Sie können Operatoren mit Ihrer Suchanfrage kombinieren, um diese noch präziser einzugrenzen. Klicken Sie auf den Suchoperator, um eine Erklärung seiner Funktionsweise anzuzeigen.
Findet Dokumente, in denen beide Begriffe in beliebiger Reihenfolge innerhalb von maximal n Worten zueinander stehen. Empfehlung: Wählen Sie zwischen 15 und 30 als maximale Wortanzahl (z.B. NEAR(hybrid, antrieb, 20)).
Findet Dokumente, in denen der Begriff in Wortvarianten vorkommt, wobei diese VOR, HINTER oder VOR und HINTER dem Suchbegriff anschließen können (z.B., leichtbau*, *leichtbau, *leichtbau*).
Diese Studie untersucht die entscheidende Rolle digitaler Geräte bei der Beteiligung der Bewohner an der städtischen Hochwasserbekämpfung, insbesondere im Kontext der jüngsten Flutkatastrophen in China. Durch die Anpassung des Community Capital Framework (CCF) und den Einsatz eines Gradient Boosting Decision Tree (GBDT) Modells werden die nichtlinearen und interaktiven Auswirkungen der Informations- und Kommunikationstechnologie (IKT) auf das Engagement der Gemeinschaft aufgedeckt. Wichtige Ergebnisse zeigen, dass digitale Geräte die Bereitschaft der Einwohner, sich an Maßnahmen zur Hochwasserbekämpfung zu beteiligen, erheblich beeinflussen, wobei Faktoren wie die Nützlichkeit und Kompetenz bei der Verwendung dieser Geräte wichtige Prädiktoren sind. Die Studie hebt außerdem zwei unterschiedliche Interaktionsmuster hervor: Verstärkung, bei der digitale Geräte die Auswirkungen anderer unterstützender Faktoren verstärken, und Kompensation, bei der digitale Werkzeuge die geringen Auswirkungen schwächerer Faktoren abmildern. Darüber hinaus wird die relative Bedeutung verschiedener Hauptstädte ermittelt, wobei nachbarschaftliche Beziehungen, IKT-Unterstützung und individuelle Fähigkeiten die einflussreichsten sind. Die Ergebnisse unterstreichen die Bedeutung der Integration digitaler Werkzeuge in Strategien zum Hochwassermanagement, um die Katastrophenvorsorge, die Reaktionseffizienz und die Ergebnisse der Wiederherstellung zu verbessern.
KI-Generiert
Diese Zusammenfassung des Fachinhalts wurde mit Hilfe von KI generiert.
Abstract
In recent years, digital devices became crucial for residents to participate in flood emergency response, as it enables rapid information flow, resource mobilization, and coordination, thus improving response efficiency. Yet the influence of digital devices on resident participation in flood disaster response has not been well explored. This study adapted the community capital framework (CCF) to explore how digital devices impact residents’ participation willingness during urban flood response by using the gradient boosting decision tree (GBDT) model. Data were collected from 1,351 respondents in Zhengzhou City, China, in 2023. The study revealed that: (1) Digital devices significantly influence residents’ participation willingness in flood response, especially in terms of information acquisition and sharing. (2) Indicators of digital devices, such as their usefulness and residents’ proficiency in using them, are key predictors of participation willingness, and their relationships are primarily nonlinear. (3) Digital devices demonstrate both reinforcement and compensation patterns in the interaction effects on residents’ participation willingness. Interaction mechanisms underscore the pivotal role of digital technologies in participatory flood emergency response. These findings can help stakeholders better assess the benefits of enhancing digital device support in flood emergency responses.
1 Introduction
Climate change has emerged as a paramount challenge, with extreme weather events such as floods, heatwaves, and wildfires intensifying and posing significant risks to human life (Zhou et al. 2024). Urban flooding, in particular, has become a severe concern due to its impact on socioeconomic and natural systems. China has experienced numerous flood disasters due to heavy precipitation, including the 2016 Yangtze River floods, the 2020 southern region rains, the 2021 Zhengzhou rainfall, and the 2024 Fujian downpours. These events underscore the importance of the United Nations’ Sustainable Development Goal 11, which calls for global action on climate change and disaster risk reduction (Bexell and Jönsson 2017; UNEP 2017).
Recent studies highlight that participatory governance can play a crucial role in flood disaster response, since it provides local insights that enhance understanding of community needs and expectations (Li et al. 2024). Resident participation facilitates conflict resolution and trust-building among residents, enhances the efficiency of community reconstruction, and complements the formal disaster management processes at the community level (UNDRR 2015; Jiang et al. 2024). In practice, resident participation is influenced by various factors, including individual abilities, social support, government direction, and socioeconomic characteristics. Gammoh et al. (2023) found self-efficacy is the most important factor that influence flood preparedness behavior in Jordan, and effective risk communication, training, and trust in the government can help improve residents’ self-efficacy and overall flood preparedness. Cui and Li (2020) identified six critical forms of social capital within urban communities and revealed the essential role of social capitals such as collective efficacy in disaster response. Government support for resident participation in flooding is primarily evident in its comprehensive management of flood emergency coordination. However, these studies provide only a partial explanation of resident participation in flood responses, necessitating a more comprehensive framework for interpretation.
Anzeige
While traditional frameworks like the community capital framework (CCF) offer a valuable lens for analyzing the factors that influence residents’ flood response, their application in the digital age reveals significant limitations. For instance, a recent study by Jiang et al. (2024) insightfully adapted the CCF to consider crucial elements like government support, social ties, and socioeconomic status. However, this analysis, while comprehensive in traditional terms, largely overlooks what may now be the most dynamic component of community response: digital connectivity. In an era where smartphones and social media are ubiquitous, omitting the role of digital devices is no longer a minor oversight but a critical gap that potentially misrepresents the modern disaster response landscape. These devices and the networks they enable function as a form of “digital capital,” a resource that intersects with and fundamentally alters the traditional forms of capital identified by the CCF.
