Breakout Summary Report
ARM/ASR User and PI Meeting
LASSO CACTI Deep Convection Scenario
24 June 2020
2:00 PM - 4:00 PM
167
Gustafson and Vogelmann
24 June 2020
2:00 PM - 4:00 PM
167
Gustafson and Vogelmann
Breakout Description
The Large-Eddy Simulation (LES) ARM Symbiotic Simulation and Observation (LASSO) activityis growing to include simulations of deep convection in addition to the shallow-convection LES runs
currently available. The new scenario will focus on deep convection during the CACTI field campaign in
Argentina with a primary focus on convective initiation and upscale growth. As such, the simulations will
be much more ambitious than the current shallow convection LES runs.
After a short update on the status of the current, shallow-convection scenario, the primary focus of this
breakout will be to share the current plans for the new LASSO scenario and get community feedback to
refine the path forward. Of particular importance will be a discussion of the science drivers, as these guide
both the model configuration to be used along with how the observations will be employed within the
LASSO data bundles. Potential model configurations and outputs will be discussed. We also seek to engage
those involved with the CACTI observation data sets to evaluate options for how to best use the available
measurements.
Main Discussion
The session started with an update on LASSO shallow-convection activity since the previous ARM/ASR PIMeeting. We announced that 17 new shallow-convection cases will be released this summer, at which
point the shallow-convection scenario will be put on hiatus to enable focus on the new CACTI scenario
and on development of an ACE-ENA scenario in 2021. As part of the new release this year, a new highfrequency
observation data set will be added to the data bundles. This adds the LASSO-relevant
observations at near-native sampling frequencies, which has been requested by multiple users and now
naturally falls out of the automated LASSO workflow. A new product using the COGS photogrammetrybased
cloud mask and skill scores will also be released that will provide a much more realistic
representation of the cloud vertical extent than the KAZRARSCL cloud mask currently used to retrieve
cloud mask profiles.
The bulk of the session focused on the new CACTI deep-convection scenario. We anticipate the scenario
details to be fleshed out by the end of 2020 and the LES runs to take into 2021 before we will be able to
release the new product. Science motivations for the new scenario were proposed based on convective
initiation and early upscale-growth of deep convection. This then shaped the proposed modeling and
evaluation approach that was presented. The Q&A session was very active. This indicates strong interest
in this scenario and enabled acquiring valuable input on many aspects of LASSO-CACTI plans.
Key Findings
LES of deep convection will require much bigger domains than for the current shallow-convectionscenario. A rough estimate is that we will be able to afford approximately 20 LES simulations on ARM’s
cumulus cluster if it has 12,000 cores. These 20 simulations would likely be apportioned as 2 LES per
case for 10 case dates. The maximum domain size we could reasonably afford with 100-m grid spacing
would be about 225 km wide. This will work well with the focus on the early portion of the convective
lifecycle prior the large storm development that would overwhelm a domain of this size.
Ensembles will be employed for this scenario to determine the best forcing option(s) to use for each
case date. Ensembles will be used with kilometer-scale grid spacings to identify which forcing option
generates the best storm characteristics. This will then be used to down-sample the forcing options to
use for the LES.
Users were engaged in discussing possible outputs of the model including the frequency and resolution
of data saved. One option suggested was to coarsen the raw model output to the model’s effective
resolution and making this one option for users to download. This would reduce the data size by roughly
a factor of 25.
Issues
RELAMPAGO data are ready to use through the NCAR archive including DOIs. However, details regardingtheir use will need to be worked out such that appropriate credit is given to the data providers and not
buried due to the blending of data into the LASSO data bundles.
Needs
The issue of how to encourage and disseminate community feedback was discussed. Options suggestedduring the session were Github, Confluence, and/or Slack. Gustafson will investigate options and will be
implementing a communications plan in the near future. Assistance will be needed from ARM
Communications and the ARM Data Center to ensure security and policy requirements are met.
The rough estimate of storage space needed to deliver the CACTI scenario is more than 1 PB. The ADC
should be prepared to handle this large amount of data in 2021. We will need to work out how to
disseminate this quantity of data to users and enable users to interact with the data on ARM’s clusters,
either directly or via a client-server analysis methodology. Jupyter notebooks could be a powerful tool
for users to work with the data on ARM’s clusters.
Decisions
This session focused on disseminating proposed approaches for the CACTI scenario. Final decisions willbe made later this year after solicitation of further community feedback (discussed below with possibly
a fall session or other telecon to present and discuss our final tentative decisions).
Future Plans
Development of the ACE-ENA scenario will begin while the LASSO team waits for the CACTI simulationsto complete. Some work will begin late in 2020, with the bulk of this maritime-cloud scenario occurring
in 2021.