Migration patterns and health risk assessment of polycyclic aromatic compounds in typical coal fire source.

The Wuda coal fire in Inner Mongolia, China, is a global catastrophic event. It emits a huge volume of organic pollutants, including polycyclic aromatic compounds (PACs), which are widely concerning due to their physiological toxicity and environmental persistence. However, there is no systematic study on the enrichment and migration patterns of PACs emitted from coal fires. Here, we compared samples from coal fire sponges and surrounding soil, and analyzed 47 PACs using GC × GC-TOFMS. Data analysis showed that the average content of 16 polycyclic aromatic hydrocarbons (16PAHs) in the coal fire sponge was 15400.65 ng/g, which is about 4.2 times higher than that in the surrounding soil. Meanwhile, 31 PACs were detected at levels far exceeding those of 16PAHs. The distribution pattern of PACs showed that coal fire sources are more likely to produce and store 16PAHs while surrounding soils are more likely to be enriched with PAH derivatives. The cancer risk assessment revealed a significant cancer risk in both the coal fires and the surrounding soil. The formation mechanism of oxygenated PAHs was also explored, and it was found that coal fires emit 16PAHs and alkylated PAHs, which oxidize to form oxygenated PAHs during migration to surrounding soils. The value of naphthaldehyde/alkylated naphthalene (< 2) can be referenced as characteristic markers of coal fire pollution. This provides a new perspective on the sources of PACs in the current environment.

Evidence of increasing wildfire damage with decreasing property price in Southern California fires.

Across the Western United States, human development into the wildland urban interface (WUI) is contributing to increasing wildfire damage. Given that natural disasters often cause greater harm within socio-economically vulnerable groups, research is needed to explore the potential for disproportionate impacts associated with wildfire. Using Zillow Transaction and Assessment Database (ZTRAX), hereafter "Zillow", real estate data, we explored whether lower-priced structures were more likely to be damaged during the most destructive, recent wildfires in Southern California. Within fire perimeters occurring from 2000-2019, we matched property price data to burned and unburned structures. To be included in the final dataset, fire perimeters had to surround at least 25 burned and 25 unburned structures and have been sold at most seven years before the fire; five fires fit these criteria. We found evidence to support our hypothesis that lower-priced properties were more likely to be damaged, however, the likelihood of damage and the influence of property value significantly varied across individual fire perimeters. When considering fires individually, properties within two 2003 fires-the Cedar and Grand Prix-Old Fires-had statistically significantly decreasing burn damage with increasing property value. Occurring in 2007 and later, the other three fires (Witch-Poomacha, Thomas, and Woolsey) showed no significant relationship between price and damage. Consistent with other studies, topographic position, slope, elevation, and vegetation were also significantly associated with the likelihood of a structure being damaged during the wildfire. Driving time to the nearest fire station and previously identified fire hazard were also significant. Our results suggest that further studies on the extent and reason for disproportionate impacts of wildfire are needed. In the meantime, decision makers should consider allocating wildfire risk mitigation resources-such as fire-fighting and wildfire structural preparedness resources-to more socioeconomically vulnerable neighborhoods.

Forest fire susceptibility assessment under small sample scenario: A semi-supervised learning approach using transductive support vector machine.

Forest fires threaten global ecosystems, socio-economic structures, and public safety. Accurately assessing forest fire susceptibility is critical for effective environmental management. Supervised learning methods dominate this assessment, relying on a substantial dataset of forest fire occurrences for model training. However, obtaining precise forest fire location data remains challenging. To address this issue, semi-supervised learning emerges as a viable solution, leveraging both a limited set of collected samples and unlabeled data containing environmental factors for training. Our study employed the transductive support vector machine (TSVM), a key semi-supervised learning method, to assess forest fire susceptibility in scenarios with limited samples. We conducted a comparative analysis, evaluating its performance against widely used supervised learning methods. The assessment area for forest fire susceptibility lies in Dayu County, Jiangxi Province, China, renowned for its vast forest cover and frequent fire incidents. We analyzed and generated maps depicting forest fire susceptibility, evaluating prediction accuracies for both supervised and semi-supervised learning methods across various small sample scenarios (e.g., 4, 8, 12, 16, 20, 24, 28, and 32 samples). Our findings indicate that TSVM exhibits superior prediction accuracy compared to supervised learning with limited samples, yielding more plausible forest fire susceptibility maps. For instance, at sample sizes of 4, 16, and 28, TSVM achieves prediction accuracies of approximately 0.8037, 0.9257, and 0.9583, respectively. In contrast, random forests, the top performers in supervised learning, demonstrate accuracies of approximately 0.7424, 0.8916, and 0.9431, respectively, for the same small sample sizes. Additionally, we discussed three key aspects: TSVM parameter configuration, the impact of unlabeled sample size, and performance within typical sample sizes. Our findings support semi-supervised learning as a promising approach compared to supervised learning for forest fire susceptibility assessment and mapping, particularly in scenarios with small sample sizes.

