Unmasking the sky: high-resolution PM2.5 prediction in Texas using machine learning techniques.

BACKGROUND: Although PM2.5 (fine particulate matter with an aerodynamic diameter less than 2.5 µm) is an air pollutant of great concern in Texas, limited regulatory monitors pose a significant challenge for decision-making and environmental studies. OBJECTIVE: This study aimed to predict PM2.5 concentrations at a fine spatial scale on a daily basis by using novel machine learning approaches and incorporating satellite-derived Aerosol Optical Depth (AOD) and a variety of weather and land use variables. METHODS: We compiled a comprehensive dataset in Texas from 2013 to 2017, including ground-level PM2.5 concentrations from regulatory monitors; AOD values at 1-km resolution based on images retrieved from the MODIS satellite; and weather, land-use, population density, among others. We built predictive models for each year separately to estimate PM2.5 concentrations using two machine learning approaches called gradient boosted trees and random forest. We evaluated the model prediction performance using in-sample and out-of-sample validations. RESULTS: Our predictive models demonstrate excellent in-sample model performance, as indicated by high R2 values generated from the gradient boosting models (0.94-0.97) and random forest models (0.81-0.90). However, the out-of-sample R2 values fall within a range of 0.52-0.75 for gradient boosting models and 0.44-0.69 for random forest models. Model performance varies slightly across years. A generally decreasing trend in predicted PM2.5 concentrations over time is observed in Eastern Texas. IMPACT STATEMENT: We utilized machine learning approaches to predict PM2.5 levels in Texas. Both gradient boosting and random forest models perform well. Gradient boosting models perform slightly better than random forest models. Our models showed excellent in-sample prediction performance (R2 > 0.9).

A review of machine learning applications in life cycle assessment studies.

Life Cycle Assessment (LCA) is a foundational method for quantitative assessment of sustainability. Increasing data availability and rapid development of machine learning (ML) approaches offer new opportunities to advance LCA. Here, we review current progress and knowledge gaps in applying ML techniques to support LCA, and identify future research directions for LCAs to better harness the power of ML. This review analyzes forty studies reporting quantitative assessment with a combination of LCA and ML methods. We found that ML approaches have been used for generating life cycle inventories, computing characterization factors, estimating life cycle impacts, and supporting life cycle interpretation. Most of the reviewed studies employed a single ML method, with artificial neural networks (ANNs) as the most frequently applied approach. Both supervised and unsupervised ML techniques were used in LCA studies. For studies using supervised ML, training datasets were derived from diverse sources, such as literature, lab experiments, existing databases, and model simulations. Over 70 % of these reviewed studies trained ML models with less than 1500 sample datasets. Although these reviewed studies showed that ML approaches help improve prediction accuracy, pattern discovery and computational efficiency, multiple areas deserve further research. First, continuous data collection and compilation is needed to support more reliable ML and LCA modeling. Second, future studies should report sufficient details regarding the selection criteria for ML models and present model uncertainty analysis. Third, incorporating deep learning models into LCA holds promise to further improve life cycle inventory and impact assessment. Finally, the complexity of current environmental challenges calls for interdisciplinary collaborative research to achieve deep integration of ML into LCA to support sustainable development.

Residential Proximity to Biorefinery Sources of Air Pollution and Respiratory Diseases in New York State.

Understanding potential health risks associated with biofuel production is critical to sustainably combating energy insecurity and climate change. However, the specific health impacts associated with biorefinery-related emissions are not yet well characterized. We evaluated the relationship between respiratory emergency department (ED) visits (2011-2015) and residential exposure to biorefineries by comparing 15 biorefinery sites to 15 control areas across New York (NY) State. We further examined these associations by biorefinery types (e.g., corn, wood, or soybean), seasons, and lower respiratory disease subtypes. We measured biorefinery exposure using residential proximity in a cross-sectional study and estimation of biorefinery emission via AERMOD-simulated modeling. After controlling for multiple confounders, we consistently found that respiratory ED visit rates among residents living within 10 km of biorefineries were significantly higher (rate ratios (RRs) range from 1.03 to 3.64) than those in control areas across our two types of exposure indices. This relationship held across biorefinery types (higher in corn and soybean biorefineries), seasons (higher in spring and winter), air pollutant types (highest for NO2), and respiratory subtypes (highest for emphysema). Further research is needed to confirm our findings.

Identifying and evaluating school environmental health indicators.

