Flexible Bayesian Ensemble Machine Learning Framework for Predicting Local Ozone Concentrations

Weather and the City: Machine Learning for Predicting and Attributing Fine Scale Air Quality to Meteorological and Urban Determinants

Urban air quality persists as a global concern, with critical health implications. This study employs a combination of machine learning (gradient boosting regression, GBR) and spatial analysis to better understand the key drivers behind air pollution and its prediction and mitigation strategies. Focusing on New York City as a representative urban area, we investigate the interplay between urban characteristics and weather factors, showing that urban features, including traffic-related parameters and urban morphology, emerge as crucial predictors for pollutants closely associated with vehicular emissions, such as elemental carbon (EC) and nitrogen oxides (NOx). Conversely, pollutants with secondary formation pathways (e.g., PM2.5) or stemming from nontraffic sources (e.g., sulfur dioxide, SO2) are predominantly influenced by meteorological conditions, particularly wind speed and maximum daily temperature. Urban characteristics are shown to act over spatial scales of 500 × 500 m2, which is thus the footprint needed to effectively capture the impact of urban form, fabric, and function. Our spatial predictive model, needing only meteorological and urban inputs, achieves promising results with mean absolute errors ranging from 8 to 32% when using full-year data. Our approach also yields good performance when applied to the temporal mapping of spatial pollutant variability. Our findings highlight the interacting roles of urban characteristics and weather conditions and can inform urban planning, design, and policy.

Equity and modeling in sustainability science: Examples and opportunities throughout the process

Equity is core to sustainability, but current interventions to enhance sustainability often fall short in adequately addressing this linkage. Models are important tools for informing action, and their development and use present opportunities to center equity in process and outcomes. This Perspective highlights progress in integrating equity into systems modeling in sustainability science, as well as key challenges, tensions, and future directions. We present a conceptual framework for equity in systems modeling, focused on its distributional, procedural, and recognitional dimensions. We discuss examples of how modelers engage with these different dimensions throughout the modeling process and from across a range of modeling approaches and topics, including water resources, energy systems, air quality, and conservation. Synthesizing across these examples, we identify significant advances in enhancing procedural and recognitional equity by reframing models as tools to explore pluralism in worldviews and knowledge systems; enabling models to better represent distributional inequity through new computational techniques and data sources; investigating the dynamics that can drive inequities by linking different modeling approaches; and developing more nuanced metrics for assessing equity outcomes. We also identify important future directions, such as an increased focus on using models to identify pathways to transform underlying conditions that lead to inequities and move toward desired futures. By looking at examples across the diverse fields within sustainability science, we argue that there are valuable opportunities for mutual learning on how to use models more effectively as tools to support sustainable and equitable futures.

Tree-structured parzen estimator optimized-automated machine learning assisted by meta-analysis for predicting biochar-driven N2O mitigation effect in constructed wetlands

Biochar is a carbon-neutral tool for combating climate change. Artificial intelligence applications to estimate the biochar mitigation effect on greenhouse gases (GHGs) can assist scientists in making more informed solutions. However, there is also evidence indicating that biochar promotes, rather than reduces, N2O emissions. Thus, the effect of biochar on N2O remains uncertain in constructed wetlands (CWs), and there is not a characterization metric for this effect, which increases the difficulty and inaccuracy of biochar-driven alleviation effect projections. Here, we provide new insight by utilizing machine learning-based, tree-structured Parzen Estimator (TPE) optimization assisted by a meta-analysis to estimate the potency of biochar-driven N2O mitigation. We first synthesized datasets that contained 80 studies on global biochar-amended CWs. The mitigation effect size was then calculated and further introduced as a new metric. TPE optimization was then applied to automatically tune the hyperparameters of the built extreme gradient boosting (XGBoost) and random forest (RF), and the optimum TPE-XGBoost obtained adequately achieved a satisfactory prediction accuracy for N2O flux (R2 = 91.90%, RPD = 3.57) and the effect size (R2 = 92.61%, RPD = 3.59). Results indicated that a high influent chemical oxygen demand/total nitrogen (COD/TN) ratio and the COD removal efficiency interpreted by the Shapley value significantly enhanced the effect size contribution. COD/TN ratio made the most and the second greatest positive contributions among 22 input variables to N2O flux and to the effect size that were up to 18% and 14%, respectively. By combining with a structural equation model analysis, NH4+-N removal rate had significant negative direct effects on the N2O flux. This study implied that the application of granulated biochar derived from C-rich feedstocks would maximize the net climate benefit of N2O mitigation driven by biochar for future biochar-based CWs.