The existing literature confirms that digital devices are pivotal in modern emergencies. They function as conduits for real-time information, platforms for communication, and tools for coordination, which can reduce panic and enhance residents’ sense of self-efficacy (Leong et al. 2015; Yang et al. 2024). However, this optimistic view often presents a simplistic and linear relationship, failing to capture the complex, multifaceted nature of digital technology in a crisis. The influence of digital devices is not universally positive; they can just as easily become vectors for misinformation and information overload, creating confusion and anxiety that may paralyze resident action rather than facilitate it (Wehn and Evers 2015).
This complexity points to a more profound research gap: the literature remains sparse regarding the mechanisms through which digital engagement influences resident participation. It is unclear how digital capital interacts with other forms of community capital, such as the trust inherent in neighborly relationships or the authority of government directives. Does digital information amplify, substitute for, or even conflict with these other inputs? Furthermore, the effects are unlikely to be linear; the impact of receiving one official alert is vastly different from being inundated with thousands of social media posts. Understanding these nonlinear dynamics and interactive mechanisms—how individuals, communities, and governments leverage, filter, and react to digitally mediated information in a concerted or fragmented effort—is a critical and underexplored frontier in disaster research.
In this context, our study aimed to examine the relationship between digital devices and residents’ participation willingness in flood response. To guide the research, the following three questions were proposed.
Anzeige
(1) How important are digital devices in influencing residents’ participation willingness in flood response, compared to other factors?
(2) Are there any prevailing patterns of nonlinear relationships between digital devices and residents’ participation willingness in flood response?
(3) How do digital devices interact with individual, social, and governmental factors in shaping residents’ participation willingness during flood response?
To address these research questions, this study applied the CCF to examine how digital devices influence residents’ willingness to participate in flood response. Drawing on survey data from 1,351 respondents in Zhengzhou City, China, we employed a gradient boosting decision tree (GBDT) model to capture the complex, nonlinear relationships and interactions involved. This approach allows us to move beyond the question of whether information and communication technology (ICT) matters, and instead explore how digital devices—together with individual, neighborly, and governmental factors—shape participation willingness.
2 Literature Review
The community capital framework (CCF) classifies community resources into seven capitals, offering a comprehensive tool for assessing community well-being and developmental potential. This study further developed the CCF model to examine how digital devices shape residents’ participation willingness, aiming to enhance community flood resilience.
2.1 Community Capital Framework
The CCF is a comprehensive analytical tool for communities, classifying community resources and capabilities into various forms of capital to assess overall well-being and developmental potential. The essence of the CCF is to identify and quantify the critical elements that foster the sustainable development of communities (Flora et al. 2005) thereby achieving sustainability in the community’s economic, social, and environmental aspects. This framework offers a methodology for understanding community development and has become a vital theoretical foundation for community research and practice.
The CCF encompasses seven types of capital: natural, cultural, human, social, political, financial, and built capitals (Flora et al. 2005). Natural capital refers to the natural resources and environmental assets. Cultural capital involves the shared values, traditions, languages, and cultural practices that contribute to social cohesion and identity. Human capital represents the skills, knowledge, education, and health of individuals. Social capital refers to the networks, relationships, and trust between individuals and groups within the community. Political capital involves the ability of community members and leaders to influence decision-making processes, access resources, and advocate for the community’s needs at various levels of governance. Financial capital relates to the financial resources. Built capital includes the physical infrastructure and facilities (such as housing, transportation, and utilities) that enable communities to function and grow efficiently. These capitals are interdependent and collectively influence the operation and development of communities (Paul et al. 2020).
According to Callaghan and Colton (2008), the CCF not only provides a theoretical framework for the multidimensional analysis of community capital but also guides the development and assessment of community development strategies in practice. It helps communities identify their strengths and areas for improvement, facilitates the efficient allocation and utilization of resources, and strengthens community cohesion and resilience to external impacts.
2.2 Extended Community Capital Framework (CCF) in Participatory Flood Responses
Disaster emergency management involves preparation before emergencies, response during emergencies, and recovery after emergencies (Zhu and Li 2021). Flood emergency response, a pivotal aspect of flood management, necessitates timely action to prevent incalculable losses (Guo et al. 2023). Recent studies showed that the effectiveness of flood emergency response is closely correlated with the level of resident participation (UNDRR 2015; Guo et al. 2024). In the context of community emergency response, resident participation involves any process that enables community residents to identify, decide, and address issues. Within the domain of flood emergency management, the application of the CCF facilitates a deeper comprehension of community residents’ capabilities and levels of participation in responding to disasters such as floods.
Human capital plays a crucial role in promoting participatory flood response efforts. This form of capital includes not only leadership skills during the flood response phase but also the expertise, knowledge, and socioeconomic factors of residents, such as age (Flora et al. 2005; Paul et al. 2020). Psychology highlights self-efficacy as a key motivator for behavioral change (Bandura 1978). Individuals with enhanced capabilities often have higher self-efficacy, believing in their ability to contribute effectively to flood management. This increased belief in their potential leads to greater participation in emergency response activities (Jiang et al. 2024).
Social capital acts as a critical enabler of participatory flood response. It facilitates trust, information sharing, and mutual support among neighbors, which are essential for community collaboration during emergencies (Bobbio 2019). This enhanced community cohesion boosts residents’ participation willingness in collective flood management. Research by Lo et al. (2015) further confirmed that increased trust levels between residents and stakeholders within communities significantly enhance participation in flood-related activities.
Political capital is pivotal in shaping participatory flood response through government facilitation of community organization and resource mobilization (Flora et al. 2005). Effective government support for resident engagement in disaster risk management enhances participation. This is achieved through robust flood monitoring, efficient control projects, and strong coordination in relief efforts. The government’s proactive promotion of disaster prevention strategies not only boosts public awareness but also builds trust with citizens. Furthermore, policy implementation stimulates resident involvement, increasing community engagement and resilience in flood management (Wang et al. 2022).
