Modeling spatial variation of gaseous air pollutants and particulate matters in a Metropolitan area using mobile monitoring data

Mapping environmental noise of Guangzhou based on land use regression models

BACKGROUND

Environmental noise seriously affects people's health and life quality, but there is a scarcity of noise exposure data in metropolitan cities and at nighttime, especially in developing countries.

OBJECTIVE

This study aimed to assess the environmental noise level by land use regression (LUR) models and create daytime and nighttime noise maps with high-resolution of Guangzhou municipality.

METHODS

A total of 100 monitoring sites were randomly selected according to population density. The Equivalent continuous A-weighted sound pressure level (LAeq) was measured on weekdays from December 2022 to May 2023 during daytime [morning (7:30-12:00), afternoon (13:00-16:30), evening (17:30-22:00)], and nighttime (23:00-7:00) with 30 min of measurement at each site in winter/spring and summer/autumn. The LUR model was constructed by a supervised forward stepwise method to predict the noise exposure level via introducing the geographic predictor variables. A ten-fold cross-validation method was utilized to assess the performance of LUR models.

RESULTS

A total of 800 times of measurements were conducted and the average equivalent continuous LAeq of monitoring sites was 65.79 ± 7.45 dB(A). Urban areas exhibited higher noise levels than suburban areas (66.95 ± 7.37 dB(A) vs. 63.08 ± 6.94 dB(A), P < 0.001). Further, the noise level during the day was also significantly higher than during the night (67.18 ± 6.50 dB(A) vs. 61.60 ± 7.59 dB(A), P = 0.001). Four LUR models were developed with adjusted R2 ranging from 0.54 to 0.76, and the R2 of the ten-fold cross-validation varied from 0.61 to 0.79. Points of interests (POIs), traffic-related variables, and land use were important predictive factors in LUR models. Noise maps of daytime and nighttime with a resolution of 50 × 50 m were created, respectively.

CONCLUSION

Our results revealed that daytime and nighttime environmental noise exceeded the recommended values from the World Health Organization in Guangzhou. POIs, traffic-related variables, and land use were the main influencing factors of environmental noise level.

Spatial and spatiotemporal modelling of intra-urban ultrafine particles: A comparison of linear, nonlinear, regularized, and machine learning methods

BACKGROUND

Machine learning methods are proposed to improve the predictions of ambient air pollution, yet few studies have compared ultrafine particles (UFP) models across a broad range of statistical and machine learning approaches, and only one compared spatiotemporal models. Most reported marginal differences between methods. This limits our ability to draw conclusions about the best methods to model ambient UFPs.

OBJECTIVE

To compare the performance and predictions of statistical and machine learning methods used to model spatial and spatiotemporal ambient UFPs.

METHODS

Daily and annual models were developed from UFP measurements from a year-long mobile monitoring campaign in Quebec City, Canada, combined with 262 geospatial and six meteorological predictors. Various road segment lengths were considered (100/300/500 m) for UFP data aggregation. Four statistical methods included linear, non-linear, and regularized regressions, whereas eight machine learning regressions utilized tree-based, neural networks, support vector, and kernel ridge algorithms. Nested cross-validation was used for model training, hyperparameter tuning and performance evaluation.

RESULTS

Mean annual UFP concentrations was 13,335 particles/cm3. Machine learning outperformed statistical methods in predicting UFPs. Tree-based methods performed best across temporal scales and segment lengths, with XGBoost producing the overall best performing models (annual R2 = 0.78-0.86, RMSE = 2163-2169 particles/cm3; daily R2 = 0.47-0.48, RMSE = 8651-11,422 particles/cm3). With 100 m segments, other annual models performed similarly well, but their prediction surfaces of annual mean UFP concentrations showed signs of overfitting. Spatial aggregation of monitoring data significantly impacted model performance. Longer segments yielded lower RMSE in all daily models and for annual statistical models, but not for annual machine learning models.

CONCLUSIONS

The use of tree-based methods significantly improved spatiotemporal predictions of UFP concentrations, and to a lesser extent annual concentrations. Segment length and hyperparameter tuning had notable impacts on model performance and should be considered in future studies.

Do machine learning methods improve prediction of ambient air pollutants with high spatial contrast? A systematic review

BACKGROUND & OBJECTIVE

The use of machine learning for air pollution modelling is rapidly increasing. We conducted a systematic review of studies comparing statistical and machine learning models predicting the spatiotemporal variation of ambient nitrogen dioxide (NO2), ultrafine particles (UFPs) and black carbon (BC) to determine whether and in which scenarios machine learning generates more accurate predictions.

