Pub Date : 2026-03-01Epub Date: 2025-10-23DOI: 10.1016/j.apr.2025.102784
Hamidreza Abediasl , Masoud Aliramezani , Charles Robert Koch , Mahdi Shahbakhti
Stringent regulations on real-driving emissions have been introduced to reduce the effect of tailpipe vehicular emissions on environmental pollution. The need to monitor emissions in real-driving conditions and across different driving cycles has underscored the importance of models for estimating emission rates. In this study, machine learning is employed to model commonly regulated tailpipe emissions (CO, UHC, NOx) based on real-time data obtained through on-board diagnostics (OBD) of vehicles. The models are trained using real-world tailpipe emission data and engine/vehicle operation data collected from three vehicles with various powertrains, including conventional gasoline engine, hybrid electric, and plug-in hybrid electric, under different ambient temperatures. Emphasis is placed on developing models capable of effectively estimating emissions during the cold phase of operation, which accounts for a significant portion of vehicular emissions, particularly in cold climates. The models are subsequently integrated into an intelligent fleet management system to enable real-time estimation of emissions using OBD data received from Internet of Things (IoT) modules installed on fleet vehicles.
{"title":"Machine learning-based estimation of vehicular emissions using on-board diagnostics data for intelligent fleet management","authors":"Hamidreza Abediasl , Masoud Aliramezani , Charles Robert Koch , Mahdi Shahbakhti","doi":"10.1016/j.apr.2025.102784","DOIUrl":"10.1016/j.apr.2025.102784","url":null,"abstract":"<div><div>Stringent regulations on real-driving emissions have been introduced to reduce the effect of tailpipe vehicular emissions on environmental pollution. The need to monitor emissions in real-driving conditions and across different driving cycles has underscored the importance of models for estimating emission rates. In this study, machine learning is employed to model commonly regulated tailpipe emissions (CO, UHC, NOx) based on real-time data obtained through on-board diagnostics (OBD) of vehicles. The models are trained using real-world tailpipe emission data and engine/vehicle operation data collected from three vehicles with various powertrains, including conventional gasoline engine, hybrid electric, and plug-in hybrid electric, under different ambient temperatures. Emphasis is placed on developing models capable of effectively estimating emissions during the cold phase of operation, which accounts for a significant portion of vehicular emissions, particularly in cold climates. The models are subsequently integrated into an intelligent fleet management system to enable real-time estimation of emissions using OBD data received from Internet of Things (IoT) modules installed on fleet vehicles.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102784"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135677","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2025-10-13DOI: 10.1016/j.apr.2025.102788
Baojun Yang , Lin Zhang , Danchen Yang , HuiYing Ma , Haifan Xie , Jiahui Hou , Ling Qiu , Tian Gao
High-density cities face severe PM2.5 particulate pollution issues, where landscape patterns significantly influence PM2.5 concentrations. Currently, there is a notable lack of comprehensive research on the integrated effects of urban blue-green-gray space landscape patterns on PM2.5. To address this research gap, this study examines the influence of blue-green-gray spatial configurations on PM2.5 concentrations in 410 high-density urban blocks in Xi’an during its peak pollution period. Using spatial autocorrelation, random forest regression, and correlation analysis, the research delineates the spatio-temporal distribution of PM2.5, constructs a customized landscape pattern index system, and proposes optimization strategies for urban planning. The results reveal that: (1)PM2.5 levels peak in January, with elevated nocturnal concentrations, and exhibit significant spatial clustering in the western, southwestern, and southern districts of Xi’an. (2)A set of nine primary landscape indices focused on green and gray spaces, including AI-Green, NP-Green, PD-Green, TE-Green, SHAPE (MN)-Green, CA-Gray, PD-Gray, SPLIT-Gray and SHAPE (MN)-Gray effectively characterize the impact of the urban landscape on air quality. (3) In particular, indicators such as AI-Green, TE-Green, and SHAPE (MN)-Green are significantly negatively correlated with PM2.5, while metrics such as NP-Green, PD-Green, and CA-Gray show positive associations, with CA-Gray exhibiting a particularly strong link. These findings suggest that urban planning should prioritize enhancing the aggregation and connectivity of green spaces, refining the configuration of built-up areas, and promoting a decentralized distribution of gray spaces. Such strategic spatial configurations can meaningfully lower PM2.5 concentrations, providing a scientifically grounded framework for improving air quality and public health in Xi’an and other high-density urban environments.