Integrating meteorological and geospatial data for forest fire risk assessment.

Understanding the factors that cause fire is crucial for minimizing the fire risk. In this research, a comprehensive approach was adopted to recognize factors influencing forest fires. Golestan National Park (GNP) was considered as a representative area with a humid climate in this study. Initially, using the Multi-Criteria Evaluation Method, a fire risk map was created by analyzing natural and human factors, and vulnerable areas were identified. Then, the relationship between key elements such as meteorological conditions, Land Surface Temperature (LST), and precipitation, with the occurrence of fire in different years was investigated. CHIRPS and Landsat data were utilized to assess LST changes and precipitation. 23-year changes in fire occurrence areas in GNP were acquired using MODIS products. The results of the data analysis showed that the highest number of fires occurred in forest areas, and in the fire risk prediction map, the extremely high-risk class is completely consistent with the ground truth data. The assigned weights, derived from expert opinions, highlight the substantial significance of elevation, and distance from roads and settlements. Additionally, the effectiveness of the model in providing reliable forecasts for fire risks in GNP is highlighted by the ROC curve with an AUC value of 0.83. Forest fires within GNP exhibit a distinct seasonality, predominantly occurring from July to December. During the warmer months, by coinciding with summer excursions, human activities may contribute to the ignition of fires. In 2013 and 2014, rising fire incidents correlated with elevated temperatures, hinting at a potential connection. GNP fires showed an upward trend with higher monthly LST and a downward trend with increased annual precipitation. The results showed that there is a relationship between LST, precipitation, and the occurrence of fire in GNP. Approximately 176.15 ha of GNP's forest areas have been destroyed by fires over the last two decades. This research demonstrated that there is a dynamic interaction between environmental conditions and fire incidents. By considering these factors, managers and environmental planners can develop effective strategies for managing and preventing forest fire risks.

[Fire resistance of 15 main economic tree species in Liangshan Prefecture, Sichuan, China.] / 凉山州15ç§ä¸»è¦ç»æµŽæ ‘ç§çš„é˜²ç«æ€§èƒ½.

Liangshan Prefecture is one of the three major forest areas in Sichuan Province and one of the three major disaster areas of forest fire. We measured the physicochemical properties and combustion performances of different organs (leaves and branches) of 15 main economic tree species in Liangshan, and analyzed the bioecology characteristics, silviculture characteristics and value characteristics of different tree species. We investigated the fire resistance of different tree species to screen out fire-resistant species suitable for economic forest development in Liangshan Prefecture, and improve the biological fire prevention ability. The seven physicochemical properties and combustion performances indices of 15 tree species showed significant differences. Except for crude ash and lignin, the weights of moisture content, caloric value, ignition point, crude fat, and crude fibre of leaves were higher than those of branches. Crude fibre index of leaves (9.6%) and the crude ash index of branches (9.9%) were the highest weight indices of the two organs, respectively. Based on the fire resistance, we divided all the species into three classes, i.e., class Ⅰ (excellent fire-resistance trees) Juglans regia and Morus alba; class Ⅱ (better fire-resistant trees) Sapium sebiferum, Mangifera indica, Phyllanthus emblica, Eriobotrya japonica, Ligustrum lucidum, Castanea mollissima, and Punica granatum; class Ⅲ (poor fire-resistant trees) Pinus armandii, Illicium simonsii, Morella rubra, Sapindus mukorossi, Olea europaea and Camellia oleifera. J. regia and M. alba had fireproof solid performance and could be used as the preferred species for fireproof economic forest in Liangshan region. It was suggested that to use class Ⅰ to Ⅱ fire-resistant tree species built the main fireproof isolated forest belt, and pay attention to fire prevention after planting class Ⅲ tree species in a large area.

Multiple social and environmental factors affect wildland fire response of full or less-than-full suppression.