Studies showed that financial capital, due to its substitutability, plays a pivotal role in laying the foundation for the transformation or enhancement of other types of capital during floods (Lax and Krug 2013). Amidst dynamic environmental pressures, financial capital’s ease of conversion into other assets, sustains other forms of capital, significantly influencing households’ relative capacity to adapt to flood disasters (Azad and Pritchard 2022). Moreover, substantial financial capital may indicate that certain groups possess insurance, emergency funds, and superior flood-resistant infrastructure, endowing them with a relative advantage and the capability to participate (Jiang et al. 2024).
Built capital plays an indispensable role in emergency events like floods. Telecommunication infrastructure is particularly crucial for emergency planning in affected areas and is vital for enhancing the emergency response efficiency of communities and cities. In recent years, information and communication technology (ICT), exemplified by such infrastructure, has proven its potential in urban management, addressing issues related to smart cities (Jiang 2021). Through ICT, residents can access flood-related information, make emergency calls, and learn about flood protection, influencing their understanding of flood disasters (Ruslanjari et al. 2023; Yang et al. 2024). Furthermore, residents use ICT devices to form volunteer mutual aid groups through platforms like WeChat groups, which have promoted community communication and enhanced community resilience (Bakhtiari et al. 2023).
Information and communication technology has demonstrably enhanced disaster governance through improved early warning systems and resource coordination, yet its capacity to mobilize resident participation in flood response remains inadequately theorized. Prevailing research frames ICT as a logistical aid, overlooking its potential to directly motivate participation and dynamically interact with community capitals—human, social, political, and financial—within flood-risk contexts. This gap is compounded by the absence of empirical scrutiny on nonlinear threshold effects. Our study addressed these limitations by extending the CCF through the integration of ICT as a constitutive dimension that actively interfaces with existing capitals. This theoretical advancement enables granular analysis of how digital tools directly amplify participation willingness—ultimately informing context-sensitive strategies for resilient flood governance.
2.3 A Conceptual Framework
To overcome the limitations of previous studies, we adapted and expanded the CCF model to frame how digital devices influence residents’ participation willingness (Fig. 1). A key focus of this framework is the role of ICT support. It shows that digital devices not only have a direct impact on residents’ participation willingness, but also interact with other flood-related contextual factors. This interaction, further, significantly shapes residents’ participation willingness in flood response.
The methodology section outlines the research approach employed in this study, which aimed to explore the impact of digital devices on residents’ participation willingness in flood response activities. It includes a detailed description of the study area and data collection methods, the variables considered in the analysis, and the specific techniques used to model and interpret the data.
3.1 Study Area and Data Collection
Zhengzhou City presents a critical case for studying digital participation in flood response due to its unique convergence of disaster extremity and hyper-connected society. The 2021 catastrophe—where single-day rainfall (624.1 mm) nearly matched the annual average (632 mm)—overwhelmed conventional response systems, thereby creating conditions where digital platforms became essential survival tools. Crucially, as a national transportation hub with high smartphone penetration, the city’s digitally saturated environment enabled massive citizen-led innovation: residents spontaneously coordinated rescues via Weibo/Douyin (the Chinese version of TikTok) and crowdsourced aid through viral spreadsheets, demonstrating bottom-up participation mechanisms otherwise invisible in routine disasters. Moreover, China’s centralized social media ecosystem allowed clear observation of information cascades—a methodological advantage for analyzing nonlinear device effects compared to fragmented digital landscapes. While context-specific, these conditions offer transferable insights for flood-vulnerable megacities with high digital adoption, particularly where formal systems may falter during extreme events.
The questionnaire survey for this study took place from 28 June to 11 July 2023. To ensure a representative sample, this study used a multi-stage stratified proportional sampling technique. Initially, 26 primary sampling units, known as residential committees or townships, were randomly selected within the five primary urban districts in Zhengzhou that were affected by the flood disaster, including Jinshui, Erqi, Guancheng, Zhongyuan, and Huiji Districts (Fig. 2). Within these 26 primary sampling units, a further 38 neighborhoods were identified as secondary sampling units. Using structured face-to-face interviews, we engaged residents in the survey process, with research assistants inviting participants from each sampling unit to take part in the interviews. The recruited participants must be local residents who personally experienced the 2021 flood, temporarily excluding non-local residents on business trips or tourism and local residents outside the area. Ultimately, we collected over 1,500 questionnaires. After the exclusion of responses from areas outside the main urban district of Zhengzhou and the removal of incomplete surveys, a final total of 1,351 valid questionnaires were collected.
Fig. 2
The study area and distribution of the survey participants
The selection of variables in this study was meticulously tailored to capture the multifaceted impact of digital devices on residents’ participation willingness in urban flood response. Data from structured interviews were preprocessed by converting responses on five-point Likert scales to ordinal values (1–5). The variables are categorized and defined as follows (see also the Appendix1).
The dependent variable in this study is the residents’ participation willingness, which is operationalized as a categorical variable ranging from 1 (low) to 5 (high). This variable gauges the extent to which the residents are willing to engage in activities related to urban flood response. The mean participation willingness score is 3.745, indicating a moderate level among the respondents.
3.2.2 Information and Communication Technology (ICT) Support
Of particular interest are the core independent variables that measure the supportive role of ICT in resident participation: the number of digital devices, proficiency in the use of digital devices, and usefulness of digital devices.
The number of digital devices serves as an indicator of the penetration and accessibility of ICT within the community, ranging from 1 to 7. This variable is measured by counting the total number of digital devices, primarily mobile phones, that each respondent uses. The variable gauges residents’ overall ability to use modern network technology, with users having more digital devices having greater potential to access effective flood-related information. The average number of digital devices is 2.078, reflecting the prevalence of digital tools among the residents and their potential to access information and communicate during emergencies.
Proficiency in the use of digital devices encompasses the skills and confidence levels of residents in using ICT for various purposes. This composite indicator includes the duration of daily usage of electronic devices, the stability of Internet signals, and residents’ self-assessed proficiency in using electronic devices, evaluating residents’ ability to obtain information through digital devices from three dimensions. The mean proficiency score is 0.708, indicating a moderate level of competence in using digital devices for flood response activities.
