Breakout Summary Report

ARM/ASR User and PI Meeting

2 - 6 May 2016

Session Title: Update on the LES ARM Symbiotic Simulation and Observation (LASSO) project
Session Date: 4 May 2016
Session Time: 7:30 PM - 9:00 PM
Number of Attendees: 0
Summary Authors: William Gustafson and Andrew Vogelmann

Breakout Description

Many of you have questions about the ARM facility’s plans and progress toward implementing a routine LES modeling component to supplement the extensive observations made at the Southern Great Plains megasite. The LASSO project will have time during a plenary session to present a short overview of the plans. This evening session will allow us to go into more detail. We will have talks from the LASSO team regarding various aspects of the LASSO pilot project, e.g., the modeling workflow, model forcing generation, and metrics/diagnostics calculated as part of the planned data products. We plan ample time for questions and comments—your feedback will be an important part of the session.

Main Discussion

This breakout session focused on plans for the LASSO endeavor and providing an update on work done during the first year of the LASSO Pilot Project. Presenters talked about the overall LASSO vision and current state of the project, a new cloud classification product under development in support of LASSO, the data bundle concept, and a brief introduction to data discovery and analysis capabilities that could be provided through using NoSQL database technology. The session was meant to be a means for community members to learn more about LASSO and ways that they will be able to benefit from it.

Approximately 40 people attended this breakout with active participation, multiple comments, and questions, taking the session into overtime.

Issues

It was questioned how accurately the LES needs to reproduce cirrus clouds given ARM’s primary focus on shallow convection. This was noted as a valid question. Benefits of accurately simulating the cirrus include capturing their radiative feedback to the shallow clouds and enabling better simulated fields for use with instrument simulators that require the full radiative state of the troposphere.

Generating accurate forcings is one of the highest priorities of the LASSO pilot project. Three methods are being actively pursued right now: the constrained variational analysis (VARANAL), numerical-weather-prediction-derived forcing based on ECMWF analyses, and multiscale data assimilation. Testing to date with VARANAL has been with the 1-D, profile-based version. There was some discussion during the breakout of using the new 3-D VARANAL. More information will be needed to understand the nuances of the 3-D version to see how it can improve over the 1-D version.

A good discussion occurred around the issue of what to do with simulations that are representative of realistic clouds, yet do not coincide with the conditions observed on the corresponding day. For example, not all ensemble members are expected to reproduce the observed model state, which is the primary reason for using the ensembles in the first place. Some attendees felt there was value in keeping all the simulations as there could be value in them for understanding model bias as well as for providing additional statistics for certain types of studies. Other attendees questioned whether it is worth the space to save these simulations, and discarding them could be a way to conserve archive resources. No clear consensus was reached.

Another question was raised about the representativeness of the point measurements being used for the metrics, particularly the vertically pointing instruments of cloud variables. The response was to look forward to the scanning radar data, LWP from the BF array, and stereo photogrammetry. However, it was noted that, with the current metrics, caution should be used in over-interpreting the skill scores from the point measurements.

Needs

The LASSO team needs input to gain a better understanding of the required LES output frequency to adequately address anticipated researcher needs. This is particularly important for outputting time series of model volumes in 4-D. This will be the most expensive aspect of the LASSO storage needs.

One attendee noted that there could be value in saving model restart files to enable users to restart the model without having to run the entire time period of an event.

Future Plans

A suggestion was made to quantify the climatology of the LES results over time. This climatology would be useful for determining when atypical simulations occur where the results should be questioned and possibly discarded.

Action Items

The next year will be spent testing the value of forcings that include the new profiling instruments that will soon become available. The basic testing infrastructure is now in place, which will enable more efficient and systematic testing of a larger number of cases.

Prototype demonstrations are being developed with the Cassandra NoSQL database to evaluate its potential for enabling users to easily find simulations of interest, and potentially even do complicated analyses with the data. One attendee noted that this might be a great way to also interact with other ARM data beyond the initial focus on LASSO.