ABSTRACT Recent tsunamigenic earthquakes in Japan have highlighted the emerging fire hazard triggered by tsunami inundation and its impact on tsunami vertical evacuation (TVE) structures. This new type of fire following earthquake, referred to as “tsunami fires,” may be a potential universal hazard that tsunami‐prone countries face; however, it has not been considered in earthquake‐related risk management. As part of initiatives to evolve regional fire‐following‐earthquake risk assessments, this study focuses on modeling the uncertainty of tsunami fire occurrences. An analysis of 92 tsunami fires following the 2011 Tohoku and 2024 Noto Peninsula earthquakes reveals the following. (1) In areas of high tsunami intensity where numerous wooden houses and automobiles are washed away, tsunami fires may mostly start from tsunami debris that is washed away and accumulated by the tsunami or may start while tsunami debris is being transported by the tsunami and then accumulated while burning; conversely, in areas of low tsunami intensity where few wooden houses and automobiles are washed away, tsunami fires may mostly start from electricity‐related sources in tsunami‐inundated buildings and automobiles. (2) The probability of tsunami fire ignition is approximately 2–6 fires on average per 10,000 tsunami‐inundated buildings, depending on the building washout ratio, and varies from approximately 1/3 to 3 times that of the average within one sigma due to unobserved regional differences. Recommendations to prevent losses of life in TVE structures caused by tsunami fires are presented. These findings will help us better understand and prepare for tsunami fires as earthquake‐related cascading hazards.

Abstract The construction of large interchange tunnels addresses urban traffic congestion but introduces complex operational safety challenges, with fire being among the most severe hazards. To mitigate tunnel fire risks and enhance emergency response capabilities, this study proposes a rapid prediction model for tunnel fire development based on Long Short-Term Memory (LSTM) and Fully Connected Neural Network (FCNN) models. The model requires only basic inputs—including heat release rate (HRR), fire location, longitudinal ventilation, and tunnel geometry—to quickly forecast key fire parameters, such as ceiling temperature, temperature at 2 m below the ceiling, and visibility over time. Given the high cost and data scarcity of full-scale tunnel fire experiments, an extensive fire scenario database was generated using Fire Dynamics Simulator (FDS) simulations under varying fire locations, HRRs, and ventilation conditions. Time-series data extracted from these simulations were used to train and validate the neural networks. Comparative analysis showed that the LSTM model achieved a coefficient of determination (R 2 ) of 0.8952, outperforming the FCNN model (R 2 = 0.8721) in capturing temporal fire dynamics. The predictive capability of the neural network model was further validated against theoretical ceiling temperature models derived from the FDS data, confirming its feasibility for practical application in guiding fire rescue and ventilation strategies. It should be noted that the model's performance is inherently dependent on the accuracy and scope of the underlying simulation data, and its reliability may decrease when extrapolating beyond the trained parameter ranges.

Sealing tunnel portals is widely recognized as a pivotal strategy for mitigating fire hazards in tunnel safety management. Nevertheless, the interplay between fire source locations—both longitudinally and transversely—and its impact on flame behavior and ceiling temperature profiles within enclosed structures has not been fully elucidated. Utilizing a 1:15 reduced-scale rectangular tunnel model, this research investigates how varying the fire source position affects the maximum ceiling temperature under enclosed scenarios. Dimensionless parameters, including the longitudinal dimensionless distance D and transverse dimensionless distance Z′, were derived through dimensional analysis. Observations indicate that as the fire approaches the enclosed end, the flame initially leans toward the boundary, peaking in inclination at D = 0.73, and subsequently exhibits a “wall-attached combustion” pattern due to wall confinement. While lateral displacement of the fire source pushes the high-temperature zone toward the corresponding side wall, the longitudinal temperature rise follows a non-monotonic pattern: declining continuously in in Region I (0 ≤ D ≤ 0.73) and rebounding in Region II (0.73 < D < 1). Based on these findings, a dimensionless prediction model incorporating heat release rate (HRR), transverse offset, and longitudinal fire location was developed. Furthermore, a thermal accumulation factor was introduced to refine the predictive model in Region II. The results offer theoretical insights to support fire protection design and risk assessment in enclosed tunnels.

