Air Pollution Model Research Articles

Air Quality-Driven Traffic Management Using High-Resolution Urban Climate Modeling Coupled with a Large Traffic Simulation

This study presents a framework for integrating traffic simulation with high-resolution air pollution modeling to design adaptive traffic management policies aimed at reducing urban air pollution. Building on prior work that establishes the coupling of the MATSim traffic model with the PALM-4U urban climate model, this second part focuses on implementing a feedback loop to inform traffic management decisions based on simulated air pollution concentration levels. The research explores how traffic volumes and atmospheric conditions, such as boundary layer dynamics, influence air quality throughout the day. In an artificial case study of Berlin, a time-based toll is introduced, aimed at mitigating concentration peaks in the morning hours. The toll scheme is tested in two simulation scenarios and evaluated regarding the effectiveness of reducing air pollution levels, particularly NO2 during the morning hours. The case study results serve to illustrate the framework’s capabilities and highlight the potential of integrating traffic and environmental models for adaptive policy design. The presented approach provides a model for responsive urban traffic management, effectively aligning transportation policies with environmental goals to improve air quality in urban settings.

Development and evaluation of rapid, national-scale outdoor air pollution modelling and exposure assessment: Hybrid air dispersion exposure system (HADES)

Development and evaluation of rapid, national-scale outdoor air pollution modelling and exposure assessment: Hybrid air dispersion exposure system (HADES)

The INGENIOUS project: towards understanding air pollution in homes.

This paper provides an overview of the INGENIOUS (UnderstandING the sourcEs, traNsformations and fates of IndOor air pollUtantS) project, aiming to better understand air pollution in homes. Although our homes are the microenvironment in which we spend most of our time, we know relatively little about the sources, transformation processes and fates of indoor air pollutants, or our exposure to them. INGENIOUS aims to address this knowledge gap by delivering: an indoor emissions inventory for UK homes; comprehensive air pollutant measurements in 310 homes in Bradford using a combination of low cost-sensors and more advanced air quality instrumentation; an analysis of the impact of indoor air pollution on outdoor air quality and vice versa using mobile measurements; insight into future indoor air quality using detailed air pollution models; identification of indoor air pollutants that warrant further toxicological study; and better understanding of the barriers and facilitators for behaviour that drives improved indoor air quality. Median daily PM2.5 and CO2 concentrations varied from 7.8 μg m-3 and 666 ppm in the summer, to 16.4 μg m-3 and 857 ppm in the winter respectively in our sampled homes. Peak daily PM2.5 concentrations above 150 μg m-3 were frequently observed across all seasons, and were driven by cooking. Cooking activities also generated high concentrations of volatile organic compounds during emissions measurements, such as harmful aldehydes (up to ∼50 ppb), and alcohols (up to ∼600 ppb) from a chicken stir-fry. Our sampled homes displayed a wide variation in indoor pollutant concentrations, with a strong link to behaviour, including frequency and type of cooking activities, and use of ventilation.

Temporal trends and predictive modeling of air pollutants in Delhi: a comparative study of artificial intelligence models

Air pollution monitoring and modeling are the most important focus of climate and environment decision-making organizations. The development of new methods for air quality prediction is one of the best strategies for understanding weather contamination. In this research, different air quality parameters were forecasted, including Carbon Monoxide (CO), Nitrogen Monoxide (NO), Nitrogen Dioxide (NO2), Ozone (O3), Sulphur Dioxide (SO2), Fine Particles Matter (PM2.5), Coarse Particles Matter (PM10), and Ammonia (NH3). Hourly datasets were collected for air quality monitoring stations near Delhi, India, from November 25, 2020 to January 24, 2023. In this context, five intelligent models were developed, including Long Short-Term Memory (LSTM), Bidirectional Long-Short Term Memory (Bi-LSTM), Gated Recurrent Unit (GRU), Multilayer Perceptron (MLP), and Extreme Gradient Boosting (XGBoost). The modelling results revealed that Bi-LSTM model had the best predictability performance for forecasting CO with (R2 = 0.979), NO with (R2 = 0.961), NO2 with (R2 = 0.956), SO2 with (R2 = 0.955), PM10 with (R2 = 0.9751) and NH3 with (R2 = 0.971). Meanwhile, GRU and LSTM models performed better in forecasting O3 and PM2.5 with (R2 = 0.9624) and (R2 = 0.973), respectively. The current research provides illuminating visuals highlighting the potential of deep learning to comprehend air quality modeling, enabling improved environmental decisions.

