Lagrangian LES of Cloud Organization from EUREC4A

Motivation

Shallow cumulus clouds over subtropical oceans contribute to uncertainty of cloud feedbacks and climate predictions. To better understanding clouds, their patterns of organization and the underlying ocean, the EUREC4A field experiment took place in January and February 2020 near Barbados.

Since trade cumulus organization is observed to develop within air masses as they follow the trade winds, we simulate the evolution of the cloudy marine boundary layer (MBL) along quasi-Lagrangian trajectories that follow air masses towards Barbados.

Our simulations will aim to resolve both mesoscale organization of clouds on ~100 km scales and turbulence and convection within the MBL. While we refer to these simulations as large eddy simulations (LES), though they often will not resolve all of large eddies in the subcloud layer and might better be classified as "coarse LES" or "near grey zone" simulations following Honnert et al (2020).

EUREC4A Field Campaign

Summary

This poster presents initial results from the effort, which is part of a larger Grey Zone Intercomparison that will also include regional model simulations of cases from EUREC4A.

We have made initial simulations along quasi-Lagrangian trajectories that arrive near Barbados on Feb 2, 5 and 9.
These simulations generally capture the boundary layer evolution seen along trajectories in ERA and some of the in situ and remote sensing observations gathered during EUREC4a, though those comparisons are ongoing.
A general purpose tool, Lagtraj, that generates forcings for LES and single column model simulation --- both Lagrangian and Eulerian --- has been developed.
Moving forward, we will try to understand the processes driving organization in these case studies. Among other approaches, we will use some of the same tools used to study deep convective aggregation. In SAM, the cases will also be simulated with water isotopes to look for possible isotopic signals associated with organization and cold pools.

Simulation Design and Forcings

Simulations follow air masses for ~3 days towards Barbados based on winds at 950 hPa.
LES use periodic boundary conditions
Initial soundings and forcings (large-scale subsidence and horizontal advection, sea surface temperature) averaged over a 2x2° region in IFS or ERA5.
Model domains: Up to ~50km square (Larger in future)
Grid spacing: Δx=Δy=100m, Δz~20m
Models: DALES, SAM (Coming soon: MONC)

Path of quasi-Lagrangian trajectories arriving in
the intensive observation area east of Barbados on Feb 2
along with a satellite image of cloud cover that day.
Each trajectory is timed to arrive during one of the
flight circles flown by the German HALO aircraft that
day. — MODIS image of cloud cover in the subtropical North Atlantic on 2 Feb 2020, along with quasi-Lagrangian trajectories based IFS winds at 950 hPa. Each trajectory is timed to match a flight of the German HALO aircraft around a circular path east of Barbdaos.

Feb 9 Case Study

We simulated the evolution of MBL and cloud along a trajectory arriving at the HALO circle on 9 Feb at 12Z. Cold pools and precipitation were observed on this day, and a SAM simulation with a 57.6 km square domain reproduces this behavior qualitatively, though some convection reaches deeper than observed.

Comparison of SAM simulations and HALO dropsonde
measurements of velocity, temperature and moisture at 12
UTC on 9 February 2020 near Barbados. — SAM simulations show reasonable agreement with HALO dropsondes and ERA5 when the trajectory crosses the HALO circle at 12Z on 9 Feb.

Visualizations of spatial distribution of cloud and
precipitation in SAM simulations along the trajectory
that arrives in the HALO circle at 12 UTC on 9 February 2020 near Barbados. — SAM simulations show reasonable agreement with HALO dropsondes and ERA5 when the trajectory crosses the HALO circle at 12Z on 9 Feb.

Feb 2 Case Study

Comparison of DALES simulations and HALO dropsonde
measurements of velocity, temperature and moisture at 12
UTC on 2 February 2020 near Barbados. — Simulations of the Feb 2 case study have been made with DALES in 51.2 km square domains. These also agree with HALO dropsondes and IFS when crossing the HALO circle.

Constructing Forcings w/Lagtraj

YAML configuration files for each step:
Input: ERA5, to be extended with observations
Internals: xarray datasets, numba-optimised (e.g. regression, interpolation to height)
Features: Unit testing (pytest), versioning, standardised output (CF-compliant, based on ERA5 format)
Lagtraj forcings were used for the Feb 9 simulations at left.

References

Brown et al, 2020. A highly scalable met office NERC cloud model. arXiv preprint arXiv:2009.12849.
Heus et al, 2010. Formulation of the Dutch Atmospheric Large-Eddy Simulation (DALES) and overview of its applications. Geoscientific Model Development, 3(2), pp.415-444.
Khairoutdinov and Randall, 2003. Cloud resolving modeling of the ARM summer 1997 IOP: Model formulation, results, uncertainties, and sensitivities. Journal of Atmospheric Sciences, 60(4), pp.607-625.
Honnert et al, 2020. The Atmospheric Boundary Layer and the "Gray Zone" of Turbulence: A critical review. Journal of Geophysical Research: Atmospheres, 125(13), p.e2019JD030317.
Stevens et al, 2021. EUREC4A. Earth System Science Data Discussions, pp.1-78.
Narenpitak et al, 2021. From Sugar to Flowers: A Transition of Shallow Cumulus Organization During ATOMIC. Earth and Space Science Open Archive. https://doi.org/10.1002/essoar.10506770.1

Acknowledgements

PB was supported by NSF award AGS-1938108 and thanks Marat Kharioutdinov for maintaining SAM.

Photograph of shallow cumulus clouds with a variety of
cloud top heights and a mix of stratiform and convective
cloud cover as well as clear sky