Wildland fire incident commanders make wildfire response decisions within an increasingly complex socio-environmental context. Threats to human safety and property, along with public pressures and agency cultures, often lead commanders to emphasize full suppression. However, commanders may use less-than-full suppression to enhance responder safety, reduce firefighting costs, and encourage beneficial effects of fire. This study asks: what management, socioeconomic, environmental, and fire behavior characteristics are associated with full suppression and the less-than-full suppression methods of point-zone protection, confinement/containment, and maintain/monitor? We analyzed incident report data from 374 wildfires in the United States northern Rocky Mountains between 2008 and 2013. Regression models showed that full suppression was most strongly associated with higher housing density and earlier dates in the calendar year, along with non-federal land jurisdiction, regional and national incident management teams, human-caused ignitions, low fire-growth potential, and greater fire size. Interviews with commanders provided decision-making context for these regression results. Future efforts to encourage less-than-full suppression should address the complex management context, in addition to the biophysical context, of fire response.

Bridging scales for landscape-level wildfire adaptation: A case study of the Kittitas Fire Adapted Communities Coalition.

Federal-level strategies or guidance for addressing wildfire risk encourage adaptation activities that span progressively larger scales, often focusing on landscape-level action that necessitates coordination between decision-makers and socially diverse communities. Collaborative organizations are increasingly explored as one approach for coordinating local efforts that address wildfire risk and adaptation, offering a platform for scaling and adjusting federal and state guidance that align with the needs of local landscapes. We conducted semi-structured interviews with members and supporters of the Kittitas Fire Adapted Communities Coalition (KFACC) and later facilitated two workshops at the behest of the organization. The goal of our interviews and workshops were to better understand how organizations such as KFACC emerge, function, and evolve in complex social and ecological landscapes, with a focus on their role in addressing landscape-level wildfire adaptation. We use an existing theoretical analogy of fire adaptation that crosses institutional and physical scales to help conceive of lessons from in-depth analysis of KFACC functioning. We found that KFACC originated from a need to establish a shared local mission for fire adaptation and a recognition that federal and state initiatives surrounding wildfire management needed further contextualization to be effective among diverse local social conditions. Later organizational foci included identifying key audiences for targeted adaptation efforts, including the identification of key messages and communities where specific mitigation actions might be needed. KFACC members were effective in strategically advocating for fire adaptation resources and policies at broader scales that might increase adaptation within Kittitas County, including caveats to local planning efforts designed for wildfire risk reduction. Likewise, the organization had begun to focus on tailoring mitigation efforts to different communities in the landscape as an effective means of catalyzing sustained, realistic fire adaptation actions. We suggest that organizations like KFACC are well-positioned to act as "board hoppers" who can integrate community-based needs into wildfire management, but caution that the functioning and 'niche' of such organizations may require strategic development or regular reflection on organizational goals.

Global population exposure to landscape fire air pollution from 2000 to 2019.

Wildfires are thought to be increasing in severity and frequency as a result of climate change1-5. Air pollution from landscape fires can negatively affect human health4-6, but human exposure to landscape fire-sourced (LFS) air pollution has not been well characterized at the global scale7-23. Here, we estimate global daily LFS outdoor fine particulate matter (PM2.5) and surface ozone concentrations at 0.25° × 0.25° resolution during the period 2000-2019 with the help of machine learning and chemical transport models. We found that overall population-weighted average LFS PM2.5 and ozone concentrations were 2.5 µg m-3 (6.1% of all-source PM2.5) and 3.2 µg m-3 (3.6% of all-source ozone), respectively, in 2010-2019, with a slight increase for PM2.5, but not for ozone, compared with 2000-2009. Central Africa, Southeast Asia, South America and Siberia experienced the highest LFS PM2.5 and ozone concentrations. The concentrations of LFS PM2.5 and ozone were about four times higher in low-income countries than in high-income countries. During the period 2010-2019, 2.18 billion people were exposed to at least 1 day of substantial LFS air pollution per year, with each person in the world having, on average, 9.9 days of exposure per year. These two metrics increased by 6.8% and 2.1%, respectively, compared with 2000-2009. Overall, we find that the global population is increasingly exposed to LFS air pollution, with socioeconomic disparities.

Costs of preventing and supressing wildfires in Victoria, Australia.

