Background: Forest, grass, and peat fires release approximately 2 petagrams of carbon into the atmosphere each year, influencing weather, climate, and air quality.Objective: We estimated the annual global mortality attributable to landscape fire smoke (LFS).Methods: Daily and annual exposure to particulate matter ≤ 2.5 μm in aerodynamic diameter (PM2.5) from fire emissions was estimated globally for 1997 through 2006 by combining outputs from a chemical transport model with satellite-based observations of aerosol optical depth. In World Health Organization (WHO) subregions classified as sporadically affected, the daily burden of mortality was estimated using previously published concentration–response coefficients for the association between short-term elevations in PM2.5 from LFS (contrasted with 0 μg/m3 from LFS) and all-cause mortality. In subregions classified as chronically affected, the annual burden of mortality was estimated using the American Cancer Society study coefficient for the association between long-term PM2.5 exposure and all-cause mortality. The annual average PM2.5 estimates were contrasted with theoretical minimum (counterfactual) concentrations in each chronically affected subregion. Sensitivity of mortality estimates to different exposure assessments, counterfactual estimates, and concentration–response functions was evaluated. Strong La Niña and El Niño years were compared to assess the influence of interannual climatic variability.Results: Our principal estimate for the average mortality attributable to LFS exposure was 339,000 deaths annually. In sensitivity analyses the interquartile range of all tested estimates was 260,000–600,000. The regions most affected were sub-Saharan Africa (157,000) and Southeast Asia (110,000). Estimated annual mortality during La Niña was 262,000, compared with 532,000 during El Niño.Conclusions: Fire emissions are an important contributor to global mortality. Adverse health outcomes associated with LFS could be substantially reduced by curtailing burning of tropical rainforests, which rarely burn naturally. The large estimated influence of El Niño suggests a relationship between climate and the burden of mortality attributable to LFS.
Ambient air pollution is associated with numerous adverse health impacts. Previous assessments of global attributable disease burden have been limited to urban areas or by coarse spatial resolution of concentration estimates. Recent developments in remote sensing, global chemical-transport models, and improvements in coverage of surface measurements facilitate virtually complete spatially resolved global air pollutant concentration estimates. We combined these data to generate global estimates of long- term average ambient concentrations of fine particles (PM2.5) and ozone at 0.1° × 0.1° spatial resolution for 1990 and 2005. In 2005, 89% of the world’s population lived in areas where the World Health Organization Air Quality Guideline of 10 μg/m3 PM2.5 (annual average) was exceeded. Globally, 32% of the population lived in areas exceeding the WHO Level 1 Interim Target of 35 μg/m3; driven by high proportions in East (76%) and South (26%) Asia. The highest seasonal ozone levels were found in North and Latin America, Europe, South and East Asia, and parts of Africa. Between 1990 and 2005 a 6% increase in global population-weighted PM2.5 and a 1% decrease in global population- weighted ozone concentrations was apparent, highlighted by increased concentrations in East, South and Southeast Asia and decreases in North America and Europe. Combined with spatially resolved population distributions, these estimates expand the evaluation of the global health burden associated with outdoor air pollution.
Land use regression (LUR) is a promising technique for predicting ambient air pollutant concentrations at high spatial resolution. We expand on previous work by modeling oxides of nitrogen and fine particulate matter in Vancouver, Canada, using two measures of traffic. Systematic review of historical data identified optimal sampling periods for NO and N02. Integrated 14-day mean concentrations were measured with passive samplers at 116 sites in the spring and fall of 2003. Study estimates for annual mean NO and NO2 ranged from 5.4-98.7 and 4.8-28.0 ppb, respectively. Regulatory measurements ranged from 4.8-29.7 and 9.0-24.1 ppb and exhibited less spatial variability. Measurements of particle mass concentration (PM2.5) and light absorbance (ABS) were made at a subset of 25 sites during another campaign. Fifty-five variables describing each sampling site were generated in a Geographic Information System (GIS) and linear regression models for NO, NO2, PM2.5, and ABS were built with the most predictive covariates. Adjusted R(2) values ranged from 0.39 to 0.62 and were similar across traffic metrics. Resulting maps show the distribution of NO to be more heterogeneous than that of NO2, supporting the usefulness of this approach for assessing spatial patterns of traffic-related pollution.
Background: During the summer of 2003 numerous fires burned in British Columbia, Canada.Objectives: We examined the associations between respiratory and cardiovascular physician visits and hospital admissions, and three measures of smoke exposure over a 92-day study period (1 July to 30 September 2003).Methods: A population-based cohort of 281,711 residents was identified from administrative data. Spatially specific daily exposure estimates were assigned to each subject based on total measurements of particulate matter (PM) ≤ 10 μm in aerodynamic diameter (PM10) from six regulatory tapered element oscillating microbalance (TEOM) air quality monitors, smoke-related PM10 from a CALPUFF dispersion model run for the study, and a SMOKE exposure metric for plumes visible in satellite images. Logistic regression with repeated measures was used to estimate associations with each outcome.Results: The mean (± SD) exposure based on TEOM-measured PM10 was 29 ± 31 μg/m3, with an interquartile range of 14–31 μg/m3. Correlations between the TEOM, smoke, and CALPUFF metrics were moderate (0.37–0.76). Odds ratios (ORs) for a 30-μg/m3 increase in TEOM-based PM10 were 1.05 [95% confidence interval (CI), 1.03–1.06] for all respiratory physician visits, 1.16 (95% CI, 1.09–1.23) for asthma-specific visits, and 1.15 (95% CI, 1.00–1.29) for respiratory hospital admissions. Associations with cardiovascular outcomes were largely null.Conclusions: Overall we found that increases in TEOM-measured PM10 were associated with increased odds of respiratory physician visits and hospital admissions, but not with cardiovascular health outcomes. Results indicating effects of fire smoke on respiratory outcomes are consistent with previous studies, as are the null results for cardiovascular outcomes. Some agreement between TEOM and the other metrics suggests that exposure assessment tools that are independent of air quality monitoring may be useful with further refinement.
Language modelling provides a step towards intelligent communication systems by harnessing large repositories of written human knowledge to better predict and understand the world. In this paper, we present an analysis of Transformer-based language model performance across a wide range of model scales -from models with tens of millions of parameters up to a 280 billion parameter model called Gopher. These models are evaluated on 152 diverse tasks, achieving state-of-the-art performance across the majority. Gains from scale are largest in areas such as reading comprehension, fact-checking, and the identification of toxic language, but logical and mathematical reasoning see less benefit. We provide a holistic analysis of the training dataset and model's behaviour, covering the intersection of model scale with bias and toxicity. Finally we discuss the application of language models to AI safety and the mitigation of downstream harms.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.