Constraining ammonia dairy emissions during NASA DISCOVER-AQ California: surface and airborne observation comparisons with CMAQ simulations

Abstract:

Agricultural ammonia (NH₃) emissions are highly uncertain in current bottom-up inventories. Ammonium nitrate is a dominant component of fine aerosols in agricultural regions such as the Central Valley of California, especially during winter. Recent high resolution regional modeling efforts in this region have found significant ammonium nitrate and gas-phase NH₃ biases during summer. We compare spatially-resolved surface and boundary layer gas-phase NH₃ observations during NASA DISCOVER-AQ California with Community Multi-Scale Air Quality (CMAQ) regional model simulations driven by the EPA NEI 2008 inventory to constrain wintertime NH₃ model biases. We evaluate model performance with respect to aerosol partitioning, mixing and deposition to constrain contributions to modeled NH₃ concentration biases in the Central Valley Tulare dairy region. Ammonia measurements performed with an open-path mobile platform on a vehicle are gridded to 4 km resolution hourly background concentrations. A peak detection algorithm is applied to remove local feedlot emission peaks. Aircraft NH₃, NH₄⁺ and NO₃^- observations are also compared with simulations extracted along the flight tracks. We find NH₃ background concentrations in the dairy region are underestimated by three to five times during winter and NH₃ simulations are moderately correlated with observations (r = 0.36). Although model simulations capture NH₃ enhancements in the dairy region, these simulations are biased low by 30-60 ppbv NH₃. Aerosol NH₄⁺ and NO₃^- are also biased low in CMAQ by three and four times respectively. Unlike gas-phase NH₃, CMAQ simulations do not capture typical NH₄+ or NO₃- enhancements observed in the dairy region. In contrast, boundary layer height simulations agree well with observations within 13%. We also address observational constraints on simulated NH₃ deposition fluxes. These comparisons suggest that NEI 2008 wintertime dairy emissions are underestimated by a factor of three to five. We test sensitivity to emissions by increasing the NEI 2008 NH₃ emissions uniformly across the dairy region and evaluate the impact on modeled concentrations. These results are applicable to improving predictions of ammoniated aerosol loading and highlight the value of mobile platform spatial NH₃ measurements to constrain emission inventories.