Breakout Summary Report

ARM/ASR User and PI Meeting

13 - 17 March 2017

Session Title: LASSO
Session Date: 13 March 2017
Session Time: 4:00 PM - 6:00 PM
Number of Attendees: 65
Summary Authors: William Gustafson and Andrew Vogelmann

Breakout Description

We will provide details about results from the LASSO Pilot Phase as it nears its end and begins transitioning to its operational phase. Particular emphasis will be on familiarizing potential users with the LASSO design, anticipated products that will become available for users, means for users to interface with LASSO, and plans for transition to operations. Additionally, a short presentation will be invited for each of the three primary user types to facilitate discussion on potential science applications and user needs. The three user types identified so far are: (1) Observationalists, to aid in retrieval development and sampling strategies; (2) Theoreticians, to use co-variation of properties to test relationships across cases, and (3) Modelers, to use LASSO results to help jump-start the process for their in-depth modeling studies.

Main Discussion

This session focused on 1) presenting the current status of the Large-Eddy Simulation (LES) Atmospheric Radiation Measurement (ARM) Symbiotic Simulation and Observation (LASSO) pilot project and 2) encouraging researchers to discover ways that LASSO could enhance their research. Toward these ends, a series of talks were presented aimed at stimulating discussion around LASSO capabilities and ways others have considered using LASSO for their research.

The session began with an overview presentation by William Gustafson, the project principal investigator (PI), about what LASSO entails. This was followed by three talks about how LASSO can benefit different research areas. Pavlos Kollias and Sonja Drueke presented on how LES output from LASSO can be used as a virtual environment to test radar scan strategies and for developing tools for estimating entrainment. Graham Feingold presented on how an LES library spanning many cases can enable the detailed separation of signal from noise for covarying processes, such as aerosol impacts on clouds and the meteorological environment impacting cloud liquid water. And, Vince Larson presented on using the LASSO data bundle library to smartly choose LES forcings and cases to provide information necessary for parameterization development. The session then had presentations aimed at some of the technical details surrounding product development within LASSO. Andy Vogelmann, the LASSO Co-PI, presented on the status of data bundle development and efforts that have been conducted since reporting at last year’s ARM/ASR PI Meeting. Bhargavi Krishna introduced enhancements to the Bundle Browser to simplify discovery of LASSO data within the ARM Data Archive. Zhijin Li presented on the LASSO data assimilation plans. Gustafson closed the session with a quick overview of what will be done during the remainder of the LASSO pilot project and encouraged community members to provide feedback to the LASSO team and ARM leadership to ensure LASSO will meet community needs. Discussion was interspersed throughout the session.

Key Findings

LASSO has forced us to address difficult, but previously neglected, issues such as how well a single profile provides cloud fraction representative of the larger Southern Great Plains (SGP) region.

LASSO provides a useful way to do observing system simulation experiments (OSSEs), such as for developing retrieval methodologies for entrainment. This was demonstrated by Drueke.

The current periodic LES domain configuration being used for LASSO will not meet the needs of all applications. Some users will need a nested LES approach to better represent heterogeneity of the cloud and meteorological state around SGP. The periodic domain works well for applications needing a semi-steady-state representation of the cloud field at SGP, while applications targeting spatial variability require nesting.

The range of forcings provided by LASSO provide a wider range of conditions that users can take advantage of for their research, such as for the large ensemble of LES performed by Feingold.

Larson gave a provocative talk based on the following premise: observations are good for detecting model errors but are not adequate for fixing these errors—LES is needed for diagnosing specific causes of errors and correcting them. When using single-column models (SCMs) to develop parameterizations, Larson stated that “one can never expect SCMs to outperform an LES.” This resulted in a discussion of the applicability of LES and SCM for developing model parameterizations. The overall message from the participants is that one needs to carefully use the model and consider its benefits and limitations when using the LES fields. For example, details surrounding handling of surface heterogeneity in the LES could impact its representativeness of reality. LES modeling is one piece of an overall tool set required for developing accurate parameterizations. The talk was effective in forcing participants to think about how to appropriately use LES and understand its benefits and limitations.

Needs

LASSO needs to better capture the inversion to accurately simulate the shallow clouds. Toward this end, optimal assimilation of ARM profiles is important. However, this is difficult due to the smoothed nature of the AERIoe retrieval. Research into how to assimilate the AERI radiances directly could benefit MSDA and the subsequent forcing generation for LASSO.

Thijs Heus expressed a desire to be able to search LASSO data bundles based on properties of the observations and/or simulations -- for example, all cases where liquid water path is greater than a given value.

Vince Larson and Maike Ahlgrimm expressed a need for budget terms in the LES output to aid process understanding and parameterization development. The LASSO team will look into which budget terms should be archived from the LES.

Steve Klein requested that satellite images be included as part of the data bundles to help users easily identify the cloud conditions for each case. This was partially done in the Alpha 1 LASSO release, but a decision had not been made regarding whether this would be done in the final version of the data bundles.

Yolanda Serra asked about plans for adapting LASSO for other locations and about the inclusion of different types of instrument simulators, e.g., a GPS simulator. More information will be needed to discern the usefulness of additional simulators and what simulators are available for LES-scale models. By providing sufficient output from the LES, users should be able to apply their own simulators as needed.

A concern was shared about the LASSO data volume becoming large and cumbersome to work with. The ability to work on ARM’s clusters should help address this concern for users who cannot easily download the data bundles. This could also enable users to build a library of software routines for the community to share.