An Advanced Hybrid Model Based On Stochastic - Eulerian Numerical Approach: Application To Atmospheric Pollution

In this paper, we propose for the first time to the best of our knowledge, extend the application of a stochastic Eulerian numerical approach based on the Extended Kalman Filter (EKFE.N.M.) to address the limitations of the Eulerian air pollution model CHIMERE. This approach integrates a comprehensive set of processes, including advection, turbulence, chemical reactions, emissions, and deposition, to model the dynamics of pollutant mass concentration. The EKF technique is employed to transform nonlinear dynamic problems into a succession of locally linearized ones, which are then used to estimate system states and adjust pollutant concentrations based on measured data. This stochastic approach is tested through two scenarios: one without external forces or control terms, and another that incorporates external factors like temperature, wind speed, and nitrogen dioxide as ozone precursors. A comparison of the obtained results with those from the standard CHIMERE model and studies from the literature demonstrates the accuracy and effectiveness of the proposed method.

Environmental Benefits of Turbo Roundabout: A Case Study of Noise and Air Pollution Modelling in the Village of Košťany Nad Turcom, Slovakia

Abstract Road transport represents a significant source of environmental pollution by air pollutants and noise. An important task is therefore to find ways to reduce these negative impacts. One option is also to change the conditions of the road transport infrastructure, for example by re-arranging the spatial layout of intersections and thus changing the driving mode of vehicles. The task of this study is to analyze the impact of the potential reconstruction of the intersection on the noise and air pollution load in its surroundings using modelling with the CadnaA program. This is a proposal to change the intersection (PK) to a turbo roundabout (TOK), with a specific structural layout and driving mode. As part of the study, a traffic survey was carried out to obtain detailed information about the volume and directions of road traffic at the intersection. The traffic volume at the intersection is specific to a high proportion of trucks, which also significantly affected the (un)change in the environmental load by noise and emissions. From the modeled variants, a reduction in noise pollution was observed, but no significant positive change in air pollution was demonstrated at TOK. At the observation point MP1, an increase in noise by 2 dB was recorded, and at MP4 a decrease of 3.6 dB. PM10 concentrations did not decrease at any observation point and increased most at MP1 by 7 µg/m3 according to the model.

Air pollution modelling of accidents involving hazardous substances

Abstract Air pollution modeling for accidents involving hazardous substances is a critical aspect of emergency response and environmental protection. This modeling aims to predict the dispersion and impact of toxic substances released into the atmosphere during accidents, such as industrial spills, transport accidents, chemical plant incidents or acts of terrorism. Such dangers are among the challenges which the modern society is facing and has to overcome. In this study we propose a systematic approach for organizing monitoring, collecting data from mobile sensors, creating situation development scenarios, modelling the spread of potential toxic-element pollution in dependence with the current meteorological conditions. The UNI-DEM model for simulation of the pollutants’ dispersion and transport in the air, as well as models for spread of leaking gas and liquid vapors are involved in this study. Finally, depending on the comprehensive analysis of the modeling results, an adequate response to such severe situations can be proposed. Assessment of safety, healthcare and environmental impacts should be the final result and the most socially important application of our work. The calculated pollutant concentrations can be compared with regulatory standards or health-based guidelines to assess potential health risks to exposed populations and environmental impacts on ecosystems. The importance of this information for emergency response planning, public health interventions, and environmental management strategies is beyond dispute.

Association of Tailpipe-Related and Nontailpipe-Related Air Pollution Exposure with Cognitive Decline in the Chicago Health and Aging Project.

