Future PM2 Research Articles

Updating On-Road Vehicle Emissions for China: Spatial Patterns, Temporal Trends, and Mitigation Drivers.

Vehicle emissions in China have been decoupled from rapid motorization owing to comprehensive control strategies. China's increasingly ambitious goals for better air quality are calling for deep emission mitigation, posing a need to develop an up-to-date emission inventory that can reflect the fast-developing policies on vehicle emission control. Herein, large-sample vehicle emission measurements were collected to update the vehicle emission inventory. For instance, ambient temperature correction modules were developed to depict the remarkable regional and seasonal emission variations, showing that the monthly emission disparities for total hydrocarbon (THC) and nitrogen oxide (NOX) in January and July could be up to 1.7 times in northern China. Thus, the emission ratios of THC and NOX can vary dramatically among various seasons and provinces, which have not been considered well by previous simulations regarding the nonlinear atmospheric chemistry of ozone (O3) and fine particulate matter (PM2.5) formation. The new emission results indicate that vehicular carbon monoxide (CO), THC, and PM2.5 emissions decreased by 69, 51, and 61%, respectively, during 2010-2019. However, the controls of NOX and ammonia (NH3) emissions were not as efficient as other pollutants. Under the most likely future scenario (PC [1]), CO, THC, NOX, PM2.5, and NH3 emissions were anticipated to reduce by 35, 36, 35, 45, and 4%, respectively, from 2019 to 2025. These reductions will be expedited with expected decreases of 56, 58, 74, 53, and 51% from 2025 to 2035, which are substantially promoted by the massive deployment of new energy vehicles and more stringent emission standards. The updated vehicle emission inventory can serve as an important tool to develop season- and location-specific mitigation strategies of vehicular emission precursors to alleviate haze and O3 problems.

PM2.5 Exposure and Associated Premature Mortality to 2100 in China Under Climate and Socioeconomic Change Scenarios

AbstractReducing PM2.5‐related premature mortality is essential for health‐related sustainable development. China, one of the most populated and PM2.5 polluted developing countries in the world, is striving to be in the vanguard for meeting this target. However, the global chemical transport methods for future PM2.5 projections are difficult to use and computationally expensive and may import measurement uncertainty into regional exposure assessments, thus bringing challenges to policy making. Here, we proposed an integrated PM2.5 projection model framework based on regional land use, emission, climate and population simulations. The ambient PM2.5 exposure and associated premature mortality to 2100 in China at a scale of 10 × 10 km were projected and compared under different development pathways. Ambient PM2.5 exposure is expected to peak in recent decades (2030–2060) with mean values ranging from 32.72 to 35.11 μg/m3 for different pathways, while associated premature mortality are projected to decrease (2273.9–778.59) (in thousands) over time (2030–2100). The change in the emission scenario with significant CH4 and NMVOC increases could lead to the greatest increase in average PM2.5 exposure (4.03 μg/m3), while the decrease (−0.90 μg/m3) was linked to BC, SO2, CH4, and NMVOC decreases. Meanwhile, premature deaths decrease (15–226,424) for most projection periods when land use, emissions, and population data were separately replaced with RCP2.6‐SSP1 data. Land use impacts in socioeconomic change scenarios could be moderate in certain regions. Therefore, the sustainable development pathway of the RCP2.6‐SSP1 scenario should be prioritized in China for future development considering both environmental protection and health sustainability.

