Abstract

Much of our understanding of atmospheric chemistry has been propelled forward by attempting realistic simulations of composition and relating the results (and errors) to measurements. Typically, photochemical models are driven offline using wind and temperature analyses obtained by data assimilation into a numerical weather prediction model. In this talk, deficiencies in tracer simulations driven by analysed winds will be identified using the global and material conservation properties of fluids.

Errors in global conservation of tracers arise from discretisation of the mass continuity equation, even when the tracer advection scheme is conservative by design (given a flow obeying continuity exactly). For example, in ozone hole simulations the error in global ozone burden can exceed the photochemical change by a factor of 3 when operational analyses are spatially truncated. The reasons behind the non-conservation and its effects on the global distribution are explored. Recommendations are given regarding wind field truncation and choice of vertical coordinate.

Advection schemes inevitably introduce numerical mixing but it is hard to quantify this effect with realistic wind fields. Simulations of the dispersal of SO2 from the Sarychev volcano are compared with IASI satellite data for column amounts. It is shown that it is possible to back out both the implicit mixing rate in the tracer model and the mixing rate affecting the plume dispersal in the real atmosphere. The mixing timescale (in the upper troposphere) is surprisingly long (>20 days), but is supported by independent evidence from a range of altitudes on the ICARTT Lagrangian experiment. It exceeds the implicit mixing timescale at the resolution that it is currently feasible to achieve using a global chemical transport model.