Adaptive Methods are not yet widely accepted in atmospheric modelling. However, they can help to solve some of the problems with multi-scale phenomena. Atmospheric modelling on the other hand, is mostly concerned with large scale computations, requiring parallel comptuing architectures. Adaptivity and parallel computing in general do not go well together.

This presentation will give a brief introduction to atmospheric transport phenomena. We will utilize the semi-Lagrangian discretization method (SLM) for the equations involved. SLM are well suited for adaptivity and also for parallelization. Space-filling curves, dating back to the late 19th century and motivated by pure mathematics problems, help to solve the load balancing and partitioning problem for parallel unstructured grids. Modelling the transport of air masses in the arctic atmosphere will show, how modern mathematical methods can greatly improve the understanding of environmental phenomena.