well-being. Additionally, the results revealed an interaction effect between seasons and temperature changes. Compared to spring, temperature increases in the summer were associated with a significantly higher level of negative emotional well-being (β=–.091, 95% CI –0.159 to –0.023, P=.009), as shown in Figure S1 in Multimedia Appendix 1.

    Table 1. Moderations of seasons to determine the relationship between extreme hot days and emotional well-being (N=24,156).
    Variablesβ (95% CI)z ValueP value
    Extreme hot days (binary)–0.581 (–0.957 to –0.205)–3.029.002a
    Seasons

    		Spring (reference group)—b——

    		Summer–0.866 (–1.083 to –0.649)–7.817<.001a

    		Autumn–1.152 (–1.407 to –0.897)–8.859<.001a

    		Winter–0.570 (–0.830 to –0.309)–4.287<.001a
    Summer×extreme hot days–0.135 (–0.580 to 0.311)–0.593.55
    Autumn×extreme hot days0.289 (–0.260 to 0.838)1.032.30
    Winter×extreme hot days–0.231 (–0.933 to 0.472)–0.644.52

    aSignificant P values.

    bNot applicable.

    Table 2. Moderations of seasons to determine the relationship between temperature changes and emotional well-being (N=24,156).
    Variablesβ (95% CI)z ValueP value
    Temperature changes (°C)–0.085 (–0.129 to –0.042)–3.875<.001a
    Seasons

    		Spring (reference group)—b——

    		Summer–0.800 (–1.000 to –0.601)–7.846<.001a

    		Autumn–1.125 (–1.355 to –0.894)–9.567<.001a

    		Winter–0.581 (–0.822 to –0.340)–4.724<.001a
    Summer×temperature changes–0.091 (–0.159 to –0.023)–2.625.009a
    Autumn×temperature changes0.067 (–0.002 to 0.135)1.898.06
    Winter×temperature changes0.044 (–0.025 to 0.112)1.240.22

    aSignificant P values.

    bNot applicable.

    Climate Change–Related Expressed Emotional Well-Being by Regional Vulnerability

    After controlling for different population vulnerability indexes, the impact of extreme hot days and temperature changes remained significantly associated with emotional well-being (Tables 3 and 4). For vulnerability factors, population density was identified as a significant factor in explaining emotional well-being in both models, controlling for the presence of extreme hot days and hotter temperatures. A higher population density was associated with more negative emotional well-being (βmodel 1=–.256, 95% CI –0.427 to –0.084, P=.003; βmodel 1s=–.191, 95% CI –0.345 to –0.038, P=.02). The exposure index was significantly associated with the expressed emotional well-being in the model with temperature changes, where a higher exposure index was associated with more negative emotional well-being (βmodel 2s=–.186, 95% CI –0.342 to –0.030, P=.02). A follow-up moderation analysis, incorporating interactive terms between expressed emotional well-being and population vulnerability indexes, identified the effect of a significant interaction between population density and temperature change on emotional well-being (β=–.068, 95% CI –0.118 to –0.018, P=.007), as shown in Figure S2 in Multimedia Appendix 1. Specifically, on hotter days (mean + 1SD), regions with a higher population density were associated with more negative emotional well-being than regions with a lower population density, while on colder days (mean – 1SD), regions with a higher population density displayed higher emotional well-being than regions with a lower population density (H2 supported).

    Table 3. Moderations of regional vulnerability indicators to determine the relationship between extreme hot days and emotional well-being (N=24,156).
    Variablesβ (95% CI)az ValueP value
    Model 1 with population densityb

    		Extreme hot days (1)–0.730 (–0.986 to –0.474)–5.585<.001c

    		Population density, high vs low (2)–0.256 (–0.427 to –0.084)–2.920.003c

    		(1)×(2)0.025 (–0.316 to 0.367)0.144.89
    Model 2 with exposure indexd

    		Extreme hot days (1)–0.533 (–0.816 to –0.250)–3.690<.001c

    		Exposure index, high vs low (3)–0.084 (–0.258 to 0.089)–0.952.34

    		(1)×(3)–0.268 (–0.622 to 0.085)–1.487.14
    Model 3 with sensitivity indexd

    		Extreme hot days (1)–0.823 (–1.051 to –0.595)–7.065<.001c

    		Sensitivity index, high vs low (4)–0.032 (–0.202 to 0.139)–0.366.72

    		(1)×(4)0.247 (–0.093 to 0.588)1.424.16
    Model 4 with adaptability indexd

    		Extreme hot days (1)–0.880 (–1.208 to –0.552)–5.263<.001c

    		Adaptability index, high vs low (5)–0.171 (–0.355 to 0.012)–1.830.07

    		(1)×(5)0.239 (–0.144 to 0.622)1.224.22

    aThe coefficients (β) and their 95% CIs were estimated using generalized linear regression.

    bPopulation density data for each province were extracted from the National Bureau of Statistics. A higher population density indicates that the region’s population density is higher than the median of all provinces.

    cSignificant P values.

    dRegional-level exposure, sensitivity, and adaptability indexes have been calculated and validated by Zhang et al [54].