MSAFormer: A Transformer-Based Model for PM2.5 Prediction Leveraging Sparse Autoencoding of Multi-Site Meteorological Features in Urban Areas

The accurate prediction of PM2.5 concentration, a matter of paramount importance in environmental science and public health, has remained a substantial challenge. Conventional methodologies for predicting PM2.5 concentration often grapple with capturing complex dynamics and nonlinear relationships inherent in multi-station meteorological data. To address this issue, we have devised a novel deep learning model, named the Meteorological Sparse Autoencoding Transformer (MSAFormer). The MSAFormer leverages the strengths of the Transformer architecture, effectively incorporating a Meteorological Sparse Autoencoding module, a Meteorological Positional Embedding Module, and a PM2.5 Prediction Transformer Module. The Sparse Autoencoding Module serves to extract salient features from high-dimensional, multi-station meteorological data. Subsequently, the Positional Embedding Module applies a one-dimensional Convolutional Neural Network to flatten the sparse-encoded features, facilitating data processing in the subsequent Transformer module. Finally, the PM2.5 Prediction Transformer Module utilizes a self-attention mechanism to handle temporal dependencies in the input data, predicting future PM2.5 concentrations. Experimental results underscore that the MSAFormer model achieves a significant improvement in predicting PM2.5 concentrations in the Haidian district compared to traditional methods. This research offers a novel predictive tool for the field of environmental science and illustrates the potential of deep learning in the analysis of environmental meteorological data.

Using the COVID-19 lockdown to identify atmospheric processes and meteorology influences on regional PM2.5 pollution episodes in the Beijing-Tianjin-Hebei, China

A major tool for curtailing the spread of COVID-19 pandemic in China was a nationwide lockdown implemented by the Chinese authorities, which also led to significant reductions in anthropogenic emissions and fine particulate matter (PM2.5) concentrations. However, the lockdown measures did not prevent high PM2.5 pollution episodes (EPs). Three severe EPs were identified in the Beijing-Tianjin-Hebei (BTH) region during the lockdown. The integrated process rate (IPR) analysis tool in the Community Multiscale Air Quality (CMAQ) model was employed to identify and quantify the contributions of individual atmospheric processes to PM2.5 formation during the EPs in four representative cities (Beijing, Tianjin, Shijiazhuang, and Baoding) of the BTH. The influence of meteorological conditions on the accumulation and dispersion of PM2.5 during the EPs was also elucidated. The results of the IPR showed that emissions and aerosol processes were the dominant sources of net surface PM2.5 in Beijing and Tianjin, constituting a total of 86.2% and 92.9%, respectively, while emissions, horizontal transport, and aerosol processes dominated the net surface PM2.5 formation in Shijiazhuang and Baoding. In addition, the three pollution episodes in Beijing and Tianjin were primarily driven by local emissions, while the pollution events in Shijiazhuang and Baoding were attributed to the combined local emissions and regional transport. Also, the EPs were driven by low planetary boundary layer heights, low vertical export of PM2.5 from the boundary layer to the free troposphere, and substantial horizontal import, especially in Shijiazhuang and Baoding. Furthermore, the elevated PM2.5 concentrations during the EPs were greatly enhanced by unfavorable meteorological conditions, with the first two EPs being characterized with low ventilation coefficient (VC) values in the four representative cities. This study improves the understanding of buildup of PM2.5 during the pollution episodes, and the results provide insights for designing more effective emissions control strategies to mitigate future PM2.5 pollution episodes.

STF-Net: An improved depth network based on spatio-temporal data fusion for PM2.5 concentration prediction

Accurate PM2.5 concentration prediction is essential and important for air pollution assessment. The air pollutant concentration (e.g. PM2.5 concentration) monitoring network consists of numerous stations deployed in different geographic locations, which generate multiple time series with dynamic spatio-temporal correlations. However, much existing work mainly focus on the observations themselves and ignore the dynamic spatio-temporal correlations among them, which is necessary for the accurate PM2.5 concentration prediction. On the one hand, geospatially distributed stations can provide data and additional hints for spatio-temporal dependency modeling of regional PM2.5 concentrations. On the other hand, the stations with high impact can be screened according to the correlation among values, which can reduce the computational complexity and improve the performance of the model. Inspired by the two insights mentioned above, in this paper, we design an improved depth network based on spatio-temporal data fusion model (short for, STF-Net) to predict future PM2.5 concentrations. Specifically, STF-Net is proposed based on a multilevel recurrent neural network (RNN) and temporal convolutional network that considers multiple stations’ values and other critical external data. And it is made up of two key components: (1) a data fusion component is mainly modeling the dynamic spatial correlation between the station-level data; (2) a prediction component aims to learn the temporal pattern of time series. The experiments with two real-world datasets demonstrate our method outperforms six baseline methods.