Land managers around the world are increasingly under pressure to demonstrate that the actions being used to moderate wildfire risk are effective and cost-efficient. However, little research to date has focused on determining cost-efficiency of management actions or identified the factors which increase the costs of performing such actions. Here, we aimed to identify the key drivers of cost for fuel management (prescribed burning, mulching, and slashing), fuel breaks, and suppression using data from the state of Victoria, Australia. We utilise generalised additive models to understand how environmental factors, terrain, location, and management decisions influence the cost of implementing wildfire management efforts. These models show that cost per unit declines as the area treated or the area of the fire increases for all four management approaches. Therefore, preventative, and responsive management actions represent economies of scale that reduce in cost with larger treatments. We also found that there were regional differences in the cost of fuel management and fuel breaks, potentially related to the structure of resourcing treatments in each region and the availability of land on which it is feasible to implement management. Cost of suppression per fire increased with the number of fire fighters and when there were more fires occurring concurrently in the landscape. Identifying the key drivers of cost for preventative and responsive management actions could enable managers to allocate resources to these actions more efficiently in future. Understanding drivers of cost-efficiency could be critical for adapting management to shifts in wildfire risk, particularly given climate change will alter the window in which it is safe to apply some preventative fuel management actions and reduce suppression effectiveness.

Perceptions about massive environmental impacts: a Brazilian study case.

The year 2019 brought three such impacts of high socio-environmental proportions in Brazil: the dam collapse in Brumadinho, oil spills on the coast, and fires in the Amazon. We investigated the Brazilian population's perceptions of the country's overall environmental situation, the degree to which Brazilians felt affected by these impacts considering personal and social factors, and the entities they held responsible for these disasters. Through Facebook's social media networks, we disseminated structured online surveys for Brazilian citizens above 18 years. Educational background explained how much the 775 respondents felt affected by the three evaluated events. Age was an explanatory factor for the degree to which the respondents felt affected by the dam collapse, and proximity to the disasters, while income levels were for the dam collapse and the fires in the Amazon. The government, criminal activity, and private companies were considered to be the main responsible for these three impacts. This perception reflects the series of changes in the country's environmental laws and protections that threaten biodiversity and the environment.

Community Fire Risk Reduction: Longitudinal Assessment for HomeSafe Fire Prevention Program in Canada.

(1) Background: Residential fires represent the third leading cause of unintentional injuries globally. This study aims to offer an overview and a longitudinal evaluation of the HomeSafe program implemented in Surrey in 2008 and to assess its effectiveness in mitigating fire-related outcomes. (2) Methods: Data were collected over a 12-year period (2008-2019). Assessed outcomes comprised frequency of fire incidents, residential fires, casualties, functioning smoke alarms, and contained fires. The effectiveness of each initiative was determined by comparing the specific intervention group outcome and the city-wide outcome to the pre-intervention period. (3) Results: This study targeted 120,349 households. HomeSafe achieved overwhelming success in decreasing fire rates (-80%), increasing functioning smoke alarms (+60%), increasing the percentage of contained fires (+94%), and decreasing fire casualties (-40%). The study findings confirm that the three most effective HomeSafe initiatives were firefighters' visits of households, inspections and installations of smoke alarms, and verifications of fire crew alarms at fire incidents. Some initiatives were less successful, including post-door hangers (+12%) and package distribution (+15%). (4) Conclusions: The HomeSafe program effectively decreased the occurrence and magnitude of residential fires. Lessons learned should be transferred to similar contexts to implement an evidence-based, consistent, and systematic approach to sustainable fire prevention initiatives.

Wildfire-related PM2.5 and health economic loss of mortality in Brazil.

BACKGROUND: Wildfire imposes a high mortality burden on Brazil. However, there is a limited assessment of the health economic losses attributable to wildfire-related fine particulate matter (PM2.5). METHODS: We collected daily time-series data on all-cause, cardiovascular, and respiratory mortality from 510 immediate regions in Brazil during 2000-2016. The chemical transport model GEOS-Chem driven with Global Fire Emissions Database (GFED), in combination with ground monitored data and machine learning was used to estimate wildfire-related PM2.5 data at a resolution of 0.25°â€ˆ× 0.25°. A time-series design was applied in each immediate region to assess the association between economic losses due to mortality and wildfire-related PM2.5 and the estimates were pooled at the national level using a random-effect meta-analysis. We used a meta-regression model to explore the modification effect of GDP and its sectors (agriculture, industry, and service) on economic losses. RESULTS: During 2000-2016, a total of US$81.08 billion economic losses (US$5.07 billion per year) due to mortality were attributable to wildfire-related PM2.5 in Brazil, accounting for 0.68% of economic losses and equivalent to approximately 0.14% of Brazil's GDP. The attributable fraction (AF) of economic losses due to wildfire-related PM2.5 was positively associated with the proportion of GDP from agriculture, while negatively associated with the proportion of GDP from service. CONCLUSION: Substantial economic losses due to mortality were associated with wildfires, which could be influenced by the agriculture and services share of GDP per capita. Our estimates of the economic losses of mortality could be used to determine optimal levels of investment and resources to mitigate the adverse health impacts of wildfires.