{"title":"Unraveling the impacts of blue-green-gray landscape patterns on PM2.5 in high-density urban areas: A case study of Xi’an","authors":"Baojun Yang , Lin Zhang , Danchen Yang , HuiYing Ma , Haifan Xie , Jiahui Hou , Ling Qiu , Tian Gao","doi":"10.1016/j.apr.2025.102788","DOIUrl":"10.1016/j.apr.2025.102788","url":null,"abstract":"<div><div>High-density cities face severe PM<sub>2.5</sub> particulate pollution issues, where landscape patterns significantly influence PM<sub>2.5</sub> concentrations. Currently, there is a notable lack of comprehensive research on the integrated effects of urban blue-green-gray space landscape patterns on PM<sub>2.5</sub>. To address this research gap, this study examines the influence of blue-green-gray spatial configurations on PM<sub>2.5</sub> concentrations in 410 high-density urban blocks in Xi’an during its peak pollution period. Using spatial autocorrelation, random forest regression, and correlation analysis, the research delineates the spatio-temporal distribution of PM<sub>2.5</sub>, constructs a customized landscape pattern index system, and proposes optimization strategies for urban planning. The results reveal that: (1)PM<sub>2.5</sub> levels peak in January, with elevated nocturnal concentrations, and exhibit significant spatial clustering in the western, southwestern, and southern districts of Xi’an. (2)A set of nine primary landscape indices focused on green and gray spaces, including AI-Green, NP-Green, PD-Green, TE-Green, SHAPE (MN)-Green, CA-Gray, PD-Gray, SPLIT-Gray and SHAPE (MN)-Gray effectively characterize the impact of the urban landscape on air quality. (3) In particular, indicators such as AI-Green, TE-Green, and SHAPE (MN)-Green are significantly negatively correlated with PM<sub>2.5</sub>, while metrics such as NP-Green, PD-Green, and CA-Gray show positive associations, with CA-Gray exhibiting a particularly strong link. These findings suggest that urban planning should prioritize enhancing the aggregation and connectivity of green spaces, refining the configuration of built-up areas, and promoting a decentralized distribution of gray spaces. Such strategic spatial configurations can meaningfully lower PM<sub>2.5</sub> concentrations, providing a scientifically grounded framework for improving air quality and public health in Xi’an and other high-density urban environments.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102788"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135726","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2025-11-04DOI: 10.1016/j.apr.2025.102809
Meltem Apaydın Üstün , Can Burak Özkal
Understanding and predicting odor nuisance in industrial areas is vital for public health and quality of life. In Çorlu, an industrial city with unique topography, we analyzed citizen-reported odor complaints collected via the Geographic Information System-integrated mobile application Çorlu KODER (October 2020–August 2022). Using machine learning models incorporating meteorological factors like mixed-layer height, temperature, pressure, and humidity, along with seasonal and diurnal variations, we addressed significant class imbalance in the dataset. Ensemble methods such as Random Forest and Adaptive Boosting combined with synthetic minority oversampling and edited nearest neighbors achieved macro-averaged mean absolute error scores of 0.232 and 0.276. Our findings demonstrate the potential of machine learning for proactive odor prediction, aiding urban management in improving air quality and community well-being.
