Abstract The availability of data on ambient ammonia exposure is limited in both spatial and temporal resolutions, especially in rural areas, despite its potential health impacts and rapidly increasing emissions. In this study, a land‐use regression (LUR)‐based spatiotemporal modeling approach was applied to predict agricultural‐related ammonia concentrations at a fine spatial scale, extending the method's application in terms of the species of pollutant and the type of region. In this approach, spatiotemporal variations are primarily explained by a linear combination of temporal basis functions, with weights determined by LUR. We developed the model using ammonia concentrations we collected in 2022 and 2023 from 10 monitoring sites within Urbana‐Champaign, Illinois, along with geographic covariates focused on agricultural features. Analysis of the model structure showed that livestock farms and populated areas are associated with ammonia peaks in summer and winter, whereas croplands are more closely associated with peaks in spring and fall. Cross‐validated predictions showed high accuracies at most monitoring sites ( R 2 ≥ 0.69 at 90% of the sites), though accuracy varied by location, with lower accuracy linked to unique time trend in ammonia concentrations at a location. Finally, we mapped the seasonal predictions of ammonia concentrations across the study area at 30‐m spatial resolution. In contrast, alternative models using simpler formulation or a smaller selection of covariates could not achieve an acceptable level of performance in identical evaluations. These results demonstrate the potential of LUR‐based models as valuable tools for investigating agriculture‐related air pollution exposure and highlight their broader applicability.
Abstract Atmospheric fine particulate matter (PM 2.5 ) is one of the most important risk factors for various respiratory and cardiovascular diseases. We develop a two‐phase data‐driven framework that uses geostationary Aerosol Optical Depth (AOD) retrievals to predict hourly PM 2.5 at 0.01° × 0.01° resolution and support exposure assessment across multiple spatial and temporal contexts. We use a U‐Net‐like partial convolutional neural network framework inspired by image inpainting to address missing Aerosol Optical Depth (AOD) data from the GOES‐R geostationary satellite. To fill large spatiotemporal gaps in the GOES‐R record, we implement an iterative gap‐filling approach that expands contextual information for successive predictions, maintaining strong agreement with collocated AERONET observations ( R = 0.75) while increasing paired observations from 430,603 to 728,115. A random forest is used to predict surface PM 2.5 concentrations using our spatiotemporally contiguous gap‐filled AOD. The resulting PM 2.5 estimates over 2019–2023 in the top 50 most populous counties over the CONUS capture the magnitude and spatial gradient as measured by the EPA Air Quality System (AQS) and PurpleAir low‐cost sensor network (daily R 2 = 0.85, mean bias = −0.11 μg/m 3 ). We use these predictions to assess population exposure and estimate PM 2.5 ‐attributable mortality, finding that Black, Hispanic, and Indigenous populations experienced 1.7%–5.5% higher PM 2.5 exposure than White populations across these counties. Additional wildfire case studies in less densely monitored rural regions demonstrate the framework's utility for characterizing short‐term exposure events despite substantial gaps in satellite observations. These findings highlight the value of gap‐filled geostationary AOD for high‐resolution exposure assessment.
Abstract Climate change is intensifying the frequency and severity of extreme heat events. Despite the rising frequency of compound heatwaves (CHWs)—concurrent daytime and nighttime heat—evidence regarding cause‐specific risks and socioeconomic disparities remains limited. This study aimed to examine associations between CHWs and emergency hospital admissions and to identify vulnerable subpopulations. This study analyzed 2,010,769 emergency hospital admissions in the summer (June–September, 2011–2019) using the South Korean National Health Insurance Service database. Heatwaves were classified as daytime, nighttime, or compound based on district‐specific 90th percentile thresholds of the daily maximum and minimum temperatures. A time‐stratified case‐crossover design with a distributed lag model was applied to estimate odds ratios for emergency hospital admissions associated with each heatwave type. Subgroup analyses were conducted including age, sex, household income, population density, and admission cause. Furthermore, we tested the potential impacts of income by incorporating an interaction term between income and each subgroup. CHWs were associated with emergency admissions, with particularly strong associations for genitourinary diseases. The risks were highest among the elderly, male, residents of highly populated areas, and low‐income groups. Furthermore, the association was most pronounced for genitourinary diseases, where the low‐income group showed the highest risk. Daytime heatwaves showed higher risks than nighttime heatwaves, which had weaker but disease‐specific associations. This study showed the health risks of CHWs that disproportionately affect genitourinary conditions and low‐income populations, emphasizing the need for climate adaptation planning that prioritize vulnerable groups and heat‐sensitive diseases.
Abstract Global warming intensifies non‐optimal temperature exposure (consisting of high and low temperature), which poses severe threats to human health. As the world's most populous country located in temperate and subtropical zones, China is highly vulnerable to temperature‐related mortality, with distinct yearly and population disparities in health risks. To investigate the impact of non‐optimal temperature exposure on China's all‐causes mortality during 1990–2023 and project future trends (2024–2050) under different climate change scenarios. Integrated CN05.1 gridded observational data, CMIP6 climate data (SSP2‐4.5 and SSP5‐8.5 scenarios), SimmEX extreme temperature indices data, and GBD 2023 disease burden data. We applied Spearman's correlation analysis, Joinpoint regression, Age‐Period‐Cohort analysis, and restricted cubic spline models for trend quantification and future projections. Non‐optimal temperature‐related deaths increased, while age‐standardized mortality rate declined significantly. Low‐temperature exposure remained the primary risk, but high‐temperature‐related health impacts intensified notably. The elderly (≥80 years) and individuals with cardio‐cerebrovascular or respiratory diseases were the most susceptible. Climate warming showed significant correlations with most cause‐specific mortalities. Future projections revealed divergent disease risk trends under the two emission scenarios, with some diseases facing elevated risks. Non‐optimal temperature exposure exerts heterogeneous impacts on China's all‐causes mortality, with high‐temperature risks becoming increasingly prominent amid global warming. Targeted prevention and control strategies tailored to age, disease types, and emission scenarios are crucial to mitigate temperature‐related health burdens.
Abstract This study introduces Disease Vector Intelligence (DVI), an explainable Machine Learning (ML) framework developed within the EYWA (EarlY WArning system for mosquito‐borne diseases) ecosystem. DVI integrates big Earth Observation (EO), socioeconomic, and estimated mosquito abundance data to predict the presence or absence of West Nile Virus (WNV) cases, expressed as a risk score at a high spatiotemporal resolution. The SHAP (SHapley Additive exPlanations) method is employed for a local‐level feature importance analysis to uncover key drivers with a high impact on the prediction. The framework consists of two ML models: one intended to be used as an Early Warning System (EWS) predicting at a regional scale, and the other identifying areas susceptible to WNV transmission at a 2 × 2 km 2 grid scale, independent of human‐defined administrative boundaries. Both models were trained on 11 years of historical WNV case data and validated in Greece. SHAP analysis on both models revealed temperature as a crucial driver of WNV transmission. The fine spatial resolution of the second model uncovered micro‐scale key drivers, such as elevation and land cover type, that the lower resolution model missed. DVI models could be used complementarily by health authorities to aid their decision‐making process and to provide critical insights into important drivers that influence the transmission of WNV.
