Reader TTS

ABSTRACT The global public interest in a natural disaster event will help disaster-stricken areas obtain post-disaster international relief and assistance. However, knowledge gaps still

exist in regard to global online social responses and their socioeconomic influencing factors. We used big social media data regarding the 2013 Super Typhoon Haiyan to explore global online

social responses and to investigate the socioeconomic factors influencing this behavior based on the Geographical Detector (Geodetector) model and geographically weighted regression (GWR)

model. The results show that global online social responses have little relation with geographical distance and follow the disaster’s development. In addition to the most response in the

disaster-affected countries, Western countries and neighboring countries have more online social response to the disaster than other regions. Among all the influencing factors, economic

factors have the strongest effect on public interest both before and after the typhoon’s landfall. Our findings indicate that online social users are of great potential for volunteers and

donors. SIMILAR CONTENT BEING VIEWED BY OTHERS CROSS-PLATFORM ANALYSIS OF PUBLIC RESPONSES TO THE 2019 RIDGECREST EARTHQUAKE SEQUENCE ON TWITTER AND REDDIT Article Open access 31 January

2022 TWITTER DATA FROM THE 2019–20 AUSTRALIAN BUSHFIRES REVEALS PARTICIPATORY AND TEMPORAL VARIATIONS IN SOCIAL MEDIA USE FOR DISASTER RECOVERY Article Open access 08 October 2022 EVENT

TRIGGERS AND OPINION LEADERS SHAPE CLIMATE CHANGE DISCOURSE ON WEIBO Article Open access 24 February 2025 INTRODUCTION Natural disasters frequently occur around the world and bring massive

casualties, economic losses, and social disruptions to countries and international society (Shen et al., 2018; United Nations Office For Disaster Risk Reduction (2020)). The synergies

between disaster risk reduction and sustainable development require global efforts (Aitsi-Selmi et al., 2016). Especially for developing countries and less developed regions, international

humanitarian assistance and foreign governmental aid are particularly essential for disaster risk reduction and mitigation (Cook et al., 2018). In addition to governmental rescuing relief

and financial and economic assistance, donations from the foreign private sector are critical resources for disaster-stricken countries (Coppola, 2020). Evidence has confirmed the

significant impact of disasters on global public interest and responses (David et al., 2016; Tan and Maharjan, 2018; Kam et al., 2021). Online social responses are the main reflection of

public interest, especially on a global scale. Due to the unbelievably rapid progress in information and communication technologies and personal mobile devices, online social networks have

become a vital and primary channel of information communication and dissemination (Tang et al., 2021). Before, during, and after the occurrence of a disaster event, global social media users

are highly active in posting, discussing, and forwarding the situation (Kumar, 2020). In this manner, information related to a local disaster event can be enhanced and immediately

disseminated to and spread through international society. Consequently, a local natural disaster, especially a catastrophic disaster, will provoke global social responses and attract more

significant public interest through online social networks (Ruan et al., 2022). Thus, aid agencies and humanitarian organizations are increasingly aware of the significance of communication

during natural disasters, which is conducive to not only the effective implementation of relief but also timely social responses, and it can stimulate the enthusiasm of donors even in remote

countries (Kam et al., 2021). Understanding global social responses to disaster events and the influencing factors will help integrate global society, enterprises, nongovernmental

organizations and other social forces to participate in humanitarian assistance. This is of great significance and key to post-disaster reconstruction for impoverished countries.

Nevertheless, research on global public interest in or online social responses to a disaster is still a challenging issue, and it is even more difficult on a global scale. Existing research

on online social responses follows three narratives: the emotions of social media users, the themes of social media posts, and the Twitter communication mechanism. In a study on users’

emotions after a disaster, Chen et al. (2020) explored the emotions and Twitter forwarding patterns of affected and nonaffected areas before and after the disaster, revealing the significant

impact of public emotional expression on the release and redistribution of disaster-related information. Later, scholars further studied the spatial and temporal distribution

characteristics of people’s negative emotions caused by disasters. Garske et al. (2021) used the global and local Moran’s I to analyze Twitter data with geographic references and negative

emotions and found that all negative emotions were clustered in the local space during a disaster. Gruebner et al. (2018) used an ordinary least squares regression model to evaluate the

association between pre-disaster and post-disaster discomfort rates and to detect spatial clusters of negative emotions in various administrative regions of New York City. In addition, some

scholars have combined machine learning methods to assess regional post-disaster reconstruction recovery by studying the emotions and perspectives of social media users after a disaster (Yan

et al., 2020; Contreras et al., 2022). In research on the posting themes of social media users, García-Ramírez et al. statistically analyzed the posting themes of social media users only in

the disaster response stage (García-Ramírez et al., 2021). Brandt et al. analyzed the posting themes of social media users during the pre-disaster, disaster occurrence, and post-disaster

short-term and long-term periods (Brandt et al., 2019). Zhang and Cheng used a machine learning method to classify social media data, explored the change process of the public’s discussion

topics at different stages during disasters, and analyzed the differences in people’s discussion content under different emotions (Zhang and Cheng, 2021). Some scholars have also studied

different themes of online social response when disasters occur. Correlation analyses demonstrate that there are differences in themes between different genders and between black and white

groups when extreme disasters occur (Yuan et al., 2020; Zhu and Liu, 2021). Regarding the transmission mechanism of Twitter, Takahashi et al. investigated the use of Twitter when and after

Typhoon Haiyan hit the Philippines and explored the external factors (using time and geographical location) and internal factors affecting the use of social media (Takahashi et al., 2015).

However, the area covered by this study covered only affected countries. Kam et al. (2021) established a model to simulate global attention to disaster events by using the relative search

activities of users of Google products, revealing that Western countries play a dominant role in global attention to disasters. However, there is still a lack of understanding of global

online social responses to disaster events and their influencing factors. Applications of social media data in the global public interest to disaster events have not been well investigated

due to the lack of typical datasets. To fill this knowledge gap, this study takes the 2013 Super Typhoon Haiyan, a historic deadly disaster in the western Pacific, as a representative case

to investigate the spatiotemporal changes in global social responses and to attribute their socioeconomic influencing factors in various countries. DATA AND METHODS OVERVIEW OF TYPHOON

HAIYAN Super Typhoon Haiyan is one of the most representative extreme disaster events. As the most powerful typhoon (maximum sustained winds near the center of 315 km/h) ever recorded in the

western Pacific, it swept across the Philippines and other regions (Fig. 1) and caused severe losses to the Philippines. According to official statistics, the direct economic loss suffered

by the Philippines was approximately 8.96 billion Philippine pesos (approximately 1.63 billion current US dollars), and more than 16 million people were affected by the disaster (NDRRMC

Philippines, 2013). Due to its historic intensity and damage power, Super Typhoon Haiyan attracted global attention and public interest, and information flooded Twitter (David et al., 2016;

Shen et al., 2021). The track of Typhoon Haiyan is shown in Fig. 3. Typhoon Haiyan generated and developed rapidly in the South Pacific Ocean on November 4, 2013. It was officially upgraded

to a tropical storm. On November 8, Typhoon Haiyan made landfall along the coast of Guiuan in the Philippines with extremely destructive winds, causing devastating damage to the country.