    Table 4. Moderations of regional vulnerability indicators to determine the relationship between temperature changes and emotional well-being (N=24,156).
    Variablesβ (95% CI)az ValueP value
    Model 1s with population densityb

    		Temperature changes (1)–0.061 (–0.097 to –0.024)–3.261<.001c

    		Population density, high vs low (2)–0.191 (–0.345 to –0.038)–2.438.02c

    		(1)×(2)–0.068 (–0.118 to –0.018)–2.687.007c
    Model 2s with exposure indexd

    		Temperature changes (1)–0.119 (–0.163 to –0.075)–5.283<.001c

    		Exposure index, high vs low (3)–0.186 (–0.342 to –0.030)–2.333.02c

    		(1)×(3)0.033 (–0.020 to 0.087)1.223.22
    Model 3s with sensitivity indexd

    		Temperature changes (1)–0.083 (–0.115 to –0.051)–5.129<.001c

    		Sensitivity index, high vs low (4)0.046 (–0.107 to 0.199)0.594.55

    		(1)×(4)–0.035 (–0.085 to 0.016)–1.342.18
    Model 4s with adaptability indexd

    		Temperature changes (1)–0.090 (–0.133 to –0.047)–4.109<.001 a

    		Adaptability index, high vs low (5)–0.123 (–0.288 to 0.042)–1.461.14

    		(1)×(5)–0.009 (–0.062 to 0.044)–0.333.74

    aThe coefficients (β) and their 95% CIs were estimated using generalized linear regression.

    bPopulation density data for each province were extracted from the National Bureau of Statistics. A higher population density indicates that the region’s population density is higher than the median of all provinces.

    cSignificant P values.

    dRegional-level exposure, sensitivity, and adaptability indexes have been calculated and validated by Zhang et al [54].

    Mediation of Three Cognitive Responses

    The mediation analysis using SEM, as depicted in Figure 2, confirmed the mediation of the three proposed conceptual factors in the relationship between extreme hot days and emotional well-being. This mediation model accounted for the covariance between social affiliations and thinking styles, as well as the covariance between social affiliations and somatosensory experiences, suggested by model modification indices in Mplus. The results identified a direct effect of extreme hot days on emotional well-being (direct effect=–0.189, 95% CI –0.319 to –0.050, P=.007). The indirect effect of social affiliations ranked at the top (indirect effect=–0.445, 95% CI –0.537 to –0.347, P<.001), followed by the analytical-intuitive thinking style (indirect effect=–0.100, 95% CI –0.126 to –0.073, P<.001) and somatosensory experiences (indirect effect=0.022, 95% CI 0.005-0.041, P=.02). Specifically, SEM showed that when there was extreme heat, people showed fewer social affiliations, expressed more bodily sensation experiences, and deployed a more intuitive thinking style about climate change. However, although fewer social affiliations and a more intuitive-leaning thinking style were associated with lower emotional well-being, more somatosensory experiences were positively associated with emotional well-being (H3 partially supported).

    ‎
    Figure 2. Mediation paths from SEM. Fit indices: CFI=0.997, GFI=0.999, and RMSEA=0.046. The number on each arrow indicates the association between two variables in the model. *P<.05, **P<.01, and ***P<.001. CFI: comparative fit index; GFI: goodness-of-fit index; RMSEA: root mean square error of approximation; SEM: structural equation modeling.

    We also modeled the mediation pathways between temperature changes and emotional well-being. The model explanation power with temperature changes was lower than the model with extreme hot days (direct effect=–0.030, 95% CI –0.050 to –0.012, P=.002; total indirect effect=–0.067, 95% CI –0.082 to –0.050, P<.001), as shown in Figure S3 in Multimedia Appendix 1.

    Impacts of Climate Change on a Pro-Environmental Tendency

    When no controls were considered (model 1 in Table 5), extreme hot days were associated with a decreased pro-environmental tendency (odds ratio [OR] 0.802, 95% CI 0.747-0.861, P<.001). After controlling for emotional well-being (model 2 in Table 6), the impact of extreme hot days weakened (OR 0.903, 95% CI 0.822-0.994, P=.04; H4 supported). When controlling for the three additional cognitive factors (model 3 in Table 7), the impact of extreme hot days disappeared. All three cognitive responses remained positively associated with a pro-environmental tendency, among which, social affiliation was the most prominent predictor (OR 1.432, 95% CI 1.409-1.456, P<.001). A more analytical-leaning thinking style (OR 1.018, 95% CI 1.015-1.022) and somatosensory experiences (OR 1.038, 95% CI 1.025-1.051) were marginally associated with a higher chance of acting pro-environmentally. When adding regional vulnerability factors into the model (model 4 in Table 8), the results for extreme hot days, emotional responses, and cognitive responses remained as robust as those for model 3. Additionally, high sensitivity was associated with a lower probability of acting pro-environmentally (OR 0.756, 95% CI 0.685-0.835, P<.001), while high adaptability was associated with a higher chance of being pro-environment (OR 1.254, 95% CI 1.130-1.393).

    Table 5. Logistic regression estimates for variables in model 1 for predicting a pro-environmental tendencya (N=24,156).
    VariablesModel 1 (crude model) estimates

    		ORb (95% CI)P valuePseudoR2
    Meteorological indicators

    		Extreme hot vs not extreme hot0.802 (0.747-0.861)<.0010.002
    Emotional responses

    		Emotional well-being—c——
    Cognitive responses

    		Social affiliations———

    		Analytical-intuitive thinking style———

    		Somatosensory experiences———
    Regional vulnerability factors

    		Population density, high vs low———

    		Exposure index, high vs low———

    		Sensitivity index, high vs low———

    		Adaptability index, high vs low———

    aThe pro-environmental tendency is a binary outcome. If a post includes keywords and phrases from any of the seven pro-environmental categories curated and validated by Kalamas et al [53], it is labeled as 1 (see Table S1 in Multimedia Appendix 1).

    bOR: odds ratio.

    cNot applicable.