{"title":"Predicting odor nuisance levels using meteorological data and citizen complaints records: A machine learning approach","authors":"Meltem Apaydın Üstün , Can Burak Özkal","doi":"10.1016/j.apr.2025.102809","DOIUrl":"10.1016/j.apr.2025.102809","url":null,"abstract":"<div><div>Understanding and predicting odor nuisance in industrial areas is vital for public health and quality of life. In Çorlu, an industrial city with unique topography, we analyzed citizen-reported odor complaints collected via the Geographic Information System-integrated mobile application Çorlu KODER (October 2020–August 2022). Using machine learning models incorporating meteorological factors like mixed-layer height, temperature, pressure, and humidity, along with seasonal and diurnal variations, we addressed significant class imbalance in the dataset. Ensemble methods such as Random Forest and Adaptive Boosting combined with synthetic minority oversampling and edited nearest neighbors achieved macro-averaged mean absolute error scores of 0.232 and 0.276. Our findings demonstrate the potential of machine learning for proactive odor prediction, aiding urban management in improving air quality and community well-being.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102809"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135699","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2025-11-04DOI: 10.1016/j.apr.2025.102815
Bianca Wernecke , Caradee Y. Wright , Kristy Langerman , Angela Mathee , Nada Abdelatif , Marcus A. Howard , Nkosana Jafta , Christiaan Pauw , Shumani Phaswana , Kareshma Asharam , Ishen Seocharan , Hendrik Smith , Rajen N. Naidoo
Domestic fuel use contributes significantly to household air pollution levels and to the disease burden in low-income households in South Africa. The link between residential fuel stacking and switching, and respiratory health, mediated by household air pollution, remains underexplored, posing challenges to transition to cleaner fuels. This study identified socio-economic determinants of fuel use patterns in two low-income communities of KwaZamokuhle and eMzinoni in South Africa. It also examined the impacts of these patterns on household air pollution levels and respiratory health outcomes. Over half of households relied on dirty fuels across all needs. Average household PM2.5 levels exceeded national daily standards (40 μg/m3). Education level and employment status were significant factors in determining fuel choice, with employed participants less likely to rely on dirty fuels. Town-specific characteristics also influenced household fuel use patterns. In terms of health, 9.5 % of participants had obstructive airways disease and 26.9 % tested positive for inhalant allergens. Heating fuels were strongest predictor of obstructive airways disease (>75 %) whereas cooking fuels were the main predictor of allergen sensitivity (∼75 %). The stepwise introduction of cleaner fuels predicted better respiratory health outcomes. The findings of this study suggest that even the partial adoption of cleaner fuels has health benefits and supports the formulation of context-specific mitigation efforts aiming to address negative health effects associated with household air pollution.
{"title":"Multiple fuel use in low-income communities: socio-economic determinants and impacts on household air pollution and respiratory health in South Africa","authors":"Bianca Wernecke , Caradee Y. Wright , Kristy Langerman , Angela Mathee , Nada Abdelatif , Marcus A. Howard , Nkosana Jafta , Christiaan Pauw , Shumani Phaswana , Kareshma Asharam , Ishen Seocharan , Hendrik Smith , Rajen N. Naidoo","doi":"10.1016/j.apr.2025.102815","DOIUrl":"10.1016/j.apr.2025.102815","url":null,"abstract":"<div><div>Domestic fuel use contributes significantly to household air pollution levels and to the disease burden in low-income households in South Africa. The link between residential fuel stacking and switching, and respiratory health, mediated by household air pollution, remains underexplored, posing challenges to transition to cleaner fuels. This study identified socio-economic determinants of fuel use patterns in two low-income communities of KwaZamokuhle and eMzinoni in South Africa. It also examined the impacts of these patterns on household air pollution levels and respiratory health outcomes. Over half of households relied on dirty fuels across all needs. Average household PM<sub>2.5</sub> levels exceeded national daily standards (40 μg/m<sup>3</sup>). Education level and employment status were significant factors in determining fuel choice, with employed participants less likely to rely on dirty fuels. Town-specific characteristics also influenced household fuel use patterns. In terms of health, 9.5 % of participants had obstructive airways disease and 26.9 % tested positive for inhalant allergens. Heating fuels were strongest predictor of obstructive airways disease (>75 %) whereas cooking fuels were the main predictor of allergen sensitivity (∼75 %). The stepwise introduction of cleaner fuels predicted better respiratory health outcomes. The findings of this study suggest that even the partial adoption of cleaner fuels has health benefits and supports the formulation of context-specific mitigation efforts aiming to address negative health effects associated with household air pollution.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102815"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135706","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2025-10-24DOI: 10.1016/j.apr.2025.102795
Benjamin Drummond , Ailish Graham , Lucy Neal , Pedro Molina Jiménez , Richard J. Pope , Carly Reddington
Wildfires can be important drivers of poor air quality. Numerical atmosphere models are routinely used to estimate pollutant concentrations emitted from a wide range of sources, including from wildfires. Such models often take the Eulerian field or Lagrangian particle method for representing mass and transport in the atmosphere. Using the Saddleworth Moor and Winter Hill fires that occurred in North West England in 2018 as a case study, we compared these two methods consistently within the same model framework. We also explored the impact of model spatial resolution on predicted concentrations and health impacts. In the Eulerian simulations, as the horizontal resolution was made finer (from 12 km to 1 km) the horizontal spread of the downwind wildfire pollution decreased substantially, leading a smaller geographical area and population being impacted by the smoke. The estimated number of people exposed to poor air quality due to wildfire from the 1 km Eulerian simulation was 30% lower than from the 12 km Eulerian simulation. A health impact assessment found a similar relative decrease for the estimated excess mortality due to short-term PM2.5 exposure when going from 12 km to 1 km horizontal resolution. Estimated air quality impacts were also found to be sensitive to horizontal resolution for the Lagrangian simulations but to a lesser degree (10% decrease from 12 km to 1 km). We recommend that model spatial resolution should be considered as a source of uncertainty for wildfire air quality impact assessments, particularly when an Eulerian model is used.