Children's health, attendance, and academic performance may be affected by school environmental hazards. While prior studies evaluated home environment and health, few have evaluated indicators of school in-/outdoor environment and health. This study addresses this knowledge gap by systematically reviewing and evaluating outdoor and indoor indicators of school environment and student's health and performance in New York State (NYS). We also evaluate statistical methodologies to address highly correlated indicators and integrate multiple exposures. Multiple school environmental indicators were identified from various existing NYS datasets. We summarized data sources, completeness, geographic and temporal coverage, and data quality for each indicator. Each indicator was evaluated by scientific basis/relevance, analytic soundness/feasibility, and interpretation/utility, and validated using objective NYS data. Finally, advanced variable selection methods were described and discussed. We have identified and evaluated multiple school environmental health indicators. It was found that mold and moisture problems, ventilation problems, ambient ozone, and PM2.5 levels are among the top priorities of school environmental issues/indicators in NYS, which were also consistent while using NYS data. Choice of best variable selection method should be made based on the research questions and data characteristics. The school environmental health indicators identified, and variable selection methods evaluated, in this study could be used by other researchers to help school officials and policy makers initiate prevention programs.

Spatially and Temporally Explicit Life Cycle Environmental Impacts of Soybean Production in the U.S. Midwest.

Understanding spatially and temporally explicit life cycle environmental impacts is critical for designing sustainable supply chains for biofuel and animal sectors. However, annual life cycle environmental impacts of crop production at county scale across mutiple years are lacking. To address this knowledge gap, this study used a combination of Environmental Policy Integrated Climate and process-based life cycle assessment models to quantify life cycle global warming (GWP), eutrophication (EU) and acidification (AD) impacts of soybean production in nearly 1000 Midwest counties yr-1 over 9 years. Sequentially, a machine learning approach was applied to identify the top influential factors among soil, climate, and farming practices, which drive the spatial and temporal heterogeneity of life cycle environmental impacts. The results indicated that significant variations existed in life cycle GWP, EU, and AD among counties and across years. Life cycle GWP impacts ranged from -11.4 to 22.0 kg CO2-eq kg soybean-1, whereas life cycle EU and AD impacts varied by factors of 302 and 44, respectively. Nitrogen application rates, temperature in March and soil texture were the top influencing factors for life cycle GWP impacts. In contrast, soil organic content and nitrogen application rate were the top influencing factors for life cycle EU and AD impacts.

Assessing Health Impacts of Conventional Centralized and Emerging Resource Recovery-Oriented Decentralized Water Systems.

Energy shortage and climate change call for sustainable water and wastewater infrastructure capable of simultaneously recovering energy, mitigating greenhouse gas emissions, and protecting public health. Although energy and greenhouse gas emissions of water and wastewater infrastructure are extensively studied, the human health impacts of innovative infrastructure designed under the principles of decentralization and resource recovery are not fully understood. In order to fill this knowledge gap, this study assesses and compares the health impacts of three representative systems by integrating life cycle and microbial risk assessment approaches. This study found that the decentralized system options, such as on-site septic tank and composting or urine diverting toilets, presented much lower life cycle cancer and noncancer impacts than the centralized system. The microbial risks of decentralized systems options were also lower than those of the centralized system. Moreover, life cycle cancer and noncancer impacts contributed to approximately 95% of total health impacts, while microbial risks were associated with the remaining 5%. Additionally, the variability and sensitivity assessment indicated that reducing energy use of wastewater treatment and water distribution is effective in mitigating total health damages of the centralized system, while reducing energy use of water treatment is effective in mitigating total health damages of the decentralized systems.

How community vulnerability factors jointly affect multiple health outcomes after catastrophic storms.

BACKGROUND: While previous studies uncovered individual vulnerabilities to health risks during catastrophic storms, few evaluated the population vulnerability which is more important for identifying areas in greatest need of intervention. OBJECTIVES: We assessed the association between community factors and multiple health outcomes, and developed a community vulnerability index. METHODS: We retained emergency department visits for several health conditions from the 2005-2014 New York Statewide Planning and Research Cooperative System. We developed distributed lag nonlinear models at each spatial cluster across eight counties in downstate New York to evaluate the health risk associated with Superstorm Sandy (10/28/2012-11/9/2012) compared to the same period in other years, then defined census tracts in clusters with an elevated risk as "risk-elevated communities", and all others as "unelevated". We used machine-learning techniques to regress the risk elevation status against community factors to determine the contribution of each factor on population vulnerability, and developed a community vulnerability index (CVI). RESULTS: Overall, community factors had positive contributions to increased community vulnerabilities to Sandy-related substance abuse (91.35%), injuries (70.51%), cardiovascular diseases (8.01%), and mental disorders (2.71%) but reversely contributed to respiratory diseases (-34.73%). The contribution of low per capita income (max: 22.08%), the percentage of residents living in group quarters (max: 31.39%), the percentage of areas prone to flooding (max: 38.45%), and the percentage of green coverage (max: 29.73%) tended to be larger than other factors. The CVI based on these factors achieved an accuracy of 0.73-0.90 across outcomes. CONCLUSIONS: Our findings suggested that substance abuse was the most sensitive disease susceptible to less optimal community indicators, whereas respiratory diseases were higher in communities with better social environment. The percentage of residents in group quarters and areas prone to flooding were among dominant predictors for community vulnerabilities. The CVI based on these factors has an appropriate predictive performance.