    Table 6. Logistic regression estimates for variables in model 2 for predicting a pro-environmental tendencya (N=24,156).
    VariablesModel 2 (with emotional responses) estimates

    		ORb (95% CI)P valuePseudo R2
    Meteorological indicators

    		Extreme hot vs not extreme hot0.903 (0.822-0.994).040.356
    Emotional responses

    		Emotional well-being1.415 (1.403-1.427)<.001—c
    Cognitive responses

    		Social affiliations———

    		Analytical-intuitive thinking style———

    		Somatosensory experiences———
    Regional vulnerability factors

    		Population density, high vs low———

    		Exposure index, high vs low———

    		Sensitivity index, high vs low———

    		Adaptability index, high vs low———

    aThe pro-environmental tendency is a binary outcome. If a post includes keywords and phrases from any of the seven pro-environmental categories curated and validated by Kalamas et al [53], it is labeled as 1 (see Table S1 in Multimedia Appendix 1).

    bOR: odds ratio.

    cNot applicable.

    Table 7. Logistic regression estimates for variables in model 3 for predicting a pro-environmental tendencya (N=24,156).
    VariablesModel 3 (with additional cognitive responses) estimates

    		ORb (95% CI)P valuePseudo R2
    Meteorological indicators

    		Extreme hot vs not extreme hot1.024 (0.922-1.138).660.413
    Emotional responses

    		Emotional well-being1.223 (1.211-1.235)<.001—c
    Cognitive responses

    		Social affiliations1.432 (1.409-1.456)<.001—

    		Analytical-intuitive thinking style1.018 (1.015-1.022)<.001—

    		Somatosensory experiences1.038 (1.025-1.051)<.001—
    Regional vulnerability factors

    		Population density, high vs low———

    		Exposure index, high vs low———

    		Sensitivity index, high vs low———

    		Adaptability index, high vs low———

    aThe pro-environmental tendency is a binary outcome. If a post includes keywords and phrases from any of the seven pro-environmental categories curated and validated by Kalamas et al [53], it is labeled as 1 (see Table S1 in Multimedia Appendix 1).

    bOR: odds ratio.

    cNot applicable.

    Table 8. Logistic regression estimates for variables in model 4 for predicting a pro-environmental tendencya (N=24,156).
    VariablesModel 4 (with additional regional vulnerability factors) estimates

    		ORb (95% CI)P valuePseudoR2
    Meteorological indicators

    		Extreme hot vs not extreme hot1.010 (0.908-1.122).860.414
    Emotional responses

    		Emotional well-being1.224 (1.212-1.236)<.001—c
    Cognitive responses

    		Social affiliations1.433 (1.410-1.456)<.001—

    		Analytical-intuitive thinking style1.019 (1.015-1.022)<.001—

    		Somatosensory experiences1.038 (1.025-1.051)<.001—
    Regional vulnerability factors

    		Population density, high vs low0.951 (0.864-1.045).30—

    		Exposure index, high vs low1.001 (0.906-1.105).99—

    		Sensitivity index, high vs low0.756 (0.685-0.835)<.001—

    		Adaptability index, high vs low1.254 (1.130-1.393)<.001—

    aThe pro-environmental tendency is a binary outcome. If a post includes keywords and phrases from any of the seven pro-environmental categories curated and validated by Kalamas et al [53], it is labeled as 1 (see Table S1 in Multimedia Appendix 1).

    bOR: odds ratio.

    cNot applicable.


    Principal Findings

    Climate change is increasingly relevant and concerning due to the ongoing emergence of heat waves and extreme weather events as the new normal [62]. The impact of climate change on individuals extends beyond physical, psychological, cognitive, and behavioral responses. By combining social media data and meteorological data, this study elucidated how extreme meteorological events and changes affected the public’s emotional well-being and the potential pathways of the impacts through three cognitive responses.

    First, we found that extreme hot days and hotter temperatures are associated with more negative emotional well-being in climate change discourse. This result was robust after controlling for seasonal and regional vulnerability factors. We also found that extreme cold days are positively associated with expressed emotional well-being, while extreme wet days have a weak association with negative emotional well-being. Although cold weather is often linked to lower moods or increased feelings of isolation, when discussing climate change, the prevailing perception tends to focus on rising temperatures. One possible explanation is that on extremely cold days, people may experience relief from long-standing patterns of increased temperature and express more positive sentiments. In contrast, extreme wet days, characterized by rain and overcast skies, may limit outdoor activities and contribute to feelings of lethargy or negative emotions. However, these two weather indicators do not appear to be associated with the proposed mediators, such as social connections and sensory experiences. Overall, this finding is consistent with previous research that has demonstrated the adverse effects of heat stress on mental health outcomes [4,6,63]. However, in comparison, our chosen outcome of emotional well-being may offer insights into a more immediate and nuanced impact of climate change on individuals’ mood states. Over the long term, if these effects persist and accumulate, they have the potential to contribute to the development of clinical mental disorders. Moreover, the results showed that the impact of extreme hot days on emotional well-being is not uniform across seasons and regions and varies depending on the level of vulnerability to heat risk. Specifically, extreme hot days in the summer and regions with a higher population density are associated with more negative emotional well-being than other seasons and regions. These findings confirm previous evidence that heat waves during summer can cause excessive adverse health consequences [64] and negative sentiments [63], and a high urban land use density (including residential buildings) contributes to increased affective symptoms [65]. Together, these findings suggest that the impact of climate change is not only influenced by objective meteorological conditions but also interacts with environmental (eg, average temperature, air pollution) and social (eg, population density) contexts. Individuals with high-risk exposures and population density or in the summer are more vulnerable to heat, in terms of emotional well-being.