Quantifying the premature mortality and economic loss from wildfire-induced PM2.5 in the contiguous U.S.

Emissions from wildfires worsen air quality and can adversely impact human health. This study utilized the fire inventory from NCAR (FINN) as wildfire emissions, and performed air quality modeling of April-October 2012, 2013, and 2014 using the U.S. Environmental Protection Agency CMAQ model under two cases: with and without wildfire emissions. This study then assessed the health impacts and economic values attributable to PM2.5 from fires. Results indicated that wildfires could lead annually to 4000 cases of premature mortality in the U.S., corresponding to $36 billion losses. Regions with high concentrations of fire-induced PM2.5 were in the west (e.g., Idaho, Montana, and northern California) and Southeast (e.g., Alabama, Georgia). Metropolitan areas located near fire sources, exhibited large health burdens, such as Los Angeles (119 premature deaths, corresponding to $1.07 billion), Atlanta (76, $0.69 billion), and Houston (65, $0.58 billion). Regions in the downwind of western fires, although experiencing relatively low values of fire-induced PM2.5, showed notable health burdens due to their large population, such as metropolitan areas of New York (86, $0.78 billion), Chicago (60, $0.54 billion), and Pittsburgh (32, $0.29 billion). Results suggest that impacts from wildfires are substantial, and to mitigate these impacts, better forest management and more resilient infrastructure would be needed.

Incorporating fire-smartness into agricultural policies reduces suppression costs and ecosystem services damages from wildfires.

In southern Europe, land abandonment and an unbalanced investment toward fire suppression instead of prevention has gradually increased wildfire risk, which calls for a paradigm change in fire management policies. Here we combined scenario analysis, fire landscape modelling, and economic tools to identify which land-use policies would reduce the expected wildfire-related losses in the Transboundary Biosphere Reserve 'Gerês-Xurés' (Spain-Portugal). To do so, we applied the least-cost-plus-net-value-change approach and estimated net changes in wildfire damages based on their implications for the 2010-2050 period and five ecosystem services: agriculture, pasture, timber, recreation and climate regulation. Four land-use scenarios were considered: (1) Business as Usual (BAU); (2) fire-smart, fostering more fire-resistant (less flammable) and/or fire-resilient landscapes (fire-smart); (3) High Nature Value farmlands (HNVf), wherein the abandonment of extensive agriculture is reversed; and (4) a combination of HNVf and fire-smart. HNVf is the best scenario for suppression cost savings, but it generates the lowest net present value of societal benefits from climate regulation. In fact, the most efficient scenario with the lowest societal discounted net suppression costs and change on ecosystem services damages is the HNVf + fire-smart scenario, as it also generates suppression cost savings from agricultural expansion, and lead to a significant reduction in damages on timber and recreational benefits. Therefore, reverting land abandonment through recultivation and promoting fire-resistant tree species is the most efficient way to reduce wildfire hazard. In this sense, payments for ecosystem services should reward farmers and landowners for their role in wildfire prevention. This study improves the understanding of the financial and societal benefits derived from reducing fire suppression spending and ecosystem services damage by undertaking fire-smart land-use strategies, which can be essential to enhance local stakeholders' support for Payments of Ecosystem Services policies for wildfire prevention.

Risk assessment of methanol storage tank fire accident using hybrid FTA-SPA.

Fire accidents in storage tanks are of great importance due to the difficulty in extinguishing and ease of spread to nearby products. This study aimed to introduce a framework based on FTA-based Set Pair Analysis (SPA) established via experts' elicitation to identify and assess the risk of storage tank fire. In the quantitative FTA of a system, sufficient data are only sometimes available to calculate the failure probability of the system appertains to study. Thus, the obtained result of the SPA added new value to the Basic Events (BEs) and estimated top event. To illustrate the applicability of the proposed approach, a fault tree of the methanol storage tank fire is performed and analyzed BEs. According to the obtained results, the fire accident was computed by 48 BEs, and the occurrence probability value of the top event was estimated 2.58E-1/year. In addition, the most crucial paths that led to the fire accident are listed in this study. The proposed approach established in the present study can assist decision-makers in determining where to take preventative or appropriate action on the storage tank system. Moreover, it can be adjusted for various systems with limited manipulation.