Modeling PM2.5 During Severe Atmospheric Pollution Episode in Lagos, Nigeria: Spatiotemporal Variations, Source Apportionment, and Meteorological Influences

AbstractIn 2021, the World Health Organization ranked Nigeria among the most polluted nations in the world, an indication of a deteriorating air quality, especially in the major urban areas of the country, which might pose adverse human health impacts. In this study, the Integrated Source Apportionment Method (ISAM) tool in the Community Multiscale Air Quality (CMAQ) model (CMAQ‐ISAM) was employed to quantify the contributions of eight emissions sectors to fine particulate matter (PM2.5) and its major components in Lagos during a prolonged severe atmospheric pollution episode (APE) in January 2021. The influence of meteorological conditions on the formation and dispersion of PM2.5 during the APE was also elucidated. Spatially, elevated PM2.5 concentrations were found in the northwestern region of Lagos, an urban area with larger anthropogenic emissions. Residential and industry were the two major sources of PM2.5. Residential contributed the most to total PM2.5 (∼40 μg/m3), followed by industry (∼20 μg/m3). High concentrations of secondary inorganic aerosols (SIA) at the northwest and upper northern areas of Lagos were majorly attributed to residential and industry sectors. In addition, sulfate accounted for the largest fraction of PM2.5, with residential, industry, and energy being its major sources. Residential, industry, and on‐road sectors dominated the contributions to nitrate, while residential and industry were the major contributors to ammonium. Furthermore, the elevated PM2.5 concentrations during the APE were greatly enhanced by unfavorable meteorological conditions. This study provides insights for designing effective emissions control strategies to mitigate future severe PM2.5 pollution episode in Lagos.

A deep spatio-temporal learning network for continuous citywide air quality forecast based on dense monitoring data

As urban air pollution becomes a severe environmental and societal issue globally, there is an increasing need on making air quality forecasts to prevent health and capital loss. Currently, most of the studies only provide air pollution forecasts on single or multiple sites, rather than citywide information. Due to complex urban structure and spatial transportation of air pollutants, the distribution of citywide air quality can be highly varying, which makes it extremely challenging to make accurate forecasts. Here, we address this challenge by exploring the potential of combining the dense air quality monitoring data with the deep learning technique. By using the fine-granular PM2.5 concentrations data collected from 417 micro monitoring stations in Lanzhou City, China, we proposed an end-to-end method to predict future citywide PM2.5 concentrations at 500 m × 500 m × 1 h resolution. This method consists of a spatio-temporal transformation component and a spatio-temporal prediction component, of which the former is responsible for converting PM2.5 concentration data into spatial-maps, while the spatio-temporal prediction network (i.e., Air-PredNet in our study) based on ConvLSTM in the latter is responsible for capturing the spatio-temporal characterizations in the spatial-map sequences and finally generating citywide air quality forecasts. The results show that our method can accurately predict the spatial distribution and variations of PM2.5 concentration with an RMSE of 2.12 μg/m3 for 1-h prediction and an RMSE of 4.09 μg/m3 for 6-h prediction. In addition, we also proved the robustness of Air-PredNet based on the prediction performance for O3 and NO2. Our results demonstrate the potential and feasibility of dense monitoring data combined with advanced data science methods for citywide urban air quality forecast, which supports the urban atmospheric environment fine management and decision-making.