ABSTRACT Timber lounge bridges, as important cultural heritage structures with roofed and partially enclosed corridors, are highly vulnerable to fire because of their combustible materials, intricate timber structures, and complex environments. However, fire risk assessment is challenging because of information uncertainty, limited data availability, and complex interdependencies among risk factors. To address these issues, this study establishes a fire risk assessment framework based on Fuzzy Bayesian Networks (FBN). In this framework, a hierarchical indicator system includes ignition sources, structural vulnerabilities, and emergency response capacities while explicitly modelling risk failure propagation. Prior probabilities are derived from expert judgments through triangular fuzzy numbers and similarity aggregation, and conditional relationships are quantified through the noisy-OR model. The proposed model was validated through a case study of the Liuzhai Bridge in Zhejiang Province, China. The model estimates a fire risk probability of 17% and identifies key sensitive factors and major risk failure propagation pathways. The results align well with on-site risk characteristics and demonstrate the effectiveness of the model through multi-faceted validation. This study shows that fuzzy probabilistic modelling provides a practical framework for early warning, risk prediction, and targeted mitigation strategies in timber heritage preservation.

The dead fine fuel moisture content (DFFMC) directly affects forest fire occurrence and spread. Accurate DFFMC prediction is key to estimating forest fire risk and behavior. The well-fitting machine learning (ML)-based meteorological factor regression models are a focus of DFFMC prediction modeling. Nevertheless, this method’s reliance on a considerable amount of training data and limited extrapolation hinders its potential for extensive implementation in practice. To improve the prediction accuracy of the model in the context of limited training data volumes and interspecies and spatial extrapolated predictions, this study proposed a novel DFFMC prediction method based on a knowledge-embedded neural network (KENN). By integrating the partial differential equation (PDE) of the meteorological response of forest fuel moisture content into a multilayer perceptron (MLP), the KENN utilizes prior physical knowledge and posterior observational data to determine the relationship between meteorology and moisture content. Data from Mongolian oak, white birch, and larch were collected to evaluate model performance. Compared with three representative ML algorithms for DFFMC prediction—random forest (RF), long short-term memory networks (LSTM), and MLP—the KENN can efficiently reduce training data volume requirements and improve extrapolation prediction accuracy within the investigated fire season, thereby enhancing the usability of ML-based DFFMC prediction methods.

Quantifying the impacts of typical environmental factors on oil depot fire and explosion risk

Integrated multi-hazard assessment for climate-resilient watershed management: A transferable prioritization framework from Nepal's Mid-Hills.

Remote sensing diagnosis of Forest fire risk based on state-trend characteristics using machine learning models

Developing lightweight multifunctional materials with both strong electromagnetic microwave (EMW) absorption and robust flame retardancy is essential for addressing electromagnetic interference and fire risks in modern electronics. In this work, ZIF‐67@NH 2 ‐MIL‐88B(Fe)/MXene‐modified melamine foam (ZIF@MIL/MXene/MF) composite aerogels were constructed to achieve synergistic dielectric loss and magnetic loss mechanisms. The optimized ZIF@MIL/MXene/MF‐5 exhibited outstanding EMW absorption, with a minimum reflection loss (RL min ) of −55.82 dB and an effective absorption bandwidth of 4.52 GHz, along with a radar cross‐section reduction of 15.27 dB·m 2 . The superior absorption originates from conductive network pathways, abundant heterogeneous interfaces for polarization loss, dipole relaxation, and magnetic loss from CoFe alloy nanoparticles, combined with excellent impedance matching. Additionally, the composite demonstrated excellent fire safety, with peak heat release rate, total heat release, and total smoke production reduced by 34.0%, 64.6%, and 87.9%, respectively, owing to catalytic graphitization and the formation of a compact char barrier. These results demonstrate an effective strategy for designing lightweight multifunctional aerogels for next‐generation electromagnetic protection and fire‐safe engineering applications.

Small- and medium-sized industrial complexes that frequently use fire pits are continually exposed to fire hazards due to worker negligence or equipment malfunctions. In this paper, we propose a multimodal fire alarm system that simultaneously recognizes flames, fire pits, smoke, and worker movements to assess fire risk in real time and provide timely warnings. The proposed system incorporates a complex sensor module that includes an RGB optical camera, a thermal imaging sensor, a temperature and humidity sensor, a gas sensor, and a dust sensor. A deep learning model performs fusion analysis of RGB and thermal images to learn flame and heat patterns. Moreover, the system compares gas and dust sensor outputs to identify risk factors such as smoke, dust, and gas leaks. Further, it monitors worker behavior patterns in real time, enabling the early detection of careless approaches or dangerous behaviors around fire pits. A sensor fusion algorithm distinguishes between fire-induced and normal heat through RGB-thermal image comparison. Recurrent neural network filtering is employed to ensure time-series consistency, minimizing false positives and false negatives. Experimental results confirm that the proposed system maintained high detection accuracy and stability under various working environments and fire tool usage conditions. Thus, the proposed system successfully enables real-time fire risk assessment and alarm functions.

