A Physically Constrained Deep-Learning Fusion Method for Estimating Surface NO2 Concentration from Satellite and Ground Monitors DOI Creative Commons
Jia Xing, Bok H. Baek, Siwei Li

и другие.

Environmental Science & Technology, Год журнала: 2024, Номер unknown

Опубликована: Ноя. 20, 2024

Accurate estimation of atmospheric chemical concentrations from multiple observations is crucial for assessing the health effects air pollution. However, existing methods are limited by imbalanced samples observations. Here, we introduce a novel deep-learning model-measurement fusion method (DeepMMF) constrained physical laws inferred transport model (CTM) to estimate NO2 over Continental United States (CONUS). By pretraining with spatiotemporally complete CTM simulations, fine-tuning satellite and ground measurements, employing optimization strategy selecting proper prior emission, DeepMMF delivers improved estimates, showing greater consistency daily variation alignment (with NMB reduced −0.3 −0.1 compared original simulations). More importantly, effectively addressed sample imbalance issue that causes overestimation (by 100%) downwind or rural in other methods. It achieves higher R2 0.98 lower RMSE 1.45 ppb surface observations, overperforming approaches, which show values 0.4–0.7 RMSEs 3–6 ppb. The also offers synergistic advantage adjusting corresponding emissions, agreement changes (−10% −20%) reported NEI between 2019 2020. Our results demonstrate great potential data better support pollution exposure forecasting.

Язык: Английский

Impact of Digitization and Artificial Intelligence on Carbon Emissions Considering Variable Interaction and Heterogeneity: An Interpretable Deep Learning Modeling Framework DOI

Gongquan Zhang,

Shenglin Ma, Mingxing Zheng

и другие.

Sustainable Cities and Society, Год журнала: 2025, Номер unknown, С. 106333 - 106333

Опубликована: Март 1, 2025

Язык: Английский

Процитировано

7
Application of the Lasso regularisation technique in mitigating overfitting in air quality prediction models DOI Creative Commons
Abbas Pak,

Abdullah Kaviani Rad,

Mohammad Javad Nematollahi

и другие.

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Янв. 2, 2025

Язык: Английский

Процитировано

4
Review on the Application of Remote Sensing Data and Machine Learning to the Estimation of Anthropogenic Heat Emissions DOI Creative Commons
Lei Feng, Danyang Ma, Min Xie

и другие.

Remote Sensing, Год журнала: 2025, Номер 17(2), С. 200 - 200

Опубликована: Янв. 8, 2025

Anthropogenic heat is the generated by human activities such as industry, construction, transport, and metabolism. Accurate estimates of anthropogenic are essential for studying impacts on climate atmospheric environment. Commonly applied methods estimating include inventory method, energy balance equation building model simulation method. In recent years, rapid development computer technology availability massive data have made machine learning a powerful tool fluxes assessing its effects. Multi-source remote sensing also been widely used to obtain more details spatial temporal distribution characteristics heat. This paper reviews main approaches emissions. The typical algorithms abovementioned three introduced, their advantages limitations evaluated. Moreover, progress in application discussed well. Based big techniques, research feature engineering fusion will bring about major changes analysis modeling More in-depth this issue recommended provide important support curbing global warming, mitigating air pollution, achieving national goals carbon peak neutrality strategy.

Язык: Английский

Процитировано

4
Machine learning for ecological analysis DOI

Zhengyang Yu,

Chongfeng Bu, Yanjie Li

и другие.

Chemical Engineering Journal, Год журнала: 2025, Номер unknown, С. 160780 - 160780

Опубликована: Фев. 1, 2025

Язык: Английский

Процитировано

1
Harmonizing low-cost and regulatory air quality monitoring networks with interpretable semi-supervised learning: Reducing exposure misclassification in underrepresented communities DOI
Dié Tang,

Tan Mi,

Xi Zheng

и другие.

