Hypothesis
CMAQ v5.0's new dust tracking and wind-blown-dust parameterisation would reproduce observed soil and trace-metal concentrations across the continental United States.
What failed
Simulating all of 2006 against surface networks, the model was acceptable in the western US (mean bias about ±0.5 μg/m³) but consistently overestimated soil concentrations in the eastern US (0.5–1.5 μg/m³) in every season, and badly overestimated some urban trace metals — Al by 380%, Ti 370%, Si 470% in the autumn.
Broken assumption
Adding element tracking and a wind-blown-dust scheme is enough to make modelled dust and trace metals match observations.
Why
Underestimated night-time mixing in urban areas appears to contribute. Removing anthropogenic fugitive dust emissions and wind-blown dust lowered concentrations but did not remove the overestimate, so the authors conclude other errors remain in the system.
From the published paper: Appel et al. Geoscientific model development, 2013. ↗CC BY