Previous studies have found that the higher the level of quality of life (QoL) or well-being is, the more desirable behavior people may take, including disaster preparedness actions. However, other variables like socio-demographics and risk perception might have varying effects on both of them. Therefore, comparisons between socio-demographic groups might be of help to disentangle this complex mechanism, but no studies have investigated their relationships. This study revisited surveys in Taiwan, China and Kinki, Japan and took a simple QoL measurement as an indicator of well-being. Multi-group analysis provided the analysis of the relationships between QoL, trust in government, disaster experience (EX), disaster risk perception (RP), and preparedness behavior (PB) for each dataset and for each grouping, here, by gender and by marital status. As a result, among others, the effect of QoL on PB was only significant in the female group in Taiwan, while RP and EX differently affected PB in different gender groups in Kinki. Moreover, no significant differences were found between the marital status groups in the associations between QoL, RP, and PB in Taiwan. These results imply the importance of group-specific approaches in intervention measures. Considering and addressing the limitations of this study, such as a cross-sectional design and the omission of potential variables, future studies may further explore the associations in different groups or with new variables such as self-efficacy.

Road networks are a critical infrastructure system for the sustainable functioning of cities. However, they are frequently disrupted by urban flooding, leading to increased travel times and hindering emergency responses. This study proposed a novel dynamic flood-response simulation framework for urban transportation to evaluate the impacts of rainstorms and flooding on traffic systems, focusing on coupling the Integrated Hydrology and Hydrodynamics Urban Flood Model (IHUM) and the Simulation of Urban MObility (SUMO) model. The results obtained from Xiaoguwei Island, Guangzhou City, indicate that a 2-h rainstorm of a 2-year return period can affect traffic for over 4.5 h. During a 100-year return period rainstorm, average travel speed declines by 54%, while the emergency response time, for example, for police services, increases from 4.83 to 14.52 min. These findings highlight the significant impacts of flooding on urban traffic networks, assisting local authorities and stakeholders to proactively identify vulnerable network segments and prioritize targeted interventions for enhancing transportation system resilience to floods.

This study examined public support for disaster-related measures (risk awareness, memory transmission, prevention facilities, and information networks) in the long-term recovery phase of a disaster-affected municipality. Focusing on Rikuzentakata City, which was severely impacted by the 2011 Great East Japan Earthquake and Tsunami, a stated-preference survey was conducted and analyzed using discrete choice models. Three research questions were addressed: RQ1 on the main effects of disaster-related measures, RQ2 on the interaction effects among these measures, and RQ3 on the interaction effects between disaster-related measures and economic-livelihood measures (ELMs). Stand-alone disaster-related measures and their within-domain combinations reduced public support, indicating dilution effects. By contrast, support increased when they were integrated with complementary ELMs. The policy implications are threefold. First, disaster-related measures should be implemented with caution in the medium- to long-term recovery process, as pursuing them alone may generate dilution effects. Second, integrated policy packages that combine disaster-related measures with complementary ELMs should be prioritized. Third, shrinking-city municipalities can better overcome fiscal and human resource constraints and enhance sustainability by leveraging complementarities across policy domains. Overall, the findings provide quantitative evidence that disaster-related policy must be reconceptualized as part of a broader policy portfolio. This insight has broader relevance for hazard-prone regions worldwide and offers international implications for long-term disaster governance in line with the Sendai Framework, the Sustainable Development Goals, and the Paris Agreement.

Climate change has intensified extreme rainfall events, challenging progress toward SDG 11's urban resilience targets. Current assessment methods often neglect dynamic recovery processes and regional precipitation disparities. We propose a three-phase framework combining interpretable machine learning (ML) and factorial experiments, using the Prep_shock index that integrates standardized rainfall intensity, capital exposure, and historical probability, to evaluate the dynamic resilience of 220+ Chinese cities from 2019 to 2022. Key findings reveal that: (1) The Prep_shock index effectively eliminates north-south precipitation biases, identifying Shandong coastal cities and Yangtze River Delta city clusters (36.2%) as high-resilience areas, in contrast to Henan Province. COVID-19 exacerbated systemic risks in megacities, undermining their capital protection capacities. (2) Spatial diagnostics classify 75.6% of the cities into Quadrant III (the balanced resilience category), with recovery times decreasing from the west to the east. Super-large cities like Zhengzhou (2021) exhibited critical recovery deficiencies (Quadrant IV). (3) Interpretable ML models (XGBoost/EBM) identify redundancy as the dominant resilience driver—robustness governs baseline resilience, while recovery relies on emergency support (for example, hospital beds density and fiscal inputs) and redundant infrastructure (for example, road network density). (4) Factorial experiments reveal optimization trade-offs: simultaneous enhancement of rapidity and redundancy diminishes their individual benefits, necessitating context-specific prioritization. The study advances dynamic resilience assessment methods and proposes quadrant-specific strategies for tailored urban adaptation.