PM2.5 Concentration Prediction Method Based on Temporal Attention Mechanism and CNN-LSTM

Accurately predicting PM2.5 concentration can effectively avoid the harm caused by heavy pollution weather to human health. In view of the non-linearity, time series characteristics, and the problem of large multi-step prediction errors in PM2.5 concentration data, a method combining Long Short-term Memory Network and Convolutional Neural Network with Time Pattern Attention mechanism (TPA-CNN-LSTM) is proposed. The method uses historical PM2.5 concentration data, historical meteorological data, and surrounding station data to predict the future 6-hour PM2.5 concentration of air quality monitoring stations. Firstly, CNN is used to obtain the spatial characteristics between multiple stations, secondly, LSTM is added after CNN to extract the temporal changes of non-linear data, and finally, to capture the key features of temporal information, Temporal Pattern Attention mechanism (TPA) is added. TPA can automatically adjust weights based on the input of each time step, and select the most relevant time step for prediction, thereby improving the accuracy of the model. An example analysis is conducted on the measured data of Beijing's air quality stations in 2018, and compared with other mainstream algorithms. The results show that the proposed model has higher prediction accuracy and performance.

Source apportionment of PM2.5 and the impact of future PM2.5 changes on human health in the monsoon-influenced humid subtropical climate

Local emission sources and long-range transport of air pollutants are the main causes of PM2.5 pollution in developing metropolises. A modeling system was used to determine the main contributors of PM2.5 and future changes in PM2.5 concentrations in Hanoi, one of the fast-growing and most polluted capital cities in the world. Additionally, BenMAP was used to estimate the number of premature deaths caused or avoided due to PM2.5 concentration changes. October and December 2017 as well as April and June 2018 were selected for analysis. The source apportionment approach revealed that the boundary conditions (BCON), were the primary source of PM2.5, accounting for 24–59.7%. Agriculture was a substantial contributor of PM2.5 within Hanoi in June and October, accounting for 27.8 and 18.8%, respectively. Projected emissions for 2030 from our previous work, which were estimated based on the action plans and scenarios proposed by the government, were used as input into the air quality modeling system to analyze changes in the concentrations of the total PM2.5 and its major components owing to emissions changes. It is shown that, the concentrations of PM2.5 can increase by 0.5 and 1.6 μg/m3 in April and December, and decrease by 3.2 and 1.5 μg/m3 in June and October, respectively. Using BenMAP , it is estimated that these changes can cause 45 and 95 premature deaths in April and December, but prevent 198 and 88 premature deaths in June and October respectively, in Hanoi.

A novel air pollution prediction system based on data processing, fuzzy theory, and multi-strategy improved optimizer.

PM2.5 is an important air pollution index, which has been widely concerned. An excellent PM2.5 prediction system can effectively help people protect their respiratory tract from injury. However, due to the strong uncertainty of PM2.5 data, the accuracy of traditional point prediction and interval prediction method is not satisfactory, especially for interval prediction, which is usually difficult to achieve the expected interval coverage (PINC). In order to solve the above problems, a new hybrid PM2.5 prediction system is proposed, which can quantify the certainty and uncertainty of future PM2.5 at the same time. For point prediction, a multi-strategy improved multi-objective crystal algorithm (IMOCRY) is proposed; the chaotic mapping and screening operator are added to make the algorithm more suitable for practical application. At the same time, the combined neural network based on unconstrained weighting method further improves the point prediction accuracy. For interval prediction, a new strategy is proposed, which uses the combination of fuzzy information granulation and variational mode decomposition to process the data. The high-frequency components are extracted by the VMD method, and then quantified by FIG method. By this way, the fuzzy interval prediction results with high coverage and low interval width are obtained. Through 4 groups of experiments and 2 groups of discussions, the advanced nature, accuracy, generalization, and fuzzy prediction ability of the prediction system are all satisfactory, which verified the effect of the system in practical application.