This paper examines the state of afforestation within a forest clearing servicing a hard-to-reach section of an overhead power line located near the “Malo Buchino” tunnel, part of the “Struma” highway. The study includes data acquisition through aerial photogrammetric surveying using a multispectral sensor integrated into a UAV. The spectral behavior of the foliage, part of the vegetation encroaching upon the right-of-way zone servicing the overhead power line, has been analyzed. The data from the study assist in assessing the risk of forest fires caused by the proximity of vegetation to the power conductor, as well as in planning maintenance activities for forest clearings that service such types of technical infrastructure.

Due to the high flammability of materials used at all stages of the textile industry, fires are frequently observed in textile-based businesses. A factory fire can cause significant financial losses and may even lead to loss of life. Therefore, this study examines textile fires. Specifically fire risks encountered in cotton yarn manufacturing businesses were investigate and a Risk Priority Number (RPN) based risk analysis was conducted. The causes of fire incidents in workplaces and the resulting risks were examined. The stages of a fire initial, steady-state, and hot smoke were analyzed using the fuzzy FMEA method to assess control management. The study concluded that the danger of flame tongues, toxic gases in the smoke, and the risk of spreading during the initial phase are highly critical. The scale used in traditional FMEA is based on absolute values, and the lack of historical data presents challenges for occupational safety professionals conducting risk analyses. Fuzzy logic can provide more realistic results without relying on absolute values.

Forest fires are escalating in frequency, intensity, and economic impact under a warming climate, stressing the need for high‑resolution, proactive fire risk mapping. This paper proposes an end‑to‑end framework that fuses multi‑sensor remote sensing, meteorological reanalysis, topography, vegetation dynamics, human activity proxies, and fire history with deep learning models for spatially explicit fire risk prediction at daily to weekly lead times. We review the state of the art, describe a modular data engineering pipeline, compare candidate model families (CNN–LSTM/Temporal Convolution, Vision Transformers, U‑Net, Graph Neural Networks), and outline rigorous spatiotemporal validation protocols to avoid leakage. We present a complete experimental blueprint—feature engineering, class imbalance handling, metrics (AUC‑PR, TSS, brier, reliability), ablations, and uncertainty quantification—so that researchers and agencies can reproduce, adapt, and deploy in heterogeneous biomes. A reference implementation and open data recipe are described to facilitate translation to operations.

Elderly welfare institutions in Taiwan have experienced multiple severe fire incidents, with smoke inhalation accounting for the majority of fatalities. Hot smoke can rapidly propagate through interconnected ceiling spaces, complicating evacuation for residents with limited mobility who depend heavily on caregiving staff and external responders. Field inspections conducted in this study indicate that 82% of residents require assisted evacuation, underscoring the critical role of early detection, staff-mediated response, and effective smoke control. Drawing on disaster management theory, this study examines key determinants of fire safety performance in elderly welfare institutions, where caregiving staff are primarily trained in medical care rather than fire safety. A total of 64 licensed institutions in Tainan City were investigated through on-site inspections, structured checklist-based surveys, and statistical analyses of fire protection systems. In addition, a comparative review of building and fire safety regulations in Taiwan, the United States, Japan, and China was conducted to contextualize the findings. Using the defense-in-depth framework, this study proposes a three-layer fire safety strategy comprising (1) prevention of fire occurrence, (2) rapid fire detection and early suppression, and (3) containment of fire and smoke spread. From a sustainability perspective, this study conceptualizes fire safety in elderly welfare institutions as a problem of risk governance, illustrating how defense-in-depth can be operationalized as a governance-oriented framework for managing fire and smoke risks, safeguarding vulnerable older adults, and sustaining the resilience and continuity of long-term care systems in an aging society.

