Interpreting Highly Variable Indoor PM2.5 in Rural North China Using Machine Learning

Understanding the low-temperature drying process of sludge with machine learning in a sewage-source heat pump drying system

Heat pump drying technology based on sewage heat source is an eco-friendly sludge drying method. It can effectively reduce the pollution of natural water bodies by waste heat while reducing energy consumption. However, the drying characteristics of sludge in this case remain unclear. Here, we proposed and constructed a novel sewage-source heat pump sludge low-temperature drying system, and combine sludge drying theory with machine learning algorithms to model and analyze the drying process. The results revealed that the performance of the machine learning models was significantly better than that of the existing numerical simulation models. After Bayesian optimization, the XGBoost model showed superior prediction effect. In addition, the interpretable analysis of the model indicated that in the constant rate stage, the drying rate is mainly influenced by external air parameters, with temperature being the most critical factor. When temperature exceeds 50 °C, the effect of relative humidity becomes significant. During the falling rate stage, the dominant factors begin to gradually shift from external air parameters to internal characteristics within the sludge itself, with dry basis moisture content becoming the new key factor. When air velocity exceeds 1.5 m/s, the response of drying rate to air velocity is significantly influenced by changes in dry basis moisture content. The results of this study indicated that it is feasible to use machine learning models for predicting and explaining the low-temperature drying process of sludge. This provides valuable insights for the application of machine learning models in the development and management of drying strategies.

Natural factor-based spatial prediction and source apportionment of typical heavy metals in Chinese surface soil: Application of machine learning models

Predicting the natural distribution of heavy metals (HMs) in soil is important to understand the potential risk of pollution. However, suitable technologies are still lacking for wide scale due to the large spatial heterogeneity. In this study, we developed machine learning models for predicting natural contents of five typical HMs in soil, including As, Cd, Cr, Hg and Pb. It was found that the optional random forest (RF) model had the best performance with the R2 up to 0.64. Based on this model, potential distribution of the five HMs explored that elevated contents were mainly concentrated in the southwest and south central of China. Feature analysis illustrated that importance of natural factors followed the order of geological attributes > soil properties > climatic conditions > ecological functions. In particular, lithology of the parent material dominated the content of metals, with the contributions of 18-25%. Moreover, soil properties of pH, cation exchange capacity, profile depth of soil and vegetation coverage had different influences on HMs, due to the variability in the properties of different HMs. This study developed a mapping relationship between natural factors and soil HMs by data science method, which may provide instructive information for pollution control and planning decisions.

Unintended side effect of the coal-to-gas policy in North China Plain: Migration of the sources and health risks of ambient PAHs

Ongoing coal-to-gas (CTG) largely cut down both coal consumption and associated PM2.5. However, a knowledge gap still existed in CTG impacts on the other energy and organic pollutant emissions. Coupling on-site investigation with statistical yearbooks, we provided a more realistic energy evolutions before (BCTG), during (DCTG), and after (ACTG) the CTG for Hebei Province. Together, we examined the impacts of CTG derived energy conversion on PM2.5-bound PAHs at urban (UA)/suburban rural (SRA)/remote rural (RRA) sites in winter 2022. As expected, the consumptions of coal and natural gas (NG) far decreased and increased from BCTG to ACTG, respectively. Accidentally, biomass usage rose by 60.7%, and rural CTG acted as a main driver. Specially, SRA's NG-shortage and coal-stove demolition should be the main inducements, and RRA's coal-sale ban was another trigger in the early stage of CTG. ∑18PAHs and ∑8TPAHs stand for the sum of 18 PAHs and 8 toxic PAHs, respectively. ∑18PAHs (ng/m3) presented as SRA (81.8) > RRA (46.4) > UA (19.4). Biomass burning (BB) and NG combustion (NGC) contributed most to∑18PAHs of 31.0% and 23.1% at SRA, resulting in the highest ∑18PAHs, ∑18PAHs/PM2.5, and ∑8TPAHs/PM2.5, and incremental lifetime cancer risk values. Also, NGC has become the second largest contributor at UA. Variations in both diagnostic ratios and source-depend isomers further proved the prominence of NGC related PAHs at UA vs. SRA. Notably, RRA was least affected by the CTG, coal combustion (CC, 40.4%) and BB (32.6%) still occupied the top positions. In short, CTG gave rise to an upsurge in biomass usage, and the incremental PAHs emissions from BB vs. NGC. This study underlined that the priorities should be given to rural NG guarantee and subsidy retention, and biomass prohibition for further air quality improvement.