Typhoon Haiyan entered the central South China Sea on November 9, and its intensity significantly weakened. Affected by the circulation and the northeast monsoon, Typhoon Haiyan turned to

the north on November 10 and attacked the Hainan Province of China. On November 11, due to the further influence of topography and wind direction, the intensity of Typhoon Haiyan rapidly

weakened, and it gradually dissipated. DATA AND PREPROCESSING Tweets were obtained from the Twitter platform using disaster hashtags and web crawler scripts based on the method and data

provided in previous publications (Murzintcev and Cheng, 2017; Shen et al., 2021; Zhang et al., 2021). The resulting global tweets related to Typhoon Haiyan from November 4 to 20, 2013,

totaled 234,042. The data contain the location attributes or mobile phone location addresses of social media users and were then geo-decoded into points through ArcGIS 10.7 software. In

total, there are 113 countries included in the following attributional analysis of this study (Fig. 2). Countries were filtered out based on two criteria: (1) tweets from a country are

missing or number less than 30; and (2) the subsequent acquisition of relevant data on influencing factors. In addition, due to a very low penetration rate of Twitter users, China was

excluded from this study. Therefore, the countries in the attribution analysis include 8 South American countries, 10 countries in North America, 27 countries in Asia, 33 European countries,

33 countries in Africa, and two Oceania countries (Australia and New Zealand). The influencing factors and data sources in this study are shown in Table 1. In the era of globalization, the

interpretation of online social responses to disasters is quite complex. They interact with political power, surrounding social relations and cultural significance and the process of global

interdependence (Kam et al., 2021). Therefore, this paper explores the political, economic, social, cultural, natural and demographic factors that affect the differences in online social

responses to a disaster in different countries. All of the specific factors and corresponding indices are 2013 values. (1) Political dimension: The government effectiveness can be used to

measure the quality of public and civil services, independence from political pressure, the quality of policy formulation and implementation, and the credibility of the government’s

commitment to such policies around the world (Sadaf et al., 2018). People in high government effectiveness countries have more willingness to express their response to natural disasters. (2)

Economic dimension: The level of economic development can reflect people’s capability to aid foreign countries in a country. The economic development of a country can be measured by per

capita GDP. If a country is more connected to other countries, its citizens are more interested in international events. Therefore, exports of goods and services are selected to measure the

degree of trade openness. (3) Social dimension: Transparent countries advocate the construction of a harmonious and mutually supportive society (Raschky and Schwindt, 2012). The corruption

perceptions index (CPI) can be used to measure the social development of a country and the position of the national public toward foreign aid. Citizens of countries with higher levels of

social development may have more empathy and willingness to respond to foreign disasters. (4) Cultural dimension: Culture affects people’s willingness to call for help and their views on

aid, and it is an important background for people to understand and respond to disasters (Hoffman and Oliver-Smith, 2002). The level of education explains the level of public awareness of

disasters, which is measured by the higher education enrollment rate here (Zhang and Cheng, 2021). In addition, different religions have different understandings of humanitarianism, which

will also affect people’s response to disasters to some extent. Here, religious influence is measured by the proportion of mainstream religious beliefs (i.e., Christianity, Islam, Buddhism,

Hinduism) and the proportion of nonreligious beliefs in a country. (5) Natural dimension: The risks of natural disasters in a country will affect the sensitivity and attention of domestic

people to extreme disaster events. The expected death rate, the affected population rate and the GDP loss rate can be used to measure the comprehensive natural disaster risk level of each

country. In addition, distance is an obstacle to generously responding to others’ needs or caring actions, and it is one of the determinants of online social responses to disaster events.

The geographical distance between countries and disaster places is expressed by the Euclidean distance from different country capitals to Manila calculated with ArcGIS. (6) Demographic

dimension: The social media user base of a country will affect people’s response to disasters. Due to the lack of precise accounts of Twitter users in each country in 2013, a country’s total

population is employed as a proxy index to indicate the number of social media users. RESEARCH WORKFLOW The specific research flow of this study is shown in Fig. 3. In this study, we used

the number of disaster-related tweets posted over a period to represent the online social response to a disaster. Hence, the daily online social response was obtained by calculating the

number of daily tweets issued in a region using the zonal statics tool in ArcGIS 10.7. We calculated the daily global and country-wide online social response to analyze the spatiotemporal

evolution of online social responses around the world. Based on its temporal changes, the research period is divided into two stages: before and after the typhoon lands. Then, the online

social response changes before and after the typhoon lands are mapped and revealed by calculating the number of tweets and the relative and absolute differences in the two stages. This study

focuses on investigating the influencing factors of online social responses. First, we selected indicators combined with the literature, collected data (details in Section 2.3), conducted a

single-factor analysis of the differences in online social responses to disasters in various countries using the Geographical Detector (Geodetector), and then selected the leading factors

for the geographically weighted regression (GWR) model for multifactor analysis. Finally, the spatial differentiation of influencing factor coefficients and the variation in the influencing

factor coefficients before and after typhoon landing are discussed. STANDARD DEVIATION ELLIPSE In this paper, the standard deviation ellipse is used to quantitatively describe the spatial

distribution characteristics of global social media users’ attention to extreme disaster events. This method can explain the spatial distribution characteristics of geographical elements

from global and spatial perspectives (Lefever, 1926). Its calculation formula is as follows: $$\begin{array}{*{20}{c}} {SDE_x = \sqrt {\frac{{\mathop {\sum }\nolimits_{i = 1}^n \left( {x_i -

\bar{x}} \right)^2}}{n}} } \end{array}$$ (1) $$\begin{array}{*{20}{c}} {SDE_y = \sqrt {\frac{{\mathop {\sum }\nolimits_{i = 1}^n \left( {y_i - \bar{y}} \right)^2}}{n}} } \end{array}$$ (2)