Usefulness of digital devices is a critical measure of how effectively residents can use ICT to access flood-related information, communicate with rescue personnel, and form community mutual aid groups. This composite indicator aggregates six aspects: information access, emergency communication, mutual aid formation, civic feedback, crisis response capacity, and community cohesion. Overall, this variable comprehensively considers the supportive role of digital devices in community participation for flood emergency responses. The average usefulness score is 0.745, suggesting that digital devices play a significant role in empowering residents to participate in flood response actions.
The processing of the above two continuous variables in this study involved assigning an equal weight of 1/n to each sub-indicator (where n is the number of sub-indicators within the composite indicator), calculating the product of the respondent’s rating for each sub-indicator and its corresponding weight, summing these products to obtain the total score for the variable, and standardizing the scores to a range of [0,1] to obtain the final scores.
3.2.3 Individual Capabilities
Individual capabilities are widely acknowledged as a key influence on residents’ participation willingness in community initiatives (Wang et al. 2022). This study encompassed two indicators: leadership, which reflects the extent that residents take on leading and coordinating roles during participation; and helping others, which measures the extent to which residents aid their fellow community members. The mean leadership score is 2.231, and the mean helping others score is 2.588, suggesting a general tendency towards active community engagement.
3.2.4 Neighborly Relationships
Neighborly relationships represent the social networks of residents and directly influence individuals’ ability and decisions to participate in flood response (Himes-Cornell et al. 2018). The indicators in this category include information sharing, strength of connection, mutual aid, and mutual trust, all of which are measured on a categorical scale to reflect the frequency and quality of these interactions. The mean scores for information sharing, strength of connection, mutual aid, and mutual trust are 3.369, 3.292, 3.733, and 3.793, respectively, indicating a robust social fabric that can contribute to collective action during emergencies.
3.2.5 Government Support
Government support is critical in the context of flood response, encompassing the assistance and resources provided by the government. This study focused on two aspects: monitoring capability, which evaluates the government’s ability to monitor and assess flood situations in real-time; and organizing capability, which assesses the government’s efficiency in coordinating disaster relief efforts. The mean monitoring capability score is 2.890, and the mean organizing capability score is 3.226, reflecting the perceived effectiveness of government initiatives in disaster management.
3.2.6 Socio-Demographic Attributes
Socioeconomic factors, including age, economic conditions, and marital status, can influence residents’ participation willingness to a certain extent. This study included variables like gender, age, income, education, professional status, presence of elderly, and children.
3.3 Methods
This subsection introduces the methods for modeling the effects of various factors on participation willingness. The aim was to quantitatively model the complex impacts of five capitals on participation willingness. We concentrated on leveraging GBDT algorithms and the Shapley Additive Explanations (SHAP) methodology for investigations in this study.
3.3.1 Gradient Boosting Decision Tree (GBDT)
Gradient boosting decision tree is an ensemble learning algorithm that iteratively trains decision trees to minimize loss parameters. This model offers two notable advantages: first, GBDT maintains high predictive accuracy while exploring the nonlinear relationships between the dependent variable and each feature, providing an appropriate framework to uncover the subtle patterns and correlations between residents’ participation willingness and various influencing factors, effectively capturing complex nonlinear relationships and interactions within the data. Second, GBDT’s ability to calculate the relative importance of predictor variables is significant for this study’s exploration of the role ICT can play in enhancing residents’ participation willingness in flood response activities (Liu et al. 2024). The computations are detailed in Eqs. 1 and 2 (Friedman 2001; Jiang et al. 2024):
In the equation, F(x) denotes the predicted value for the sample x, which is the weighted sum of the predictions from M decision trees. fm(x) represents the predicted value for the sample by the m-th decision tree, and h(x;am) is the decision tree-based classifier, where \(\beta_{m}\) denotes the weight of the m-th decision tree. After iterative optimization of the model, a learning rate \(\xi (0 \, < \xi \le 1)\) is introduced to determine the contribution of each decision tree and to control the overfitting issue. Furthermore, \(F_{m} \left( x \right)\) is the final model, and \(F_{{m{-}1}} \left( x \right)\) is the model at the (m-1)-th iteration.
In this study, the GBDT model was implemented in Python, with key parameters including the learning rate, maximum tree depth, and the number of trees. Using grid search combined with five-fold cross-validation, the optimal parameter set was identified as a learning rate of 0.02, a maximum tree depth of 2, and 200 trees for model fitting. After finalizing the hyperparameters, the dataset was randomly split into a training set (70%) and a test set (30%). The model achieved a minimum root mean squared error (RMSE) of 0.71 on the test set.
3.3.2 Shapley Additive Explanations (SHAP)
This study used the SHAP methodology to assess the contributions of 18 input variables to the GBDT model’s predictions. Shapley Additive Explanations, an explainable AI technique employing game-theoretic approaches, is widely acknowledged as a key method for elucidating the outputs of machine learning models. The core idea of SHAP values is the concept of attributing a numerical value to the features of a specific data point within a given predictive model, indicating the feature’s contribution to the model’s predictive outcome. Operating at the local explanation level, it offers insights into the impact of each factor at the level of individual samples. This approach provides a more nuanced understanding of the association between each factor and residents’ willingness to engage in flood emergency responses, beyond mere reliance on global importance scores (Zou et al. 2025). The computation of SHAP values for a model’s specific prediction is depicted in Eq. 3:
In the equation, n represents the total number of factors, \(N\backslash \, \left\{ j \right\}\) denotes the set of all possible combinations of factors excluding j, S is a subset of factors within the set \(N\backslash \, \left\{ j \right\}\). \(f \, \left( s \right)\) is the model output for the factors in S, \(f\left( {SU\{ j\} } \right)\) is the model output when the factors in S are combined with factor j. \(\frac{|S|!(n - |S| - 1)!}{{n!}}\) is the weight of S.
The GBDT model’s prediction can be represented as the sum of the SHAP values for each feature, plus a constant baseline value, as depicted in Eq. 4:
where φ0 represents the mean of the model’s output across all samples.