Evidence suggests that long-term exposure to air pollution may increase the risk of dementia and related cognitive outcomes. A major source of air pollution is automotive traffic, which is modifiable by technological and regulatory interventions. We examined associations of four traffic-related air pollutants with rates of cognitive decline in a cohort of older adults. We analyzed data from the Chicago Health and Aging Project (CHAP), a longitudinal (1993-2012) community-based cohort study of older adults that included repeated assessments of participants' cognitive performance. Leveraging previously developed air pollution models, we predicted participant-level exposures to the tailpipe pollutants oxides of nitrogen () and nitrogen dioxide (), plus the nontailpipe pollutants copper and zinc found in coarse particulate matter [PM with aerodynamic diameter to () and , respectively], over the 3 y prior to each participant's baseline assessment. Using generalized estimating equations, we estimated covariate-adjusted associations of each pollutant with rates of cognitive decline. We probed the robustness of our results via several sensitivity analyses, including alterations to the length of the exposure assessment window and exploring the influence of pre- and post-baseline selection bias. Using data from 6,061 participants, estimated associations of these pollutant exposures with cognitive decline were largely inconsistent with large adverse effects. For example, a standard deviation () increment in corresponded to a slightly slower rate of cognitive decline [e.g., mean difference in change in global score, 0.010 standard unit/5 y, 95% confidence interval (CI): , 0.036]. The results of most of our sensitivity analyses were in generally similar to those of our main analyses, but our prebaseline selection bias results suggest that our analytic results may have been influenced by differential survivorship into our study sample. In this large prospective cohort study, we did not observe compelling evidence that long-term TRAP exposure is associated with cognitive decline. https://doi.org/10.1289/EHP14585.

From Space to Health: The Impact of Earth Observation Research on the Smart4COV19 Telemedicine project

Abstract This study investigates the potential of Earth observation research in improving air quality management and supporting telemedicine initiatives in response to the COVID-19 pandemic. Utilizing ground stations in Sofia and Burgas, we collected hourly measurements of NO2, CO, PM10, and PM2.5. Satellite data from TROPOMI-S5p and ground-based air quality observations were integrated to assess the spatial distribution of surface particulate matter concentrations. Our results demonstrate the feasibility of leveraging satellite-derived atmospheric chemistry data to enhance air pollution modeling and urban-scale air quality management. Furthermore, the technology developed for Burgas has the potential for expansion to other Bulgarian cities and replication in different urban centers. Overall, this research highlights the importance of Earth observation research in addressing critical environmental and public health challenges. Telemedicine via smartphones can help manage these risks by offering assistance to patients with mild symptoms, thereby minimizing their exposure to COVID-19 patients. The study also presents the findings of models used to convert TROPOMI – S5p aerosol data into PM concentrations in Burgas, Bulgaria.

Statistical inference and neural network training based on stochastic difference model for air pollution and associated disease transmission

A polluted air environment can potentially provoke infections of diverse respiratory diseases. The development of mathematical models can study the mechanism of air pollution and its effect on the spread of diseases. The key is to characterize the intrinsic correlation between the disease infection and the change in air pollutant concentration. In this paper, we establish a coupled discrete susceptible–exposed–infectious–susceptible (SEIS) model with demography to characterize the transmission of disease, and the change in the concentration of air pollutants is described in the form of the Beverton–Holt (BH) model with a time-varying inflow rate of air pollutants. Considering the periodic variation characteristics of data, time-varying parameters are defined as specific functional forms. We estimate the change point at which the parameters switch and the parameter values within the switching interval based on Bayesian statistical theory. The data fitting of the model can reflect the seasonal peaks and annual growth trends of values of air quality index (AQI) and the number of influenza-like illnesses (ILI) cases. However, the bias in data fitting indicates a more complex correlation pattern between disease and pollutant concentration changes. To explore unknown mechanisms, we propose the extended transmission-dynamics-informed neural network (TDINN) algorithm by combining deep learning with difference equations and obtain the curves of the transmission rate and inflow rate functions over time. The results show that neural network models can help us determine time-varying parameters in the model, thereby better reflecting the trend of data changes.

Development of over 30-years of high spatiotemporal resolution air pollution models and surfaces for California

