August 19, 2022
Research Highlight

Understanding How Model Setup Affects Representations of Atmospheric Properties

Initial conditions and setup of large-eddy simulations influence representations of atmospheric properties important for larger-scale models

A photograph of measurement facilities with a blue sky and white puffy clouds above

Researchers analyzed simulations of shallow convective clouds over the north-central Oklahoma.

(Image: Atmospheric Radiation Measurement user facility |

The Science                                

Many efforts to improve scientific understanding of the convective boundary layer structure heavily rely on numerical models. It is imperative to understand the realism, potential limitations of various configurations of these models, including large-eddy simulations (LESs). LES grid spacing, domain size, inhomogeneity of surface properties, and external forcing all affect the boundary layer depth and its turbulent properties across a range of spatial scales. All simulation configurations produce a progressive increase in characteristic sizes of vertical velocity to temperature and moisture structures, confirming the generality of previously limited findings of earlier studies. Simulations that use a larger domain with diverse surface properties, however, develop mesoscale patterns that significantly affect variability of modeled temperature and moisture fields.

The Impact

LES models are a key tool for studying the atmospheric boundary layer and developing the simplified mathematical representations, known as parameterizations, that represent it in global models. There are multiple commonly used LES configurations, but quantifying differences in their simulated turbulence is difficult as atmospheric fields are highly variable in time and space. The systematic analysis of characteristic structure sizes in different fields quantitatively conducted in this study demonstrates the consequences of various model setup options. The study highlights persistent differences among characteristic structure sizes in moisture, temperature, and vertical wind fields, which need to be taken into the account in the development of scale-aware parameterizations for models with kilometer-scale grid spacing.


Atmospheric properties in the convective boundary layer vary over a wide range of spatial scales. They are commonly studied using LESs in various configurations. This study examines how boundary layer depth and the distribution of variability across scales are affected by LES grid spacing, domain size, inhomogeneity of surface properties, and external forcing. Two commonly used model setups were analyzed: a semi-idealized configuration that uses a periodic domain, flat surface, and prescribed homogeneous surface heat fluxes and a more complicated nested configuration employing a larger domain and realistic initial and boundary conditions, including an interactive land surface model with representative topography, vegetation, and soil types. A progressive increase in the size of dominant features from vertical velocity to temperature to moisture fields was systematically reproduced in all simulation configurations. The dependence of variability on the particular atmospheric quantity complicates development of scale-aware parameterizations for models with grid spacing on the order of 1 km. In simulations using a larger domain with heterogeneous surface properties, the development of internal mesoscale patterns significantly affects variance distributions. The sizes of boundary layer structures also strongly depend on the LES grid spacing. In the case of heterogeneous surface and topography, structure size depends on the location of the subdomain inside a larger computational domain.

PNNL Contact

Jerome Fast, Pacific Northwest National Laboratory,

Mikhail Ovchinnikov, Pacific Northwest National Laboratory,


Measurements collected by the Hi-SCALE field campaign were supported by the Atmospheric Radiation Measurement (ARM)  Climate Research Facility. Funding for this work was provided by the Department of Energy (DOE), Office of Science Biological and Environmental Research program as part of the Atmospheric System Research program. Computing resources for the simulations were provided by the National Energy Research Scientific Computing Center, a DOE Office of Science user facility, the ARM Data Center Computing Facility, and the Environmental Molecular Science Laboratory’s Cascade computational cluster.

Published: August 19, 2022

M. Ovchinnikov, J. D. Fast, L. K. Berg, W. I. Gustafson, J. Chen, K. Sakaguchi, and H. Xiao. 2022. “Effects of horizontal resolution, domain size, boundary conditions, and surface heterogeneity on LES of a convective boundary layer.” Monthly Weather Review, 150. [DOI: 10.1175/MWR-D-21-0244.1]