In the formulas, _x__i_ and _y__i_ are the coordinates of the _i_th subregion. \(\bar{x}\) and \(\bar{y}\) represent the coordinates of the center of gravity of subregion _i_. _n_ is the

total number of subregions. GEOGRAPHICAL DETECTOR There are a large number of factors selected in attribution analysis, and multicollinearity among factors easily occurs; leading factors and

invalid factors cannot be filtered out. Therefore, this paper utilized the Geodetector model (Wang et al., 2010) to explain the driving forces influencing online social responses to

disaster events in various countries. Its calculation formula is as follows: $$q = 1 - \frac{{\mathop {\sum}\nolimits_{h = 1}^L {N_h\sigma _h^2} }}{{N\sigma ^2}}$$ (3) In the formula, _q_ is

the measurement factor of the explanatory power of the independent variable, and the range is [0,1]. _L_ is the independent variable. _N_ and _N__h_ are the number of samples in the whole

study area and layer _h_, respectively. σ_h_2 and σ2 are the variances in the Y value of layer _h_ and the whole study area, respectively. GEOGRAPHICALLY WEIGHTED REGRESSION In this paper,

before fitting the GWR model, it is necessary to use spatial autocorrelation to test whether there is an agglomeration of an attribute in space. Moran’s I index is generally used to describe

the spatial characteristics of the distribution of an attribute in the study area (Moran, 1950). Its calculation formula is as follows: $$\begin{array}{*{20}{c}} {I = \frac{{n\mathop

{\sum}\nolimits_{i = 1}^n {\mathop {\sum}\nolimits_{j = 1}^n {W_{ij}\left( {x_i - \bar{x}} \right)\left( {x_j - \bar{x}} \right)} } }}{{\mathop {\sum}\nolimits_{{{{\mathrm{i}}}} = 1}^n

{\mathop {\sum}\nolimits_{j = 1}^n {W_{ij}} \mathop {\sum}\nolimits_{i = 1}^n {\left( {x_i - \bar{x}} \right)^2} } }}} \end{array}$$ (4) In the formula, _x__i_ is the observed value of

country _i_. _W__ij_ is the spatial weight matrix. GWR is a local form of regression that models the spatial heterogeneity of relations (Brunsdon et al., 1996). With the change in local

geographical position in space, the estimated parameters are also different. The calculation formula is as follows: $$\begin{array}{*{20}{c}} {y_i = \beta _0\left( {u_i,v_i} \right) +

\mathop {\sum }\limits_{k = 1}^p \beta _k\left( {u_i,v_i} \right)x_{ik} + \varepsilon _i} \end{array}$$ (5) In the formula, _y__i_ is the dependent variable of sample _i_. (_u__i_, _v__i_)

are the latitude and longitude coordinates of sample _i_. β0(_u__i_, _v__i_) is the regression constant of sample _i_. βk(_u__i_, _v__i_) is the coefficient of the _k_th independent variable

of sample _i_. _p_ is the number of independent variables. _x__ik_ is the _k_th independent variable of sample _i_. ε_i_ is the random error of the model. RESULTS TEMPORAL EVOLUTION OF

GLOBAL ONLINE SOCIAL RESPONSES Figure 4 shows the temporal evolution of the total number of global daily tweets related to Typhoon Haiyan from November 4 to 20. The number of tweets

increased significantly from November 4 to 8 and decreased from November 9 to 20, with a slight increase from November 10 to 11. The trend of global online social response followed the

development of Typhoon Haiyan. Typhoon Haiyan generated and developed rapidly in the South Pacific Ocean on November 4, 2013, and was soon upgraded to a tropical storm and quickly attracted

widespread interest from social media users around the world. On November 8, Typhoon Haiyan landed along the coast of Guiuan, the Philippines, with the highest wind speed, caused thousands

of casualties and heavy economic losses, and attracted the highest global interest. From November 9 to 20, the intensity of Typhoon Haiyan gradually weakened, it moved throughout the

Philippines, and global attention to it gradually decreased. There are two reasons for the increase in global online social responses from November 10 to 11. First, the weakening Typhoon

Haiyan attacked the Hainan Province of China during this period and caused minor casualties, which again attracted global attention. Second, as disaster relief work was in full swing after

the disaster, global social media users increased their attention and actively provided humanitarian relief (Zhang and Cheng, 2021). In short, the response of global social media users to

extreme disaster events will change with the intensity of disasters and the severity of damage caused in the process. Moreover, we carried out a clustering analysis using the Euclidean

distance and the connection between groups to obtain the time clustering tree diagram (Supplementary Fig. S1). Based on Supplementary Fig. S1, online social responses were divided into four

categories: November 4–7, November 8, November 9–13, and November 14–20. These four types are closely related to the evolutionary process of Typhoon Haiyan. November 4–7 is the development

stage, November 8 is the peak stage, November 9–13 is the extinction stage, and November 14–20 is the complete dissipation stage. Therefore, in the following analysis, the response of social

media users to disasters around the world was divided into two stages: before (November 4–8) and after (November 9–20) Typhoon Haiyan’s landing. SPATIAL PATTERN CHANGES IN GLOBAL ONLINE

SOCIAL RESPONSES The online social responses around the world to Typhoon Haiyan are roughly the same in the spatial distribution before and after the typhoon’s landing (Fig. 5). Users in the

Philippines, the main affected country, had the strongest responses to Typhoon Haiyan. Additionally, users in developed countries in Europe and North America paid more attention to Typhoon

Haiyan, mainly because citizens in developed countries had greater capabilities, resources, and capital for disaster relief than those in developing countries. After the Philippines, users

in the U.S. had the greatest interest in Typhoon Haiyan. The reason may be that the Philippines was once a colony of the U.S., and there were more common political and economic values, as

well as stronger social and cultural connections, between these two countries and their citizens. Due to geographical proximity, neighboring countries (e.g., Australia, Vietnam, and

Indonesia) paid relatively strong attention to Typhoon Haiyan. Africa and West Asia had the fewest responses to Typhoon Haiyan. By calculating the relative and absolute differences in tweets

before and after Typhoon Haiyan’s landing, we analyzed the spatial distribution of online social response differences around the world. As shown in Fig. 6, most countries had more responses

to disaster events after the typhoon made landfall. Among them, Bhutan, Botswana, Luxembourg, Mongolia, Uruguay, and Nigeria saw increases of 39 times, 15 times, 12.5 times, 12 times, 11.6

times, and 11.4 times, respectively, compared to the countries before Typhoon Haiyan’s landing. However, some small countries saw decreases in responses. For instance, in Andorra, Guyana,