California’s diverse geography and meteorological conditions necessitate models capturing fine-grained patterns of air pollution distribution. This study presents the development of high-resolution (100 m) daily land use regression (LUR) models spanning 1989–2021 for nitrogen dioxide (NO2), fine particulate matter (PM2.5), and ozone (O3) across California. These machine learning LUR algorithms integrated comprehensive data sources, including traffic, land use, land cover, meteorological conditions, vegetation dynamics, and satellite data. The modeling process incorporated historical air quality observations utilizing continuous regulatory, fixed site saturation, and Google Streetcar mobile monitoring data. The model performance (adjusted R2) for NO2, PM2.5, and O3 was 84 %, 65 %, and 92 %, respectively.Over the years, NO2 concentrations showed a consistent decline, attributed to regulatory efforts and reduced human activities on weekends. Traffic density and weather conditions significantly influenced NO2 levels. PM2.5 concentrations also decreased over time, influenced by aerosol optical depth (AOD), traffic density, weather, and land use patterns, such as developed open spaces and vegetation. Industrial activities and residential areas contributed to higher PM2.5 concentrations. O3 concentrations exhibited no significant annual trend, with higher levels observed on weekends and lower levels associated with traffic density due to the scavenger effect. Weather conditions and land use, such as commercial areas and water bodies, influenced O3 concentrations.To extend the prediction of daily NO2, PM2.5, and O3 to 1989, models were developed for predictors such as daily road traffic, normalized difference vegetation index (NDVI), Ozone Monitoring Instrument (OMI)–NO2, monthly AOD, and OMI-O3. These models enabled effective estimation for any period with known daily weather conditions.Longitudinal analysis revealed a consistent NO2 decline, regulatory-driven PM2.5 decreases countered by wildfire impacts, and spatially variable O3 concentrations with no long-term trend. This study enhances understanding of air pollution trends, aiding in identifying lifetime exposure for statewide populations and supporting informed policy decisions and environmental justice advocacy.

EV-miRNA associated with environmental air pollution exposures in the MADRES cohort.

Air pollution is a hazardous contaminant, exposure to which has substantial consequences for health during critical periods, such as pregnancy. MicroRNA (miRNA) is an epigenetic mechanism that modulates transcriptome responses to the environment and has been found to change in reaction to air pollution exposure. The data are limited regarding extracellular-vesicle (EV) miRNA variation associated with air pollution exposure during pregnancy and in susceptible populations who may be disproportionately exposed. This study aimed to identify EV-miRNA expression associated with ambient, residential exposure to PM2.5, PM10, NO2, O3 and with traffic-related NOx in 461 participants of the MADRES cohort, a low income, predominantly Hispanic pregnancy cohort based in Los Angeles, CA. This study used residence-based modeled air pollution data as well as Nanostring panels for EVmiRNA extracted with Qiagen exoRNeasy kits to evaluate 483 miRNA in plasma in early and late pregnancy. Average air pollution exposures were considered separately for 1-day, 1-week, and 8-week windows before blood collection in both early and late pregnancy. This study identified 63 and 66 EV-miRNA significantly associated with PM2.5 and PM10, respectively, and 2 miRNA associated with traffic-related NOX (False Discovery Rate-adjusted P-value < .05). Of 103 unique EV-miRNA associated with PM, 92% were associated with lung conditions according to HMDD (Human miRNA Disease Database) evidence. In particular, EV-miRNA previously identified with air pollution exposure also associated with PM2.5 and PM10 in this study were: miR-126, miR-16-5p, miR-187-3p, miR200b-3p, miR486-3p, and miR-582-3p. There were no significant differences in average exposures in early vs late pregnancy. Significant EV-miRNAs were only identified in late pregnancy with an 8-week exposure window, suggesting a vulnerable timeframe of exposure, rather than an acute response. These results describe a wide array of EV-miRNA for which expression is affected by PM exposure and may be in part mediating the biological response to ambient air pollution, with potential for health implications in pregnant women and their children.

Achieving deep transport energy demand reductions in the United Kingdom

The transport sector is a crucial yet challenging area to decarbonize, given its heavy reliance on fossil fuel usage, carbon-intensive infrastructure and car-centric lifestyles. It remains the largest contributor to local air pollution in cities yet has the potential to improve people's physical and mental health. This research investigated the potential contribution of transport energy demand reduction to climate change mitigation and improving public health. Using a comprehensive bottom-up modelling framework, the Transport Energy and Air pollution Model (TEAM), this study provides an integrated assessment of the impacts of deep mobility-related energy demand reductions, including lifecycle carbon emissions, local air pollution and health impacts. Using a sociotechnical scenario approach and the UK as a case study, this research reveals that energy demand reductions of up to 61 % by 2050 compared to baseline levels are achievable and can enhance citizens' quality of life. Business as usual approaches which rely on a technical transition miss the legislated carbon budgets and result in higher energy demand in 2050. More comprehensive scenarios deliver a reduction of up to 72 % in total lifecycle carbon emissions by 2050, with approximately half of the reduction achieved through mode shifting and avoiding travel, while the other half comes from vehicle energy efficiency, electrification, and downsizing of the vehicle fleets. The research shows that it can lead to significant co-benefits such as improved local air pollution and public health. The feasibility and practicality of policy measures and integrated strategies identified for achieving deep transport-energy demand reductions are discussed.

