Kriging-Based Land-Use Regression Models That Use Machine Learning Algorithms to Estimate the Monthly BTEX Concentration

Geospatial artificial intelligence for estimating daytime and nighttime nitrogen dioxide concentration variations in Taiwan: A spatial prediction model

Nitrogen dioxide (NO2) is a major air pollutant primarily emitted from traffic and industrial activities, posing health risks. However, current air pollution models often underestimate exposure risks by neglecting the bimodal pattern of NO2 levels throughout the day. This study aimed to address this gap by developing ensemble mixed spatial models (EMSM) using geo-artificial intelligence (Geo-AI) to examine the spatial and temporal variations of NO2 concentrations at a high resolution of 50m. These EMSMs integrated spatial modelling methods, including kriging, land use regression, machine learning, and ensemble learning. The models utilized 26 years of observed NO2 measurements, meteorological parameters, geospatial layers, and social and season-dependent variables as representative of emission sources. Separate models were developed for daytime and nighttime periods, which achieved high reliability with adjusted R2 values of 0.92 and 0.93, respectively. The study revealed that mean NO2 concentrations were significantly higher at nighttime (9.60 ppb) compared to daytime (5.61 ppb). Additionally, winter exhibited the highest NO2 levels regardless of time period. The developed EMSMs were utilized to generate maps illustrating NO2 levels pre and during COVID restrictions in Taiwan. These findings could aid epidemiological research on exposure risks and support policy-making and environmental planning initiatives.

Evaluating long-term and high spatiotemporal resolution of wet-bulb globe temperature through land-use based machine learning model

BACKGROUND

The increase in global temperature and urban warming has led to the exacerbation of heatwaves, which negatively affect human health and cause long-term loss of work productivity. Therefore, a global assessment in temperature variation is essential.

OBJECTIVE

This paper is the first of its kind to propose land-use based spatial machine learning (LBSM) models for predicting highly spatial-temporal variations of wet-bulb globe temperature (WBGT), which is a heat stress indicator used to assess thermal comfort in indoor and outdoor environments, specifically for the main island of Taiwan.

METHODS

To develop spatiotemporal prediction models for both the working period and noon period, we calculated the WBGT of each weather station from 2001 to 2019 using temperature, humidity, and solar radiation data. These WBGT estimations were then used as the dependent variable for developing the spatiotemporal prediction models. To enhance model performance, we used innovative approaches that combined SHapley Additive exPlanations (SHAP) values for the selection of non-linear variables, along with machine learning algorithms for model development.

RESULTS

When incorporating temperature along with other land-use/land cover predictor variables, the performance of LBSM models was excellent, with an R2 value of up to 0.99. The LBSM models explained 98% and 99% of the spatial-temporal variations in WBGT for the working and noon periods, respectively, within the complete models. In the temperature-excluded models, the explained variances were 94% and 96% for the working and noon periods, respectively.

IMPACT

WBGT is a common method used by many organizations to access the impact of heat stress on human beings. However, limited studies have mentioned the association between WBGT and health impacts due to the absence of spatiotemporal databases. This study develops a new approach using land-use-based spatial machine learning (LBSM) models to better predict the fine spatial-temporal WBGT levels, with a 50-m × 50-m grid resolution for both working time and noontime. Our proposed methodology could be used in future studies aimed at evaluating the potential long-term loss of work productivity due to the effects of global warming or urban heat island.

An ensemble mixed spatial model in estimating long-term and diurnal variations of PM2.5 in Taiwan