Release of Organochlorine Pollutants from Forest Fires: 1. Emission Factors and Revisiting Their Emissions in the Himalayan Regions

Worldwide forest fires have occurred frequently in recent years, a result of which may be the emission of so-called "legacy" organochlorine pollutants (OCPs) accumulated in forests. However, few studies have measured the emission factors (EFs) of the toxicity of the OCPs from forest fires. In this study, the EFs of vegetation burning were observed in forests along the altitudinal gradient from 1000 to 4200 m, and the EFs of ∑DDTs (dechlorodiphenylthrichloroethanes), HCB (hexachlorobenzene), ∑HCHs (hexachlorocyclohexanes), and ∑PCBs (polychlorinated biphenyls) were 2050 ± 1175, 379 ± 409, 48 ± 51, and 65 ± 59 ng/kg, respectively. Re-evaporation was the primary mechanism of the emission of OCP from forest fires. The masses of HCB, β-HCH, o,p'-DDD, p,p'-DDD, and PCB-28 in smoke increased 3-7 times compared with those in unburnt vegetation, suggesting the formation of these pollutants by the pyrolysis of biomass or other pollutants. Based on the observed EFs, previously estimated quantities of fire-emitted OCPs in the Himalayan regions were revisited. The DDT emissions from the Himalayan forest fires increased ∼70% compared with the previous estimation (from 19 to 32 kg/year). This highlighted that the EF observations could decrease the uncertainties of estimating OCP emissions from forest fires, which is helpful in revealing the potential roles of forest fires on global POP cycling.

Exploring the relationship between personal exposure to multiple water-soluble components and ROS in size-resolved PMs in solid fuel combustion households

Water-soluble species are the main components of particulate matters (PMs), which have important impacts on visibility, climate change and human health. Here, personal exposure (PE) to size-resolved PMs from housewives using different solid fuels (biomass and coal) was collected during winter in rural Yuncheng city, Fenwei Plain, China. The concentrations of water-soluble organic carbon (WSOC) and reactive oxygen species (ROS) were higher in the biomass group than coal group, whereas the concentrations of water-soluble inorganic ions and water-soluble nitrogen were higher in the coal group than biomass group. Almost all measured water-soluble components in both groups showed a pattern of increasing concentration with decreasing particle size, with more than 50% of WSOC and water-soluble total nitrogen (WSTN) enriched in PM0.25. The Pearson correlation result was in general agreement with the relationship between water-soluble components and ROS found by random forest model. There was a strong positive correlation between ROS and WSOC in PMs in the coal group, especially in PMs <0.25 μm, which may be due to the emission of a large number of transition metals chelated with WSOC from coal combustion. The contribution of Cl- and F- to ROS was greater in the biomass group. NO2- in both coal and biomass groups had a decent positive effect on ROS generation. The strongest positive linear correlation (R = 0.95) between ROS and K+ in total suspended particulates in the biomass group, whereas there was almost no contribution of K+ to ROS when particle size was distinguished or in random forest model, which reflects the specificity of K+ in inducing ROS. The present study provides new insights for a deeper exploration of the relationship between water-soluble components and oxidative potential in PE PMs from domestic combustion sources.