Computational modeling of air pollutants for aquatic risk: Prediction of ecological toxicity and exploring structural characteristics

Assessing the aquatic toxicity originating from air pollutants is essential in sustaining water resources and maintaining the ecosystem's safety. Quantitative structure-activity relationship (QSAR) models provide a computational tool for predicting pollutant toxicity, facilitating the identification/evaluation of the contaminants and identifying responsible structural fragments. One-vs-all (OvA) QSAR is a tailored approach to address multi-class QSAR problems. The study aims to determine five distinct levels of aquatic hazard categories for airborne pollutants using OvA-QSAR modeling containing 254 air contaminants. This QSAR analysis reveals the critical descriptors of air pollutants to target for molecular modification. Various factors, including the selection of relevant mechanistic descriptors, data quality, and outliers, determine the reliability of QSAR models. By employing feature selection and outlier identification approaches, the robustness and accuracy of our QSAR models were significantly increased, leading to more reliable predictions in chemical hazard assessment. The results revealed that models using the Random Forest algorithm performed the best based on the selected descriptors, with internal and external validation accuracy ranging from 71.90% to 97.53% and 76.47%–98.03%, respectively. This study indicated that the aquatic risk of air contaminants might be attributed predominantly to their sp3/sp2 carbon ratio, hydrogen-bond acceptor capability, hydrophilicity/lipophilicity, and van der Waals volumes. These structures can be critical in developing innovative strategies to mitigate or avoid the chemicals' harmful effects. Supporting air quality improvement, this study contributes to the rapid implementation of measures to protect aquatic ecosystems affected by air pollution.

Exposure to PM2.5 air pollution and mortality in Northern Ireland

Objective and ApproachThis study assesses the relationship between PM2.5 air pollution exposure and mortality in Northern Ireland. It draws on a rich array of individual-level mortality and Census data from the Northern Ireland Longitudinal Study linked to finely grained neighbourhood-level modelled air pollution data. ResultsCox proportional hazards models reveal associations between mortality risk and PM2.5 exposure, though the magnitude of the effects depends on the specification. Without adjustment for measured characteristics, an interquartile range increase in the five-year moving average of PM2.5 exposure is associated with a relatively large increase in the hazard of mortality. This effect is reduced after adjusting for individual-level demographic, socioeconomic and health-related factors as well as household-level factors; however, the hazard ratio remains above one and comparable with estimates from other studies. In further analysis focusing on changes in exposure resulting from address changes, propensity score matching is used to assess whether, among those with similar levels of pre-move exposure, those who move to areas with different pollution levels face subsequently different mortality risks. The results show a higher (lower) mortality risk for those moving to a higher (lower) pollution neighbourhood. Conclusions and ImplicationsThis study provides evidence of mortality effects of PM2.5 from a setting in which annual mean PM2.5 concentrations are low by international standards (averaging roughly 8 micrograms per cubic metre during the study period) and offers timely evidence to inform the delivery of Northern Ireland’s first-ever Clean Air Strategy.

New NOx and NO2 vehicle emission curves, and their implications for emissions inventories and air pollution modelling

Emissions of NOx and primary NO2 from road transport sources are highly influential in NO2 exposure at both local and regional scales; quantifying these accurately is therefore an important but challenging component of emissions inventory and air pollution model development. Results are presented from an urban air pollution model, after creation of new speed-emissions curves for NOx through the combination of available vehicle drive cycles and nearly 500,000 UK-based remote sensing measurements of exhaust emissions. Vehicle power-based relationships are applied to 1 Hz drive cycle datasets, with random sampling of the outputs allowing generation of the new curves. These demonstrate significantly higher emissions than those predicted by existing curves for most Euro VI HGVs, and among successive petrol and diesel passenger cars; this may be partly explained by relatively low UK ambient temperatures, as well as an underestimation of the level of tampering with HGV SCR systems. Implementation of the curves in a detailed emissions inventory for London, UK in 2019 leads to substantially improved air pollution model performance for NOx/NO2; normalised mean bias reduces in magnitude, changing from −0.18 to +0.01 for NOx and −0.12 to +0.01 for NO2. The curves developed are widely applicable, and the novel approach outlined has the potential to improve source apportionment and future model predictions under differing policy scenarios, produce better exposure estimates for health-related studies and revise NOx emissions budgets for compliance with the NEC Directive, all of which are important for the development of mitigation policies.