    Adding to the existing literature, our study indicates the proposed mediation paths of three cognitive responses of individuals. Specifically, the findings indicate that extreme hot days and hotter temperatures may trigger intuitive-leaning thinking, a less affiliative attitude, and a higher level of somatosensory experience, the former two being associated with more negative emotional well-being. Heat as a stress can comprise people’s cognitive function, which triggers the more cognitive-saving intuitive strategy in thinking [66]. This also confirms previous findings that analytical thinking displays more positive emotions by gaining more control and self-efficacy with rational and logical reasoning [15,67]. Furthermore, the findings are consistent with the literature on social affiliations, which suggests that individuals with high affiliative motives tend to experience positive emotions and exhibit better mental health outcomes [19-21]. Conversely, heat stress, as an immediate and vivid threat, can shift people’s attention to “self” and defensive states and hence make them less socially affiliated. It aligns with previous findings indicating that heat stress can increase individuals’ hostile cognition and aggression [68] and block people’s capacity to care for others (eg, cognitive empathy) [69]. It is noteworthy that although somatosensory experiences were positively associated with extreme heat events, they were unexpectedly found to be positively associated with emotional well-being, contrasting with our initial expectations. This suggests that somatic symptoms induced by heat may not necessarily be linked to more negative emotional well-being. This observation can be supported by the content of several sampled posts with high somatosensory experiences (see Table S2 in Multimedia Appendix 1). A nuanced examination of the post contents found that individuals who shared their somatosensory experiences online may have adapted their observations around their own lives or the internet to make sense of their experiences and proposed environmental protection measures to mitigate the impact. Consequently, expressing somatosensory experiences through social media platforms can serve as a coping method for people dealing with the impacts of climate change and is linked to positive emotional well-being.

    We also identified the negative impact of extreme hot days and hotter temperatures on people’s pro-environmental tendencies, which aligns with previous research in highlighting that extreme weather, such as high temperatures and heavy rain, can make people more reluctant to engage in pro-environmental behaviors [70]. However, this effect weakened when controlling for emotional well-being. This indicates that part of the negative association between extreme heat/temperature increase and pro-environmental tendencies may be attributed to the effect of negative emotional well-being. We found that more negative emotional well-being is associated with lower pro-environmental tendencies. This finding contradicts previous studies with coping mechanisms, highlighting that negative emotions increase pro-environmental behaviors as individuals take actions to mitigate the negative impacts caused by climate change [29,71]. This relationship is particularly evident at low-to-moderate levels of climate anxiety, which were found to be associated with increased pro-environmental behavior. However, our results identified an alternative perspective with positive emotional coping strategies. Positive emotional adaptive coping strategies aim to evoke positive feelings to counteract negative emotions [72] and form a positive self-reinforcing feedback loop [73]. This strategy of regulating emotions, also framed as meaning-focused coping, involves individuals adapting their beliefs, values, and goals to maintain well-being [74]. Combining the existing literature with this study suggests a U-shaped correlation between emotional response to climate change and pro-environmental behaviors, rather than a linear one. This finding supports the broader theoretical framework of an inverted U-shaped correlation. The positive emotions arising from meaning-focused coping align with Fredrickson’s broaden-and-build theory [75], which posits that positive emotions can enhance creative problem solving, social cooperation, and resilience—all essential for sustained environmental action. This insight can inform future interventions and theory conceptualization by highlighting that rather than assuming that inducing negative emotions consistently results in increased action, it is also beneficial to leverage positive emotions to establish effective coping strategies and create a self-reinforcing cycle. In our study, we also found that three cognitive strategies, namely engaging in higher analytical thinking, social affiliations to the community, and sharing somatosensory experiences, can help enhance emotional well-being. Adopting these cognitive strategies is also associated with a greater likelihood of advocating pro-environmental behaviors. This is evidenced by several sampled posts that demonstrated co-occurrences between the three cognitive factors and pro-environmental advocacy (Figure S4 in Multimedia Appendix 1). Although the mediation models suggest that cognitive variables would mediate the relationship between extreme weather indices and emotional well-being, it is important to note that the results do not establish a causal relationship. Based on dual process theory [76], it is possible that emotional and cognitive variables mutually influence and jointly contribute to behavioral responses, which can, in turn, foster positive emotions and coping strategies related to climate change. As mentioned earlier in this section, this creates a self-reinforcing positive cycle. Future studies can further explore this relationship in an experimental setting to investigate specific causal relationships.

    Implications

    This study has several implications for climate change communication and intervention. First, the findings suggest that social media can be a valuable platform for monitoring and assessing the public’s emotional and cognitive responses to climate change, as well as their pro-environmental tendencies. By analyzing the psycholinguistic cues and contents of social media posts, researchers and practitioners can gain insights into the psychological impact of climate change across populations and regions. The results of this study indicate that future interventions should focus on leveraging positive emotions to develop effective coping strategies. This approach can help create a self-reinforcing cycle in addressing the risks associated with climate change. Second, the findings suggest that affiliative thoughts, more analytic thinking, and expression of somatosensory experiences can be important cognitive strategies to regulate emotional well-being and motivate pro-environmental behaviors of individuals. Climate change communicators and educators can actively use more analytical thinking and affiliative language and encourage sharing personal somatosensory experiences to enhance meaning-focused coping abilities and further enhance the public’s emotional well-being and pro-environmental tendencies. For example, a study examining communication strategies found that environmental groups tend to use more affiliative language, such as emphasizing collective identity and efficacy to engage people in collective action [40]. Third, the findings suggest that climate change communication and intervention should address the regional disparities and vulnerabilities in the impact of climate change to increase the resilience and adaptation of the most vulnerable communities. For instance, in regions characterized by a high population density, a potential strategy is to leverage individuals with high social capital and stronger social networks in the community to enhance adaptation and spread innovations in combating climate change [77].

