Air quality in the United States has dramatically improved, yet exposure to air pollution is still associated with 100000−200000 deaths annually. Reducing the number of deaths effectively, efficiently, and equitably relies on attributing them to specific emission sources, but so far, this has been done for only highly aggregated groups of sources, or a select few sources of interest. Here, we estimate mortality in the United States attributable to all domestic, human-caused emissions of primary PM 2.5 and secondary PM 2.5 precursors. We present detailed sourcespecific attributions in four alternate groupings relevant for identifying promising ways to reduce mortality. We find that nearly half of the deaths can be attributed to just five activities, all in different sectors. Around half of the deaths can be attributed to fossil fuel combustion, with the remainder attributable to combustion of nonfossil fuels, agricultural processes, and other noncombustion processes. Both primary and secondary PM 2.5 are important, including PM 2.5 from currently unregulated precursor pollutants such as ammonia. We suggest improvements in air quality can be realized by continued reductions of emissions from traditionally important sources and by novel strategies for reducing emissions from sources of emerging relative importance and research focus. Such changes can contribute to improved health outcomes and other environmental goals.

Agriculture is a major contributor to air pollution, the largest environmental risk factor for mortality in the United States and worldwide. It is largely unknown, however, how individual foods or entire diets affect human health via poor air quality. We show how food production negatively impacts human health by increasing atmospheric fine particulate matter (PM2.5), and we identify ways to reduce these negative impacts of agriculture. We quantify the air quality–related health damages attributable to 95 agricultural commodities and 67 final food products, which encompass >99% of agricultural production in the United States. Agricultural production in the United States results in 17,900 annual air quality–related deaths, 15,900 of which are from food production. Of those, 80% are attributable to animal-based foods, both directly from animal production and indirectly from growing animal feed. On-farm interventions can reduce PM2.5-related mortality by 50%, including improved livestock waste management and fertilizer application practices that reduce emissions of ammonia, a secondary PM2.5 precursor, and improved crop and animal production practices that reduce primary PM2.5 emissions from tillage, field burning, livestock dust, and machinery. Dietary shifts toward more plant-based foods that maintain protein intake and other nutritional needs could reduce agricultural air quality–related mortality by 68 to 83%. In sum, improved livestock and fertilization practices, and dietary shifts could greatly decrease the health impacts of agriculture caused by its contribution to reduced air quality.

Each year, millions of premature deaths worldwide are caused by exposure to outdoor air pollution, especially fine particulate matter (PM2.5). Designing policies to reduce these deaths relies on air quality modeling for estimating changes in PM2.5 concentrations from many scenarios at high spatial resolution. However, air quality modeling typically has substantial requirements for computation and expertise, which limits policy design, especially in countries where most PM2.5-related deaths occur. Lower requirement reduced-complexity models exist but are generally unavailable worldwide. Here, we adapt InMAP, a reduced-complexity model originally developed for the United States, to simulate annual-average primary and secondary PM2.5 concentrations across a global-through-urban spatial domain: “Global InMAP”. Global InMAP uses a variable resolution grid, with horizontal grid cell widths ranging from 500 km in remote locations to 4km in urban locations. We evaluate Global InMAP performance against both measurements and a state-of-the-science chemical transport model, GEOS-Chem. Against measurements, InMAP predicts total PM2.5 concentrations with a normalized mean error of 62%, compared to 41% for GEOS-Chem. For the emission scenarios considered, Global InMAP reproduced GEOS-Chem pollutant concentrations with a normalized mean bias of 59%–121%, which is sufficient for initial policy assessment and scoping. Global InMAP can be run on a desktop computer; simulations here took 2.6–8.4 hours. This work presents a global, open-source, reduced-complexity air quality model to facilitate policy assessment worldwide, providing a screening tool for reducing air pollution-related deaths where they occur most.

We present alternative methods for estimating spatial surrogates and temporal factors for ammonia (NH 3 ) emissions from chemical fertilizer usage (CFU), in the USA, at spatial and temporal scales used to simulate regional air quality and deposition of reactive nitrogen to ecosystems. The newly developed Improved Spatial Surrogate (ISS) method incorporates year-specific fertilizer sales data, high resolution and year-specific crop maps, and local crop nitrogen demands to allocate NH 3 emissions at 4 km × 4 km grid cells. Results are compared with the commonly used gridded emission estimates by the Sparse Matrix Operator Kernel Emissions (SMOKE) preprocessor. NH 3 emissions over Central Illinois in the USA, estimated at the 4 km × 4 km grid level in SMOKE and ISS methods, exhibit differences between À10% and 120%, with 58% of the grid cells exhibiting more than ±10% difference. Application of the ISS method for a larger domain over the Midwest USA, at 4 km × 4 km, reflected similar differences. We also employed the Denitrification Decomposition (DNDC) model to develop daily temporal factors of NH 3 emissions from CFU using multi-site and multi-year analyses. Ratio of temporal factors estimated by SMOKE and DNDC methods is 0.54 ± 2.35, with DNDC identifying daily emission peaks 2.5-8 times greater than SMOKE. Identified emission peaks will be useful for future air quality modeling efforts to understand particulate matter episodes, as well as trends in regional particulate matter formation and nitrogen deposition for Midwest USA, using the proposed NH 3 emissions inventory.

The objective of this research is to quantify NH 3 flux above an intensively managed cornfield in the Midwestern United States to improve understanding of NH 3 emissions and evaluations of new and existing emission models. A relaxed eddy accumulation (REA) system was deployed above a corn canopy in central Illinois, USA (40 • 3 46.209 N, 88 • 11 46.0212 W) from May through September 2014 (day of year 115-273) to measure NH 3 fluxes due to chemical fertilizer application. NH 3 flux was measured in fourhour periods during mornings and afternoons. Mean atmospheric NH 3 concentration during the complete measurement period was 2.6 ± 2.0 g m −3. Larger upward fluxes of gaseous NH 3 were measured during the first 30 days after fertilization, with variations observed throughout the field campaign. Measured NH 3 fluxes ranged from −246.0 ng m −2 s −1 during wintertime background measurements to 799.6 ng m −2 s −1 within two weeks of fertilization (where negative flux indicates deposition). Mean positive flux was 233.3 ± 203.0 ng m −2 s −1 in the morning and 260.0 ± 253.3 ng m −2 s −1 in the afternoon while mean negative flux was −45.3 ± 38.6 ng m −2 s −1 in the morning and −78.35 ± 74.9 ng m −2 s −1 in the afternoon. NH 3 volatilization during the first 21 days after fertilization accounted for 79% of total nitrogen loss during the growing season. Such measurements are critical to improve understanding of agricultural NH 3 emissions in managed agricultural ecosystems dominated by rotations of highly fertilized corn and moderately to lightly fertilized soybeans, such as the plot studied herein. These measurements are also important to improve understanding of how managed agricultural ecosystems impact air quality, and contribute to the global nitrogen cycle, and to evaluate current NH 3 emission models.