Integrating Augmented In Situ Measurements and a Spatiotemporal Machine Learning Model To Back Extrapolate Historical Particulate Matter Pollution over the United Kingdom: 1980-2019

Historical PM2.5 data are essential for assessing the health effects of air pollution exposure across the life course or early life. However, a lack of high-quality data sources, such as satellite-based aerosol optical depth before 2000, has resulted in a gap in spatiotemporally resolved PM2.5 data for historical periods. Taking the United Kingdom as an example, we leveraged the light gradient boosting model to capture the spatiotemporal association between PM2.5 concentrations and multi-source geospatial predictors. Augmented PM2.5 from PM10 measurements expanded the spatiotemporal representativeness of the ground measurements. Observations before and after 2009 were used to train and test the models, respectively. Our model showed fair prediction accuracy from 2010 to 2019 [the ranges of coefficients of determination (R2) for the grid-based cross-validation are 0.71-0.85] and commendable back extrapolation performance from 1998 to 2009 (the ranges of R2 for the independent external testing are 0.32-0.65) at the daily level. The pollution episodes in the 1980s and pollution levels in the 1990s were also reproduced by our model. The 4-decade PM2.5 estimates demonstrated that most regions in England witnessed significant downward trends in PM2.5 pollution. The methods developed in this study are generalizable to other data-rich regions for historical air pollution exposure assessment.

Investigation of the Binding Fraction of PFAS in Human Plasma and Underlying Mechanisms Based on Machine Learning and Molecular Dynamics Simulation

More than 7000 per- and polyfluorinated alkyl substances (PFAS) have been documented in the U.S. Environmental Protection Agency's CompTox Chemicals database. These PFAS can be used in a broad range of industrial and consumer applications but may pose potential environmental issues and health risks. However, little is known about emerging PFAS bioaccumulation to assess their chemical safety. This study focuses specifically on the large and high-quality data set of fluorochemicals from the related environmental and pharmaceutical chemicals databases, and machine learning (ML) models were developed for the classification prediction of the unbound fraction of compounds in plasma. A comprehensive evaluation of the ML models shows that the best blending model yields an accuracy of 0.901 for the test set. The predictions suggest that most PFAS (∼92%) have a high binding fraction in plasma. Introduction of alkaline amino groups is likely to reduce the binding affinities of PFAS with plasma proteins. Molecular dynamics simulations indicate a clear distinction between the high and low binding fractions of PFAS. These computational workflows can be used to predict the bioaccumulation of emerging PFAS and are also helpful for the molecular design of PFAS to prevent the release of high-bioaccumulation compounds into the environment.

Interpretation and Prediction of the CO2 Sequestration of Steel Slag by Machine Learning

The utilization of steel slag for CO2 sequestration is an effective way to reduce carbon emissions. The reactivity of steel slag in CO2 sequestration depends mainly on material and process parameters. However, there are many puzzles in regard to practical applications due to the different evaluations of process parameters and the lack of investigation of material parameters. In this study, 318 samples were collected to investigate the interactive influence of 12 factors on the carbonation reactivity of steel slag by machine learning with SHapley Additive exPlanations (SHAP). Multilayer perceptron (MLP), random forest, and support vector regression models were built to predict the slurry-phase CO2 sequestration of steel slag. The MLP model performed well in terms of prediction ability and generalization with comprehensive interpretability. The SHAP results showed that the impact of the process parameters was greater than that of the material parameters. Interestingly, the iron ore phase of steel slag was revealed to have a positive effect on steel slag carbonation by SHAP analysis. Combined with previous literature, the carbonation mechanism of steel slag was proposed. Quantitative analysis based on SHAP indicated that steel slag had good carbonation reactivity when the mass fractions of "CaO + MgO", "SiO2 + Al2O3", "Fe2O3", and "MnO" varied from 50-55%, 10-15%, 30-35%, and <5%, respectively.