    Limitations

    Despite the potential theoretical and practical contributions of this study, some limitations and directions for future research should be noted. First, this study relied on social media data from Weibo, and hence, the users may not be representative of the general population or users of other social media platforms. Future research could replicate and extend the study on other platforms (eg, X). Second, the study used a correlational design, which does not allow for causal inference or temporal ordering of the variables. Future research could use longitudinal data or experimental designs to establish the causal relationships and mechanisms between meteorological factors, emotional and cognitive responses, and pro-environmental behaviors. Third, due to the data privacy of Weibo, our study did not capture individuals’ sociodemographic data, and hence, these variables were not controlled in the modeling, which may limit the understanding and implications. Fourth, the study focused on a limited set of meteorological factors, emotional and cognitive responses, and pro-environmental behaviors. Future research could explore other meteorological factors, such as wind speed, humidity, and atmospheric pressure, as well as other emotional and cognitive responses, such as hope and moral reasoning, to capture the multifaceted impact of climate change. Although this study offers valuable insights by leveraging data from China—a country notable for its vast geographical diversity and wide range of meteorological and regional vulnerability characteristics—there are inherent limitations regarding the generalizability of the findings to other countries or regions. Differences in climate patterns, topography, socioeconomic factors, and infrastructures worldwide may influence the applicability of the results beyond the Chinese context. Therefore, caution should be exercised when extrapolating these findings to areas with distinct weather conditions or geographic features. Nonetheless, the methodological framework developed in this study provides a robust and adaptable approach that could be effectively applied or modified for similar analyses in other regions, potentially enhancing comparative studies and contributing to a broader understanding of meteorological impacts on emotional well-being worldwide.

    Conclusion

    In summary, this study investigated how extreme meteorological events and meteorological changes affect the public’s emotional and cognitive responses to climate change, as well as their pro-environmental tendencies, using social media, meteorological data, and regional vulnerability data. The results showed that extreme hot days and hotter temperatures are robustly associated with more negative emotional well-being, an intuitive-leaning thinking style, a less affiliative attitude, and a lower inclination toward pro-environmental behaviors. The results also revealed regional and seasonal variations in the impact of climate change. Three cognitive responses, namely thinking style, social affiliations, and somatosensory experiences, showed a positive mediated impact on the relationship between climate change-related events and emotional well-being. The findings have implications for climate change communication and intervention, suggesting the need to consider the psychological impact of climate change on different groups and regions and to promote analytical thinking, affiliative language, and positive somatosensory experiences to enhance the public’s mental well-being, coping abilities and pro-environmental behaviors.

    Acknowledgments

    The authors want to sincerely thank Ms Yuyi Chen for providing support in coding pro-environment-related posts. We did not use any generative artificial intelligence (AI) tools provided by OpenAI or other AI companies during any stage of this manuscript. This work was supported by the Hong Kong Research Grants Council’s General Research Fund (project number 17613724).

    Data Availability

    All data used in this study are publicly available open source data. The details of data collection are reported in the Data Collection section in Multimedia Appendix 1.

    Conflicts of Interest

    None declared.

    Multimedia Appendix 1

    Data collection, weather indicators, vulnerability indicators, and cognitive response measurement.