METHODS

Web of Science and Scopus were searched up to June 13, 2024. All records were screened by two independent reviewers. Differences in the coefficient of determination (R2) and Root Mean Square Error (RMSE) between best statistical and machine learning methods were compared across categories of methodological elements.

RESULTS

A total of 38 studies with 46 model comparisons (30 for NO2, 8 for UFPs and 8 for BC) were included. Linear non-regularized methods and Random Forest were most frequently used. Machine learning outperformed statistical models in 34 comparisons. Mean differences (95% confidence intervals) in R2 and RMSE between best machine learning and statistical models were 0.12 (0.08, 0.17) and 20% (11%, 29%) respectively. Tree-based methods performed best in 12 of 17 multi-model comparisons. Nonlinear or regularization regression methods were used in only 12 comparisons and provided similar performance to machine learning methods.

CONCLUSION

This systematic review suggests that machine learning methods, especially tree-based methods, may be superior to linear non-regularized methods for predicting ambient concentrations of NO2, UFPs and BC. Additional comparison studies using nonlinear, regularized and a wider array of machine learning methods are needed to confirm their relative performance. Future air pollution studies would also benefit from more explicit and standardized reporting of methodologies and results.

Characterizing sector-oriented roadside exposure to ultrafine particles (PM0.1) via machine learning models: Implications of covariates influences on sectors variability

Ultrafine particles (UFPs; PM0.1) possess intensified health risk due to their smaller size and unique spatial variability. One of major emission sources for UFPs is vehicle exhaust, which varies based on the traffic composition in each type of roadside sector. The current challenge of epidemiological UFPs study is limited characterization ability due to expensive instruments. This study assessed the UFPs particle number concentrations (UFPs PNC) exposure dose for typical healthy adults and children at three different roadside sectors, including industrial roadside (IN), residential roadside (RS), and urban background (UB). Furthermore, this study also developed and utilized machine learning (ML) algorithms that could accurately characterize the UFPs exposure dose and explain the covariates effects on the model outputs, representing the intra-urban variability of UFPs between sectors. It was found that the average inhaled UFPs dose for healthy adults and children during off-peak season (warm period) were 1.71 ± 0.19 × 1010; 1.28 ± 0.22 × 1010; 1.09 ± 0.18 × 1010 #/hour and 1.33 ± 0.15 × 1010; 0.99 ± 0.17 × 1010; 0.86 ± 0.14 × 1010 #/hour at IN, RS, UB. Inhaled UFPs were mainly deposited in tracheobronchial (TB) respiratory fraction for adults (67.7%) and in alveoli (ALV) fraction for children (67.5%). Among three ML algorithms implemented in this study, XGBoost possessed the highest UFPs PNC exposure dose estimation performances with R2 = 0.965; 0.959; 0.929 & RMSE = 0.79 × 108; 0.54 × 108; 0.15 × 105 #/hour at IN, RS, and UB which then followed by multiple linear regression (MLR), and random forest (RF). Furthermore, SHAP analysis from the XGBoost model has successfully pointed out the spatial variability of each roadside sector by quantifying the approximated contributions of covariates to the model's output. Findings in this study highlighted the potential use of ML models as an alternative for preliminary particle exposure source apportionment.

Capturing Exposure Disparities with Chemical Transport Models: Evaluating the Suitability of Downscaling Using Land Use Regression

High resolution exposure surfaces are essential to capture disparities in exposure to traffic-related air pollution in urban areas. In this study, we develop an approach to downscale Chemical Transport Model (CTM) simulations to a hyperlocal level (∼100m) in the Greater Toronto Area (GTA) under three scenarios where emissions from cars, trucks and buses are zeroed out, thus capturing the burden of each transportation mode. This proposed approach statistically fuses CTMs with Land-Use Regression using machine learning techniques. With this proposed downscaling approach, changes in air pollutant concentrations under different scenarios are appropriately captured by downscaling factors that are trained to reflect the spatial distribution of emission reductions. Our validation analysis shows that high-resolution models resulted in better performance than coarse models when compared with observations at reference stations. We used this downscaling approach to assess disparities in exposure to nitrogen dioxide (NO2) for populations composed of renters, low-income households, recent immigrants, and visible minorities. Individuals in all four categories were disproportionately exposed to the burden of cars, trucks, and buses. We conducted this analysis at spatial resolutions of 12, 4, 1 km, and 100 m and observed that disparities were significantly underestimated when using coarse spatial resolutions. This reinforces the need for high-spatial resolution exposure surfaces for environmental justice analyses.

