Natural disasters cause significant casualties, economic losses, and social disruptions globally. International humanitarian assistance is crucial, particularly for developing countries, with private sector donations playing a vital role. Online social responses reflect public interest and are amplified by information and communication technologies. Social media facilitates rapid information dissemination, connecting local disasters to global audiences and stimulating international aid. Understanding these responses and their influencing factors is essential for coordinating global humanitarian efforts and post-disaster reconstruction. Existing research focuses on user emotions, posting themes, and Twitter communication mechanisms, but lacks a comprehensive understanding of global online responses and their drivers. This study addresses this gap by analyzing the spatiotemporal dynamics of global social media responses to Typhoon Haiyan and identifying key socioeconomic influencing factors.

Prior research on online social responses to disasters has explored three main areas: user emotions, posting themes, and Twitter communication mechanisms. Studies on user emotions examined the impact of emotional expression on information dissemination and identified spatial clusters of negative emotions. Research on posting themes analyzed the evolution of discussion topics across disaster stages and identified thematic differences among various demographic groups. Studies on Twitter communication mechanisms focused on the factors affecting social media usage during and after disasters, primarily within affected countries. However, a comprehensive investigation of global online social responses and their influencing factors remains limited, highlighting the need for this study.

This study used 234,042 tweets related to Typhoon Haiyan collected from November 4 to 20, 2013. Tweets were geo-decoded to identify the location of users. The study area included 113 countries, excluding China due to low Twitter penetration. Eleven influencing factors were considered, categorized into political, economic, social, cultural, natural, and demographic dimensions. Data sources included the World Bank, CNKI, TIKenya, Wikipedia, and EM-DAT. The study used two main methods: 1. **Spatiotemporal analysis:** Daily tweet counts were calculated to analyze the temporal evolution of global and country-level online social responses. Standard deviation ellipses were used to describe the spatial distribution of attention. The study period was divided into two stages: before and after Typhoon Haiyan's landfall. 2. **Attributional analysis:** The Geographical Detector (Geodetector) was employed for single-factor analysis to identify leading factors influencing online social responses. The Geographically Weighted Regression (GWR) model was then used for multiple-factor analysis to investigate spatial heterogeneity in factor coefficients. Spatial autocorrelation was assessed using Moran's I before fitting the GWR model.

The temporal analysis revealed that global online social responses followed the development of Typhoon Haiyan, peaking on November 8th (landfall) and declining thereafter. A secondary peak was observed on November 10-11, likely due to the typhoon's impact on Hainan, China, and ongoing relief efforts. Spatial analysis showed the highest response in the Philippines (the affected country), followed by Western countries and neighboring countries. Developed countries showed higher response levels than developing countries, possibly due to greater capacity for disaster relief. The US showed a particularly high response, potentially due to historical ties with the Philippines. The Geodetector analysis revealed that economic factors (export value and per capita GDP) had the strongest influence on online social responses both before and after landfall, followed by demographic factors (population). Political, social, and cultural factors had weaker effects. The GWR model showed that export value and population had predominantly positive influences on online social responses globally. However, the CPI and the proportion of nonreligious beliefs showed mixed effects, both positive and negative in different regions. The GDP loss rate's influence was also spatially heterogeneous, showing positive effects in some regions and negative effects in others. Post-landfall analysis showed similar patterns, with the addition of government effectiveness as a significant factor.

The findings demonstrate that online social responses to natural disasters transcend geographical boundaries, exceeding the influence of physical distance. The dominance of economic factors highlights the capacity and willingness of wealthier nations to engage in international aid. The strong response from Western countries reflects their economic strength and established connections with the disaster-affected region. The relatively high response from neighboring countries points to the proximity factor in typhoon-related events. The study's findings are consistent with previous research highlighting the role of Western countries in global responses to disasters. The results have significant implications for humanitarian aid organizations and governments, highlighting the potential of leveraging online social networks for mobilizing volunteers and donors.

This study demonstrates the global reach of online social responses to natural disasters and the pivotal role of economic factors in driving these responses. Western countries exhibit strong engagement, indicating their potential as a source of volunteers and donations for international aid efforts. Future research should explore the dynamics of online social responses across diverse disaster types and geographical contexts.

This study has some limitations. Data limitations resulted in the exclusion of some countries from the analysis, potentially affecting the GWR results. The selection of influencing factors might have involved omitted variables. Data sources like Wikipedia introduced potential uncertainty. The use of government effectiveness as a political factor is also subject to definitional and methodological debates. The unexpectedly weak global response to the typhoon's impact on Hainan, China requires further investigation.