The 2d- and 3d-version of the Chemical Lagrangian Model of the Stratosphere (CLaMS) are used for the interpretation of the tracer distributions observed during the CRISTA-1 experiment in 1994 and during the SOLVE-THESEO-2000 campaign. Comparing the experimental data with CLaMS simulations, the coupling between the large-scale horizontal deformations and stratospheric mixing is studied. Mixing in CLaMS occurs owing to adaptive regridding of the air parcels applied after each advection time step (6-24 hours) calculated in terms of the isentropic (2d version) or cross-isentropic (3d version) trajectories and is driven by the horizontal deformations in the flow measured by the finite time Lyapunov exponent.

Mixing parameters in CLaMS, are adjusted on the CRISTA observations by using the probability density function technique (PDF) quantifying the statistics of N₂O fluctuations. The PDF derived from CRISTA observations at 700 K and on the horizontal scales of the order 200 km is characterized by a Gaussian core and non-Gaussian tails indicating filamentary structures typical for 2d turbulence. The PDFs obtained from CLaMS simulations strongly depend on the critical deformation switching mixing on but only weakly on the model resolution tested between 45 and 200 km. The results indicate that scale collapse followed by significant mixing has to be expected in flow regions where horizontal scales of the order 200 km are shrunk or elongated by the factor less than 0.5 or greater than 2, respectively, on time scales of the order 10 hours.

To quantify mixing in terms of the diffusivity, CLaMS simulations are compared with in situ aircraft observations at the edge of the northern polar vortex during the SOLVE/THESEO-2000 campaign. Using the concept of effective diffusion to represent vertical mixing processes in the isentropic (2d) simulations of the stratosphere, the lateral (across the wind) effective diffusion coefficient was estimated to be of the order 10³ m²s^-1. Mixing in CLaMS generalizes the idea of the (bulk) effective diffusivity to a more realistic inhomogeneous (i.e. driven by spatial and time dependent flow deformations) and anisotropic (i.e. dependent on the wind direction) mixing.