Particulate and gaseous air pollutants exceed WHO guideline values and have the potential to damage human health in Faisalabad, Metropolitan, Pakistan

First-ever measurements of particulate matter (PM2.5, PM10, and TSP) along with gaseous pollutants (CO, NO2, and SO2) were performed from June 2019 to April 2020 in Faisalabad, Metropolitan, Pakistan, to assess their seasonal variations; Summer 2019, Autumn 2019, Winter 2019-2020, and Spring 2020. Pollutant measurements were carried out at 30 locations with a 3-km grid distance from the Sitara Chemical Industry in District Faisalabad to Bhianwala, Sargodha Road, Tehsil Lalian, District Chiniot. ArcGIS 10.8 was used to interpolate pollutant concentrations using the inverse distance weightage method. PM2.5, PM10, and TSP concentrations were highest in summer, and lowest in autumn or winter. CO, NO2, and SO2 concentrations were highest in summer or spring and lowest in winter. Seasonal average NO2 and SO2 concentrations exceeded WHO annual air quality guide values. For all 4 seasons, some sites had better air quality than others. Even in these cleaner sites air quality index (AQI) was unhealthy for sensitive groups and the less good sites showed Very critical AQI (> 500). Dust-bound carbon and sulfur contents were higher in spring (64 mg g-1) and summer (1.17 mg g-1) and lower in autumn (55 mg g-1) and winter (1.08 mg g-1). Venous blood analysis of 20 individuals showed cadmium and lead concentrations higher than WHO permissible limits. Those individuals exposed to direct roadside pollution for longer periods because of their occupation tended to show higher Pb and Cd blood concentrations. It is concluded that air quality along the roadside is extremely poor and potentially damaging to the health of exposed workers.

A comprehensive review of the development of land use regression approaches for modeling spatiotemporal variations of ambient air pollution: A perspective from 2011 to 2023

Land use regression (LUR) models are widely used in epidemiological and environmental studies to estimate humans' exposure to air pollution within urban areas. However, the early models, developed using linear regressions and data from fixed monitoring stations and passive sampling, were primarily designed to model traditional and criteria air pollutants and had limitations in capturing high-resolution spatiotemporal variations of air pollution. Over the past decade, there has been a notable development of multi-source observations from low-cost monitors, mobile monitoring, and satellites, in conjunction with the integration of advanced statistical methods and spatially and temporally dynamic predictors, which have facilitated significant expansion and advancement of LUR approaches. This paper reviews and synthesizes the recent advances in LUR approaches from the perspectives of the changes in air quality data acquisition, novel predictor variables, advances in model-developing approaches, improvements in validation methods, model transferability, and modeling software as reported in 155 LUR studies published between 2011 and 2023. We demonstrate that these developments have enabled LUR models to be developed for larger study areas and encompass a wider range of criteria and unregulated air pollutants. LUR models in the conventional spatial structure have been complemented by more complex spatiotemporal structures. Compared with linear models, advanced statistical methods yield better predictions when handling data with complex relationships and interactions. Finally, this study explores new developments, identifies potential pathways for further breakthroughs in LUR methodologies, and proposes future research directions. In this context, LUR approaches have the potential to make a significant contribution to future efforts to model the patterns of long- and short-term exposure of urban populations to air pollution.

Ambient air particulate total lung deposited surface area (LDSA) levels in urban Europe

This study aims to picture the phenomenology of urban ambient total lung deposited surface area (LDSA) (including head/throat (HA), tracheobronchial (TB), and alveolar (ALV) regions) based on multiple path particle dosimetry (MPPD) model during 2017-2019 period collected from urban background (UB, n = 15), traffic (TR, n = 6), suburban background (SUB, n = 4), and regional background (RB, n = 1) monitoring sites in Europe (25) and USA (1). Briefly, the spatial-temporal distribution characteristics of the deposition of LDSA, including diel, weekly, and seasonal patterns, were analyzed. Then, the relationship between LDSA and other air quality metrics at each monitoring site was investigated. The result showed that the peak concentrations of LDSA at UB and TR sites are commonly observed in the morning (06:00-8:00 UTC) and late evening (19:00-22:00 UTC), coinciding with traffic rush hours, biomass burning, and atmospheric stagnation periods. The only LDSA night-time peaks are observed on weekends. Due to the variability of emission sources and meteorology, the seasonal variability of the LDSA concentration revealed significant differences (p = 0.01) between the four seasons at all monitoring sites. Meanwhile, the correlations of LDSA with other pollutant metrics suggested that Aitken and accumulation mode particles play a significant role in the total LDSA concentration. The results also indicated that the main proportion of total LDSA is attributed to the ALV fraction (50 %), followed by the TB (34 %) and HA (16 %). Overall, this study provides valuable information of LDSA as a predictor in epidemiological studies and for the first time presenting total LDSA in a variety of European urban environments.

