Meso-scale contribution to air–sea turbulent fluxes at GCM scale

Abstract

Parametrizations of sea surface turbulent fluxes used in general circulation models (GCMs) assume horizontal homogeneity of atmospheric properties at the grid-cell scale. The present study assesses the contribution of the meso-scale (i.e., subgrid) to the grid-scale surface fluxes, for GCM resolution ranging from 20 to 200 km, and thus quantifies the associated GCM surface flux error. A coarse-graining method allows for an a priori analysis of the subgrid information. It is based on an atmospheric reference dataset produced by the convection-permitting operational model AROME. The meso-scale relative contribution to GCM-scale fluxes exceeding 10% is shown to have large regional patterns, with large values (up to 90%) and high frequency of occurrence (up to 76% of the time). These meso-scale motions are not necessarily due to convective activity but also occur frequently under dynamical perturbation conditions. Contributions to surface fluxes, at both the GCM scale and the meso-scale, are disentangled through a Reynolds decomposition. It is found that temperature and humidity meso-scale heterogeneities do not contribute much to the GCM-scale fluxes. The subgrid dynamical processes are the main meso-scale contribution and consist of two parts. The first one represents the wind magnitude and wind direction heterogeneities and corresponds to the so-called gustiness approach. It is clarified that the contribution of the gustiness wind in the transfer coefficients cannot be neglected. The second part is the contribution of the wind speed subgrid variance, which contributes up to 10% of the GCM-scale momentum flux.