4 Results
The results of this study reveal feature importance and nonlinear relationships among determinants. And within the CCF, interactions between usefulness of digital devices and capital resources constitute a key analytical focus.
4.1 Variables’ Relative Importance
As shown in Table 1 and Fig. 3, this study encompassed five categories of independent variables: ICT support, individual abilities, neighborly relationships, government support, and socioeconomic attributes. Based on the ranking of feature importance, neighborly relationships (31.30%), ICT support (25.59%), and individual abilities (25.15%) exhibit the strongest explanatory power, highlighting the pivotal roles that these three factors play in facilitating resident participation in flood disaster management. This reflects the main influence of social, built, and human capitals in encouraging resident participation in urban flood responses.
Table 1
Feature importance
Feature
Relative Importance
Rank
Helping others
0.2039
1
Usefulness of digital devices
0.1532
2
Mutual trust
0.1369
3
Proficiency in the use of digital devices
0.0834
4
Monitoring capability
0.0716
5
Mutual aid
0.0687
6
Leadership
0.0476
7
Organizing capability
0.0441
8
Age
0.0380
9
Information sharing
0.0358
10
Education
0.0256
11
Strength of connection
0.0251
12
Number of digital devices
0.0193
13
Income
0.0127
14
Children
0.0110
15
Profession
0.0090
16
Gender
0.0076
17
Elderly
0.0065
18
Fig. 3
Relative importance and Shapley Additive Explanations (SHAP) summary plot
A closer look reveals that the individual ability indicator, specifically helping others (20.39%), is the most significant, highlighting the pivotal role of individual abilities in the participation of residents. Next is the usefulness of digital devices (15.32%), indicating that the availability of digital devices is essential for residents to access information and maintain communication, which significantly influences residents’ behavior patterns and willingness to participate. Mutual trust among community residents has also received a high relative importance score (13.69%), further confirming the critical role of community trust in the disaster response phase. Although the government’s monitoring capability and organizing capability are lower in the importance ranking, their role is indispensable, and is equally vital for disaster management. In contrast, individual socioeconomic attributes, such as profession and gender, have a relatively minor impact on the resident participation.
With regard to the summary plot for SHAP (Fig. 3), the low values of helping others, usefulness of digital devices, and mutual trust (blue dots) are mainly on the left side, while the high values (red dots) are mostly on the right. This means that helping others, usefulness of digital devices, and mutual trust values are positively associated with participation willingness. Similar results are observed for monitoring capability, mutual aid, and so on. Low organizing capability and information sharing values are mainly distributed on the right, which means that they are generally negatively related to participation willingness.
4.2 Nonlinear Relationships between Key Independent Variables and Residents’ Participation Willingness
Figure 4 shows that there is an approximately positive linear relationship between digital devices and residents’ participation willingness in flood response. Greater usefulness of digital devices increases residents’ participation willingness. Specifically, the positive impact of usefulness of digital devices on resident participation initially increases gradually and then more rapidly (Fig. 4a). At lower levels of usefulness of digital devices (below 0.2), its role in enhancing residents’ participation willingness is negligible. This is likely because, at these lower levels, the application of digital devices has not yet reached a threshold capable of significantly altering residents’ behavior. When usefulness of digital devices reaches 0.2, there is a marked increase in the level of residents’ participation willingness. This turning point suggests that, at a certain level, the support from digital devices has become sufficiently widespread and ingrained to significantly change residents’ behavioral patterns. This may be because residents can use digital devices to access almost all flood-related information, making the real-time situation of floods completely transparent at the individual level.
Fig. 4
Nonlinear links: Information and communication technology (ICT) support and participation willingness. SHAP Shapley Additive Explanations
Like usefulness of digital devices, proficiency in the use of digital devices also has a pronounced effect on the propensity of residents to engage in flood response activities (Fig. 4b). Unlike usefulness of digital devices, it is at the 0.25 threshold of proficiency in the use of digital devices that a notable increase in residents’ participation willingness is observed. This suggests that residents may have elevated expectations for the overall user experience of digital devices, including the reliability of their functions and the stability of the devices themselves. It is only when these experiences surpass a certain tipping point that residents become sufficiently satisfied and, consequently, more inclined to engage in urban flood response efforts.
A U-shaped relationship is observed between the number of digital devices and participation willingness (Fig. 3c) although the effect size is modest. Willingness decreases as device count increases from 1 to 4, but shows a positive trend beyond 4 devices. This pattern may reflect threshold dynamics where, once basic informational needs are met, additional devices potentially enable access to diversified services and richer information streams, which could enhance engagement motivation. However, alternative explanations warrant consideration—including potential confounding by unmeasured variables (for example, technological proficiency correlating with device ownership).
Individual abilities, including helping others and leadership, significantly enhance residents’ participation willingness. Specifically, helping others has a nonlinear positive effect on participation willingness (Fig. 5a). When the level is below 3, there is a significant increase in residents’ willingness. This indicates that during the early stages of a flood, when community members feel particularly helpless, the belief that they can provide assistance to others leads to a strong desire to engage in emergency response efforts. After reaching the level of 3, although the participation willingness continues to rise, the rate of increase slows down significantly. This may be due to the diminishing marginal effects of helping abilities on willingness.
Fig. 5
Nonlinear links: Individual abilities, neighborly relationships, government support, and participation willingness. SHAP Shapley Additive Explanations
Leadership has a significantly positive effect on participation, with a clear inflection point observed (Fig. 5b). When leadership is below a threshold of 2, there is a slight dampening effect on participation willingness. However, when it exceeds a threshold of 2, there is a notable increase in residents’ participation willingness. This may be because individuals with strong leadership can more effectively mobilize and organize community resources, thereby boosting residents’ confidence and willingness to engage in emergency responses.
Regarding neighborly relationships (Fig. 5c–f), the overall impact of mutual trust, along with mutual aid, is positively and nonlinearly correlated with residents’ participation willingness. The relationship between community information sharing and network connectivity is intricate.