This study analyzes the relationship between fires and climate extremes in the Caatinga biome from 2012 to 2023 by integrating Fire Radiative Power (FRP) from VIIRS (S-NPP and NOAA-20), Vapor Pressure Deficit (VPD) and air temperature from ERA5, drought indices (SPI-1 and SPI-6), and heatwave events from the Xavier database. Daily percentiles of maximum (CTX90pct) and minimum (CTN90pct) temperatures were used to characterize heatwaves. Spatial and temporal dynamics of fire patterns were identified using the HDBSCAN algorithm, an unsupervised Machine Learning clustering method applied in three-dimensional space (latitude, longitude, and time). A marked seasonality was observed, with fire activity peaking from August to November, especially in October, when FRP reached ~1000 MW/h. The years 2015, 2019, 2021, and 2023 exhibited the highest fire intensities. A statistically significant upward trend in cluster frequency was detected (+1094.96 events/year; p < 0.001). Cross-correlations revealed that precipitation deficits (SPI) preceded FRP peaks by about four months, while VPD and air temperature exerted immediate positive effects. FRP correlated positively with heatwave frequency (r = 0.62) and negatively with SPI (r = −0.69). These findings highlight the high vulnerability of the Caatinga to compound drought and heat events, indicating that fire management strategies should account for both antecedent drought conditions, monitored through SPI, and real-time atmospheric dryness, measured by VPD, to effectively mitigate fire risks.

ABSTRACT Building engineering fire risk management faces growing challenges owing to rapid urbanisation and industrialisation. In dense urban areas, high-rise clusters, multi-functional zones, and interconnected infrastructure increase the complexity of fire spread, posing major threats to public safety. Accurate prediction of fire evolution is essential for effective risk mitigation. However, current methods often rely on subjective and static approaches, which are inefficient and lack scalability. Integrating knowledge graph with machine learning improves prediction accuracy and practical applicability. This study analyses 510 Chinese building engineering fire accidents (2000–2024), constructing a knowledge graph to integrate diverse urban data. Three models, logical convolutional neural network (L-CNN), random forest, and k-nearest neighbours, were applied for fire evolution prediction. The L-CNN model achieved the highest accuracy (83.29%) and lowest variance (0.27), demonstrating superior adaptability in complex fire scenarios. Further sensitivity analysis was conducted to rigorously assess the robustness of the predictive model outputs. This research supports early warning systems and emergency decision-making, advancing data-driven fire safety management.

The paper presents estimates of vegetation fire impact zones in the Far Eastern region of Russia. The study is based on taking into account data on the burning rate in modern conditions, as well as the distances from the fires to the nearest settlements, since most fires are anthropogenic in nature. The actual burning rate of vegetation in the Far Eastern Federal District was estimated by the relative number of fires per unit area of the subjects, on average per season, based on long-term data. Territorial units were grouped into zones by the degree of fire risks, dependent on different distances of the areas from settlements and roads. The largest number of fires in all subjects of the Far Eastern Federal District occurs within 9 km from the settlement. In Buryatia – within 15 km. In the Republic of Sakha (Yakutia), the dependence of fires on proximity to populated areas is associated with a complex of natural and anthropogenic factors. There is a direct relationship between proximity to populated areas and the frequency of fires. Regional peculiarities of anthropogenic impact on fire risks have been established. The southern regions of the Far East are subject to the greatest anthropogenic impact; they are: the Primorsky Territory, Amur and Jewish Autonomous regions.

Ensuring the preservation and sustainable functioning of forest stands is of key importance in countering the worsening climate change. By maintaining ecological balance, forest ecosystems ensure the stability of natural processes. In addition to industrial exploitation of forests, forest fires, massive outbreaks of insects, and the spread of diseases are the main factors in reducing forested areas. The Siberian silkworm (Dendrolimus sibiricus Tschetv.) and the Ussuri polygraph (Polygraphus proximus Blandford). An urgent area is the search for correlations between damage caused by entomological pests and the burning of forests, which will improve the prediction of fire danger in forest plantations and in the formation of sound fire protection measures. The study used data on forest fires for 2000-2023 using a well-known method for calculating the burn rate, as well as weather information (according to correlation analysis, a weak relationship was revealed – the coefficient is 0.23). The analysis of damage to forest stands was carried out according to the data of the Federal Forestry Agency of Russia "Rosleskhoz". Based on the analysis of quantitative indicators, it was revealed that in plantations affected by the Siberian silkworm (D. sibiricus T.), the probability of fires increases significantly 2 years after damage, and in fir plantations affected by the polygraph of the Ussuriysky (P. proximus B.), mainly in the 8th year. It is assumed that there is a tendency to increase the frequency and burning in damaged plantations at the current level of forest protection. To reduce the negative consequences, a set of preventive and economic measures developed by the forestry sector is necessary. The results obtained contribute to an improved understanding of the impact of entomological pests on forest fire risk and can be applied in the management of forest resources in the subtaiga zone of Central Siberia.

