Impacts of Emission Inventory Resolution and Regional Mesh Refinement on Aerosols, Ozone, and Clouds in E3SMv3

Emission inventory, emission spatial resolution, and model resolution affect atmospheric composition and clouds, but their distinct contributions remain uncertain. We assess these contributions using five E3SMv3 simulations (2005-2014), combining CMIP6 and CMIP7 emissions, approximately 100 km global and 25 km regionally refined grids, and emission resolutions of approximately 2° and 0.25°. On the refined grid, finer emissions reduce Continental United States (CONUS) PM 2.5 root mean squared error against satellite-derived estimates from 3.66 to 3.19 µg m -3 and urban mean bias from -3.3 to -2.3 µg m -3 , although regional overestimation remains. CMIP7 emissions reduce the global tropospheric ozone burden by 1.6%. Regional refinement improves 2010 surface maximum daily 8-hour average ozone, reducing mean bias from -5.2 to -1.8 ppb with CMIP6 emissions and from -4.2 to -3.0 ppb with CMIP7, while increasing CONUS tropospheric column bias. Refinement also reduces summertime shortwave cloud radiative effect bias off California from above 60 to 30-50 W m -2 . Finer emissions alter regional shortwave cloud radiative effects by 4-8 W m -2 , reducing bias over Baja California and the southeastern coast but increasing it near New York. Over Baja California, cloud fraction and liquid water path increase despite lower droplet number, indicating an important contribution from cloud amount and condensate changes, although the mechanisms remain unresolved. Emission and model refinement therefore require joint evaluation across air-quality and cloud diagnostics.