Comparatively, the impact of mutual trust among community residents on their participation willingness in flood emergency response is the most pronounced (Fig. 5c). Trust levels at various degrees differentially affect participation willingness, exhibiting a “slow then rapid” pattern of positive enhancement. When trust levels are below 3, the enhancement of residents’ participation willingness is relatively slow. Above 3, the willingness level increases more markedly. This indicates that in an environment of high trust, residents’ sense of identification with community flood management is strengthened, and they are more inclined to actively engage in response.
The influence of mutual aid on the participation willingness is particularly distinctive (Fig. 5d). Residents’ participation willingness significantly increases at lower (1–2) and higher (4–5) levels of mutual aid. This is likely because, at lower levels of mutual aid, residents are more motivated to seek additional support. At higher levels of mutual aid, residents recognize the value of stronger community ties, which in turn promotes their willingness to participate. At moderate levels (2–4), the increase in residents’ participation willingness is less pronounced.
There is a negative correlation between community information sharing and residents’ participation willingness (Fig. 5e). As the level of information sharing increases, the participation willingness declines, stabilizing at moderate to high levels (> 3), indicating a threshold effect. Contrary to anticipated positive outcomes, this pattern may reflect that heightened risk awareness from expanded information exposure potentially triggers protective avoidance behaviors—as residents comprehend evolving hazards and safety complexities (Zaw and Lim 2017), elevated risk perceptions may paradoxically diminish engagement motivation. When the information sharing level exceeds 3, the declining trend does not worsen. This phenomenon may be due to residents adapting to the ongoing flow of information.
The relationship between strength of connection and residents’ participation willingness exhibits a U-shaped correlation (Fig. 5f). Below the level of 3, the willingness tends to decrease. This may occur because at low levels of network connection, residents may perceive community participation as superficial and formal, thus reducing their willingness. Above 3, the willingness begins to increase. High levels of network connection foster mutual trust among residents, which in turn enhances their participation willingness.
Regarding government support, the influence of monitoring capability on residents’ participation willingness exhibits a U-shaped correlation (Fig. 5g). There is a slight decline in the willingness when the capability is below 2.5. This could be because at this stage, residents perceive that the government’s monitoring capability does not sufficiently provide a sense of security, thus reducing their participation willingness. When it exceeds 2.5, there is a noticeable increase in the participation willingness. This may be attributed to residents witnessing the government’s efforts and progress, which stimulates their willingness.
The government’s organizing capability exhibits a nonlinear negative correlation with residents’ participation willingness (Fig. 5h). When the government’s organizing ability is below 4, the residents’ participation willingness significantly decreases as it improves. This counterintuitive pattern, contradicting conventional assumptions, may stem from heightened institutional trust fostering dependency: residents perceiving governmental competence as sufficient may reduce their perceived need for personal action (van Heel and and van den Born 2020). When reaching 4, the willingness level stabilizes and may even slightly increase. It is likely that the government can more effectively organize disaster relief efforts, thereby enhancing residents’ sense of security and motivation to engage.
4.3 Two-Way Interaction Effects
In this subsection, we analyze the potential interaction effect between digital devices and other flood-relevant factors. We present this effect using a two-dimensional partial dependence plot (PDP), which highlights the second-order effect of two variables. Specifically, this study focused on the usefulness of digital devices, which plays a significant role in predicting residents’ participation willingness. Additionally, the usefulness of these devices demonstrates strong interaction effects on residents’ participation willingness.
First, we identified the reinforcement pattern of the interaction effect, which emphasizes how various factors collectively amplify participation willingness. Figure 6a–d vividly illustrates that the interaction between the usefulness of digital devices and key factors, including helping others, leadership, mutual trust, and mutual aid, exerts a significant and profound influence on the inclination of residents to engage in flood response. The data indicate that when these factors interact strongly, the residents’ participation willingness in response increases accordingly. This pattern underscores the idea that the usefulness of digital devices, when combined with these factors, operates in a synergistic manner, strengthening and amplifying the participation willingness.
Fig. 6
Interactions between usefulness of digital devices and other key factors on participation willingness
Second, we also uncovered a compensation pattern within the interaction effect, revealing how the usefulness of digital devices can counterbalance the influence of weaker factors. As demonstrated in Fig. 6e–h, even in cases where key factors—such as information sharing, strength of connection, monitoring capability, and organizing capacity—are comparatively low, residents’ participation willingness remains remarkably high. This resilience suggests that device usefulness may compensate for limitations in these factors. When these factors are suboptimal, usefulness of digital devices appears to sustain participation levels. Thus, within the scope of our measured constructs, digital tools potentially mitigate participation constraints arising from diminished social interaction or organizing capacity.
5 Discussions
Recent advances in ICTs have significantly enhanced residents’ involvement in flood emergency response by facilitating rapid information exchange, resource mobilization, and coordination. However, its impact on residents’ participation willingness has yet to be fully explored. This study employed the CCF framework and the GBDT model to analyze how ICT influences residents’ willingness to engage in flood response activities. We now discuss the findings.
While previous research has recognized the pivotal role of ICT in disaster management, it often overlooked the intricate interplay between ICT and resident behavior. By employing the CCF, this study delved into the nonlinear influence of ICT on residents’ willingness to engage in urban flood response, offering novel theoretical perspectives in the field of disaster management. It broadens the application of the CCF based on Jiang et al. (2024), particularly when considering ICT as a key capital in contemporary society, revealing the interplay between ICT and residents’ participation willingness. Through an empirical analysis, we not only verified the central role of ICT in boosting residents’ willingness to engage in flood emergency response, but also illustrated how ICT can enhance the sense of security and participation in preparedness activities by bolstering other community capitals. This nuanced understanding of ICT’s role in disaster response provides a valuable foundation for refining the theoretical framework of crisis management.