How much does it cost to mitigate soil erosion after wildfires?

Wildfires usually increase the hydrological and erosive response of forest areas, carrying high environmental, human, cultural, and financial on- and off-site effects. Post-fire soil erosion control measures have been proven effective at mitigating such responses, especially at the slope scale, but there is a knowledge gap as to how cost-effective these treatments are. In this work, we review the effectiveness of post-fire soil erosion mitigation treatments at reducing erosion rates over the first post-fire year and provide their application costs. This allowed assessing the treatments' cost-effectiveness (CE), expressed as the cost of preventing 1 Mg of soil loss. This assessment involved a total of 63 field study cases, extracted from 26 publications from the USA, Spain, Portugal, and Canada, and focused on the role of treatment types and materials, and countries. Treatments providing a protective ground cover showed the best median CE (895 $ Mg-1), especially agricultural straw mulch (309 $ Mg-1), followed by wood-residue mulch (940 $ Mg-1) and hydromulch (2332 $ Mg-1). Barriers showed a relatively low CE (1386 $ Mg-1), due to their reduced effectiveness and elevated implementation costs. Seeding showed a good CE (260 $ Mg-1), but this reflected its low costs rather than its effectiveness to reduce soil erosion. The present results confirmed that post-fire soil erosion mitigation treatments are cost-effective as long as they are applied in areas where the post-fire erosion rates exceed the tolerable erosion rate thresholds (>1 Mg-1 ha-1 y-1) and are less costly than the loss of on- and off-site values that they are targeted to protect. For this reason, the proper assessment of post-fire soil erosion risk is vital to ensure that the available financial, human and material resources are applied appropriately.

Application of machine learning for the low-cost prediction of soot concentration in a turbulent flame.

The second most potent forcer of climate change, soot, has severe harmful effects on both human health and the environment. Accurate numerical modeling of soot formation is extremely complex and has a high computational cost due to its dependence on many physical and chemical interactions, specifically in turbulent flames. The high computational cost of coupling chemistry, fluid dynamics, thermodynamics, and heat transfer raise the need for a novel, precise, and computationally cost-effective numerical technique for predicting soot concentrations. This study applies machine learning (ML) to predict soot formation in a turbulent flame. It has been discovered that the local soot volume fraction is correlated to the histories of gas properties strongly correlative to soot formation and oxidation. A library with the Lagrangian temporal histories of soot-containing fluid parcels is created from turbulent diffusion flame data computed using direct numerical simulation (DNS). This library is then used to train an ML algorithm to predict soot volume fraction along randomly selected trajectories (pathlines) in the domain. The prediction capability is tested over 10% of the entire dataset, and it is seen that soot volume fraction can be predicted well along the selected pathlines with low error and computational cost. To describe quantitative results, the calculated R2 in the current work is equal to 0.92, which shows good accuracy of the predictions.

An Integrated Quantitative Risk Assessment Method for Underground Engineering Fires.

Fires are one of the main disasters in underground engineering. In order to comprehensively describe and evaluate the risk of underground engineering fires, this study proposes a UEF risk assessment method based on EPB-FBN. Firstly, based on the EPB model, the static and dynamic information of the fire, such as the cause, occurrence, hazard, product, consequence, and emergency rescue, was analyzed. An EPB model of underground engineering fires was established, and the EPB model was transformed into a BN structure through the conversion rules. Secondly, a fuzzy number was used to describe the state of UEF variable nodes, and a fuzzy conditional probability table was established to describe the uncertain logical relationship between UEF nodes. In order to make full use of the expert knowledge and empirical data, the probability was divided into intervals, and a triangulated fuzzy number was used to represent the linguistic variables judged by experts. The α-weighted valuation method was used for de-fuzzification, and the exact conditional probability table parameters were obtained. Through fuzzy Bayesian inference, the key risk factors can be identified, the sensitivity value of key factors can be calculated, and the maximum risk chain can be found in the case of known evidence. Finally, the method was applied to the deductive analysis of three scenarios. The results show that the model can provide realistic analysis ideas for fire safety evaluation and emergency management of underground engineering. The proposed EPB risk assessment model provides a new perspective for the analysis of UEF accidents and contributes to the ongoing development of UEF research.