Impacts of pollution heterogeneity on population exposure in dense urban areas using ultra-fine resolution air quality data

Combining Google traffic map with deep learning model to predict street-level traffic-related air pollutants in a complex urban environment

BACKGROUND

Traffic-related air pollution (TRAP) is a major contributor to urban pollution and varies sharply at the street level, posing a challenge for air quality modeling. Traditional land use regression models combined with data from fixed monitoring stations may be unable to predict and characterize fine-scale TRAP, especially in complex urban environments influenced by various features. This study aims to estimate fine-scale (50 m) concentrations of nitrogen oxides (NO and NO₂) in Hong Kong using a deep learning (DL) structured model.

METHODS

We collected data from mobile air quality sensors on buses and crowd-sourced Google real-time traffic status as a proxy for real-time traffic emissions. Our DL model was compared with existing machine learning models to assess performance improvements. Using an interpretable machine learning method, we hierarchically evaluated the global, local, and interaction effects for different features.

RESULTS

Our DL model outperformed existing machine learning models, achieving R2 values of 0.72 for NO and 0.69 for NO₂. The incorporation of traffic status as a key predictor improved model performance by 9% to 17%. The interpretable machine learning method revealed the importance of traffic-related features and their pairwise interactions.

CONCLUSION

The results indicate that traffic-related features significantly contribute to TRAP and provide insights and guidance for urban planning. By incorporating crowd-sourced Google traffic information, we assessed traffic abatement scenarios that could inform targeted strategies for improving urban air quality.

Characterizing the distribution pattern of traffic-related air pollutants in near-road neighborhoods

In near-road neighborhoods, residents are more frequently exposed to traffic-related air pollution (TRAP), and they are increasingly aware of pollution levels. Given this consideration, this study adopted portable air pollutant sensors to conduct a mobile monitoring campaign in two near-road neighborhoods, one in an urban area and one in a suburban area of Shanghai, China. The campaign characterized spatiotemporal distributions of fine particulate matter (PM2.5) and black carbon (BC) to help identify appropriate mitigation measures in these near-road micro-environments. The study identified higher mean TRAP concentrations (up to 4.7-fold and 1.7-fold higher for PM2.5 and BC, respectively), lower spatial variability, and a stronger inter-pollutant correlation in winter compared to summer. The temporal variations of TRAP between peak hour and off-peak hour were also investigated. It was identified that district-level PM2.5 increments occurred from off-peak to peak hours, with BC concentrations attributed more to traffic emissions. In addition, the spatiotemporal distribution of TRAP inside neighborhoods revealed that PM2.5 concentrations presented great temporal variability but almost remained invariant in space, while the BC concentrations showed notable spatiotemporal variability. These findings provide valuable insights into the unique spatiotemporal distributions of TRAP in different near-road neighborhoods, highlighting the important role of hyperlocal monitoring in urban micro-environments to support tailored designing and implementing appropriate mitigation measures.

High-resolution estimation of near-surface ozone concentration and population exposure risk in China

Considering the spatial and temporal effects of atmospheric pollutants, using the geographically and temporally weighted regression and geo-intelligent random forest (GTWR-GeoiRF) model and Sentinel-5P satellite remote sensing data, combined with meteorological, emission inventory, site observation, population, elevation, and other data, the high-precision ozone concentration and its spatiotemporal distribution near the ground in China from March 2020 to February 2021 were estimated. On this basis, the pollution status, near-surface ozone concentration, and population exposure risk were analyzed. The findings demonstrate that the estimation outcomes of the GTWR-GeoiRF model have high precision, and the precision of the estimation results is higher compared with that of the non-hybrid model. The downscaling method enhances estimation results to some extent while addressing the issue of limited spatial resolution in some data. China's near-surface ozone concentration distribution in space shows obvious regional and seasonal characteristics. The eastern region has the highest ozone concentrations and the lowest in the northeastern region, and the wintertime low is higher than the summertime high. There are significant differences in ozone population exposure risks, with the highest exposure risks being found in China's eastern region, with population exposure risks mostly ranging from 0.8 to 5.