Fractions of smoke leakage into indoor space from residential solid fuel combustion in chimney stoves

Severe indoor air pollution from solid-fuel combustion is a global health concern. Although stove chimneys can expel most of the smoke to outside, unignorable amounts can remain indoors, known as indoor fugitive emissions. Quantitative analyses of indoor emission rates (IER) and indoor fugitive fractions (IFF) are limited, particularly in field settings. This study quantified the IERs and IFFs of particulate matters (PMs) from residential solid fuel combustion, covering different fuel-stove combinations in rural China. The study showed that both IERs and IFFs were not normally distributed. The median IER for PM2.5, which peaked at 860 mg/min, was 32 mg/min. IERs very significantly among different fuel and stove types, with biomass pellets and improved stoves demonstrating lower IERs than traditional biomass and coal. Approximately 27% of PM2.5 was leaked into indoor air, but this fraction ranged largely from a low percentage to 80%. Higher IFFs were observed for coals burned in traditional stoves. The median IFFs of organic carbon and elemental carbon were 26% (12%-43% as the interquartile range) and 19% (9%-40%), respectively. The chimney lifting effect significantly affected the degree of indoor leakage, with relatively low IFFs under high gas velocity conditions. Chimney exhaust and fugitive smoke had distinct size distributions, and small particles exhibited fewer leakages than coarse particles. The study provides valuable datasets for quantifying internal combustion impacts on indoor air quality and consequently human health.

Heavy metals in roadside soil along an expressway connecting two megacities in China: Accumulation characteristics, sources and influencing factors

Transportation is widely recognized as a significant contributor to heavy metal (HM) pollution in roadside soils. A better understanding of HM pollution in soils near expressways is crucial, particularly given the rapid expansion of expressway transportation in China in recent years. In this study, 329 roadside topsoil samples were collected along the Beijing-Tianjin Expressway, which connects two megacities in China. Chemical analysis showed that HM concentrations in the soil samples were generally below national limits. The mean pollution index (Pi) values for As, Cr, Cu, Ni, Pb, and Zn ranged from 0.94 to 1.01, while Cd and Hg exhibited slightly higher mean Pi values of 1.19 and 1.13, respectively. The Nemerow integrated pollution index values for all samples ranged from 0.71 to 4.97, with a mean of 1.26. This suggests a slight enrichment of HM above natural background levels, especially for Cd and Hg. Source apportionment using positive matrix factorization revealed that natural sources contributed the most to soil HMs (64.51 %), followed by agricultural sources (19.15 %), traffic sources (9.77 %), and industrial sources (6.57 %). The Shapley additive explanation analysis, based on the random forest model, identified soil organic carbon, deep soil HM content, altitude, total soil K2O, urbanization composite impact index, and total soil P as primary influencing factors. This indicates that the impact of transportation on roadside soils along the Beijing-Tianjin Expressway is currently relatively limited. The prominent influence of soil properties and altitude underscored the importance of "transport" and "receptor" in the soil HMs accumulation process at the local scale. These findings provide critical data and a scientific basis for decision-makers to develop policies for expressway design and roadside soil protection.

Predictions of the Optical Properties of Brown Carbon Aerosol by Machine Learning with Typical Chromophores

The linkages between BrC optical properties and chemical composition remain inadequately understood, with quantified chromophores explaining less than 25% of ambient aerosol light absorption. This study characterized 38 typical chromophores in aerosols collected in Xi'an, with light absorption contributions to BrC ranging from 1.6 ± 0.3 to 5.8 ± 2.6% at 365 nm. Based on these quantified chromophores, an interpretable machine learning model and the Shapley Additive Explanation (SHAP) method were employed to explore the relationships between BrC optical properties and chemical composition. The model attained high accuracy with Pearson correlation coefficients (r) exceeding 0.93 for the absorption coefficient (Absλ) and surpassing 0.57 for mass absorption efficiency (MAEλ) of BrC. It explains more than 80% of the variance in Abs and over 50% in MAE, significantly improving the understanding of BrC light absorption. Polycyclic aromatic hydrocarbons (PAHs) and oxygenated PAHs (OPAHs) with four and five rings exhibit significant positive effects on Absλ, suggesting that similar unidentified chromophores may also notably impact BrC optical characteristics. The model based on chromophore mass concentrations further simplifies studying BrC optical characteristics. This study advances understanding of the relationship between BrC composition and optical properties and guides the investigation of unrecognized chromophores.