Meteorology, human activities, and other emission sources drive diurnal cyclic patterns of air pollution. Previous studies mainly focused on the variation of PM_2.5 concentrations during daytime rather than nighttime. In addition, assessing the spatial variations of PM_2.5 in large areas is a critical issue for environmental epidemiological studies to clarify the health effects from PM_2.5 exposures. In terms of air pollution spatial modelling, using only a single model might lose information in capturing spatial and temporal correlation between predictors and pollutant levels. Hence, this study aimed to propose an ensemble mixed spatial model that incorporated Kriging interpolation, land-use regression (LUR), machine learning, and stacking ensemble approach to estimate long-term PM_2.5 variations for nearly three decades in daytime and nighttime. Three steps of model development were applied: 1) linear based LUR and Hybrid Kriging-LUR were used to determine influential predictors; 2) machine learning algorithms were used to enhance model prediction accuracy; 3) predictions from the selected machine learning models were fitted and evaluated again to build the final ensemble mixed spatial model. The results showed that prediction performance increased from 0.514 to 0.895 for daily, 0.478 to 0.879 for daytime, and 0.523 to 0.878 for nighttime when applying the proposed ensemble mixed spatial model compared with LUR. Results of overfitting test and extrapolation ability test confirmed the robustness and reliability of the developed models. The distance to the nearest thermal power plant, density of soil and pebbles fields, and funeral facilities might affect the variation of PM_2.5 levels between daytime and nighttime. The PM_2.5 level was higher in daytime compared with nighttime with little difference, revealing the importance of estimating nighttime PM_2.5 variations. Our findings also clarified the emission sources in daytime and nighttime, which serve as valuable information for air pollution control strategies establishment.

Living proximity to petrochemical industries and the risk of attention-deficit/hyperactivity disorder in children

Evidence regarding the negative neurodevelopmental effects of compound exposure to petrochemicals remains limited. We aimed to evaluate the association between exposure to petrochemical facilities and generated emissions during early life and the risk of attention-deficit/hyperactivity disorder (ADHD) development in children. We conducted a population-based birth cohort study using the 2004 to 2014 Taiwanese Birth Certificate Database and verified diagnoses of ADHD using the National Health Insurance Database. The level of petrochemical exposure in each participant's residential township was evaluated using the following 3 measurements: distance to the nearest petrochemical industrial plant (PIP), petrochemical exposure probability (accounting for monthly prevailing wind measurements), and monthly benzene concentrations estimated using kriging-based land-use regression models. We applied Cox proportional hazard models to evaluate the association. During the study period, 48,854 out of 1,863,963 children were diagnosed as having ADHD. The results revealed that residents of townships in close proximity to PIPs (hazard ratio [HR] = 1.20, 95% confidence interval [CI]: 1.16-1.23, <3 vs. ≥10 km), highly affected by petrochemical-containing prevailing winds (HR = 1.12, 95% CI: 1.08-1.16, ≥40% vs. <10%), and with high benzene concentrations (HR = 1.26, 95% CI: 1.23-1.29, ≥0.75 vs. <0.55 ppb) were consistently associated with the increased risk of ADHD development in children. The findings of the sensitivity analysis remained robust, particularly for the 2004 to 2009 birth cohort and for models accounting for a longer duration of postnatal exposure. This work provided clear evidence that living near petrochemical plants increases the risk of ADHD development in children. Further studies are warranted to confirm our findings.

A mixed spatial prediction model in estimating spatiotemporal variations in benzene concentrations in Taiwan

It is well known benzene negatively impacts human health. This study is the first to predict spatial-temporal variations in benzene concentrations for the entirety of Taiwan by using a mixed spatial prediction model integrating multiple machine learning algorithms and predictor variables selected by Land-use Regression (LUR). Monthly benzene concentrations from 2003 to 2019 were utilized for model development, and monthly benzene concentration data from 2020, as well as mobile monitoring vehicle data from 2009 to 2019, served as external data for verifying model reliability. Benzene concentrations were estimated by running six LUR-based machine learning algorithms; these algorithms, which include random forest (RF), deep neural network (DNN), gradient boosting (GBoost), light gradient boosting (LightGBM), CatBoost, extreme gradient boosting (XGBoost), and ensemble algorithms (a combination of the three best performing models), can capture how nonlinear observations and predictions are related. The results indicated conventional LUR captured 79% of the variability in benzene concentrations. Notably, the LUR with ensemble algorithm (GBoost, CatBoost, and XGBoost) surpassed all other integrated methods, increasing the explanatory power to 92%. This study establishes the value of the proposed ensemble-based model for estimating spatiotemporal variation in benzene exposure.