Enhancing spatial modeling and risk mapping of six air pollutants using synthetic data integration with convolutional neural networks

Air pollution poses significant risks to human health and the environment, necessitating effective air quality management strategies. This study presents a novel approach to air quality management by integrating an autoencoder (AE) with a convolutional neural network (CNN) algorithm in Tehran city of Iran. One of the primary and vital problems in deep learning is model complexity, and the complexity of a model is affected by data distribution, data complexity, and information volume. AE provide a helpful way to denoise input data and make building deep learning models much more efficient. The proposed methodology enables spatial modeling and risk mapping of six air pollutants, namely, particulate matter 2.5 (PM2.5), particulate matter 10 (PM10), sulfur dioxide (SO2), nitrogen dioxide (NO2), ozone (O3), and carbon monoxide (CO). For air pollution modelling, data from a spatial database containing the annual average of six pollutants from 2012 to 2022 was utilized. The model considered various parameters influencing air pollution: altitude, humidity, distance to industrial areas, NDVI (normalized difference vegetation index), population density, rainfall, distance to the street, temperature, traffic volume, wind direction, and wind speed. The risk map accuracy was assessed using the area under the receiver operating characteristic (ROC) curve for six pollutants. Among them, NO2, PM10, CO, PM2.5, O3, and SO2 exhibited the highest accuracy with values of 0.964, 0.95, 0.896, 0.878, 0.877, and 0.811, respectively, in the risk map generated by the CNN-AE model. The findings demonstrated the CNN-AE model’s impressive precision when generating the pollution risk map.

Modeling air pollution around major dumpsites in Lagos State using geospatial methods with solutions

Air pollution is gradually becoming a major health concern globally and Nigeria is especially susceptible to air pollution Available data indicates that vehicular traffics, industries, wildfire and biomass burning are significant sources of air pollutants such as Particulate Matter PM, Carbon Monoxide CO, Nitrogen Dioxide NO2, Sulphur Dioxide SO2 and Volatile Organic Compounds VOCs in the air in Lagos State. The aim of this study is to perform geospatial modelling of air pollution around Olusosun and Solous dumpsites in Lagos State. The study analyses, assesses, and predicts air pollution emission in these dumpsites and examines their possible health impact on residents. Using both ground and satellite-based data, different GIS techniques were used to model this pollution concentration. These methods include Kriging, Regression, and Landfill Gas Emissions Model (LANDGEM) which helps to analyse, model and predict the air pollutant levels around these dumpsite. The average pollution level of Particulate Matter with a diameter that is 2.5 micrometers (PM2.5) in Olusosun was 94.29 μg/m3 while those in Solous was 94.12 μg/m3 over a 24 h period. This is way above the Air Quality Guideline recommended by the WHO which is 15 μg/m3 over a 24 h period. This is similar to the high pollution level of NO2 which was 237.64 μg/m3 in Olusosun and 255.84 μg/m3 in Solous as against the WHO Air Quality Guideline of 25 μg/m3 over a 24 h period. SO2 level is also varyingly high in Olusosun and Solous. The reading of 109.27 μg/m3 in Olusosun is considerably high as compared to the reading of 43.82 μg/m3 in Solous. However, both readings are relatively higher than the Air Quality Guideline recommended by the WHO which was 40 μg/m3. The CO air quality level of 0.55 mg/m3 in Olusosun and 0.52 mg/m3 in Solous is within the accepted level of the Air Quality Guideline recommended by the WHO. This shows that CO is not a pollution threat in the major dumpsites in Lagos. This is similar to the readings of PM5 and PM10 which are within the WHO Air Quality Guidelines over a 24 h period. The Moderate Resolution Imaging Spectroradiometer (MODIS) data shows a trend of increasing concentration of PM2.5 over the 5 study months due to an increase in the quantity of waste disposal, which shows that high waste disposal is directly proportional to high level of PM2.5. It is recommended that living farther (more than 15 km) away from the dumpsites can reduce the high risk of air pollution and its related diseases for residents. Also, there is need for proper incineration and maintenance of waste to reduce the level of air and other pollutants in and around the dumpsites.