Additionally, the study used GBDT to highlight the nonlinear relationships between ICT, individual abilities, neighborly relationships, government support, and residents’ participation willingness, demonstrating the intricate mechanisms that influence residents’ participation willingness. Variables within the study all manifest distinct nonlinear relationships. Notably, participation willingness increases markedly with higher usefulness of digital devices and greater user proficiency. This aligns with established findings that ICT advancements accelerate engagement in flood response initiatives (Wehn and Evers 2015; Ohta et al. 2021). Beyond this, however, our analysis revealed impact thresholds for these factors (0.20 and 0.25, respectively), indicating that effective participation in flood emergency management requires residents to exceed baseline ICT proficiency levels. These insights further substantiate the pivotal role of ICT and similar factors in disaster management, while also highlighting the significance of factoring in these nonlinear elements within disaster response strategies.
Beyond examining the nonlinear relationships between ICT and other factors that affect residents’ participation willingness, the study highlights the combined effects of digital devices with various capitals on residents’ willingness, offering a novel perspective on their participation. In general, the interaction effects observed in this study exhibit two distinct patterns: reinforcement and compensation. First, a reinforcement effect is evident, where the usefulness of digital devices, combined with factors like helping others, leadership, mutual trust, and mutual aid, boosts residents’ participation willingness. This suggests that it is necessary to enhance the usefulness of digital devices by promoting digital literacy and accessibility while also fostering key social factors like mutual trust, leadership, and mutual aid to boost resident participation. Second, a compensation effect occurs when low levels of factors such as information sharing, strength of connection, monitoring capability, and organizing capacity still result in high participation, due to the usefulness of digital devices mitigating their low impact. Even in areas with weak community engagement factors, digital devices can compensate and maintain participation levels. Therefore, an integrated approach that combines both digital infrastructure and social initiatives is essential for optimizing community engagement and participation.
Despite these findings, the study has its limitations. First, the study inadequately considers natural and cultural capitals within the CCF. Natural capital, such as rainfall and storm frequency, may be directly related to residents’ participation willingness in flood response, encouraging them to adopt protective measures (Himes-Cornell et al. 2018). Moreover, cultural capital, including personal values, significantly impacts resident participation, with emotions, risk perception, and personal attitudes influencing individual disaster prevention behaviors (Jacob et al. 2023). Future studies should consider integrating natural and cultural capitals into the framework to provide a more comprehensive perspective on understanding and interpreting resident participation in flood responses. Second, the study’s sample is confined to residents of the main urban area of Zhengzhou City, Henan Province, which may limit the generalizability of the findings. Given the specific socioeconomic conditions and cultural backgrounds, residents’ experiences and coping strategies may significantly differ from other regions. Thus, the findings of this study should be applied to other cities or countries with caution. Third, the study’s exploration of the relationship between ICT and resident participation in urban flood response is primarily constrained to data collection and analysis at a specific time point. Future studies should consider a longer time frame to capture the dynamic impacts of technological evolution and policy alterations on residents’ participation willingness. Finally, post-event data collection introduces potential self-reporting biases. More rigorous sensitivity analyses in subsequent studies would strengthen reliability assessments.
6 Conclusion
This study sought to fill a gap in research by adapting the CCF to examine how digital devices influence residents’ participation willingness in urban flood response activities. The analysis was conducted using the GBDT model, a powerful tool for capturing complex, nonlinear relationships between variables. Data for this research were gathered from a sample of 1,351 residents in Zhengzhou City in 2023.
The findings highlight several important insights: (1) Digital devices significantly influence residents’ participation willingness in flood response, especially in terms of information acquisition and sharing. (2) Indicators of digital devices, such as their usefulness and residents’ proficiency in using them, are key predictors of participation willingness, and their relationships are primarily nonlinear. (3) Digital devices demonstrate both reinforcement and compensation patterns in the interaction effects on residents’ participation willingness. Reinforcement occurs when the presence of multiple supportive factors, such as digital access alongside social networks or government support, amplifies participation willingness. Compensation, on the other hand, refers to the ability of digital devices to mitigate the low impact of other factors, such as limited social capital or government responsiveness, thereby still enabling high participation in flood response efforts.
Overall, this research contributes to the refinement of the CCF by emphasizing the critical role of digital devices in shaping residents’ participation willingness in flood disaster management. It highlights that, in addition to individual, social, and governmental factors, digital devices serve as a powerful tool for enhancing participation, thereby improving disaster preparedness, response efficiency, and recovery outcomes. The study underscores the importance of providing adequate digital device support in flood emergency response initiatives, which can help meet the needs of residents, strengthen community resilience, and enhance governmental coordination. By incorporating these technological advances, flood disaster management can be more adaptive, inclusive, and effective in addressing the challenges posed by natural hazards.
Acknowledgments
This work was supported by the National Natural Science Foundation of China (Grant No. 42201207).
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 licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence 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 licence, visit http://creativecommons.org/licenses/by/4.0/.
Azad, M.J., and B. Pritchard. 2022. Financial capital as a shaper of households’ adaptive capabilities to flood risk in northern Bangladesh. Ecological Economics 195: Article 107381.
Bakhtiari, V., F. Piadeh, K. Behzadian, and Z. Kapelan. 2023. A critical review for the application of cutting-edge digital visualisation technologies for effective urban flood risk management. Sustainable Cities and Society 99: Article 104958.
Bandura, A. 1978. Self-efficacy: Toward a unifying theory of behavioral change. Advances in Behaviour Research and Therapy 1(4): 139–161.CrossRef
Bexell, M., and K. Jönsson. 2017. Responsibility and the United Nations’ sustainable development goals. Forum for Development Studies 44: 13–29.CrossRef
Bobbio, L. 2019. Designing effective public participation. Policy and Society 38: 41–57.CrossRef
Callaghan, E.G., and J. Colton. 2008. Building sustainable & resilient communities: A balancing of community capital. Environment, Development and Sustainability 10: 931–942.CrossRef
Cui, P., and D. Li. 2020. A SNA-based methodology for measuring the community resilience from the perspective of social capitals: Take Nanjing, China as an example. Sustainable Cities and Society 53: Article 101880.
Flora, C.B., M. Emery, S. Fey, and C. Bregendahl. 2005. Community capitals: A tool for evaluating strategic interventions and projects. North Central Regional Center for Rural Development, Ames, IA. https://aae.wisc. edu/ced/wp-content/uploads/sites/8/2014/01/204.2-Handout-Community-Capit als.pdf. Accessed 24 Aug 2024.