A hybrid methodology to quantitatively identify inorganic aerosol of PM2.5 source contribution

It is difficult to identify inorganic aerosol (IA) (primary and secondary), the main component of PM_2.5, without the significant tracers for sources. We are not aware of any studies specifically related to the IA's local contribution to PM_2.5. To effectively reduce the IA load, however, the contribution of local IA sources needs to be identified. In this work, we developed a hybrid methodology and applied online measurement of PM_2.5 and the associated compounds to (1) classify local and long-range transport PM_2.5, (2) identify sources of local PM_2.5 using PMF model, and (3) quantify local source contribution to IA in PM_2.5 using regression analysis. Coal combustion and iron ore and steel industry contributed the most amount of IA (~42.7%) in the study area (City of Taichung), followed by 32.9% contribution from oil combustion, 8.9% from traffic-related emission, 4.6% from the interactions between agrochemical applications and combustion sources (traffic-related emissions and biomass burning), and 2.3% from biomass burning. The methodology developed in this study is an important preliminary step for setting up future control policies and regulations, which can also be applied to any other places with serious local air pollution.

The Seasonality Impact of the BTEX Pollution on the Atmosphere of Arad City, Romania

Benzene, toluene, and total BTEX (benzene, toluene, ethylbenzene, and xylene) concentrations registered for one year (2016) have been determined every month for one high-density traffic area. The assessment was performed in Arad City, Romania, to evaluate these pollutants and their influence on the inhabitants’ health. The contaminants were sampled using a static sampling method and analyzed by gas chromatography coupled with mass spectrometry. Benzene was the most dominant among the BTEX compounds—the average concentrations ranged from 18.00 ± 1.32 µg m⁻³ in December to 2.47 ± 0.74 µg m⁻³ in August. The average toluene concentration over the year was 4.36 ± 2.42 µg m⁻³ (with a maximum of 9.60 ± 2.39 µg m⁻³ in November and a minimum of 1.04 ± 0.29 µg m⁻³ in May). The toluene/benzene ratio (T/B) was around 0.5, indicating substantial contributions from mobile sources (vehicles). The emission and accumulation of different aromatic compounds (especially benzene) could deteriorate the urban air quality. The lifetime cancer risk (LTCR) for benzene was found to be more than 10⁻⁵ in winter, including the inhabitants in the “probable cancer risk” category.

Comparison of Spatial Modelling Approaches on PM10 and NO2 Concentration Variations: A Case Study in Surabaya City, Indonesia

Because of fast-paced industrialization, urbanization, and population growth in Indonesia, there are serious health issues in the country resulting from air pollution. This study uses geospatial modelling technologies, namely land-use regression (LUR), geographically weighted regression (GWR), and geographic and temporal weighted regression (GTWR) models, to assess variations in particulate matter (PM₁₀) and nitrogen dioxide (NO₂) concentrations in Surabaya City, Indonesia. This is the first study to implement spatiotemporal variability of air pollution concentrations in Surabaya City, Indonesia. To develop the prediction models, air pollution data collected from seven monitoring stations from 2010 to 2018 were used as dependent variables, while land-use/land cover allocations within a 250 m to 5000 m circular buffer range surrounding the monitoring stations were collected as independent variables. A supervised stepwise variable selection procedure was applied to identify the important predictor variables for developing the LUR, GWR, and GTWR models. The developed models of LUR, GWR, and GTWR accounted for 49%, 50%, and 51% of PM₁₀ variations and 46%, 47%, and 48% of NO₂ variations, respectively. The GTWR model performed better (R² = 0.51 for PM₁₀ and 0.48 for NO₂) than the other two models (R² = 0.49–0.50 for PM₁₀ and 0.46–0.47 for NO₂), LUR and GWR. In the PM₁₀ model four predictor variables, public facility, industry and warehousing, paddy field, and normalized difference vegetation index (NDVI), were selected during the variable selection procedure. Meanwhile, paddy field, residential area, rainfall, and temperature played important roles in explaining NO₂ variations. Because of biomass burning issues in South Asia, the paddy field, which has a positive correlation with PM₁₀ and NO₂, was selected as a predictor. By using long-term monitoring data to establish prediction models, this model may better depict PM₁₀ and NO₂ concentration variations within areas across Asia.