Development of a recurrent spatiotemporal deep-learning method coupled with data fusion for correction of hourly ozone forecasts

Ambient ozone (O3) predictions can be very challenging mainly due to the highly nonlinear photochemistry among its precursors, and meteorological conditions and regional transport can further complicate the O3 formation processes. The emission-based chemical transport models (CTM) are broadly used to predict O3 formation, but they may deviate from observations due to input uncertainties such as emissions and meteorological data, in addition to the treatment of O3 nonlinear chemistry. In this study, an innovative recurrent spatiotemporal deep-learning (RSDL) method with model-monitor coupled convolutional recurrent neural networks (ConvRNN) has been developed to improve O3 predictions of CTM. The RSDL method was first used to build the ConvRNN within a 24-h scale to characterize the spatiotemporal relationships between the monitored O3 data and CTM simulations, and then incorporated the recurrent pattern to achieve 72-h multi-site forecasts based on a pilot study over the Pearl River Delta (PRD) region of China. The results showed that the RSDL method predicted O3 with high accuracy over this case study, with an increase of 27.54% in the correlation coefficient (R) average for all sites as well as an increase in R of 0.14-0.21 for all cities compared to CTM. Moreover, the regional distribution of CTM was further improved by the RSDL predictions with the data fusion technique, which greatly reduced the underpredictions of O3 concentrations, particularly in high O3-level areas (concentrations >160 μg/m3), with a 33.55% reduction in the mean absolute error (MAE).

The application of machine learning to air pollution research: A bibliometric analysis

Machine learning (ML) is an advanced computer algorithm that simulates the human learning process to solve problems. With an explosion of monitoring data and the increasing demand for fast and accurate prediction, ML models have been rapidly developed and applied in air pollution research. In order to explore the status of ML applications in air pollution research, a bibliometric analysis was made based on 2962 articles published from 1990 to 2021. The number of publications increased sharply after 2017, comprising approximately 75% of the total. Institutions in China and United States contributed half of all publications with most research being conducted by individual groups rather than global collaborations. Cluster analysis revealed four main research topics for the application of ML: chemical characterization of pollutants, short-term forecasting, detection improvement and optimizing emission control. The rapid development of ML algorithms has increased the capability to explore the chemical characteristics of multiple pollutants, analyze chemical reactions and their driving factors, and simulate scenarios. Combined with multi-field data, ML models are a powerful tool for analyzing atmospheric chemical processes and evaluating the management of air quality and deserve greater attention in future.

Socioexposomics of COVID-19 across New Jersey: a comparison of geostatistical and machine learning approaches

BACKGROUND

Disparities in adverse COVID-19 health outcomes have been associated with multiple social and environmental stressors. However, research is needed to evaluate the consistency and efficiency of methods for studying these associations at local scales.

OBJECTIVE

To assess socioexposomic associations with COVID-19 outcomes across New Jersey and evaluate consistency of findings from multiple modeling approaches.

METHODS

We retrieved data for COVID-19 cases and deaths for the 565 municipalities of New Jersey up to the end of the first phase of the pandemic, and calculated mortality rates with and without long-term-care (LTC) facility deaths. We considered 84 spatially heterogeneous environmental, demographic and socioeconomic factors from publicly available databases, including air pollution, proximity to industrial sites/facilities, transportation-related noise, occupation and commuting, neighborhood and housing characteristics, age structure, racial/ethnic composition, poverty, etc. Six geostatistical models (Poisson/Negative-Binomial regression, Poison/Negative-Binomial mixed effect model, Poisson/Negative-Binomial Bersag-York-Mollie spatial model) and two Machine Learning (ML) methods (Random Forest, Extreme Gradient Boosting) were implemented to assess association patterns. The Shapley effects plot was established for explainable ML and change of support validation was introduced to compare performances of different approaches.

RESULTS

We found robust positive associations of COVID-19 mortality with historic exposures to NO2, population density, percentage of minority and below high school education, and other social and environmental factors. Exclusion of LTC deaths does not significantly affect correlations for most factors but findings can be substantially influenced by model structures and assumptions. The best performing geostatistical models involved flexible structures representing data variations. ML methods captured association patterns consistent with the best performing geostatistical models, and furthermore detected consistent nonlinear associations not captured by geostatistical models.

SIGNIFICANCE

The findings of this work improve the understanding of how social and environmental disparities impacted COVID-19 outcomes across New Jersey.