Spatioseasonal Variations of Atmospheric Ammonia Concentrations Over the United States: Comprehensive Model‐Observation Comparison

Atmospheric ammonia plays an important role in a number of environmental issues, including new particle formation and aerosol indirect radiative forcing. Over the United States, atmospheric ammonia has seen an increasing trend due in most part to the declining SO2 and NOx emissions. We conduct the first comprehensive assessment of multiyear Goddard Earth Observing System (GEOS)‐Chem simulated ammonia concentration ([NH3]) over conterminous United States along with surface observations from all 90 National Atmospheric Deposition Program Ammonia Monitoring Network (AMoN) sites that have at least 2 years of continuous measurements. Model‐simulated [NH3] is along empirically expected lines with regard to temporal trends, seasonal variations, and spatial distribution. GEOS‐Chem‐simulated [NH3], compared to AMoN observed values, has weighted average correlation (τ) of 0.50 ± 0.15 and mean fractional bias (MFB) of −8.8 ± 56%. Most sites (63 out of 90) have −60% < MFB < +60%. The deviations from observed values vary spatially and seasonally, and there is significant wintertime underestimation (−44 ± 58%) across most of conterminous United States (except the Pacific states). The largest positive deviations occur in the Pacific states (101 ± 46%) and the largest negative deviations in the Southern Plain states (−73 ± 39%) and the Mountain states (−73 ± 84%), both in the winter months. Over the Great Plains region, GEOS‐Chem simulated [NH3] shows a much stronger dependence to emissions than AMoN observed [NH3], indicating scope for improved representation of emissions for the region. Over Southeast United States, there appears to be the strong effect of the changing emissions of SO2 and NOx in both modeled and observed [NH3]. Key Points Modeled atmospheric ammonia concentrations is in good spatiotemporal agreement with 90 surface stations across the United States from 2007‐2017 Model underestimation is dominant in winter (–44 ± 58% bias) across the United States, except the Pacific states (+101 ± 46% bias) Model‐observation correlation of seasonality is good; simulated concentrations appear more emission dominated than observations