Emissions, meteorological and climate impacts on PM2.5 levels in Southern California using a generalized additive model: Historic trends and future estimates

Annual fine particulate matter (PM2.5) mass concentrations in the South Coast Air Basin (SoCAB) of California decreased from around 30 μg/m3 to 11 μg/m3 between 2000 and 2013 but rose from 11 μg/m3 to 13 μg/m3 between 2014 and 2018, raising important questions about the effectiveness of ongoing emission control policies. A two-step generalized additive model (GAM)-least squares approach was developed to explore the effects of emissions, large-scale climate events and meteorological factors on daily PM2.5 mass concentrations from 2000 to 2019 to quantitatively link impacts of emissions and meteorological on PM2.5 and to assess factors leading to the increase. The GAM had an R2 = 0.99 and root mean square error (RMSE) = 0.7 μg/m3 for the annual average PM2.5 concentrations. The two-step method had an R2 = 0.93 and RMSE = 4.07 μg/m3 for the 98th percentile 24-hr average PM2.5 concentrations. Variations in both emissions and relative humidity were of high importance compared with other included factors. Interactions of NH3 emissions with NOx and SO2 emissions, which lead to ammonium nitrate and sulfate aerosol formation, were the most important factors. Meteorological effects on PM2.5 explained the majority of the daily PM2.5 fluctuations. Emission changes (increases in SO2 and PM2.5) led to increases in predicted PM2.5 between 2014 and 2018. Predicted future PM2.5, using projected emissions and meteorological data from model simulations of representative concentration pathway (RCP) scenarios, are around 12 μg/m3 (annual) and 30 μg/m3 (98th percentile daily), which are both close to the current National Ambient Air Quality Standards (NAAQS) for PM2.5. Meteorological impacts on the predicted PM2.5 in future years lead to variations of ±2 μg/m3 for the annual average and ±5 μg/m3 for the 98th percentile daily level. Future climate changes lead to a probable year-to-year variation that will let PM2.5 levels in some years exceed the standard.

Associations of Autism Spectrum Disorder with PM2.5 Components: A Comparative Study Using Two Different Exposure Models.

This retrospective cohort study examined associations of autism spectrum disorder (ASD) with prenatal exposure to major fine particulate matter (PM2.5) components estimated using two independent exposure models. The cohort included 318 750 mother-child pairs with singleton deliveries in Kaiser Permanente Southern California hospitals from 2001 to 2014 and followed until age five. ASD cases during follow-up (N = 4559) were identified by ICD codes. Prenatal exposures to PM2.5, elemental (EC) and black carbon (BC), organic matter (OM), nitrate (NO3-), and sulfate (SO42-) were constructed using (i) a source-oriented chemical transport model and (ii) a hybrid model. Exposures were assigned to each maternal address during the entire pregnancy, first, second, and third trimester. In single-pollutant models, ASD was associated with pregnancy-average PM2.5, EC/BC, OM, and SO42- exposures from both exposure models, after adjustment for covariates. The direction of effect estimates was consistent for EC/BC and OM and least consistent for NO3-. EC/BC, OM, and SO42- were generally robust to adjustment for other components and for PM2.5. EC/BC and OM effect estimates were generally larger and more consistent in the first and second trimester and SO42- in the third trimester. Future PM2.5 composition health effect studies might consider using multiple exposure models and a weight of evidence approach when interpreting effect estimates.

Present-day and future PM2.5 and O3-related global and regional premature mortality in the EVAv6.0 health impact assessment model

We used the EVAv6.0 system to estimate the present (2015) and future (2015–2050) global PM2.5 and O3-related premature mortalities, using simulated surface concentrations from the GISS-E2.1-G Earth system model. The PM2.5-related global premature mortality is estimated to be 4.3 and 4.4 million by the non-linear and linear models, respectively. Ischemic heart diseases are found to be the leading cause of PM2.5-related premature deaths, contributing by 35% globally. Both long-term and short-term O3-related premature deaths are estimated to be around 1 million, globally. Overall, PM2.5 and O3-related premature mortality leads to 5.3–5.4 million premature deaths, globally. The global burden of premature deaths is mainly driven by the Asian region, which in 2015 contributes by 75% of the total global premature deaths. An increase from 6.2% to 8% in the PM2.5 relative risk as recommended by the WHO leads to an increase of PM2.5-related premature mortality by 28%, to 5.7 million. Finally, bias correcting the simulated PM2.5 concentrations in 2015 leads to an increase of up to 73% in the global PM2.5-related premature mortality, leading to a total number of global premature deaths of up to 7.7 million, implying the necessity of bias correction to get more robust health burden estimates. PM2.5 and O3-related premature mortality in 2050 decreases by up to 57% and 18%, respectively, due to emission reductions alone. However, the projected increase and aging of the population leads to increases of premature mortality by up to a factor of 2, showing that the population exposed to air pollution is more important than the level of air pollutants, highlighting that the population dynamics should be considered when setting up health assessment systems.