{"title":"Air quality impacts of a major wildfire in the UK: Sensitivity to model spatial resolution and transport method","authors":"Benjamin Drummond , Ailish Graham , Lucy Neal , Pedro Molina Jiménez , Richard J. Pope , Carly Reddington","doi":"10.1016/j.apr.2025.102795","DOIUrl":"10.1016/j.apr.2025.102795","url":null,"abstract":"<div><div>Wildfires can be important drivers of poor air quality. Numerical atmosphere models are routinely used to estimate pollutant concentrations emitted from a wide range of sources, including from wildfires. Such models often take the Eulerian field or Lagrangian particle method for representing mass and transport in the atmosphere. Using the Saddleworth Moor and Winter Hill fires that occurred in North West England in 2018 as a case study, we compared these two methods consistently within the same model framework. We also explored the impact of model spatial resolution on predicted concentrations and health impacts. In the Eulerian simulations, as the horizontal resolution was made finer (from 12 km to 1 km) the horizontal spread of the downwind wildfire pollution decreased substantially, leading a smaller geographical area and population being impacted by the smoke. The estimated number of people exposed to poor air quality due to wildfire from the 1 km Eulerian simulation was 30% lower than from the 12 km Eulerian simulation. A health impact assessment found a similar relative decrease for the estimated excess mortality due to short-term PM<sub>2.5</sub> exposure when going from 12 km to 1 km horizontal resolution. Estimated air quality impacts were also found to be sensitive to horizontal resolution for the Lagrangian simulations but to a lesser degree (<span><math><mo>∼</mo></math></span>10% decrease from 12 km to 1 km). We recommend that model spatial resolution should be considered as a source of uncertainty for wildfire air quality impact assessments, particularly when an Eulerian model is used.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102795"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135715","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2025-11-08DOI: 10.1016/j.apr.2025.102814
Jianhui Bai , Zhixiang Wu , Chuan Yang , Alex B. Guenther
Based on the measurements of emissions of biogenic volatile organic compounds (BVOCs), solar radiation and meteorological variables in three representative forests from temperate to subtropical zone in China, a primary empirical model of BVOC emissions (EMBE) has been developed. During 2018 and 2019, BVOC emission fluxes were measured in a tropical rubber tree plantation in China, and EMBE was improved to a generalized empirical model of BVOC emissions (GEMBE) and evaluated. This paper presents GEMBE as a fully empirical modeling framework, developed directly from ecosystem-scale flux observations, and compares this approach with MEGAN, a widely-used model that includes both empirical and process-based components. Isoprene emission estimated with GEMBE, using an emission factor based on the average for other sites, overestimated the observation by 34.8 % and monoterpenes by 41.0 %. A summary of the BVOC emission fluxes in the four typical forests in China is reported. With site-specific flux measurements the GEMBE model can be used to calculate BVOC emissions in China. Using GEMBE, the responses of BVOC emissions to their driving factors showed that isoprene and monoterpene emissions were more sensitive to the change in PAR than to changes in other factors. The responses of BVOC emissions to their driving factors were discussed for the four representative forests. The emission factors (EFs) calculated using GEMBE and MEGAN were summarized for the typical forests and grassland. The MEGAN estimated isoprene EF at this rubber plantation is lower, and the monoterpene EF is higher. Both GEMBE and MEGAN showed reasonable agreement in simulations of temporal trends over one year and two years, and well reproduced the evident monthly and seasonal BVOC emissions. As a fully empirical model, GEMBE provides a framework for estimating regional BVOC emissions, and investigating their impact on atmospheric chemistry, with an alternative approach to more complex process-based models that require biophysical parameterization.