Seychelles, North Korea, and Papua New Guinea, responses decreased by 50%, 50%, 50%, 43% and 10%, respectively, compared with those before the typhoon’s landing. The online social responses

in Liechtenstein, Chad, Antigua and Barbuda even dropped to zero. This may be because these countries have weak national strength and are unable to provide more attention and assistance to

the outside world. In addition, some countries restrict the use of Twitter, making it impossible to truly reflect the attention of domestic social media users to Typhoon Haiyan. FACTOR

DETECTION OF ONLINE SOCIAL RESPONSES TO THE DISASTER The results of single factors affecting online social responses before Typhoon Haiyan’s landing are listed in Table 2. Overall, economic

factors have the greatest influence on the response, while natural, demographic and cultural factors have a relatively weak influence. The social factors are the weakest, while the political

factors are not significant. The impact of economic factors may be due to the social and geographical disparities in the use of Twitter, and areas with higher disaster-related Twitter usage

are generally communities with better socioeconomic conditions (Zou et al., 2018). Specifically, six factors have significant impacts on online social responses. According to the _q_

values, their explanatory power is, in descending order, export value, the GDP loss rate, total population, per capita GDP, the proportion of nonreligious beliefs, and the CPI. Among them,

export value, the CPI, the proportion of nonreligious beliefs, the GDP loss rate, and population are higher than other factors and are all significant at the 0.05 level, indicating that

these influencing factors have major explanatory power in regard to the dependent variable. Table 2 shows that the major factors do not change much after landing. Economic factors still have

the greatest influence on the response in various countries, while natural, demographic, cultural, and social factors have a weak influence. All six factors are significant at the 0.01

level. In addition, political factors have the weakest but most significant influence. Based on the _q_ value, the explanatory power of these factors is ranked as follows: export value, per

capita GDP, the GDP loss rate, population, the proportion of nonreligious beliefs, the CPI, and the DI. Comparing the dominant factors before and after the typhoon’s landing, we find that

the difference lies in whether political factors are included. Before and during the disaster, social media users focused more on reporting the progress of the disaster, the extent of

damage, property losses and other disaster situation facts. Tweets mainly reflected users’ views, attitudes, and thoughts on reports of disaster events and the official emergency response.

Therefore, the political effect is relatively weak in the five target layers. In contrast, during the post-disaster period, users focused more on disaster recovery and international

humanitarian assistance. Governments in more democratic countries tended to be more inclined to help disaster-hit countries. Moreover, the actions taken by the national government attract

the attention and assistance of the local people. Therefore, compared with the stage before and during the disaster, the dominant factor in the post-disaster stage involved political

factors. ATTRIBUTING MULTIPLE FACTORS AFFECTING ONLINE SOCIAL RESPONSES BEFORE LANDFALL After the normalization of the dominant factor and the dependent variable, the GWR model was

regressed, and the results are summarized in Table 3. Only one factor was selected as the dominant factor of each target layer to achieve dimensionality reduction of the influencing factor

to obtain more accurate research conclusions and avoid the repeatability and inefficiency of the research (Garske et al., 2021). Nevertheless, both per capita GDP and export value could

represent the economic target layer. The export value was selected since it can also reflect a country’s connection to others. The _R_2 values of the two research stages (before and after

landing) are 0.94 and 0.90, respectively, indicating that the model fitting effect is good. As shown in Table 3, the coefficients of the factors are both positive and negative, indicating

that these five influencing factors fluctuate greatly in space and have an unstable influence on the interest of social media users. Overall, from Fig. 7, export values and population have a

more positive influence on online social responses around the world. CPI and the proportion of nonreligious beliefs generally have a negative influence on the online social response in many

countries. The GDP loss rate has a complicated influence on the online social response which has a positive influence in North America, North Africa, and Russia but a negative impact in

other regions. Regarding spatial distributions, the coefficients of export value are overwhelmingly positive around the world (Fig. 7a). This pattern indicates that the economic and trade

links between countries have positive impacts on the online social responses to natural disasters. In detail, the coefficients are relatively high in North and South American countries. The

coefficients of Western Europe, West Africa, East Africa, Oceania, and Southeast Asia follow. Countries in Eastern Europe and Southern Africa have lower coefficients. In addition, exports of

goods and services have only a significant negative effect in Iceland. This exception may be due to its few economic connections to the Philippines. Figure 7b indicates that the CPI has

both positive and negative effects on online social responses in different countries. The countries with higher influencing factor coefficients are mainly distributed in Western Europe,

among which Iceland, the U.K., France, and Switzerland have the highest influencing factor coefficients, i.e., 0.13, 0.06, 0.06, and 0.05, respectively. Countries with lower coefficients of

the CPI are mainly distributed in North America and northern South America. In addition, the CPI has positive effects in 82.3% of countries, which are mainly distributed in North Africa,

Oceania and Europe. The coefficients of the proportion of nonreligious beliefs have both positive and negative effects on the social response to disasters in different countries (Fig. 7c).

From the perspective of the spatial distribution, the countries with higher coefficients of the proportion of nonreligious beliefs are mainly distributed in southern South America and

Central Africa, among which Madagascar, Lesotho, Argentina, Malawi, and Costa Rica have the highest influencing factor coefficients, i.e., 0.07 for all these countries. Countries with lower

coefficients of the proportion of nonreligious beliefs are mainly distributed in northern North America and West Africa. In addition, the proportion of nonreligious beliefs showed a negative

influence in 34.5% of countries, mainly distributed in North America, western South America, Western Europe, and West Africa. Although the proportion of nonreligious beliefs in these

countries is not low, the religious belief in these regions is mainly Christianity. This may be due to the characteristics of Christianity’s religious belief, leading to the positive effect

of the proportion of unbelief in these regions. The GDP loss rate has a positive effect on the responses in most countries (Fig. 7d), indicating that countries prone to natural disasters and

vulnerable carriers are more sensitive to extreme disaster events, and social media users will have more interest in disasters. The countries with higher coefficients of the GDP loss rate

are mainly distributed in Eastern Europe, among which Ukraine, Russia, and Lithuania have the highest influencing factor coefficients, i.e., 0.33, 0.32, and 0.31, respectively. Countries

with lower coefficients of the GDP loss rate are mainly distributed in Western Europe, among which Switzerland, the Netherlands, France and the United Kingdom have the lowest influencing

factor coefficients, i.e., −0.17, −0.15, −0.15, and −0.12, respectively. In addition, the GDP loss rate of 11.5% of countries has negative effects; the countries are mainly distributed in

Western Europe, Oceania, and Southeast Asia. It may be that these countries are close to the disaster area and have a relatively high level of natural risk; thus, users in these countries

pay more attention to calming people’s emotions rather than the disaster itself. The population has a generally positive effect on online social responses to disasters in various countries

(Fig. 7e), indicating that countries with a large population base will pay more active responses to extreme disaster events. The countries with higher coefficients of population are mainly

distributed in Western Europe, among which Iceland, the U.K., France and Switzerland have the highest influencing factor coefficients, i.e., 3.27, 1.47, 1.47, and 1.23, respectively.

