Quantifying the effects of mesoscale motion

An underlying assumption founding the creation of numerical atmospheric models for weather- and climate prediction is the existence of a separation between the mean-flow and turbulent motion, known as the spectral gap. The existence of this gap is supported by several early observational studies, though these have recently been called into question by a number of authors. If such a spectral gap exists, however, it is relatively easy to apply Reynolds averaging using a time-filter. On the other hand, if the flow does not exhibit a spectral gap the results will be dependent on the choice of separation scale. Further, the potential end-user of the observations, i.e. the weather- and climate modellers, may well be interested in including the mesoscales in their parameterisations, provided there is a systematic influence on the mean flow.

Mesoscale motion is often perceived to consist of a large number of phenomena, for example gravity waves, katabatic flows, convection and meandering to mention a few, and these have been studied in quite great detail. Unfortunately, it is not without difficulties to include these resulting detailed descriptions of these mesoscale phenomena into model parameterisations. On the contrary, one could suspect that one reason for operational models benefiting from being more diffusive than what can be justified from observations and large-eddy simulations is the presence of such unresolved and un-parameterised motion.

In this ongoing study I will attempt to quantify the collective effect of the mesoscales, rather than that of a particular phenomena. I will also discuss some of the fundamental problems involved with doing so. The software developed for the purpose is based on modern signal-processing methods and will be made publicly available upon completion of the project.