Can mesoscale models reproduce meandering motions?

The influence of meandering flow on dispersion of pollutants is frequently under-represented in dispersion models. In terms of measurements, meandering is primarily associated with time-scales between the turbulence and the applied averaging time, which is usually 1 h. The related spatial scales thus range roughly from 102 to 104 m (referred to here as submesoscales). As the state-of-the-art mesoscale models should be capable of reproducing flow features on scales larger than the turbulence, and as the meandering-generating mechanisms are not fully understood yet, it is useful to examine if the mesoscale models can reproduce meandering. For that purpose, the WRF/Chem model at 1/3 km horizontal resolution is used to simulate a weak-wind night during the CASES99 experiment. The measurements are used for detailed model verification. The model with its typical set-up fails to reproduce the variability at submesoscales and the locus of the under-representation is traced to too-strong horizontal diffusion. Reducing or removing the model diffusion allows the appearance of the submeso variability, whose spectral properties and the resulting plume behaviour agree well with the measurements. The linear correlation between the simulations with reproduced variability and the measurements is low, as is the case between two simulations with only slightly different set-up. The conclusion is that mesoscale models are able to reproduce the strength of variability and the effects of meandering, but only with reduced or removed horizontal diffusion. The question arises whether it is possible to obtain a linear correlation, i.e. to correctly reproduce individual modes at these scales at all. Copyright (C) 2010 Royal Meteorological Society