Countries with lower influencing factor coefficients are mainly distributed in North and South America, among which Chile, Peru, and Brazil have the lowest influencing factor coefficients,

i.e., −0.91, −0.86, and −0.84, respectively. ATTRIBUTING MULTIPLE FACTORS AFFECTING ONLINE SOCIAL RESPONSES AFTER LANDFALL The GWR coefficients of each influencing factor after landing are

statistically summarized in Table 4 and shown in Fig. 8. The coefficient statistics of export value, the CPI, the proportion of nonreligious beliefs, the GDP loss rate, and population are

similar to the situation before landing. The coefficient value of the GE of the increased influencing factor is positive and negative, indicating that the influencing factor also fluctuates

in space and its effect is changeable. Figure 8a shows that the influencing factor coefficient of the GE has significant spatial heterogeneity, with values ranging from approximately −0.16

to 0.10. These results indicate that most of GE has positive and negative effects on the attention of social media users to disasters. From the perspective of spatial distribution, the

countries with high influencing factor coefficients are Australia (0.10) and European countries, such as Norway (0.98), Denmark (0.96), Sweden (0.94), and Germany (0.92). Countries with

lower influencing factor coefficients are mainly distributed in American countries, among which Mexico, Ecuador, Guatemala, El Salvador and Costa Rica have the lowest five influencing factor

coefficients, i.e., −0.16, −0.15, −0.14, −0.14, and −0.14, respectively. In addition, the negative effects of GE are mainly distributed in the Western Hemisphere and Western Africa. The

export value is positively correlated with the attention of social media users to the disaster and is generally consistent with the distribution pattern of the influencing factor coefficient

of export value before the typhoon made landfall (Fig. 8b). However, the influencing factor coefficient changed. Chile, Peru, Argentina, Uruguay, and Brazil had the five highest influencing

factor coefficients, i.e., 1.24, 1.23, 1.22, 1.21 and 1.19, respectively. Iceland, Indonesia, Malaysia, Ireland, and Cambodia had the lowest five coefficients, i.e., −0.15, 0.02, 0.04, 0.05

and 0.06, respectively. The CPI is positively and negatively correlated with the attention of social media users to the disaster (Fig. 8c) and changes with the distribution pattern of the

influencing factor coefficient of the CPI before the typhoon’s landing. Canada, U.S., Mexico, Guatemala, and El Salvador had the highest influencing factor coefficients, i.e., 0.19, 0.16,

0.14, 0.13, and 0.13, respectively. Sweden, Finland, Estonia, Latvia and Norway had the lowest influencing factor coefficients, i.e., −0.07, −0.06, −0.06,−0.06, and −0.06, respectively. The

proportion of nonreligious beliefs has both positive and negative correlations to the attention of social media users to the disaster (Fig. 8d) and is basically consistent with the pattern

of the influencing factor coefficient of the proportion of nonreligious beliefs before the typhoon’s landing. However, the influencing factor coefficient changed. Iceland, Brazil, Uruguay,

Kenya, and Tanzania had the highest influencing factor coefficients, i.e., 0.025, 0.010, 0.008, 0.008, and 0.008, respectively. Cuba, Jamaica, Mexico, Guatemala, and Honduras had the lowest

influencing factor coefficients, i.e., −0.15, −0.14, −0.14, −0.13 and −0.13, respectively. The GDP loss rate is positively correlated with the attention of social media users to the disaster

(Fig. 8e) and is different from the distribution pattern of the influencing factor coefficient of the GDP loss rate before the typhoon’s landing. The influencing factor coefficient changed.

Iceland, Ireland, Canada, the U.S., and Cuba had the highest influencing factor coefficients, i.e., 1.53, 0.47, 0.35, 0.29, and 0.15, respectively. Poland, Denmark, Germany, Moldova, and

Sweden had the lowest influencing factor coefficients, i.e., −0.12, −0.10, −0.09, −0.08, and −0.07, respectively. Population is positively correlated with the attention of social media users

to the disaster (Fig. 8f) and is basically consistent with the distribution pattern of the influencing factor coefficient of population before the typhoon’s landing. However, the

influencing factor coefficient changed. Iceland, Canada, Ireland, Portugal, and the U.S. had the highest impact factor coefficients, i.e., 2.35, 1.95, 1.52, 1.11, and 1.09, respectively.

Chile, Peru, Argentina, Uruguay, and Brazil had the lowest impact factor coefficients, i.e., −0.99, −0.94, −0.94, −0.93, and −0.84, respectively. CHANGES IN INFLUENCING FACTORS BEFORE AND

AFTER THE TYPHOON’S LANDING We further compared and analyzed the changes in the five influencing factors before and after the typhoon made landfall. Figure 9 shows the influencing factor

coefficient differences (left panel) and their change directions (right panel) before and after the typhoon’s landing in corresponding countries or regions. The letters P and N indicate

positive or negative coefficients, respectively. It is found that the influencing factor coefficients of the export value, the CPI, the proportion of nonreligious beliefs, the GDP loss rate

and population increased and decreased significantly. By comparing the changes in the effect direction of the influencing factors before and after the typhoon, this study finds that the

effect direction of export value, the GDP loss rate and population basically did not change before and after the typhoon landing, and the effect direction of the CPI and proportion of

nonreligious beliefs changed in some countries. The countries with enhanced effects of export value before and after the typhoon’s landing are mainly distributed in Europe, South America,

and East and Southern Africa (Fig. 9a, b). Among these countries, Argentina, Uruguay, Panama, and Costa Rica have significantly enhanced effects, with difference values of export value being

1.14, 1.14, 1.03, and 1.02, respectively. Countries with a weakened effect of export value were mainly distributed in northern South Asia, North America, Central America, Southeast Asia,

and Oceania, among which the United States, Cuba, Guatemala, and El Salvador had significantly weakened effects, and the differences in the change in export value were −0.47, −0.16, −0.11,

and −0.11, respectively. Figure 9b shows that the effect directions of export value before and after the typhoon’s landing both have a negative effect on social media users’ attention to the

disaster in Iceland, and both have positive effects on social media users in other countries. The countries with enhanced effects of the CPI before and after the typhoon’s landing are

mainly distributed in North America and South America (Fig. 9c, d). The United States, Guatemala, El Salvador, and Cuba have significantly enhanced effects, with difference values of the CPI

being 0.21, 0.19, 0.19, and 0.18, respectively. Countries with weakened effects of the CPI are mainly distributed in the Eastern Hemisphere, among which Norway, Sweden, the Netherlands, and