Deep Learning-Based Indoor Air Quality Forecasting Framework for Indoor Subway Station Platforms.

Particulate matter (PM) of sizes less than 10 µm () and 2.5 µm () found in the environment is a major health concern. As PM is more prevalent in an enclosed environment, such as a subway station, this can have a negative impact on the health of commuters and staff. Therefore, it is essential to continuously monitor PM on underground subway platforms and control it using a subway ventilation control system. In order to operate the ventilation system in a predictive way, a credible prediction model for indoor air quality (IAQ) is proposed. While the existing deterministic methods require extensive calculations and domain knowledge, deep learning-based approaches showed good performance in recent studies. In this study, we develop an effective hybrid deep learning framework to forecast future and on a subway platform using past air quality data. This hybrid framework is an integration of several deep learning frameworks, namely, convolution neural network (CNN), long short-term memory (LSTM), and deep neural network (DNN), and is called hybrid CNN-LSTM-DNN; it has the characteristics to capture temporal patterns and informative characteristics from the indoor and outdoor air quality parameters compared with the standalone deep learning models. The effectiveness of the proposed and forecasting framework is demonstrated using comparisons with the different existing deep learning models.

Estimating the health burden of ambient fine particulate matter in Korea

Korea has been making efforts to reduce the PM2.5 concentration through policy intervention but the PM2.5 level still belongs to a lower group among OECD countries. The aim of this study is to thus to evaluate the cases of premature deaths due to PM2.5 levels in Korea using AirQ+ model. We also examined the effect of reducing the burden of disease according to the current level of PM2.5 and future reduction targets in Korea. The burden of disease analysis of PM2.5 was based on cases of premature death of chronic obstructive pulmonary disease (COPD), stroke, ischemic heart disease (IHD), acute lower respiratory disease (ALRI), and lung cancer (LC). For the scenario, the current PM2.5 conc. of 25㎍/㎥, and the future reduction targets of 17㎍/㎥ and 15㎍/㎥ were applied. Also, Statistics Korea’s population and baseline mortality data by disease were used. The premature deaths of the five diseases due to the current and future PM2.5 scenarios in Korea were calculated as 12826, 12808, and 13223, respectively. The increase of premature deaths cases due to aging in the future population offset the effect of improving PM2.5, and thus the actual number of deaths did not decrease. However, the contribution rates of PM2.5 to the premature death cases decreased to 15.8%, 12.2%, and 11% as the concentration improved. In particular, the concentration rate of PM2.5 in ALRI, COPD, and LC decreased by more than 5%. In conclusion, it could be more effective to estimate the contribution rate rather than the absolute death case for the health improvement effect due to the reduction of PM2.5 in the case of countries where aging is expected in the future. In addition, this study showed that the reduction of PM2.5 in Korea was slightly more effective in reducing premature death of respiratory disease than cardiovascular disease.