    DOCX File , 564 KB
    1. Intergovernmental Panel on Climate Change (IPCC). Climate Change 2022 – Impacts, Adaptation and Vulnerability: Working Group II Contribution to the Sixth Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge. Cambridge University Press; 2022.
    2. Palinkas LA, Wong M. Global climate change and mental health. Curr Opin Psychol. Apr 2020;32:12-16. [FREE Full text] [CrossRef] [Medline]
    3. Obradovich N, Migliorini R, Paulus MP, Rahwan I. Empirical evidence of mental health risks posed by climate change. Proc Natl Acad Sci USA. Oct 23, 2018;115(43):10953-10958. [FREE Full text] [CrossRef] [Medline]
    4. Thompson R, Lawrance EL, Roberts LF, Grailey K, Ashrafian H, Maheswaran H, et al. Ambient temperature and mental health: a systematic review and meta-analysis. Lancet Planet Health. Jul 2023;7(7):e580-e589. [FREE Full text] [CrossRef]
    5. Cianconi P, Betrò S, Janiri L. The impact of climate change on mental health: a systematic descriptive review. Front Psychiatry. Mar 6, 2020;11:74. [FREE Full text] [CrossRef] [Medline]
    6. Rocque RJ, Beaudoin C, Ndjaboue R, Cameron L, Poirier-Bergeron L, Poulin-Rheault R-A, et al. Health effects of climate change: an overview of systematic reviews. BMJ Open. Jun 09, 2021;11(6):e046333. [FREE Full text] [CrossRef] [Medline]
    7. Ojala M, Cunsolo A, Ogunbode CA, Middleton J. Anxiety, worry, and grief in a time of environmental and climate crisis: a narrative review. Annu Rev Environ Resour. Oct 18, 2021;46(1):35-58. [FREE Full text] [CrossRef]
    8. Sharpe I, Davison CM. Climate change, climate-related disasters and mental disorder in low- and middle-income countries: a scoping review. BMJ Open. Oct 14, 2021;11(10):e051908. [FREE Full text] [CrossRef] [Medline]
    9. Sun G, Qian S, Jiang Q, Liu K, Li B, Li M, et al. Hyperthermia-induced disruption of functional connectivity in the human brain network. PLoS One. Apr 8, 2013;8(4):e61157. [FREE Full text] [CrossRef] [Medline]
    10. Lõhmus M. Possible biological mechanisms linking mental health and heat-a contemplative review. Int J Environ Res Public Health. Jul 18, 2018;15(7):1515. [FREE Full text] [CrossRef] [Medline]
    11. Berry HL, Bowen K, Kjellstrom T. Climate change and mental health: a causal pathways framework. Int J Public Health. Apr 22, 2010;55(2):123-132. [FREE Full text] [CrossRef] [Medline]
    12. Jacoby J. Stimulus-organism-response reconsidered: an evolutionary step in modeling (consumer) behavior. J Consum Psychol. Jan 25, 2008;12(1):51-57. [FREE Full text] [CrossRef]
    13. Daniel K. Thinking, Fast and Slow. New York. Farrar, Straus and Giroux; 2011.
    14. Volkova EV, Rusalov VM. Cognitive styles and personality. Pers Indiv Differ. Sep 2016;99:266-271. [FREE Full text] [CrossRef]
    15. Nisbett RE, Peng K, Choi I, Norenzayan A. Culture and systems of thought: holistic versus analytic cognition. Psychol Rev. Apr 2001;108(2):291-310. [FREE Full text] [CrossRef] [Medline]
    16. Zhou X, He L, Yang Q, Lao J, Baumeister RF. Control deprivation and styles of thinking. J Pers Soc Psychol. Mar 2012;102(3):460-478. [FREE Full text] [CrossRef] [Medline]
    17. Trémolière B, Djeriouat H. Exploring the roles of analytic cognitive style, climate science literacy, illusion of knowledge, and political orientation in climate change skepticism. . J Environ Psychol. Apr 2021;74:101561. [FREE Full text] [CrossRef]
    18. Slavich GM. Social safety theory: a biologically based evolutionary perspective on life stress, health, and behavior. Annu Rev Clin Psychol. May 07, 2020;16:265-295. [FREE Full text] [CrossRef] [Medline]
    19. Crocker J, Canevello A, Breines JG, Flynn H. Interpersonal goals and change in anxiety and dysphoria in first-semester college students. J Pers Soc Psychol. Jun 2010;98(6):1009-1024. [FREE Full text] [CrossRef] [Medline]
    20. Keltner D, Kogan A, Piff PK, Saturn SR. The sociocultural appraisals, values, and emotions (SAVE) framework of prosociality: core processes from gene to meme. Annu Rev Psychol. Jan 03, 2014;65(1):425-460. [FREE Full text] [CrossRef] [Medline]
    21. Gilbert P. Affiliative and prosocial motives and emotions in mental health. Dialogues Clin Neuro. Apr 01, 2022;17(4):381-389. [FREE Full text] [CrossRef]
    22. Fredrickson BL. What good are positive emotions? Rev Gen Psychol. 1998;2(3):300-319. [FREE Full text] [CrossRef] [Medline]
    23. Huang X, Yao R, Xu T, Zhang S. The impact of heatwaves on human perceived thermal comfort and thermal resilience potential in urban public open spaces. Build Environ. Aug 2023;242:110586. [FREE Full text] [CrossRef]
    24. Hornung J. Social identities in climate action. Clim Action. Mar 11, 2022;1(1):4. [FREE Full text] [CrossRef]
    25. Raison CL, Hale MW, Williams LE, Wager TD, Lowry CA. Somatic influences on subjective well-being and affective disorders: the convergence of thermosensory and central serotonergic systems. Front Psychol. Jan 13, 2014;5:1580. [FREE Full text] [CrossRef] [Medline]
    26. McGlone F, Reilly D. The cutaneous sensory system. Neurosci Biobehav Rev. Feb 2010;34(2):148-159. [FREE Full text] [CrossRef] [Medline]
    27. Nummenmaa L, Glerean E, Hari R, Hietanen JK. Bodily maps of emotions. Proc Natl Acad Sci USA. Jan 14, 2014;111(2):646-651. [FREE Full text] [CrossRef] [Medline]
    28. Brosch T. Affect and emotions as drivers of climate change perception and action: a review. Curr Opin Behav Sci. Dec 2021;42:15-21. [FREE Full text] [CrossRef]
    29. D'Amico S, Marano A, Geraci MA, Legge E. Perceived self-efficacy and coping styles related to stressful critical life events. PLoS One. Jul 12, 2013;8(7):e67571. [FREE Full text] [CrossRef] [Medline]
