<p>Grounded in the universal human predisposition to seek social connection, this study uniquely integrates social media data with mobility patterns to provide a holistic view of public emotional dynamics during the pandemic. Drawing on 45 million geocoded tweets from SINA.com, 57 billion trip records across 335 cities from Tencent Location Big Data, and 296 million mutual following records from social media users, we construct a large-scale, daily city-level emotional metric across different stages of pandemic. Our findings reveal significant emotional autocorrelation across different ties and identify four distinct feedback mechanisms shaping emotional interactions during the pandemic: amplified positive feedback in social media ties during the warning phase; shrinking positive feedback in geographic ties during the isolation phase; and negative and asymmetric feedback in social media and mobility ties, respectively, during the normalization phase. Notably, the easing of pandemic restrictions corresponded with a surge in positive emotional interactions in geographic and mobility ties. These findings underscore the role of physical and digital social networks, spanning geographic, mobility, and social media ties, in shaping public emotional interactions throughout different pandemic stages. Understanding these dynamics can inform crisis communication strategies, mitigate public distress, and enhance social resilience, offering valuable insights for policymakers managing future large-scale disruptions.</p>