Friedman, J.H. 2001. Greedy function approximation: A gradient boosting machine. The Annals of Statistics 29(5): 1189–1232.CrossRef
Gammoh, L.A., I.G.J. Dawson, and K. Katsikopoulos. 2023. How flood preparedness among Jordanian citizens is influenced by self-efficacy, sense of community, experience, communication, trust and training. International Journal of Disaster Risk Reduction 87: Article 103585.
Guo, J., Y. Bian, M. Li, and J. Du. 2024. Assessing resilience through social networks: A case study of flood disaster management in China. International Journal of Disaster Risk Reduction 108: Article 104583.
Guo, X., J. Cheng, C. Yin, Q. Li, R. Chen, and J. Fang. 2023. The extraordinary Zhengzhou flood of 7/20, 2021: How extreme weather and human response compounding to the disaster. Cities 134: Article 104168.
Himes-Cornell, A., C. Ormond, K. Hoelting, N. Ban, J. Koehn, E. Allison, E. Larson, and D. Monson et al. 2018. Factors affecting disaster preparedness, response, and recovery using the community capitals framework. Coastal Management 46: 1–24.CrossRef
Jacob, J., P. Valois, M. Tessier, D. Talbot, F. Anctil, G. Cloutier, and J.-S. Renaud. 2023. Using the theory of planned behavior to identify key beliefs underlying flood-related adaptive behaviors in the province of Québec, Canada. Journal of Flood Risk Management 16: Article e12906.
Jiang, H. 2021. Smart urban governance in the “smart” era: Why is it urgently needed? Cities 111: Article 103004.
Jiang, H., Y. Wang, W. Ma, J. Wang, and M. Zhang. 2024. Unlocking the nonlinear nexus: Accessibility of emergency resource and resident participation in flood response. Journal of Transport Geography 118: Article 103926.
Lax, J., and J. Krug. 2013. Livelihood assessment: A participatory tool for natural resource dependent communities. Thünen Working Paper No. 7. Braunschweig: Johann Heinrich von Thünen-Institut.
Leong, C., S. Pan, P. Ractham, and L. Kaewkitipong. 2015. ICT-enabled community empowerment in crisis response: Social media in Thailand flooding 2011. Journal of the Association for Information Systems 16(3). https://doi.org/10.17705/1jais.00390.
Li, Y., Y. Tao, Q.K. Qian, E. Mlecnik, and H.J. Visscher. 2024. Critical factors for effective resident participation in neighborhood rehabilitation in Wuhan, China: From the perspectives of diverse stakeholders. Landscape and Urban Planning 244: Article 105000.
Liu, Q., J. Qiao, M. Li, and M. Huang. 2024. Spatiotemporal heterogeneity of ecosystem service interactions and their drivers at different spatial scales in the Yellow River Basin. Science of the Total Environment 908: Article 168486.
Lo, A.Y., B. Xu, F.K.S. Chan, and R. Su. 2015. Social capital and community preparation for urban flooding in China. Applied Geography 64: 1–11.CrossRef
Ohta, A., M. Nishina, C. Motohashi, N. INoue, T. Miyazaki, M. Takahashi, M. Uemura, and M. Kamei. 2021. ICT use and participation in community support activities among the elderly in Japan. International Journal of Epidemiology 50. https://doi.org/10.1093/ije/dyab168.506.
Paul, B.K., M.K. Rahman, T. Crawford, S. Curtis, M.G. Miah, M.R. Islam, and M.S. Islam. 2020. Explaining mobility using the community capital framework and place attachment concepts: A case study of riverbank erosion in the lower Meghna estuary, Bangladesh. Applied Geography 125: Article 102199.
Ruslanjari, D., E.W. Safitri, F.A. Rahman, and C. Ramadhan. 2023. ICT for public awareness culture on hydrometeorological disaster. International Journal of Disaster Risk Reduction 92: Article 103690.
UNDRR (United Nations Office for Disaster Risk Reduction). 2015. Sendai framework for disaster risk reduction 2015–2030. Geneva: UNDRR.
van Heel, B.F., and R.J.G. van den Born. 2020. Studying residents’ flood risk perceptions and sense of place to inform public participation in a Dutch river restoration project. Journal of Integrative Environmental Sciences 17: 35–55.CrossRef
Wang, T., Y. Lu, T. Liu, Y. Zhang, X. Yan, and Y. Liu. 2022. The determinants affecting the intention of urban residents to prepare for flood risk in China. Natural Hazards and Earth System Sciences 22: 2185–2199.CrossRef
Wehn, U., and J. Evers. 2015. The social innovation potential of ICT-enabled citizen observatories to increase eParticipation in local flood risk management. Technology in Society 42: 187–198.CrossRef
Yang, Y., Y. Zhang, B.X. Zhu, J. Zhou, Y. Liu, D. Gao, and J. Sauer. 2024. ICT promotes smallholder farmers’ perceived self-efficacy and adaptive action to climate change: Empirical research on China’s economically developed rural areas. Climate Services 33: Article 100431.
Zaw, T.N., and S. Lim. 2017. The military’s role in disaster management and response during the 2015 Myanmar floods: A social network approach. International Journal of Disaster Risk Reduction 25: 1–21.CrossRef
Zhou, Y., Z. Wu, Q. Liang, H. Xu, H. Wang, and W. Xue. 2024. Threshold and real-time initiation mechanism of urban flood emergency response under combined disaster scenarios. Sustainable Cities and Society 108: Article 105512.
Zhu, Y., and N. Li. 2021. Virtual and augmented reality technologies for emergency management in the built environments: A state-of-the-art review. Journal of Safety Science and Resilience 2: 1–10.CrossRef
Zou, D., Q. Li, Y. Zhou, S. Liang, and S. Zhou. 2025. Understanding factors associated with individuals’ non-mandatory activities using machine learning and SHAP interpretation: A case study of Guangzhou, China. Travel Behaviour and Society 38: Article 100894.