Denmark have significantly weakened effects, and the differences in the change in the CPI are −0.10, −0.10, −0.09, and −0.09, respectively. The countries where the effect direction of the

CPI changed from positive to negative are mainly distributed in Southern Africa before and after the typhoon’s landing (Fig. 9d). The countries where the effect direction of the CPI changed

from negative to positive were mainly distributed in southern North America and northern South America. The countries for which the CPI maintained the same direction are distributed in

Europe, Southeast Asia, Oceania, and Africa. Figure 9e, f shows that the countries with enhanced effects of the proportion of nonreligious beliefs before and after the typhoon’s landing are

mainly distributed in North America, South America, and Africa. The U.S., Iceland, Senegal and Omen have significantly enhanced effects, and the difference values of the proportion of

nonreligious beliefs are 0.02, 0.01, 0.01, and 0.003, respectively. Countries with a weakened effect of the proportion of nonreligious beliefs are mainly distributed around the world, among

which Jamaica, Mexico, Costa Rica, and Panama have a significantly weakened effect, and the difference in the change in the proportion of nonreligious beliefs are −0.20, −0.20, −0.19, and

−0.18, respectively. Figure 9f shows that the countries where the effect direction of the proportion of nonreligious beliefs changed from positive to negative are mainly distributed in

Oceania and Southeast Asia before and after the typhoon’s landing. The countries in which the proportion of nonreligious beliefs maintained the same direction are distributed in North and

South America, Europe, South Asia and Africa. Figure 9g, h shows that the countries with enhanced effects of the GDP loss rate before and after the typhoon’s landing are mainly distributed

in North and Central America, West Europe, and Northern and Southern Africa. Iceland, Ireland, Canada, and Switzerland have significantly enhanced effects, with difference values of the GDP

loss rate being 1.32, 0.31, 0.26, and 0.19, respectively. The GDP loss rate has a weak influence on the countries distributed in Southeast Asia, Europe, Northwestern Africa, and Oceania,

such as Lithuania, Ukraine, Austria and Slovenia, with changes of −0.37, −0.37, −0.34, and −0.32, respectively. Meanwhile, the effect direction of the GDP loss rate does not change in most

countries before and after the typhoon’s landing (Fig. 9h), whereas countries in Northern Europe and Eastern Europe changed from positive to negative. Figure 9i, j shows that the countries

with enhanced effects of population before and after the typhoons’ landing are mainly distributed in North and South America, South Asia, and Central Africa. Canada, the United States,

Ireland, and Cuba have significantly enhanced effects, with change values of 1.72, 1.54, 1.05, and 0.50, respectively. Countries with weakened effects of population are mainly distributed in

Southeast Asia, Oceania, Europe, and Southern and East Africa. Argentina, Uruguay, Iceland, and France have significantly weakened effects with changes of −1.18, −1.17, −0.91, and −0.83,

respectively. The effect direction of population remains consistent in most countries before and after the typhoon’s landing (Fig. 9j). However, several countries in Central America,

Argentina, and New Zealand changed from positive to negative. DISCUSSION WESTERN COUNTRIES ARE THE MAIN FOREIGN GROUP OF ONLINE SOCIAL RESPONSES TO THE NATURAL DISASTER Western countries are

the major foreign group with active online social responses to natural disasters. Similar to recent research on the global public interest in earthquakes, which found that Western countries

are dominant (Kam et al., 2021), we found that Western countries have stronger online social responses to natural disasters. This result is explained by the strong economic conditions and

high trade connections between Western countries and the Philippines. Moreover, our research reveals that countries neighboring the Philippines are a major group. This difference is rooted

in the geophysical divergence of typhoons and earthquakes since neighboring countries are more likely to be impacted by typhoons than earthquakes. DISTANCE HAS LITTLE INFLUENCE ON THE ONLINE

SOCIAL RESPONSES Geographical distance has a limited influence on online social responses. Previous research found that the distance between the country where news media are located and the

country where the disasters occur impacts the likelihood that a disaster will be covered by the media (Berlemann and Thomas, 2019). Whereas we found no global significant impact of

geographical distance between countries and the country where the disaster occurred. On one hand, the neighboring countries have more responses than African countries, but they are

comparable to Western countries. On the other hand, the timing of online social response is consistent with the evolution of Typhoon Haiyan. This nonsignificant spatial disparity and

temporal consistency reflect that online social networks have almost eliminated the spatiotemporal barriers of social responses to a disaster. ECONOMY IS THE MAIN DRIVER OF ONLINE SOCIAL

RESPONSES Overall, economic factors have more significant influence on the drivers of online social responses than other factors except the population. As shown by the Geodetector results,

economic factors have the strongest explanatory power in regard to the number of tweets posted before the typhoon’s landing, while natural, demographic and cultural factors were weaker, and

social factors were the weakest. Political factors were not significant. Economic factors had the strongest explanatory power in regard to the number of tweets posted after the typhoon’s

landing, while natural, demographic, cultural and social factors were weaker, and political factors were the weakest. Based on the GWR model before landing, export value is basically

positively correlated with the intensity of online social responses, and the factor coefficients are generally high in the west and low in the east. The coefficient of the CPI has

significant spatial heterogeneity, with positive effects in North Africa, Oceania and Europe and negative effects in North and South America, Southern Africa and Southeast Asia. The

proportion of nonreligious beliefs has a negative effect in North America, western South America, Western Europe, West Africa and Oceania and has a positive effect in eastern South America,

Eastern Europe, Asia and Southern Africa before landing. The GDP loss rate has a positive effect on social media users’ attention to disasters, and 18.58% of countries show a negative effect

before landing but 31.86% after landing, Population has a generally positive effect on disaster attention among social media users in countries, with only 18.58% (before landing) and 17.70%