Neural Networks for Geospatial Data

Analysis of geospatial data has traditionally been model-based, with a mean model, customarily specified as a linear regression on the covariates, and a Gaussian process covariance model, encoding the spatial dependence. While nonlinear machine learning algorithms like neural networks are increasingly being used for spatial analysis, current approaches depart from the model-based setup and cannot explicitly incorporate spatial covariance. We propose NN-GLS, embedding neural networks directly within the traditional Gaussian process (GP) geostatistical model to accommodate nonlinear mean functions while retaining all other advantages of GP, like explicit modeling of the spatial covariance and predicting at new locations via kriging. In NN-GLS, estimation of the neural network parameters for the nonlinear mean of the Gaussian Process explicitly accounts for the spatial covariance through use of the generalized least squares (GLS) loss, thus, extending the linear case. We show that NN-GLS admits a representation as a special type of graph neural network (GNN). This connection facilitates the use of standard neural network computational techniques for irregular geospatial data, enabling novel and scalable mini-batching, backpropagation, and kriging schemes. We provide methodology to obtain uncertainty bounds for estimation and predictions from NN-GLS. Theoretically, we show that NN-GLS will be consistent for irregularly observed spatially correlated data processes. We also provide a finite sample concentration rate, which quantifies the need to accurately model the spatial covariance in neural networks for dependent data. To our knowledge, these are the first large-sample results for any neural network algorithm for irregular spatial data. We demonstrate the methodology through numerous simulations and an application to air pollution modeling. We develop a software implementation of NN-GLS in the Python package geospaNN. Supplementary materials for this article are available online, including a standardized description of the materials available for reproducing the work.

Development and evaluation of the interactive Model for Air Pollution and Land Ecosystems (iMAPLE) version 1.0

Abstract. Land ecosystems are important sources and sinks of atmospheric components. In turn, air pollutants affect the exchange rates of carbon and water fluxes between ecosystems and the atmosphere. However, these biogeochemical processes are usually not well presented in Earth system models, limiting the explorations of interactions between land ecosystems and air pollutants from regional to global scales. Here, we develop and validate the interactive Model for Air Pollution and Land Ecosystems (iMAPLE) by upgrading the Yale Interactive Terrestrial Biosphere Model with process-based water cycles, fire emissions, wetland methane (CH4) emissions, and trait-based ozone (O3) damage. Within iMAPLE, soil moisture and temperature are dynamically calculated based on the water and energy balance in soil layers. Fire emissions are dependent on dryness, lightning, population, and fuel load. Wetland CH4 is produced but consumed through oxidation, ebullition, diffusion, and plant-mediated transport. The trait-based scheme unifies O3 sensitivity of different plant functional types (PFTs) with the leaf mass per area. Validations show correlation coefficients (R) of 0.59–0.86 for gross primary productivity (GPP) and 0.57–0.84 for evapotranspiration (ET) across the six PFTs at 201 flux tower sites and yield an average R of 0.68 for CH4 emissions at 44 sites. Simulated soil moisture and temperature match reanalysis data with high R above 0.86 and low normalized mean biases (NMBs) within 7 %, leading to reasonable simulations of global GPP (R=0.92, NMB=1.3 %) and ET (R=0.93, NMB=-10.4 %) against satellite-based observations for 2001–2013. The model predicts an annual global area burned of 507.1 Mha, close to the observations of 475.4 Mha with a spatial R of 0.66 for 1997–2016. The wetland CH4 emissions are estimated to be 153.45 Tg [CH4] yr−1 during 2000–2014, close to the multi-model mean of 148 Tg [CH4] yr−1. The model also shows reasonable responses of GPP and ET to the changes in diffuse radiation and yields mean O3 damage of 2.9 % to global GPP. iMAPLE provides an advanced tool for studying the interactions between land ecosystems and air pollutants.