Detection of atmospheric aerosols and terrestrial nanoparticles collected in a populous city in southern Brazil

The main objective of this study is to analyze hazardous elements in nanoparticles (NPs) (smaller than 100 nm) and ultrafine particles (smaller than 1 µm) in Porto Alegre City, southern Brazil using a self-made passive sampler and Sentinel-3B SYN satellite images in 32 collection points. The Aerosol Optical Thickness proportion (T550) identification was conducted using images of the Sentinel-3B SYN satellite at 634 points sampled in 2019, 2020, 2021, and 2022. Focused ion beam scanning electron microscopy analyses were performed to identify chemical elements present in NPs and ultrafine particles, followed by single-stage cascade impactor to be processed by high-resolution transmission electron microscopy. This process was coupled with energy-dispersive X-ray spectroscopy and later analysis via secondary ion mass spectrometry. Data was acquired from Sentinel-3B SYN images, normalized to a standard mean of 0.83 µg/mg, at moderate spatial resolution (260 m), and modeled in the Sentinel Application Platform (SNAP) software v.8.0. Statistical matrix data was generated in the JASP software (Jeffreys's Amazing Statistics Program) v.0.14.1.0 followed by a K-means cluster analysis. The results demonstrate the presence of between 1 and 100 nm particles of the following chemical elements: Si, Al, K, Mg, P, and Ti. Many people go through these areas daily and may inhale or absorb these elements that can harm human health. In the Sentinel-3B SYN satellite images, the sum of squares in cluster 6 is 168,265 and in cluster 7 a total of 21,583. The use of images from the Sentinel-3B SYN satellite to obtain T550 levels is of great importance as it reveals that atmospheric pollution can move through air currents contaminating large areas on a global scale.

Simulating spatio-temporal dynamics of surface PM2.5 emitted from Alaskan wildfires

Wildfire is a major disturbance agent in Arctic boreal and tundra ecosystems that emits large quantities of atmospheric pollutants, including PM2.5. Under the substantial Arctic warming which is two to three times of global average, wildfire regimes in the high northern latitude regions are expected to intensify. This imposes a considerable threat to the health of the people residing in the Arctic regions. Alaska, as the northernmost state of the US, has a sizable rural population whose access to healthcare is greatly limited by a lack of transportation and telecommunication infrastructure and low accessibility. Unfortunately, there are only a few air quality monitoring stations across the state of Alaska, and the air quality of most remote Alaskan communities is not being systematically monitored, which hinders our understanding of the relationship between wildfire emissions and human health within these communities. Models simulating the dispersion of pollutants emitted by wildfires can be extremely valuable for providing spatially comprehensive air quality estimates in areas such as Alaska where the monitoring station network is sparse. In this study, we established a methodological framework that is based on an integration of the Hybrid Single-Particle Lagrangian Integrated Trajectory (HYSPLIT) model, the Wildland Fire Emissions Inventory System (WFEIS), and the Arctic-Boreal Vulnerability Experiment (ABoVE) Wildfire Date of Burning (WDoB) dataset, an Arctic-oriented fire product. Through our framework, daily gridded surface-level PM2.5 concentrations for the entire state of Alaska between 2001 and 2015 for which wildfires are responsible can be estimated. This product reveals the spatio-temporal patterns of the impacts of wildfires on the regional air quality in Alaska, which, in turn, offers a direct line of evidence indicating that wildfire is the dominant driver of PM2.5 concentrations over Alaska during the fire season. Additionally, it provides critical data inputs for research on understanding how wildfires affect human health which creates the basis for the development of effective and efficient mitigation efforts.

Assessing the health impacts of PM2.5 and ozone pollution and their comprehensive correlation in Chinese cities based on extended correlation coefficient

The coordinated control of PM_2.5 and ozone pollution is becoming more and more important in the current and next stage of Chinese environmental pollution control. Existing studies are unable to provide sufficient quantitative assessments of the correlation of PM_2.5 and ozone pollution to support the coordinated control of the two air pollutants. This study develops a systematic method to comprehensively assess the correlation between PM_2.5 and ozone pollution, including the evaluation of the impact of two air pollutants on human health and the extended correlation coefficient (ECC) for assessing the bivariate correlation index of PM_2.5-ozone pollution in Chinese cities. According to the latest studies on epidemiology conducted in China, we take cardiovascular and cerebrovascular diseases and respiratory diseases as the ozone pollution's health burden when evaluating the health impact of ozone pollution. The results show that the health impact of PM_2.5 in China decreases by 25.9 % from 2015 to 2021, while the health impact of ozone increases by 11.8 %. The ECC of 335 cities in China shows an increasing-decreasing trend but has generally increased from 2015 to 2021. The study provides important support for an in-depth understanding of the correlation and development trend of Chinese PM_2.5 and ozone pollution by classifying the comprehensive PM_2.5-ozone correlation performances of Chinese cities into four types. China or other countries will get better environmental benefits by implementing different coordinated management approaches for different correlative types of regions based on the assessment method in this study.