(after landing) of countries having a negative effect, mainly in North and South America. Comparing the coefficient changes in influencing factors, it is found that the influencing factor

coefficients of export value, the CPI, the proportion of nonreligious beliefs and the GDP loss rate before and after Typhoon Haiyan’s landing have spatial heterogeneity. By comparing the

direction of the influencing factors, we find that the direction of the influencing factor of export value did not change before and after the typhoon’s landing. There are four types of

changes in the effect direction of the CPI: from positive to negative, from negative to positive, and no change. There are three types of change in the effect direction of the proportion of

nonreligious beliefs, that is, from positive to negative, from negative to positive, and no change. The effect direction of the GDP loss rate and population remains unchanged in most

countries. IMPLICATIONS AND LIMITATION Based on the above results and analysis, this paper reveals that the online social response to a natural disaster is beyond geographical limitation and

has a global impact. This online social response will be mainly influenced by economic factors and active in Western countries. It indicates that western active online social media users

have the most willingness and economic capacity for foreign aid. Hence, this information is of great potential for governments and humanitarian agencies to expand their pool of potential

volunteers and donors. There are limitations to this study. Due to the lack of relevant data from sources, not all the countries are obtained in our analysis. Hence it will lead to the

uncertainty of GWR’s results, whereas due to our limited knowledge, the selection of influencing factors may still involve omitted variables, and only the influence of the target layer and

the first-level factors are discussed. Besides, the religious rate data are collected from Wikipedia, which may lead to the uncertainty of our analysis. Whereas the GE is controversial

regarding its definition and algorithm, which will also introduce the analysis’ uncertainty. In addition, we noticed that Hainan Province in China, also severely affected by the typhoon, has

caused a much weaker global online social response compared to the Philippines. This is also a question worth further exploration in the future. CONCLUSION This study takes Typhoon Haiyan

in 2013 as a case study to explore the global online social responses to a natural disaster and to investigate their influencing factors. In conclusion, the global online social responses to

Typhoon Haiyan are beyond the geographical limit and consistent with the development of disaster. Except for the Philippines, Western countries and surrounding countries had higher

responses than others. Economic factors dominated the online social response. The influences of social, cultural, and demographic factors were relatively weak. Political factors had only a

weak impact during the post-disaster stage. Our findings demonstrate active online social groups in Western countries are of great potential for governments and humanitarian agencies to call

for foreign volunteers and donors. This information will deepen our understanding of online social behavior and help coordinate the humanitarian assistance of various countries and provide

a reference for the efficient and orderly implementation of international rescue work after disasters. DATA AVAILABILITY The global dataset tweets on Typhoon Haiyan and other dataset used in

this study is available at: https://doi.org/10.6084/m9.figshare.23617725. This dataset was proposed in xlsx and csv format and can be processed using Microsoft Excel software and GIS

software. REFERENCES * Aitsi-Selmi A, Blanchard K, Murray V (2016) Ensuring science is useful, usable and used in global disaster risk reduction and sustainable development: a view through

the Sendai framework lens. Palgrave Commun 2(1):1–9 Article  Google Scholar  * Berlemann M, Thomas T (2019) The distance bias in natural disaster reporting—empirical evidence for the United

States. Appl Econ Lett 26(12):1026–1032 Article  Google Scholar  * Brandt HM, Turner-McGrievy G, Friedman DB, Gentile D, Schrock C, Thomas T, West D (2019) Examining the role of Twitter in

response and recovery during and after historic flooding in South Carolina. J Public Health Manag Pract 25(5):E6 Article  PubMed  Google Scholar  * Brunsdon C, Fotheringham AS, Charlton ME

(1996) Geographically weighted regression: a method for exploring spatial nonstationarity. Geogr Anal 28(4):281–298 Article  Google Scholar  * Chen S, Mao J, Li G, Ma C, Cao Y (2020)

Uncovering sentiment and retweet patterns of disaster-related tweets from a spatiotemporal perspective: a case study of Hurricane Harvey. Telemat Inform 47:101326 Article  Google Scholar  *

Contreras D, Wilkinson S, Balan N, James P (2022) Assessing post-disaster recovery using sentiment analysis: the case of L’Aquila, Italy. Earthq Spectra 38(1):81–108 Article  Google Scholar

  * Cook ADB, Shrestha M, Zin Bo Htet (2018) An assessment of international emergency disaster response to the 2015 Nepal earthquakes. Int J Disaster Risk Reduct 31:535–547 Article  Google

Scholar  * Coppola DP (2020) Introduction to international disaster management, 4th edn. Butterworth-Heinemann. Oxford, England. https://doi.org/10.1016/B978-0-12-817368-8.00008-7 * David

CC, Ong JC, Legara EFT (2016) Tweeting supertyphoon Haiyan: evolving functions of Twitter during and after a disaster event. PLoS ONE 11(3):e0150190 Article  PubMed  PubMed Central  Google

Scholar  * García-Ramírez GM, Bogen KW, Rodríguez-Guzmán VM, Nugent N, Orchowski LM (2021) #4645Boricuas: Twitter reactions to the estimates of deaths by Hurricane María in Puerto Rico. J

Community Psychol 49(3):768–790 Article  PubMed  Google Scholar  * Garske SI, Elayan S, Sykora M, Edry T, Grabenhenrich LB, Galea S, Lowe SR, Gruebner O (2021) Space-time dependence of

emotions on Twitter after a natural disaster. Int J Environ Res Public Health 18(10):5292 Article  PubMed  PubMed Central  Google Scholar  * Gruebner O, Lowe SR, Sykora M, Shankardass K,

Subramanian SV, Galea S (2018) Spatio-temporal distribution of negative emotions in New York City after a natural disaster as seen in social media. Int J Environ Res Public Health

15(10):2275 Article  PubMed  PubMed Central  Google Scholar  * Hoffman S, Oliver-Smith A (eds) (2002) Catastrophe & culture: the anthropology of disaster. School of American Research

Press, Santa Fe * Kam J, Park J, Shao W, Song J, Kim J, Gizzi FT, Porrini D, Suh Y-J (2021) Data-driven modeling reveals the Western dominance of global public interest in earthquakes.