Understanding the variability of ground-level ozone and fine particulate matter over the Tibetan plateau with data-driven approach

The Tibetan Plateau, known as the "Third Pole", is susceptible to ground-level ozone (O3) and fine particulate matter (PM2.5) pollution due to its unique high-altitude environment. This study constructed random forest regression models using multi-source data from ground measurements and meteorological satellites to predict variations in ground-level O3 and PM2.5 concentrations and their influencing factors across seven major cities in the Tibetan Plateau over two-year periods. The models successfully reproduced O3 and PM2.5 levels with satisfactory R-squared values of 0.71 and 0.73, respectively. Results reveal combustion-related carbon monoxide (CO) and nitrogen dioxide (NO2) as the most substantial influences on O3 and PM2.5 concentrations. Solar radiation, geographical factors, and meteorological variables also played crucial roles in driving pollutant variations. Conversely, transport-related and human activity factors exhibited relatively lower significance. High O3 and PM2.5 pollution occurred during pre-monsoon and post-monsoon/winter seasons, driven by solar radiation and emissions, respectively. While CO consistently contributed across cities and seasons, key influencing factors varied locally. This study unveils the key driving forces governing air pollutant variations across the Tibetan Plateau, shedding light on complex atmospheric processes in this unique high-altitude region.

Identifying Driving Factors of Atmospheric N2O5 with Machine Learning

Dinitrogen pentoxide (N2O5) plays an essential role in tropospheric chemistry, serving as a nocturnal reservoir of reactive nitrogen and significantly promoting nitrate formations. However, identifying key environmental drivers of N2O5 formation remains challenging using traditional statistical methods, impeding effective emission control measures to mitigate NOx-induced air pollution. Here, we adopted machine learning assisted by steady-state analysis to elucidate the driving factors of N2O5 before and during the 2022 Winter Olympics (WO) in Beijing. Higher N2O5 concentrations were observed during the WO period compared to the Pre-Winter-Olympics (Pre-WO) period. The machine learning model accurately reproduced ambient N2O5 concentrations and showed that ozone (O3), nitrogen dioxide (NO2), and relative humidity (RH) were the most important driving factors of N2O5. Compared to the Pre-WO period, the variation in trace gases (i.e., NO2 and O3) along with the reduced N2O5 uptake coefficient was the main reason for higher N2O5 levels during the WO period. By predicting N2O5 under various control scenarios of NOx and calculating the nitrate formation potential from N2O5 uptake, we found that the progressive reduction of nitrogen oxides initially increases the nitrate formation potential before further decreasing it. The threshold of NOx was approximately 13 ppbv, below which NOx reduction effectively reduced the level of night-time nitrate formations. These results demonstrate the capacity of machine learning to provide insights into understanding atmospheric nitrogen chemistry and highlight the necessity of more stringent emission control of NOx to mitigate haze pollution.

Characteristics of PM2.5 bounded carbonaceous aerosols, carbon dioxide and its stable carbon isotopes (δ13C) in rural households in northwest China: Effect of different fuel combustion

In order to fully understand the carbon emission from different fuels in rural villages of China, especially in the typical atmospheric pollution areas. The characteristics of carbonaceous aerosols and carbon dioxide (CO2) with its stable carbon isotope (δ13C) were investigated in six households, which two households used coal, two households used wood as well as two households used biogas and liquefied petroleum gas (LPG), from two rural villages in Fenwei Plain from March to April 2021. It showed that the fine particulate matter (PM2.5) emitted from biogas and LPG couldn't be as lower as expected in this area. However, the clean fuels could relatively reduce the emissions of organic carbon (OC) and element carbon (EC) in PM2.5 compare to the solid fuels. The pyrolyzed carbon (OP) accounted more total carbon (TC) in coal than the other fuels use households, indicating that more water-soluble OC existed, and it still had the highest secondary organic carbon (SOC) than the other fuels. Meantime, the coal combustions in the two villages had the highest CO2 concentration of 527.6 ppm and 1120.6 ppm, respectively, while the clean fuels could effectively reduce it. The average δ13C values (-26.9‰) was much lighter than almost all the outdoor monitoring and similar to the δ13C values for coal combustion and vehicle emission, showing that they might be the main contributors of the regional atmospheric aerosol in this area. During the sandstorm, the indoor PM2.5 mass and CO2 were increasing obviously. The indoor cancer risk of PAHs for adults and children were greater than 1 × 10-6, exert a potential carcinogenic risk to human of solid fuels combustion in rural northern China. It is important to continue concern the solid fuel combustion and its health impact in rural areas.