Improving the system of preventive management of forest fire risks is a global problem in the field of forest protection. The type of fire that occurs, its intensity and further spread in the territory of forest ecosystems are determined by the reserves and humidity of forest combustible materials. One of the most important components of the accumulation of combustible materials in stands is the parameters of wood litter (forest litter). At the same time, the dynamics of accumulation and features of stands that form the structure of wood litter are not taken into account sufficiently, both in the current regulatory and legal documents when determining the class of natural fire hazard in forests and in assessing forest fire risks. In this regard, the main goal of the work is to compile a tabular model characterizing the dynamics of accumulation of wood litter in stands of different ages growing in different types of forest growth conditions in the Voronezh region. To measure the thickness of the forest litter, a field survey was carried out and temporary test sites were laid in stands of different ages. When selecting the objects of study and conducting field work, the principles of combining random and systematic sampling were used, and generally accepted methods were used. To assess the preventive fire hazard in forests, a set of silvicultural and biological factors of forest fire risk in terms of accumulation of wood litter was analyzed. It was found that the most significant criteria for the formation of the total mass of wood litter are the proportion of coniferous species in the composition (F = 18) and the crown diameter (F = 12). The age group of stands (F = 9) and the type of forest growing conditions (F = 6) also have a significant impact on the intensity of wood litter accumulation. The compiled tabular model of the dynamics of wood litter accumulation (forest litter thickness) in pure coniferous and mixed stands with different proportions of coniferous species in the composition, for stands of different age groups growing in different TLUs allows using remote sensing methods in determining the dynamics of wood litter accumulation. A direct connection between the increase in forest litter thickness and the proportion of coniferous species in the composition has been revealed, which is an additional criterion for increasing the natural fire hazard in forests. It has also been established that forest litter accumulates more slowly in dry forest conditions than in fresh ones.

Purpose The purpose of this paper is to assess the risk of post-earthquake fire (PEF) in the historic city centre of Aveiro, Portugal, by integrating seismic vulnerability analysis, fire risk assessment and GIS-based spatial modelling. Through building surveys and scenario-based analysis, the study identifies ignition risks, fire spread potential and road blockage hazards following a seismic event. It further evaluates the impact of mitigation strategies, such as automatic gas shut-off valves, to enhance urban resilience. The findings aim to support emergency planning and inform public policies for disaster risk reduction in historic urban areas. Design/methodology/approach The study combines field surveys, photographic documentation and GIS mapping to assess PEF risk in Aveiro's historic centre. Seismic vulnerability was evaluated using index-based methods for masonry and reinforced concrete buildings. Fire risk was quantified using a modified ARICA fire risk index method, while PEF scenarios were developed by correlating damage levels with ignition risks. Fire spread was modelled using an elliptical propagation model considering wind and urban density. Road blockage risk was assessed based on building collapse potential. Mitigation measures, including automatic gas shut-off valves, were simulated to evaluate their impact on fire ignition and spread. Findings The study found that 51% of buildings in Aveiro's historic centre face moderate to very high PEF risk under EMS-98 intensity VIII, with masonry structures being significantly more vulnerable than reinforced concrete ones. Fire spread analysis showed rapid escalation, affecting 2% of buildings within 15 min and 19% within 30 min. Additionally, 23% of roads were classified as high risk for blockage, potentially hindering emergency response. Implementation of automatic gas shut-off valves reduced ignition risk in some buildings but had limited overall impact due to broad risk classification intervals. Key vulnerable zones were identified for targeted mitigation and planning. Originality/value This study presents a novel, integrated approach to PEF risk assessment at the urban scale, combining seismic vulnerability, fire risk, fire spread and road blockage analyses within a GIS framework. Applied to Aveiro's historic centre, it uniquely incorporates real building data and explores the impact of mitigation strategies such as automatic gas shut-off valves. The methodology's adaptability to other European historic urban areas enhances its relevance. By identifying high-risk zones and infrastructure weaknesses, the study provides valuable insights for urban planners, policymakers and emergency services aiming to improve resilience and reduce cascading disaster risks in historic city centres.