The public's attribution of responsibility during a crisis is a central process in crisis communication, often explained by the situational crisis communication theory (SCCT). However, SCCT was developed in a pre-social media era, and its applicability in the new ecosystem of algorithm-driven, short-video platforms remains a critical theoretical gap. This study investigated how the core mechanisms of public responsibility attribution are reconfigured in the unique context of China's leading short-video platform, Douyin. Analyzing 185,148 comments following the tragic Yingcai School fire, our large language model (LLM)-based analysis answered two questions: (1) How are public attributions of responsibility structured in this emotionally charged, algorithmic environment? and (2) How do offline socioeconomic factors shape these digital crisis discourses? Our findings reveal two distinct attribution pathways, namely an anger-accountability track and a sadness-reflection track and demonstrate that critical discourse is systematically linked to regional development. This research provides a crucial empirical validation of SCCT for the short-video era and offers a data-driven guide for context-aware public administration.

Identifying the optimal intensity measures (IMs) of ground motion is a critical step in the seismic fragility analysis of slopes using probabilistic seismic demand models. In this study, we investigated various IMs of earthquake ground motions, considering both vertical and horizontal components that can trigger landslides, through numerical simulations. A set of 19 IMs was examined for three heights of slopes. The optimal IMs were assessed using the maximum permanent displacement as an engineering demand parameter. The findings reveal that acceleration-related parameters specifically sustained maximum acceleration (SMA) and root-mean-square of acceleration, and velocity-related parameters namely sustained maximum velocity (SMV) and peak ground velocity, are the most effective IMs for slopes subjected to both vertical and horizontal ground motions. For slopes subjected exclusively to horizontal ground motion, SMA is recommended as the optimal IM for lower-height slopes, while SMV is more suitable for the taller slope. In contrast, for models subjected to combined horizontal and vertical ground motions, SMA is consistently identified as the optimal IM for slope across all heights. Notably, the study concludes that peak ground acceleration, a commonly used parameter in seismic analysis, is unsuitable for the considered slopes.

Building structure characteristics are essential for seismic risk assessment and management. However, accurately classifying building structural types at the city scale remains challenging due to limited data availability, spatial heterogeneity, and the difficulty in capturing contextual dependencies. In this study, a novel graph neural network (GNN)-based framework, BSPGNN (Building Structural Type Prediction using Graph Neural Networks), was proposed, integrating spatial relationships and building geometric features to improve structural type classification. A Delaunay triangulation (DT) graph was constructed from building footprint centroids to represent spatial proximity, and node features included footprint area, height, and construction year. Experiments using a real-world building dataset from Tianjin, China demonstrated that BSPGNN significantly outperformed traditional machine learning models, such as random forest (RF) and support vector machine (SVM), particularly in capturing spatially coherent patterns. The proposed model achieved a classification accuracy of 90.25% and 85.51% on the training set and validation set respectively, showing robust performance under missing data conditions. The results highlight the potential of spatial graph-based models in advancing building structural type classification for seismic risk assessment and management.

Rural settlements have experienced a noticeable increase in extreme weather events and associated disasters in recent years. Pakistan is consistently ranked as one of the most affected regions globally, and the catastrophic floods of 2022 further underscored its vulnerability to floods, causing unprecedented human, economic, and environmental losses. This study conducted a multidimensional vulnerability assessment of flood-prone rural areas in Dera Ismail Khan, Pakistan, using a composite index approach informed by principal component analysis (PCA). Principal component analysis was employed to assign statistically robust weights to selected indicators, ensuring an objective aggregation of vulnerability components. A questionnaire with a mix of closed-ended and open-ended items was used to collect data through a household survey. The findings revealed that a substantial proportion (40%) of respondents experienced high multidimensional vulnerability, while approximately 30% exhibited moderate vulnerability. Factors such as age distribution, household income, infrastructure quality, and risk perception significantly contributed to overall vulnerability. This study developed a scalable and replicable model for assessing rural flood vulnerability, offering practical insights for policymakers, planners, and disaster management authorities.

Extreme disaster events, particularly floods, impact individuals, households, and communities differently depending on a range of factors. Some residents are more vulnerable than others to floods, hence urgent action is needed to improve preparedness of vulnerable groups to reduce potential impacts of flooding. To achieve this, it is important to assess household flood preparedness, and identify the influence of household-level attributes on flood preparedness. Using a large-scale community survey, this study investigated how key household attributes such as evacuation assistance, Internet access, and key motivational factors and sources of motivation such as personal and social networks influence flood preparedness among vulnerable households. Households were categorized into three levels of vulnerability (high, medium, and low) by combining household economic capacity, as measured by household monthly expenditure, and the presence of vulnerable family members. Six key findings emerged: (1) Highly vulnerable households showed higher resilience to floods, and flood preparedness levels are independent of household vulnerability levels; (2) Self-reported household flood preparedness is positively influenced by learning from past disaster experience; (3) Financial and time commitments, and a sense of urgency for household-level flood preparedness are key intrinsic motivational factors that influence flood preparedness; (4) Access to reliable Internet can be used as a proxy to predict the degree of household flood preparedness; (5) Higher levels of awareness and knowledge of flood preparedness were reported despite low levels of community consultation; and (6) Self-motivation is the key source of motivation for flood preparedness. The study findings will support key institutional stakeholders such as local governments to devise strategies to strengthen the flood resilience of vulnerable households and communities.