Journal of Hazardous Materials, Год журнала: 2025, Номер 491, С. 137893 - 137893

Опубликована: Март 10, 2025

Язык: Английский

Процитировано

1
Development and evaluation of rapid, national-scale outdoor air pollution modelling and exposure assessment: Hybrid air dispersion exposure system (HADES) DOI Creative Commons
Calvin Jephcote, John S. Gulliver

Environment International, Год журнала: 2025, Номер unknown, С. 109304 - 109304

Опубликована: Янв. 1, 2025

Improvements in computer processing power are facilitating the development of more detailed environmental models with greater geographical coverage. We developed a national-scale model outdoor air pollution (Hybrid Air Dispersion Exposure System - HADES) for rapid production concentration maps nitrogen dioxide (NO2) and ozone (O3) at very high spatial resolution (10m). The combines dispersion modelling satellite-derived estimates background concentrations, land cover, 3-D representation buildings, statistical calibration framework. an emissions inventory covering England Wales to implement tested its performance using data years 2018-2019 from fixed-site monitoring locations. In 10,000 Monte Carlo cross-validation iterations, hourly-annual average R2 values NO2 were 0.77-0.79 (RMSE: root mean squared error 5.3-5.7 µg/m3), 0.87-0.89 O3 (RMSE = 3.6-3.8 µg/m3) 95% confidence interval. annual was 0.80 4.9 0.86 3.2 aggregating estimates. surfaces freely available non-commercial use. these exposure assessment, all residential locations, neighbourhoods urban areas, unlikely be below 2021 World Health Organisation Quality Guidelines threshold (10 concentrations µg/m3). Rural suburban areas likely exceed peak-season 8-hour daily maximum (60

Язык: Английский

Процитировано

0
Revolutionizing air quality forecasting: Fusion of state-of-the-art deep learning models for precise classification DOI
Umesh Kumar Lilhore, Sarita Simaiya, Surjeet Dalal

и другие.

Urban Climate, Год журнала: 2025, Номер 59, С. 102308 - 102308

Опубликована: Янв. 28, 2025

Язык: Английский

Процитировано

0
bneR: A collaborative workflow for air pollution exposure modeling and uncertainty characterization using the Bayesian Nonparametric Ensemble DOI
Jaime Benavides,

Carlos Carrillo-Gallegos,

Vijay Kumar

и другие.

Journal of Environmental Management, Год журнала: 2025, Номер 375, С. 124061 - 124061

Опубликована: Янв. 29, 2025

Язык: Английский

Процитировано

0
Machine learning predicting the transport mechanisms and entrainment characteristics of negative buoyant jets DOI Creative Commons
Yaowen Xia, Wenfeng Gao, Qiong Li

и другие.

AIP Advances, Год журнала: 2025, Номер 15(1)

Опубликована: Янв. 1, 2025

Fountains injected into homogeneous fluids, characterized by combined temperature and concentration effects, are common in both natural environmental settings. In this study, the capacities of several machine learning models, including support vector regression, multi-layer perceptron, random forests, XGBoost, CatBoost, AdaBoost, LightGBM, were investigated to clarify transient flow behavior fountains. The results indicated that perceptron was superior other models as it provided improved coefficient determination, root mean squared error, absolute error. This study confirmed techniques have great potential

Язык: Английский

Процитировано

0
Environmental exposures related to gut microbiota among children with asthma: a pioneer study in Taiwan DOI Creative Commons

Aji Kusumaning Asri,

Tsunglin Liu,

Hui-Ju Tsai

и другие.

Ecotoxicology and Environmental Safety, Год журнала: 2025, Номер 291, С. 117793 - 117793

Опубликована: Янв. 31, 2025

Gut microbiota plays a crucial role in human health and can be influenced by environmental factors. While past studies have examined the impact of environment on gut microbiota, vulnerable populations often been overlooked. This study aimed to investigate association between exposures, air pollution greenspace, asthmatic children. Data were collected during recovery period for 41 eligible Air was estimated using an ensemble learning model that combined regression machine-learning algorithms, while greenspace quantified normalized difference vegetation index (NDVI) green land-cover data. The lag effects exposures assessed within defined buffer zones surrounding each child's residence. A generalized additive applied examine associations. Results revealed marginally significant negative 1-day exposure NO₂ indices, such as observed bacteria (Coef.: -1.130; 95 %CI -2.287, 0.027) bacterial richness -2.420; -4.987, 0.146). 8-day lagged average PM2.5 O₃ also showed impacts diversity. In contrast, 1-month positively associated with indices. linked specific abundances, Streptococcus. underscores need further research how factors may influence immunity children altering microbiota.

Язык: Английский

Процитировано

0