Statistically Steady State Large-Eddy Simulations

16 Using large-eddy simulations (LES) systematically has the potential to inform param17 eterizations of subgrid-scale (SGS) processes in general circulation models (GCMs), such 18 as turbulence, convection, and clouds. Here we show how LES can be run to simulate 19 grid columns of GCMs to generate LES across a cross-section of dynamical regimes. The 20 LES setup approximately replicates the thermodynamic and water budgets in GCM grid 21 columns. Resolved horizontal and vertical transports of heat and water and large-scale 22 pressure gradients from the GCM are prescribed as forcing in the LES. The LES are forced 23 with prescribed surface temperatures, but free-tropospheric temperature and moisture 24 are free to adjust, reducing the imprinting of GCM fields on the LES. In both the GCM 25 and LES, radiative transfer is treated in a unified but idealized manner (semi-gray at26 mosphere without water vapor feedback or cloud radiative effects). We show that the 27 LES in this setup reaches statistically steady states without nudging to GCM profiles. 28 The steady states provide training data for developing GCM parameterizations. The same 29 LES setup also provides a good basis for studying the cloud response to global warm30 ing. 31 Plain Language Summary 32 Clouds and their feedbacks remain one of the largest uncertainties in predictions 33 of future climate changes. High-resolution models can provide faithful simulations of clouds 34 and their underlying turbulence in limited areas, but they have primarily been used in 35 select locations, with limited success in reducing uncertainties in climate predictions. This 36 study presents a framework for driving high-resolution simulations by a global climate 37 model, which allows us to generate a library of high-resolution simulations across dif38 ferent cloud regimes. The framework leverages the potential of high-resolution models 39 to improve parameterizations of clouds and turbulence in climate models and to better 40 understand the cloud feedback mechanisms. 41