This study critically interrogates dominant models of post-disaster recovery by combining an interdisciplinary review of critical scholarship with grounded empirical analysis from Brazil. It focuses on landless and unhoused populations, as well as residents of informal settlements, to explore how disaster recovery frameworks, rather than reducing vulnerability, often reproduce spatial inequality and deepen exclusion. The research draws on a multi-site, multi-temporal mixed-methods study conducted in collaboration with the Movimento dos Atingidos por Barragens (MAB), analyzing Brazil's most severe recent climate-related disasters: landslides and floods in Petrópolis (2011, 2022, 2024) and São Sebastião (2023). Fieldwork involved participant observation, over 200 semistructured interviews with affected residents and officials, and 110 completed questionnaires. Findings reveal that Brazil's disaster governance framework embeds exclusionary dynamics, privileging legally recognized property owners while marginalizing those without formal tenure. Recovery programs often simplify complex social realities through rigid eligibility criteria, thereby silencing diverse lived experiences. As a result, recovery often becomes a prolonged, secondary disaster for the most vulnerable. The article argues that prevailing recovery models, anchored in technocratic management and depoliticized resilience discourse, fail to address the structural roots of marginalization. By centering the role of grassroots movements such as MAB, this article highlights how collective action can expose recovery injustices and foster more inclusive, participatory, and transformative approaches to disaster governance.

Deep learning models are widely used for urban flood prediction, but current research lacks a clear explanation of how indicator weight changes affect model accuracy. This study incorporated the attention mechanism between the convolutional and fully connected layers of the convolutional neural network (CNN) to enable the model to focus on critical flood-inducing factors, and employed the particle swarm optimization (PSO) to optimize the key hyperparameters (for example, the number of filters and learning rate). Furthermore, we employed Shapley additive explanation (SHAP) to analyze how flood-inducing indicator weight changes affect prediction accuracy. The model was tested on Haidian Island, China. The Nash-Sutcliffe efficiency (NSE) coefficient of the CNN model is 0.9287. After incorporating the attention mechanism into the CNN and optimizing the hyperparameters using PSO, the NSE is improved to 0.9503. The model demonstrates higher accuracy in predicting larger inundations, with the NSE for the 100-year return-period flood reaching 0.9535, compared to 0.8341 for the 5-year return period. Interpretability analysis shows that elevation is the most important flood-inducing factor, accounting for 44% of the total importance, followed by tidal levels, which account for 33%. The attention mechanism increases the weights of important flood-inducing factors (for example, elevation, tide level); after hyperparameter optimization, the model achieves more comprehensive learning, increasing the weights of the rainfall indicators that are neglected by the unoptimized model, and these weight changes improve the accuracy of the model. The research revealed the impacts of different flood-inducing factors on flooding and the influence of indicator weight changes on model accuracy.

In 2022, the United Nations launched the “Early Warnings for All” (EW4All) initiative, which aims to ensure that early warning systems are available to all inhabitants of the world by 2027. It is a response to the fact that climate-related disasters are increasing and many countries still do not have effective systems in place. For EW4All, the World Meteorological Organization (WMO) has selected 30 countries whose meteorological and hydrological services will be supported to strengthen their monitoring capacities for recurrent climate-related hazards in their country. This work aims to assess the impact of disasters in the period from January 2000 to November 2024 for the 30 EW4All countries using a composite scale representing a combination of different impact indicators. The unique feature of this work is that the different societal impacts are summarized in a single index. In this way, a comparative assessment of impacts between countries can be made, which can serve as a basis for action beyond that undertaken by the WMO and purely academic interest. This is a tool that can help decision makers to implement risk management measures. For this study, we selected the Emergency Events Database (EM-DAT) from the available global datasets because it is the only dataset with a time series long enough to fulfill the statistical criteria of this study and also uses a common disaster recording protocol for all countries.

The intensification of extreme rainfall events, driven by climate change, has exacerbated hydrological hazards in Brazil, affecting millions of people. This study investigated how previous disaster experience influences individuals' self-preparedness and key elements of community resilience, following the model by Norris et al. (2008), which includes information, social capital, community competence, and economic resources. A total of 1064 responses to a structured questionnaire survey was analyzed using the χ² test to identify statistical associations between behavioral variables and disaster experience. The results show that individuals with prior disaster experience report higher levels of perceived individual preparedness, concern about the impacts of heavy rainfall, and greater sensitivity to climate change. However, direct experience did not translate into greater confidence in institutional measures or stronger perceptions of community preparedness, revealing weaknesses in information flows and social cohesion. All interpretations are presented strictly in relation to the sample and do not imply population-level generalization. The study reinforces the relevance of experiential learning in shaping risk perception and preparedness, while highlighting contextual constraints associated with Brazil's socio-spatial inequalities and uneven governance capacities. These findings underscore the need for policies that strengthen institutional credibility, participatory risk communication, and community-level preparedness structures. The findings contribute to advancing risk management strategies in vulnerable contexts.