{"title":"Development and application of a generalized empirical model of BVOC emission (GEMBE) using observations from four Chinese forests","authors":"Jianhui Bai , Zhixiang Wu , Chuan Yang , Alex B. Guenther","doi":"10.1016/j.apr.2025.102814","DOIUrl":"10.1016/j.apr.2025.102814","url":null,"abstract":"<div><div>Based on the measurements of emissions of biogenic volatile organic compounds (BVOCs), solar radiation and meteorological variables in three representative forests from temperate to subtropical zone in China, a primary empirical model of BVOC emissions (EMBE) has been developed. During 2018 and 2019, BVOC emission fluxes were measured in a tropical rubber tree plantation in China, and EMBE was improved to a generalized empirical model of BVOC emissions (GEMBE) and evaluated. This paper presents GEMBE as a fully empirical modeling framework, developed directly from ecosystem-scale flux observations, and compares this approach with MEGAN, a widely-used model that includes both empirical and process-based components. Isoprene emission estimated with GEMBE, using an emission factor based on the average for other sites, overestimated the observation by 34.8 % and monoterpenes by 41.0 %. A summary of the BVOC emission fluxes in the four typical forests in China is reported. With site-specific flux measurements the GEMBE model can be used to calculate BVOC emissions in China. Using GEMBE, the responses of BVOC emissions to their driving factors showed that isoprene and monoterpene emissions were more sensitive to the change in PAR than to changes in other factors. The responses of BVOC emissions to their driving factors were discussed for the four representative forests. The emission factors (EFs) calculated using GEMBE and MEGAN were summarized for the typical forests and grassland. The MEGAN estimated isoprene EF at this rubber plantation is lower, and the monoterpene EF is higher. Both GEMBE and MEGAN showed reasonable agreement in simulations of temporal trends over one year and two years, and well reproduced the evident monthly and seasonal BVOC emissions. As a fully empirical model, GEMBE provides a framework for estimating regional BVOC emissions, and investigating their impact on atmospheric chemistry, with an alternative approach to more complex process-based models that require biophysical parameterization.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102814"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135705","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2025-10-24DOI: 10.1016/j.apr.2025.102798
Dharini Sahu , Shamsh Pervez , Judith C. Chow , John G. Watson , Rajan K. Chakrabarty , Aishwaryashri Tamrakar , Indrapal Karbhal , Manas Kanti Deb , Kamlesh Shrivas , Yasmeen Fatima Pervez , Sachchidanand Shukla , D.P. Bisen
This study presents a comparative analysis of gaseous pollutant emissions from solid fuel combustion under real-world household conditions and simulated experimental chamber-based conditions. Focusing on six commonly used domestic fuels in India, fuelwood (FW), dung cake (DC), coal ball (CB), agricultural residue (AR), and two mixed fuel types (M1: CB + DC, M2: FW + DC in a 10:1 ratio), the research quantifies emissions of CO2, CO, NO, NO2, SO2, CH4, and total volatile organic compounds (TVOCs). Simulated combustion chamber experiments, designed to replicate household stove operation while allowing precise emission monitoring, were conducted alongside field based real-world observations. Emission factors (EFs) and combustion efficiency metrics were assessed to understand pollutant formation mechanisms. Results showed strong correlations between combustion efficiency and emission profiles: higher modified combustion efficiency (MCE) was associated with elevated CO2 and NO2 emissions, while lower MCE resulted in higher outputs of incomplete combustion products such as CO, CH4, and TVOCs. Slightly lower EFs and higher MCEs were observed in field based real-world conditions compared to those found for simulated experimental chamber based conditions. Agricultural residues emitted the highest CH4 levels, likely due to paddy-origin biomass, whereas mixed fuels showed increased TVOC emissions, linked to their high carbon and moisture content. This comparative study emphasizes the importance of integrating field-based validation with laboratory simulations to accurately assess household air pollution, and supports targeted interventions to promote cleaner combustion practices and reduce public health risks.
