We have used 210Pb, a tracer originating from the radioactive decay of 222Rn emitted from soils, to validate different wet and dry deposition schemes in our global three‐dimensional off‐line chemical transport model, TOMCAT. We have tested two different parameterizations for the dry deposition, one with the turbulent exchange in the boundary layer and one without. When dry deposition is linked to the boundary layer scheme, the model shows the ability to resolve shallow mixing depths in winter with enhanced lifetimes and concentrations of surface aerosol. Three conceptually different wet deposition schemes have been implemented and tested in TOMCAT. In the first, where wet removal is assumed to be proportional to the local gradient of the specific humidity, the model has a global mean bias for the surface concentrations of −40%. Using this scheme, the observed surface concentrations are seriously underpredicted, and most of the aerosol burden is scavenged within the boundary layer. In the second scheme, where scavenging is parameterized proportional to the model‐derived total precipitation rate (large scale and convective), the model exhibits a +26% bias. The concentrations are overpredicted, especially in continental areas as this scheme fails to capture the coupling between vertical transport and rainout. In the third scheme, wet removal is coupled with the vertical transport inside convective clouds. The model then shows the best performance with only a −4% bias for the concentrations. Furthermore, other regional discrepancies between model and observations point to a variability in 222Rn emissions, which was not taken into account with our simple 222Rn emission scenario, and also to anomalies in our model‐derived precipitation rate. The aerosol residence times are realistic only when the wet removal schemes use the precipitation rate rather than the specific humidity. The dry subtropical and polar regions can be captured with lifetimes at 500 hPa of more than 50 days whereas in regions of high precipitation the lifetimes are less than 5 days.
Read full abstract