Exposure of agricultural workers in California to wildfire smoke under past and future climate conditions

Wildfire activity in the western U.S. has increased in frequency and severity in recent decades. Wildfire smoke emissions contribute to elevated fine particulate matter (PM2.5) concentrations that are dangerous to public health. Due to the outdoor and physically demanding nature of their work, agricultural workers are particularly vulnerable to wildfire smoke pollution. In this study, we quantify the potential exposure of agricultural workers in California to past (2004–2009) and future (2046–2051) smoke PM2.5. We find that while absolute increases in smoke PM2.5 exposure are largest in northern California, agricultural regions in the Central Valley and Central Coast may be highly vulnerable to future increases in smoke PM2.5 concentrations. We find an increase from 6 to 8 million worker smoke exposure days (+35%) of ‘smokewave’ exposure for agricultural workers across the state under future climate conditions, with the largest increases in Tulare, Monterey, and Fresno counties. Under future climate conditions, we find 1.9 million worker smoke exposure days of agricultural worker exposure to levels of total PM2.5 pollution deemed ‘Unhealthy for Sensitive Groups.’ This is a 190% increase over past climate conditions. Wildfire smoke PM2.5 contributes, on average, to more than 90% of these daily PM2.5 exceedances compared with non-fire sources of air pollution. Using the recent extreme wildfire season of 2020 as a case study, we show that existing monitoring networks do not provide adequate sampling of PM2.5 in many future at-risk wildfire regions with large numbers of agricultural workers. Policies will need to consider the changing patterns of smoke PM2.5 exposure under future climate conditions to better protect outdoor agricultural workers.

Spatial and temporal characteristics of air pollutants and their health effects in China during 2019-2020.

This work presents the temporal and spatial characteristics of the major air pollutants and their associated health risks in China from 2019 to 2020, by using the monitoring data from 367 cities. The annual average PM2.5, PM10, NO2, SO2, CO, and O3 concentrations decreased by 10.9%, 13.2%, 9.3%, 10.1%, 9.4%, and 5.5% from 2019 to 2020. National average PM2.5 concentration in 2020 met the standard of 35μg/m3, and that of O3 decreased from 2019. COVID-19 lockdown affected NO2 level dramatically, yet influences on PM2.5 and O3 were less clear-cut. Positive correlations between PM2.5 and O3 were found, even in winter in all five key regions, e.g., Jing-Jin-Ji (JJJ), FenWei Plain (FWP), Yangtze River Delta (YRD), Pearl River Delta (PRD) and Chengdu-Chongqing Region (CCR), indicating importance of secondary production for both PM2.5 and O3. Large seasonal variability of PM2.5-SO2 correlation indicates a varying role of SO2 to PM2.5 pollution in different seasons; and generally weak correlations in winter between PM2.5 and NO2 or SO2 reveal the complexity of secondary formation processes to PM2.5 pollution in winter. Multilinear regression analysis between PM2.5 and SO2, NO2 and CO demonstrates that PM2.5 is more sensitive to the change of NO2 than SO2 in JJJ, FWP, PRD and CCR, suggesting a priority of NOx emission control for future PM2.5 reduction. Furthermore, the new World Health Organization Air Quality Guidelines (WHO AQG2021) were adopted to calculate the excess health risks (ER) as well as the health-risk based air quality index (HAQIWHO) of the pollutants. Such assessment points out the severity of air pollution associated health risks under strict standards: 40.0% of days had HAQIWHO>100, while only 14.4% days had AQI>100. PM2.5ER was generally larger than O3 ER, but O3 ER in low PM2.5 region (PRD) and during summer became more serious. Notably, NO2 ER became even more important than PM2.5 due to its strict limit of WHO AQG2021. Overall, our results highlight the increasing importance of O3 in both air quality evaluation and health risk assessment, and the importance of coordinated mitigation of multiple pollutants (mainly PM2.5, O3 and NO2) in protecting the public health.