{"title":"Gaseous pollutant emissions from solid fuel combustion: Comparative study of real-world and simulated chamber-based experiments","authors":"Dharini Sahu , Shamsh Pervez , Judith C. Chow , John G. Watson , Rajan K. Chakrabarty , Aishwaryashri Tamrakar , Indrapal Karbhal , Manas Kanti Deb , Kamlesh Shrivas , Yasmeen Fatima Pervez , Sachchidanand Shukla , D.P. Bisen","doi":"10.1016/j.apr.2025.102798","DOIUrl":"10.1016/j.apr.2025.102798","url":null,"abstract":"<div><div>This study presents a comparative analysis of gaseous pollutant emissions from solid fuel combustion under real-world household conditions and simulated experimental chamber-based conditions. Focusing on six commonly used domestic fuels in India, fuelwood (FW), dung cake (DC), coal ball (CB), agricultural residue (AR), and two mixed fuel types (M1: CB + DC, M2: FW + DC in a 10:1 ratio), the research quantifies emissions of CO<sub>2</sub>, CO, NO, NO<sub>2</sub>, SO<sub>2</sub>, CH<sub>4</sub>, and total volatile organic compounds (TVOCs). Simulated combustion chamber experiments, designed to replicate household stove operation while allowing precise emission monitoring, were conducted alongside field based real-world observations. Emission factors (EFs) and combustion efficiency metrics were assessed to understand pollutant formation mechanisms. Results showed strong correlations between combustion efficiency and emission profiles: higher modified combustion efficiency (MCE) was associated with elevated CO<sub>2</sub> and NO<sub>2</sub> emissions, while lower MCE resulted in higher outputs of incomplete combustion products such as CO, CH<sub>4</sub>, and TVOCs. Slightly lower EFs and higher MCEs were observed in field based real-world conditions compared to those found for simulated experimental chamber based conditions. Agricultural residues emitted the highest CH<sub>4</sub> levels, likely due to paddy-origin biomass, whereas mixed fuels showed increased TVOC emissions, linked to their high carbon and moisture content. This comparative study emphasizes the importance of integrating field-based validation with laboratory simulations to accurately assess household air pollution, and supports targeted interventions to promote cleaner combustion practices and reduce public health risks.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102798"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135684","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2025-11-07DOI: 10.1016/j.apr.2025.102821
Tantan Tan , Gaoshan Zhang , Ke Lu , Yanpeng Li
Indoor air quality is critically influenced by microbial contaminants in settled dust, yet existing studies predominantly focus on airborne microorganisms, leaving dust-associated microbial exposure poorly characterized. This study investigated microbial contamination in dust from four university indoor environments (offices, laboratories, dormitories, and classrooms) to assess their concentration, pathogenic composition, and human exposure risks. Dust samples were analyzed via fluorescence staining, high-throughput sequencing, and the dust daily intake model (DDIM). Results revealed significant spatial heterogeneity in microbial concentrations: dormitories exhibited the highest bacterial levels (4068.88 × 104 ± 3386.55 × 104 CFU/g), while offices had the highest fungal concentrations (146.16 × 104 ± 152.93 × 104 CFU/g). Factors such as occupancy density, cleaning frequency, and indoor plant presence are strongly associated with microbial distribution. Three dominant genera (Fusobacterium, Pseudomonas, and Corynebacterium) and four dominant fungal genera (Aspergillus, Penicillium, Fusarium, and Streptomyces) were observed in all indoor dust samples, with a potential risk of association with respiratory diseases and skin infections. Exposure assessment indicated that dust ingestion dominated microbial intake, with dormitories posing the highest bacterial exposure (EDI up to 35,086 CFU/(kg·day)) and offices the highest fungal exposure (EDI up to 1263 CFU/(kg·day)). These findings highlight the urgent need for targeted interventions, such as improved ventilation, regular cleaning, and microbial monitoring, to mitigate health risks in high-exposure indoor environments. This study provides a scientific foundation for refining indoor air quality standards and safeguarding occupants in densely populated educational settings.
