Evaluating the Impact of Resolving Hourly Anthropogenic Emissions on Air Pollutant Simulations in the United States Using the MUSICAv0 Model

The capacity of chemical transport models to accurately simulate air pollutant concentrations and their diurnal changes is essential for pollution source attribution and exposure risk assessment. The Community Earth System Model (CESM) incorporates full chemistry from the Community Atmosphere Model (CAM-chem) and horizontal mesh refinement through the spectral element (SE) dynamical core, offering an innovative framework to study air pollution impacts at various spatial scales with globally consistent dynamics and chemistry. We use this CAM-chem-SE configuration featuring a ∼14km×14km refined grid for the contiguous United States (CONUS) within a 1°×1° global horizontal mesh, referred to as the Multi-Scale Infrastructure for Chemistry and Aerosols version 0 (MUSICAv0). Our analysis compares the standard MUSICAv0 CAMS-GLOB-ANT v5.17 emissions with the 2017 U.S. National Emissions Inventory (NEI) and examines the effects of replacing monthly with hourly anthropogenic emissions on simulated trace gas concentrations and their diurnal variations for July 2018. The study examines three scenarios: ‘base’ with global monthly CAMS emissions; ‘monthlyNEI,' which replaces monthly NEI over the CONUS but retains CAMS elsewhere; and ‘hourlyNEI,' which uses hourly instead of monthly NEI over CONUS. We divide the CONUS domain into West Coast, Mountain, Midwest, Southwest, Southeast, and Northeast for model evaluation. July daily averages from the ‘base’ model simulations (0:00 to 23:00, local time) compared with State and Local Air Monitoring Stations (SLAMS) measurements show high model biases in surface nitrogen dioxide (NO2) concentrations of 23-40% (1-3 ppb) in all but the Mountain region where a low bias of -18% (-1 ppb) occurs, and in surface ozone (O3) of 11-28% (6-13 ppb); and low biases of -21 to -80% (10-60 ppb) in surface carbon monoxide (CO). Modeled tropospheric vertical column densities (VCDTrop) of formaldehyde (HCHO) and NO2, calculated using TROPOspheric Monitoring Instrument (TROPOMI) satellite retrieval averaging kernels at 1:30 PM local time, show HCHO overestimates of 14-24% and NO2 underestimates of 35-52% across the six regions. Integrating NEI emissions (‘monthlyNEI’ and ‘hourlyNEI’ cases) enhances agreement with observations by improving spatial correlations and reducing model mean biases for NO2, O3, and CO surface simulations compared to the ‘base’ case. However, improvements in the ‘monthlyNEI’ and ‘hourlyNEI’ cases are region-specific; for instance, ‘monthlyNEI’ shows a lower model O3 bias on the West Coast but a higher bias in the Northeast than ‘hourlyNEI.' Likewise, the high bias in modeled HCHO VCDTrop compared with TROPOMI is reduced by approximately 1-3% compared to the ‘base’ case, but the low bias in NO2 VCDTrop worsens by 10-20%. Subsequent work will assess diurnal variations in MUSICAv0-simulated trace gas concentrations, comparing them across the three scenarios and with observational data.