Understanding How Model Numerics Bias Tracer Transport: Insight from the Age of Air in Idealized Gcms

Tracer transport continues to be a challenge in comprehensive atmospheric models. In the stratosphere, models generally differ more widely in tracer based measures of the stratospheric circulation (e.g. the age of air) than dynamical measures, such as the residual or isentropic circulation. We create a test bed to understand the impact of model numerics on tracer transport using two idealized GCMs that capture key behavior of more comprehensive models. Specifically, the two dynamical cores -- one pseudospectral, the other finite-volume -- simulate exactly the same isentropic/residual circulation, but differ greatly in their transport of age. To more tightly couple the isentropic circulation with tracer transport, we explore the mass circulation in an age coordinate. This both lays bare the differences in tracer transport between the two cores, and provides new insight into the meridional overturning of the stratosphere. We then use the “leaky pipe model” of Neu and Plumb (1999) to interpret the new formulation of the mass circulation, and to highlight the role of numerical biases in the isentropic mixing of tracers on the total tracer transport.