{"title":"Dust microbial contamination in typical indoor environments: Concentration, pathogenic composition and exposure assessment","authors":"Tantan Tan , Gaoshan Zhang , Ke Lu , Yanpeng Li","doi":"10.1016/j.apr.2025.102821","DOIUrl":"10.1016/j.apr.2025.102821","url":null,"abstract":"<div><div>Indoor air quality is critically influenced by microbial contaminants in settled dust, yet existing studies predominantly focus on airborne microorganisms, leaving dust-associated microbial exposure poorly characterized. This study investigated microbial contamination in dust from four university indoor environments (offices, laboratories, dormitories, and classrooms) to assess their concentration, pathogenic composition, and human exposure risks. Dust samples were analyzed via fluorescence staining, high-throughput sequencing, and the dust daily intake model (DDIM). Results revealed significant spatial heterogeneity in microbial concentrations: dormitories exhibited the highest bacterial levels (4068.88 × 10<sup>4</sup> ± 3386.55 × 10<sup>4</sup> CFU/g), while offices had the highest fungal concentrations (146.16 × 10<sup>4</sup> ± 152.93 × 10<sup>4</sup> CFU/g). Factors such as occupancy density, cleaning frequency, and indoor plant presence are strongly associated with microbial distribution. Three dominant genera (<em>Fusobacterium</em>, <em>Pseudomonas</em>, and <em>Corynebacterium</em>) and four dominant fungal genera (<em>Aspergillus</em>, <em>Penicillium</em>, <em>Fusarium</em>, and <em>Streptomyces</em>) were observed in all indoor dust samples, with a potential risk of association with respiratory diseases and skin infections. Exposure assessment indicated that dust ingestion dominated microbial intake, with dormitories posing the highest bacterial exposure (EDI up to 35,086 CFU/(kg·day)) and offices the highest fungal exposure (EDI up to 1263 CFU/(kg·day)). These findings highlight the urgent need for targeted interventions, such as improved ventilation, regular cleaning, and microbial monitoring, to mitigate health risks in high-exposure indoor environments. This study provides a scientific foundation for refining indoor air quality standards and safeguarding occupants in densely populated educational settings.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102821"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135716","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2025-10-24DOI: 10.1016/j.apr.2025.102789
Lucille Borlaza-Lacoste , Md. Aynul Bari , Cheng-Hsuan Lu , Philip K. Hopke
Over the past two decades, shifts in energy use and regulatory policies in New York State have shaped emissions and air quality in the New York City (NYC) metropolitan area, a densely populated and VOC-limited nonattainment region for ozone (O3). This study analyzed 24-h canister measurements from six sites, Queens, Bronx, Kings, Richmond, Elizabeth, and Chester, spanning the period of 2000–2021. Dispersion-Normalized Positive Matrix Factorization apportioned VOCs sources while accounting for atmospheric dilution, resolving twelve distinct sources dominated by aldehyde-rich factors, vehicle emissions, and industrial activities. Long-term trends from seasonal-trend decomposition and piecewise regression highlighted regulatory- and economy-driven shifts in source contributions. Significant declines in vehicle emissions, MTBE- and MEK-rich factors, and aldehydes aligned with Tier 2 and 3 fuel standards, MTBE phase-out, and MACT regulations. In contrast, natural gas, evaporative, biogenic, and background sources remained stable or increased, reflecting persistent and seasonally modulated emissions. Distinct site- and source-specific patterns in weekday/weekend and seasonal variability were also observed. These results show that while regulations have effectively reduced many anthropogenic VOCs sources, persistent emissions underscore the need for continued monitoring and adaptive control strategies in O3 nonattainment regions like NYC.
{"title":"Long-term trends reflecting regulatory impacts on VOCs sources in the New York City metropolitan area","authors":"Lucille Borlaza-Lacoste , Md. Aynul Bari , Cheng-Hsuan Lu , Philip K. Hopke","doi":"10.1016/j.apr.2025.102789","DOIUrl":"10.1016/j.apr.2025.102789","url":null,"abstract":"<div><div>Over the past two decades, shifts in energy use and regulatory policies in New York State have shaped emissions and air quality in the New York City (NYC) metropolitan area, a densely populated and VOC-limited nonattainment region for ozone (O<sub>3</sub>). This study analyzed 24-h canister measurements from six sites, Queens, Bronx, Kings, Richmond, Elizabeth, and Chester, spanning the period of 2000–2021. Dispersion-Normalized Positive Matrix Factorization apportioned VOCs sources while accounting for atmospheric dilution, resolving twelve distinct sources dominated by aldehyde-rich factors, vehicle emissions, and industrial activities. Long-term trends from seasonal-trend decomposition and piecewise regression highlighted regulatory- and economy-driven shifts in source contributions. Significant declines in vehicle emissions, MTBE- and MEK-rich factors, and aldehydes aligned with Tier 2 and 3 fuel standards, MTBE phase-out, and MACT regulations. In contrast, natural gas, evaporative, biogenic, and background sources remained stable or increased, reflecting persistent and seasonally modulated emissions. Distinct site- and source-specific patterns in weekday/weekend and seasonal variability were also observed. These results show that while regulations have effectively reduced many anthropogenic VOCs sources, persistent emissions underscore the need for continued monitoring and adaptive control strategies in O<sub>3</sub> nonattainment regions like NYC.