    30. Albrecht G. Chronic environmental change: emerging 'psychoterratic' syndromes. Int Cult Psychol. 2011:43-56. [FREE Full text] [CrossRef]
    31. Innocenti M, Santarelli G, Lombardi GS, Ciabini L, Zjalic D, Di Russo M, et al. How can climate change anxiety induce both pro-environmental behaviours and eco-paralysis? The mediating role of general self-efficacy. Int J Environ Res Public Health. Feb 10, 2023;20(4):3085. [FREE Full text] [CrossRef] [Medline]
    32. Han Z, Shen M, Liu H, Peng Y. Topical and emotional expressions regarding extreme weather disasters on social media: a comparison of posts from official media and the public. Humanit Soc Sci Commun. Nov 28, 2022;9(1):421. [FREE Full text] [CrossRef] [Medline]
    33. Pearce W, Niederer S, Özkula SM, Sánchez Querubín N. The social media life of climate change: platforms, publics, and future imaginaries. WIREs Clim Change. Dec 27, 2018;10(2):e569. [FREE Full text] [CrossRef]
    34. Roxburgh N, Guan D, Shin KJ, Rand W, Managi S, Lovelace R, et al. Characterising climate change discourse on social media during extreme weather events. Global Environ Change. Jan 2019;54:50-60. [FREE Full text] [CrossRef]
    35. Debnath R, Bardhan R, Shah DU, Mohaddes K, Ramage MH, Alvarez RM, et al. Social media enables people-centric climate action in the hard-to-decarbonise building sector. Sci Rep. Nov 17, 2022;12(1):19017. [FREE Full text] [CrossRef] [Medline]
    36. Uban A, Chulvi B, Rosso P. An emotion and cognitive based analysis of mental health disorders from social media data. Future Gener Comput Syst. Nov 2021;124:480-494. [FREE Full text] [CrossRef]
    37. Tsugawa S, Kikuchi Y, Kishino F, Nakajima K, Itoh Y, Ohsaki H. Recognizing depression from Twitter activity. 2015. Presented at: 33rd Annual ACM Conference on Human Factors in Computing Systems; April 18-23, 2015:3187-3196; Seoul, Republic of Korea. URL: https://dl.acm.org/doi/10.1145/2702123.2702280 [CrossRef]
    38. Losada D, Crestani F, Parapar J. Overview of eRisk 2019 early risk prediction on the Internet. Lect Notes Comput Sci. 2019;11696:340-357. [FREE Full text] [CrossRef]
    39. Veltri GA, Atanasova D. Climate change on Twitter: content, media ecology and information sharing behaviour. Public Underst Sci. Aug 26, 2017;26(6):721-737. [FREE Full text] [CrossRef] [Medline]
    40. Gulliver R, Fielding KS, Louis WR. Assessing the mobilization potential of environmental advocacy communication. J Environ Psychol. Apr 2021;74:101563. [FREE Full text] [CrossRef]
    41. Monzani D, Vergani L, Pizzoli SFM, Marton G, Pravettoni G. Emotional tone, analytical thinking, and somatosensory processes of a sample of Italian tweets during the first phases of the COVID-19 pandemic: observational study. J Med Internet Res. Oct 27, 2021;23(10):e29820. [FREE Full text] [CrossRef] [Medline]
    42. Ashokkumar A, Pennebaker JW. Social media conversations reveal large psychological shifts caused by COVID-19's onset across U.S. cities. Sci Adv. Sep 24, 2021;7(39):eabg7843. [FREE Full text] [CrossRef] [Medline]
    43. Jiang J, Murrugara-Llerena N, Bos MW, Liu Y, Shah N, Neves L, et al. Sunshine with a chance of smiles: how does weather impact sentiment on social media? 2022. Presented at: Sixteenth International AAAI Conference on Web and Social Media; June 6-9, 2022:393-404; Atlanta, GA. URL: https://ojs.aaai.org/index.php/ICWSM/article/view/19301 [CrossRef]
    44. Ma X, Liu W, Zhou X, Qin C, Chen Y, Xiang Y, et al. Evolution of online public opinion during meteorological disasters. Environ Hazards. Nov 07, 2019;19(4):375-397. [FREE Full text] [CrossRef]
    45. Smith TT, Zaitchik BF, Gohlke JM. Heat waves in the United States: definitions, patterns and trends. Clim Change. Jun 14, 2013;118(3-4):811-825. [FREE Full text] [CrossRef] [Medline]
    46. New M, Hewitson B, Stephenson DB, Tsiga A, Kruger A, Manhique A, et al. Evidence of trends in daily climate extremes over Southern and West Africa. J Geophys Res. Jul 21, 2006;111(D14):1-11. [FREE Full text] [CrossRef]
    47. Robinson PJ. On the definition of a heat wave. J Appl Meteorol. Apr 2001;40(4):762-775. [FREE Full text] [CrossRef]
    48. Golechha M, Shah P, Mavalankar D. Threshold determination and temperature trends analysis of Indian cities for effective implementation of an early warning system. Urban Clim. Sep 2021;39:100934. [FREE Full text] [CrossRef]
    49. Stephenson D. Definition, diagnosis, and origin of extreme weather and climate events. In: Diaz HF, Murnane RJ, editors. Climate Extremes and Society. Cambridge. Cambridge University Press; 2008:11-23.
    50. Park CL, Kubzansky LD, Chafouleas SM, Davidson RJ, Keltner D, Parsafar P, et al. Emotional well-being: what it is and why it matters. Affect Sci. Mar 15, 2023;4(1):10-20. [FREE Full text] [CrossRef] [Medline]
    51. Zhao N, Jiao D, Bai S, Zhu T. Evaluating the validity of simplified Chinese version of LIWC in detecting psychological expressions in short texts on social network services. PLoS One. Jun 20, 2016;11(6):e0157947. [FREE Full text] [CrossRef] [Medline]
    52. Valdez D, Ten Thij M, Bathina K, Rutter LA, Bollen J. Social media insights into US mental health during the COVID-19 pandemic: longitudinal analysis of Twitter data. J Med Internet Res. Dec 14, 2020;22(12):e21418. [FREE Full text] [CrossRef] [Medline]
    53. Kalamas M, Cleveland M, Laroche M. Pro-environmental behaviors for thee but not for me: green giants, green gods, and external environmental locus of control. J Bus Res. Feb 2014;67(2):12-22. [FREE Full text] [CrossRef]
    54. Zhang X, Li Y, Cheng Y, Wang Y, Wang Y, Yao X. Assessment of regional health vulnerability to extreme heat - China, 2019. China CDC Wkly. Jun 04, 2021;3(23):490-494. [FREE Full text] [CrossRef] [Medline]