Humanit Soc Sci Commun 8(1):1–9 Article  Google Scholar  * Kumar P (2020) Twitter, disasters and cultural heritage: a case study of the 2015 Nepal earthquake. J Contingencies Crisis Manag

28(4):453–465 Article  Google Scholar  * Lefever DW (1926) Measuring geographic concentration by means of the standard deviational ellipse. Am J Soc 32(1):88–94 Article  Google Scholar  *

Moran PAP (1950) Notes on continuous stochastic phenomena. Biometrika 37(1–2):17–23 Article  MathSciNet  CAS  PubMed  MATH  Google Scholar  * Murzintcev N, Cheng C (2017) Disaster hashtags

in social media. ISPRS Int J Geo-Inf 6(7):204 * NDRRMC Philippines (2013) Final report re: effects of Typhoon Yolanda (Haiyan), p 65.

http://ndrrmc.gov.ph/attachments/article/1329/FINAL_REPORT_re_Effects_of_Typhoon_YOLANDA_ (HAIYAN)_06-09NOV2013.pdf * Raschky PA, Schwindt M (2012) On the channel and type of aid: the case

of international disaster assistance. Eur J Polit Econ 28(1):119–131 Article  Google Scholar  * Ruan T, Kong Q, McBride SK, Sethjiwala A, Lv Q (2022) Cross-platform analysis of public

responses to the 2019 Ridgecrest earthquake sequence on Twitter and Reddit. Sci Rep 12(1):1634 Article  ADS  CAS  PubMed  PubMed Central  Google Scholar  * Shen S, Cheng C, Song C, Yang J,

Yang S, Su K, Yuan L, Chen X (2018) Spatial distribution patterns of global natural disasters based on biclustering. Nat Hazards 92(3):1809–1820 Article  Google Scholar  * Shen S, Huang J,

Cheng C, Zhang T, Murzintcev N, Gao P (2021) Spatiotemporal evolution of the online social network after a natural disaster. ISPRS Int J Geo-Inf 10(11):744 Article  Google Scholar  * Sadaf

R, Oláh J, Popp J, Máté D (2018) An investigation of the influence of the worldwide governance and competitiveness on accounting fraud cases: a cross-country perspective. Sustainability

10(3):588 * Takahashi B, Tandoc EC, Carmichael C (2015) Communicating on Twitter during a disaster: an analysis of tweets during Typhoon Haiyan in the Philippines. Comput Hum Behav

50:392–398 Article  Google Scholar  * Tan YJ, Maharjan R (2018) What googling trends tell us about public interest in earthquakes. Seismol Res Lett 89(2A):653–657 Article  Google Scholar  *

Tang J, Yang S, Wang W (2021) Social media-based disaster research: development, trends, and obstacles. Int J Disaster Risk Reduct 55:102095 Article  Google Scholar  * United Nations Office

For Disaster Risk Reduction (2020) Human cost of disasters: an overview of the last 20 years 2000–2019, UNDRR, pp 1–17 * Wang J, Li X, Christakos G, Liao Y, Zhang T, Gu X, Zheng X (2010)

Geographical detectors‐based health risk assessment and its application in the neural tube defects study of the Heshun Region, China. Int J Geogr Inf Sci 24(1):107–127 Article  CAS  Google

Scholar  * Yan Y, Chen J, Wang Z (2020) Mining public sentiments and perspectives from geotagged social media data for appraising the post-earthquake recovery of tourism destinations. Appl

Geogr 123:102306 Article  Google Scholar  * Yuan F, Li M, Zhai W, Qi B, Liu R (2020) Social media based demographics analysis for understanding disaster response disparity. In: Proceedings

of Construction Research Congress 2020: Computer Applications, pp 1020–1028. https://doi.org/10.1061/9780784482865.108 * Zhang T, Cheng C (2021) Temporal and spatial evolution and

influencing factors of public sentiment in natural disasters—a case study of Typhoon Haiyan. ISPRS Int J Geo-Inf 10(5):299 Article  Google Scholar  * Zhang T, Shen S, Cheng C, Su K, Zhang X

(2021) A topic model based framework for identifying the distribution of demand for relief supplies using social media data. Int J Geogr Inf Sci 35:2216–2237 * Zhu H, Liu K (2021) Temporal,

spatial, and socioeconomic dynamics in social media thematic emphases during Typhoon Mangkhut. Sustainability 13(13):7435 Article  Google Scholar  * Zou L, Lam NSN, Cai H, Qiang Y (2018)

Mining Twitter data for improved understanding of disaster resilience. Ann Am Assoc Geogr 108(5):1422–1441 Google Scholar  Download references ACKNOWLEDGEMENTS This work is supported by the

National Key Research and Development Plan of China (grant numbers 2019YFA0606901) and the National Natural Science Foundation of China (No. 42201498). AUTHOR INFORMATION AUTHORS AND

AFFILIATIONS * Key Laboratory of Environmental Change and Natural Disaster, Beijing Normal University, 100875, Beijing, China Shi Shen, Junwang Huang & Changxiu Cheng * Faculty of

Geographical Science, Beijing Normal University, 100875, Beijing, China Shi Shen, Ke Shi, Junwang Huang, Changxiu Cheng & Min Zhao * Institute of Geographic Sciences and Natural

Resources Research, Chinese Academy of Sciences, 100101, Beijing, China Ke Shi Authors * Shi Shen View author publications You can also search for this author inPubMed Google Scholar * Ke

Shi View author publications You can also search for this author inPubMed Google Scholar * Junwang Huang View author publications You can also search for this author inPubMed Google Scholar

* Changxiu Cheng View author publications You can also search for this author inPubMed Google Scholar * Min Zhao View author publications You can also search for this author inPubMed Google

Scholar CONTRIBUTIONS SS designed the study; SS and CC collected the data; SS, KS, JH, and MZ discussed the method and analysis sections; SS and KS implemented the study and drafted the

manuscript; SS, KS, CC, and MZ reviewed and revised the manuscript. CORRESPONDING AUTHORS Correspondence to Shi Shen or Ke Shi. ETHICS DECLARATIONS COMPETING INTERESTS The authors declare no

competing interests. ETHICAL APPROVAL All the data used in this article are public data collected from the Twitter platform. There are no human participants in this study. INFORMED CONSENT

Only tweet numbers and users’ public registering location information are used and analyzed, and no private information or data will be published or can be seen in this article. ADDITIONAL

INFORMATION PUBLISHER’S NOTE Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. SUPPLEMENTARY INFORMATION FIGURE S1 RIGHTS

AND PERMISSIONS OPEN ACCESS This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in

any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The

images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not

included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly

from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Reprints and permissions ABOUT THIS ARTICLE CITE THIS ARTICLE Shen, S., Shi, K.,

Huang, J. _et al._ Global online social response to a natural disaster and its influencing factors: a case study of Typhoon Haiyan. _Humanit Soc Sci Commun_ 10, 426 (2023).

https://doi.org/10.1057/s41599-023-01922-5 Download citation * Received: 01 September 2022 * Accepted: 06 July 2023 * Published: 20 July 2023 * DOI:

https://doi.org/10.1057/s41599-023-01922-5 SHARE THIS ARTICLE Anyone you share the following link with will be able to read this content: Get shareable link Sorry, a shareable link is not

currently available for this article. Copy to clipboard Provided by the Springer Nature SharedIt content-sharing initiative