Temperature influences mood: evidence from 11 years of Baidu index data in Chinese provincial capitals.

INTRODUCTION: This study explores the dynamic relationship between temperature changes and public negative emotions-specifically depression, anxiety, and loneliness. It introduces an innovative approach by integrating climate data with digital behavior metrics to provide objective insights into how environmental factors may influence mental health. METHODS: A dataset combining daily meteorological records and Baidu search indices from 31 provincial capital cities in China (2013-2023) was used. Search engine query data served as a proxy for public emotional states, avoiding social desirability bias commonly found in self-reported surveys. Panel fixed-effect models and heterogeneity analysis were employed to assess the impact of daily mean temperature (DMT) and daily temperature range (DTR). A "climate zone × season" framework was constructed to explore regional and seasonal variations. Threshold regression analysis was also conducted to identify nonlinear effects. RESULTS: The results showed that for every 1°C increase in DMT, search indices for depression, anxiety, and loneliness increased significantly by 22.71%, 18.76%, and 19.59%, respectively (p < 0.01). Conversely, a 1°C increase in DTR led to decreases of 30.35%, 31.19%, and 15.41% in these indices (p < 0.05). Threshold regression analysis revealed that the adverse effect of high temperatures on loneliness became insignificant when DTR exceeded 14°C. Heterogeneity analysis highlighted significant regional and seasonal differences, particularly during cold seasons in severely cold zones and hot seasons in warm summer-cold winter zones. DISCUSSION: The findings suggest that temperature fluctuations have a complex and regionally dependent impact on public mental health. The moderating role of climate characteristics and seasonal patterns underscores the importance of localized climate policies and mental health interventions. This study provides empirical evidence based on objective behavioral data, contributing to climate-related public health strategies and adaptive policy design.