    55. Turner BL, Kasperson RE, Matson PA, McCarthy JJ, Corell RW, Christensen L, et al. A framework for vulnerability analysis in sustainability science. Proc Natl Acad Sci USA. Jul 08, 2003;100(14):8074-8079. [FREE Full text] [CrossRef] [Medline]
    56. Debnath R, van der Linden S, Alvarez RM, Sovacool BK. Facilitating system-level behavioural climate action using computational social science. Nat Hum Behav. Feb 27, 2023;7(2):155-156. [FREE Full text] [CrossRef] [Medline]
    57. Chernick MR. Bootstrap Methods: A Guide for Practitioners and Researchers. Hoboken NJ. John Wiley & Sons; 2007.
    58. Mulaik SA, James LR, Van Alstine J, Bennett N, Lind S, Stilwell CD. Evaluation of goodness-of-fit indices for structural equation models. Psychol Bull. May 1989;105(3):430-445. [FREE Full text] [CrossRef]
    59. Yuan KH. Fit indices versus test statistics. Multivariate Behav Res. Jan 01, 2005;40(1):115-148. [FREE Full text] [CrossRef] [Medline]
    60. James G, Witten D, Hastie T, Tibshirani R. An Introduction to Statistical Learning. New York. Springer; 2013.
    61. Markham A, Buchanan E, Committee wfftAEW. Ethical decision-making and internet research: recommendations from the AoIR Ethics Working Committee (version 2.0). AoIR. 2012. URL: http://www.aoir.org/reports/ethics2.pdf [accessed 2025-09-16]
    62. Lewis SC, King AD, Perkins-Kirkpkpatrick SE. Defining a new normal for extremes in a warming world. B Am Meteorol Soc. Jun 2017;98(6):1139-1151. [FREE Full text] [CrossRef]
    63. Wang JH, Obradovich N, Zheng SQ. A 43-million-person investigation into weather and expressed sentiment in a changing climate. One Earth. Jun 2020;2(6):568-577. [FREE Full text] [CrossRef]
    64. Ebi KL, Capon A, Berry P, Broderick C, de Dear R, Havenith G, et al. Hot weather and heat extremes: health risks. Lancet. Aug 2021;398(10301):698-708. [FREE Full text] [CrossRef]
    65. Xu J, Liu N, Polemiti E, Garcia-Mondragon L, Tang J, Liu X, et al. the environMENTAL Consortium. Effects of urban living environments on mental health in adults. Nat Med. Jun 2023;29(6):1456-1467. [FREE Full text] [CrossRef] [Medline]
    66. Mazloumi A, Golbabaei F, Mahmood Khani S, Kazemi Z, Hosseini M, Abbasinia M, et al. Evaluating effects of heat stress on cognitive function among workers in a hot industry. Health Promot Perspect. 2014;4(2):240-246. [FREE Full text] [CrossRef] [Medline]
    67. Sagone E, De Caroli ME. Relationships between resilience, self-efficacy, and thinking styles in Italian middle adolescents. Procedia: Social Behav Sci. Oct 2013;92:838-845. [FREE Full text] [CrossRef]
    68. Anderson CA, Deuser WE, DeNeve KM. Hot temperatures, hostile affect, hostile cognition, and arousal - tests of a general-model of affective aggression. Pers Soc Psychol Bull. May 01, 1995;21(5):434-448. [FREE Full text] [CrossRef]
    69. Nitschke JP, Bartz JA. Neurosci Biobehav Rev. Jan 2023;144:105003. [FREE Full text] [CrossRef] [Medline]
    70. Wang Y, Wu Y, Han Z. Extreme weather and residents’ pro-environmental behaviors. Front Environ Sci. Oct 7, 2022;10:1030010. [FREE Full text] [CrossRef]
    71. Ogunbode CA, Doran R, Hanss D, Ojala M, Salmela-Aro K, van den Broek KL, et al. Climate anxiety, wellbeing and pro-environmental action: correlates of negative emotional responses to climate change in 32 countries. J Environ Psychol. Dec 2022;84:101887. [FREE Full text] [CrossRef]
    72. Folkman S. The case for positive emotions in the stress process. Anxiety Stress Coping. Jan 26, 2008;21(1):3-14. [FREE Full text] [CrossRef] [Medline]
    73. Schneider CR, van der Linden S. An emotional road to sustainability: how affective science can support pro-climate action. Emotion Rev. Aug 08, 2023;15(4):284-288. [FREE Full text] [CrossRef]
    74. Ojala M. How do children cope with global climate change? Coping strategies, engagement, and well-being. J Environ Psychol. Sep 2012;32(3):225-233. [FREE Full text] [CrossRef]
    75. Schneider CR, Zaval L, Markowitz EM. Positive emotions and climate change. Curr Opin Behav Sci. Dec 2021;42:114-120. [FREE Full text] [CrossRef]
    76. Evans JSBT, Stanovich KE. Dual-process theories of higher cognition: advancing the debate. Perspect Psychol Sci. May 2013;8(3):223-241. [FREE Full text] [CrossRef] [Medline]
    77. Wilson RS, Herziger A, Hamilton M, Brooks JS. From incremental to transformative adaptation in individual responses to climate-exacerbated hazards. Nat Clim Change. Feb 10, 2020;10(3):200-208. [FREE Full text] [CrossRef]

    ‎
    AI: artificial intelligence
    CFI: comparative fit index
    GFI: goodness-of-fit index
    LIWC: Linguistic Inquiry and Word Count
    OR: odds ratio
    PEB: Pro-Environmental Behavior questionnaire
    RMSEA: root mean square error of approximation
    SC-LIWC: Simplified Chinese version of the Linguistic Inquiry and Word Count
    SEM: structural equation modeling
    VIF: variance inflation factor

    Edited by A Mavragani; submitted 19.Dec.2024; peer-reviewed by M ten Thij, TJ Tao; comments to author 01.Jul.2025; revised version received 21.Jul.2025; accepted 19.Aug.2025; published 03.Oct.2025.

    Copyright

    ©Yucan Xu, Jiehu Yuan, Sijia Li, Qiuyan Liao. Originally published in the Journal of Medical Internet Research (https://www.jmir.org), 03.Oct.2025.

    This is an open-access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work, first published in the Journal of Medical Internet Research (ISSN 1438-8871), is properly cited. The complete bibliographic information, a link to the original publication on https://www.jmir.org/, as well as this copyright and license information must be included.