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102789"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135727","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-03-01Epub Date: 2025-10-26DOI: 10.1016/j.apr.2025.102792
Mantena Sireesha , Abdul Gaffar Sheik
This study explored the potential of four machine learning (ML) models such as Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Gated Recurrent Units (GRU), and Deep Feedforward Neural Networks (DFNN) or predicting greenhouse gas (GHG) emissions from an agricultural field. It measured carbon dioxide (CO2), methane (CH4), and nitrous oxide (N2O) over a global scenario spanning 172 years. The rigorous analysis, which included statistical comparisons and cross-validation for predicting CO2, CH4, and N2O fluxes, demonstrated that GRU, CNN, and DFNN models consistently exhibited high prediction accuracy across most sectors. Notably, the GRU model outperformed the others, achieving an R2 of 0.9985 and an RMSE of 0.0108 for N2O emissions in the Waste sector. In contrast to previous studies, this research not only predicts future GHG emissions but also identifies the relationship between these predictions and their influential variables. To achieve this, an interpretable prediction framework was utilized, incorporating explainable artificial intelligence (XAI) methods including SHapley Additive Explanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), Individual Conditional Expectation (ICE) plots, and Partial Dependence Plots (PDPs) to reveal each GHG’s contribution to overall emissions. The SHAP analysis indicated that CH4 was the dominant contributor across all sectors, with a high average SHAP value of 10,632.68 in agriculture and 3386.09 in the Waste sector, followed by N2O and CO2. Further analyses using ICE and PDP clarified the sector-specific nonlinear interactions, showing that CH4 had the greatest influence on emissions, particularly in synergy with N2O. These findings illustrate the significant potential of ML models for predicting GHG emissions in the agricultural sector.
{"title":"Understanding the agriculture sectors of greenhouse gas emissions prediction in the global scenario: Insights from explainable artificial intelligence (XAI)","authors":"Mantena Sireesha , Abdul Gaffar Sheik","doi":"10.1016/j.apr.2025.102792","DOIUrl":"10.1016/j.apr.2025.102792","url":null,"abstract":"<div><div>This study explored the potential of four machine learning (ML) models such as Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), Gated Recurrent Units (GRU), and Deep Feedforward Neural Networks (DFNN) or predicting greenhouse gas (GHG) emissions from an agricultural field. It measured carbon dioxide (CO<sub>2</sub>), methane (CH<sub>4</sub>), and nitrous oxide (N<sub>2</sub>O) over a global scenario spanning 172 years. The rigorous analysis, which included statistical comparisons and cross-validation for predicting CO<sub>2</sub>, CH<sub>4</sub>, and N<sub>2</sub>O fluxes, demonstrated that GRU, CNN, and DFNN models consistently exhibited high prediction accuracy across most sectors. Notably, the GRU model outperformed the others, achieving an R<sup>2</sup> of 0.9985 and an RMSE of 0.0108 for N<sub>2</sub>O emissions in the Waste sector. In contrast to previous studies, this research not only predicts future GHG emissions but also identifies the relationship between these predictions and their influential variables. To achieve this, an interpretable prediction framework was utilized, incorporating explainable artificial intelligence (XAI) methods including SHapley Additive Explanations (SHAP), Local Interpretable Model-Agnostic Explanations (LIME), Individual Conditional Expectation (ICE) plots, and Partial Dependence Plots (PDPs) to reveal each GHG’s contribution to overall emissions. The SHAP analysis indicated that CH<sub>4</sub> was the dominant contributor across all sectors, with a high average SHAP value of 10,632.68 in agriculture and 3386.09 in the Waste sector, followed by N<sub>2</sub>O and CO<sub>2</sub>. Further analyses using ICE and PDP clarified the sector-specific nonlinear interactions, showing that CH4 had the greatest influence on emissions, particularly in synergy with N<sub>2</sub>O. These findings illustrate the significant potential of ML models for predicting GHG emissions in the agricultural sector.</div></div>","PeriodicalId":8604,"journal":{"name":"Atmospheric Pollution Research","volume":"17 3","pages":"Article 102792"},"PeriodicalIF":3.5,"publicationDate":"2026-03-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146135730","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号