Spatial distribution of arsenic and metals suggest a high ecotoxicological potential in Puno Bay, Lake Titicaca, Peru

Spatial distribution and interpolation methods provide a summarized overview about the pollution dispersion, concerning the environment's quality. A high-altitude lake was taken as a model to assess the metalloid As and metals Cr, Cu, Ni, Pb, Zn distribution in superficial sediment and classify them according to their ecotoxicological potential in the aquatic environment. Surface sediments were collected from 11 sites along Puno Bay located at the western area of Lake Titicaca, Peru, and analyzed for pseudo total-metals. Sediment concentration data and quality were plotted using the Inverse Distance Weighting (IDW) as an interpolation method. High concentrations of As were found especially in the outer bay (81.73 mg.kg^-1). Spatial heterogeneity was evidenced for metal by the coefficient of variation, although no significative differences were observed between the two bays applying a Kruskall Wallis test (p < 0.05, df = 1). Sediment quality classification showed that most metal values were below TEL and toxicity was unlikely to occur, only As exceeded threefold PEL values, which categorized sediment as "Very Bad", indicating a rather high ecotoxicological potential to the aquatic environment. In conclusion, spatial analysis connected to interpolation methods demonstrated the superficial sediment heterogeneity in Puno Bay.

Ambient ultrafine particle (PM0.1): Sources, characteristics, measurements and exposure implications on human health

The problem of ultrafine particles (UFPs; PM_0.1) has been prevalent since the past decades. In addition to become easily inhaled by human respiratory system due to their ultrafine diameter (<100 nm), ambient UFPs possess various physicochemical properties which make it more toxic. These properties vary based on the emission source profile. The current development of UFPs studies is hindered by the problem of expensive instruments and the inexistence of standardized measurement method. This review provides detailed insights on ambient UFPs sources, physicochemical properties, measurements, and estimation models development. Implications on health impacts due to short-term and long-term exposure of ambient UFPs are also presented alongside the development progress of potentially low-cost UFPs sensors which can be used for future UFPs studies references. Current challenge and future outlook of ambient UFPs research are also discussed in this review. Based on the review results, ambient UFPs may originate from primary and secondary sources which include anthropogenic and natural activities. In addition to that, it is confirmed from various chemical content analysis that UFPs carry heavy metals, PAHs, BCs which are toxic in its nature. Measurement of ambient UFPs may be performed through stationary and mobile methods for environmental profiling and exposure assessment purposes. UFPs PNC estimation model (LUR) developed from measurement data could be deployed to support future epidemiological study of ambient UFPs. Low-cost sensors such as bipolar ion and ionization sensor from common smoke detector device may be further developed as affordable instrument to monitor ambient UFPs. Recent studies indicate that short-term exposure of UFPs can be associated with HRV change and increased cardiopulmonary effects. On the other hand, long-term UFPs exposure have positive association with COPD, CVD, CHF, pre-term birth, asthma, and also acute myocardial infarction cases.

Links between the concentrations of gaseous pollutants measured in different regions of Estonia

The factors that determine the concentrations of air pollutants (NO, NO₂, SO₂, O₃), measured in 8 monitoring stations (4 rural background, 3 urban, and 1 industrial) in Estonia, are studied applying the factor analysis. The factor analysis reveals remarkable impact of COVID-19 lockdown, effects caused by dramatic decrease in oil-shale based energy production in Estonia provoked by new socio-economic conditions such as elevated price for CO₂ emission quota, differences between rural and urban stations, maritime-continental difference for NO₂ and ozone, and specific industrial impact in case of SO₂. The multiple regression analysis to predict the ozone concentration in one rural background station at Tahkuse was performed, based on the ozone concentrations measured in other stations and the concentrations of NO, NO₂, and CO₂, recorded in the same station. It was found that the ozone concentration at Tahkuse is rather well predictable (determination coefficient, i.e., correlation coefficient squared, R ² = 0.714), using only the concentrations from another rural station at Saarejärve that is about 110 km away from Tahkuse. Adding all the available data into the list of regression analysis arguments, the model predictability is improved moderately (determination coefficient R ² = 0.795). Large model residuals above all tend to occur with the values measured and predicted at summer nights. Surprisingly, neither NO nor NO₂ concentration measured in the Tahkuse station did appear a good predictor for ozone (R ² = 0.02 and 0.05, respectively), possibly long-range transport of ozone (that has also experienced NO and/or NO₂ influence during transport) overrides the local effects of NO and/or NO₂.