RCL-Learning: ResNet and convolutional long short-term memory-based spatiotemporal air pollutant concentration prediction model

Predicting the concentration of air pollutants is an effective method for preventing pollution incidents by providing an early warning of harmful substances in the air. Accurate prediction of air pollutant concentration can more effectively control and prevent air pollution. In this study, a big data correlation principle and deep learning technology are used for a proposed model of predicting PM2.5 concentration. The model comprises a deep learning network model based on a residual neural network (ResNet) and a convolutional long short-term memory (LSTM) network (ConvLSTM). ResNet is used to deeply extract the spatial distribution features of pollutant concentration and meteorological data from multiple cities. The output is used as input to ConvLSTM, which further extracts the preliminary spatial distribution features extracted from the ResNet, while extracting the spatiotemporal features of the pollutant concentration and meteorological data. The model combines the two features to achieve a spatiotemporal correlation of feature sequences, thereby accurately predicting the future PM2.5 concentration of the target city for a period of time. Compared with other neural network models and traditional models, the proposed pollutant concentration prediction model improves the accuracy of predicting pollutant concentration. For 1- to 3-hours prediction tasks, the proposed pollutant concentration prediction model performed well and exhibited root mean square error (RMSE) between 5.478 and 13.622. In addition, we conducted multiscale predictions in the target city and achieved satisfactory performance, with the average RMSE value able to reach 22.927 even for 1- to 15-hours prediction tasks.

Current and Future Estimates of Marginal Emission Factors for Indian Power Generation.

Emission factors from Indian electricity remain poorly characterized, despite known spatial and temporal variability. Limited publicly available emissions and generation data at sufficient detail make it difficult to understand the consequences of emissions to climate change and air pollution, potentially missing cost-effective policy designs for the world's third largest power grid. We use reduced-form and full-form power dispatch models to quantify current (2017-2018) and future (2030-2031) marginal CO2, SO2, NOX, and PM2.5 emission factors from Indian power generation. These marginal emissions represent emissions changes due to small changes in demand. For 2017-2018, spatial variability in marginal CO2 emission factors range 3 orders of magnitude across India's states. There is limited seasonal and intraday variability with coal generation likely to meet changes in demand more than half the time in more than half of the states. Assuming the Government of India approximate 2030 targets, the median marginal CO2 emission factor across states decreases by approximately a factor of 2, but emission factors still span 3 orders of magnitude across states. Under 2030-2031 assumptions there is greater seasonal and intraday variability by up to factors of two and four, respectively. Estimates provide emission factors to evaluate interventions such as electric vehicles, increased air conditioning, and energy efficiency.

Ultrahigh-resolution PM2.5 estimation from top-of-atmosphere reflectance with machine learning: Theories, methods, and applications

Intra-urban pollution monitoring requires fine particulate (PM2.5) concentration mapping at ultrahigh-resolution (dozens to hundreds of meters). However, current PM2.5 concentration estimation, which is mainly based on aerosol optical depth (AOD) and meteorological data, usually had a low spatial resolution (kilometers) and severe spatial missing problem, cannot be applied to intra-urban pollution monitoring. To solve these problems, top-of-atmosphere reflectance (TOAR), which contains both the information about land and atmosphere and has high resolution and large spatial coverage, may be efficiently used for PM2.5 estimation. This study aims to systematically evaluate the feasibility of retrieving ultrahigh-resolution PM2.5 concentration at a large scale (national level) from TOAR. Firstly, we make a detailed discussion about several important but unsolved theoretic problems on TOAR-based PM2.5 retrieval, including the band selection, scale effect, cloud impact, and mapping quality evaluation. Secondly, four types and eight retrieval models are compared in terms of quantitative accuracy, mapping quality, model generalization, and model efficiency, with the pros and cons of each type summarized. Deep neural network (DNN) model shows the highest retrieval accuracy, and linear models were the best in efficiency and generalization. As a compromise, ensemble learning shows the best overall performance. Thirdly, using the highly accurate DNN model (cross-validated R2 equals 0.93) and through combining Landsat 8 and Sentinel 2 images, a 90 m and ∼4-day resolution PM2.5 product was generated. The retrieved maps were used for analyzing the fine-scale interannual pollution change inner the city and the pollution variations during novel coronavirus disease 2019 (COVID-19). Results of this study proves that ultrahigh resolution can bring new findings of intra-urban pollution change, which cannot be observed at previous coarse resolution. Lastly, some suggestions for future ultrahigh-resolution PM2.5 mapping research were given.