Large scale drag representation in simulations of 2D turbulence

Numerical simulations of forced dissipative 2D turbulence are hindered by the presence of the intrinsic dual cascade, enstrophy down- and energy up-scales. When the upscale cascade expands to the lowest modes, the ensuing energy condensation causes steepening of energy and enstrophy spectra and generation of strong coherent vortices. To circumvent these effects, the lowest modes energy is usually suppressed via introducing an energy sink in the form of linear drag or infrared hyperviscosity which do, however, distort intermodal energy and enstrophy exchange leading to violation of Kolmogorov--Kraichnan statistical laws. Here, we introduce a new concept of the large scale drag: it is conceptually similar to the two-parametric viscosity by Kraichnan and is designed to withdraw the large scale energy without distorting the inverse cascade. This new large scale drag has been used in both DNS and LES in a periodic box setting where a robust steady state was preserved for many eddy turnover times. The flow field remained structureless and in good agreement with the Kolmogorov statistics; coherent structures did not form at any time. We conclude that homogeneous, isotropic, forced, dissipative 2D turbulence in the energy subrange is not prone to coherent structure formation as long as its inverse cascade remains undisturbed.