Understanding the spatial and seasonal variation of the ground-level ozone in Southeast China with an interpretable machine learning and multi-source remote sensing

Ground-level ozone (O3) pollution poses significant threats to both human health and air quality. This study uses ground observations and satellite retrievals to explore the spatiotemporal characteristics of ground-level O3 in Zhejiang Province, China. We created data-driven machine learning models that include meteorological, geographical and atmospheric parameters from multi-source remote sensing products, achieving good performance (Pearson's r of 0.81) in explaining regional O3 dynamics. Analyses revealed the crucial roles of temperature, relative humidity, total column O3, and the distributions and interactions of precursor (volatile organic compounds and nitrogen oxides) in driving the varied O3 patterns observed in Zhejiang. Furthermore, the interpretable modeling quantified multifactor interactions that sustain high O3 levels in spring and autumn, suppress O3 levels in summer, and inhibit O3 formation in winter. This work demonstrates the value of a combined approach using satellite and machine learning as an effective novel tool for regional air quality assessment and control.

Application of machine learning in atmospheric pollution research: A state-of-art review

Machine learning (ML) is an artificial intelligence technology that has been used in atmospheric pollution research due to their powerful fitting ability. In this review, 105 articles related to ML and the atmospheric pollution research are critically reviewed. Applications of ML in the prediction of atmospheric pollution (mainly particulate matters) are systematically described, including the principle of prediction, influencing factors and improvement measures. Researchers can improve the accuracy of the prediction model through three main aspects, namely considering the geographical features of the study area into the model, introducing the physical characteristics of pollutants, matching and optimizing ML models. And by using interpretable ML tools, researchers are able to understand the mechanism of the model and gain in-depth information. Then, the state-of-art applications of ML in the source apportionment of atmospheric particulate matter and the effect of atmospheric pollutants on human health are also described. In addition, the advantages and disadvantages of the current applications of ML in atmospheric pollution research are summarized, and the application perspective of ML in this field is elucidated. Given the scarcity of source apportionment applications and human health research, standardized research methods and specialized ML methods are required in atmospheric pollution research to connect these two disciplines.

Indoor exposure to polycyclic aromatic hydrocarbons associated with solid fuel use in rural China

Polycyclic aromatic hydrocarbons (PAHs) are widespread environmental contaminants associated with various health risks including lung cancer. Indoor exposure to PAHs, particularly from the indoor burning of fuels, is significant; however, long-term large-scale assessments of indoor PAHs are hampered by high costs and time-consuming in field sampling and laboratory experiments. A simple fuel-based approach and statistical regression models were developed as a trial to predict indoor BaP, as a typical PAH, in China, and consequently spatiotemporal variations in indoor BaP and indoor exposure contributions were discussed. The results show that the national population-weighted indoor BaP concentration has decreased substantially from 46.1 ng/m3 in 1992 to 6.60 ng/m3 in 2017, primarily due to the increased use of clean energies for cooking and heating. Indoor BaP exposure contributed to > 70% of the total inhalation exposure in most cities, particularly in regions where solid fuels are widely utilized. With limited experimental observation data in building statistical models, quantitative results of the study are associated with high uncertainties; however, the study undoubtedly supports effective countermeasures on indoor PAHs from solid fuel use and the importance of promoting clean household energy usage to improve household air quality.