Breakout Summary Report

ARM/ASR User and PI Meeting

Breakout Description

The emergence of global storm-resolving models signifies the advent of a new generation of weather and climate models. However, uncertainties remain with these models in simulating clouds and convection as parameterizations are still needed for processes operating at scales that are smaller than the grid resolution of a few kilometers. This session aims to discuss how to best use ARM measurements with modeling approaches to better understand cloud and convective processes. Novel approaches with a view to future opportunities will be a particular focus. Discussions will also focus on methods for improving the representation of clouds and convection in kilometer-grid-scale models, especially the newly developed E3SM global storm-resolving model. The session will include talks and promote discussions on current pressing issues of model physics, improved methods to compare ARM observations with kilometer-scale model outputs, and the need for observations to diagnose model performance and improve physics parameterizations at local, regional, and global scales.

Main Discussion

This session invited two2 long talks (15 to 20 minutes) and seven short talks (around 6 minutes or up to 4 slides). After each talk, we had a Q/A and short discussions.

All these talks aimed to address how to best use ARM measurements with modeling approaches to better understand cloud and convective processes and improve their representation in kilometer-grid-scale models. Talks and discussions focused on current pressing issues of model physics, improved methods to compare ARM observations with kilometer-scale model outputs, and the need for observations to diagnose model performance and improve physics parameterizations at local, regional, and global scales.

Key Findings

The two long talks targeted on the performance of kilometer-scale models (and LES models) in simulating mesoscale convective systems and mesoscale variabilities in boundary- layer cloud transitions, e.g., cold air outbreak (CAO) events. The remaining challenges include spatiotemporal displacements, large case-to-case variabilities, lack of constraints on model prognostics, and comparability of simulated and observed fields, especially on subgrid-scale turbulence and ice nucleation processes. Several remedies were suggested, which include observational-based case libraries, ensemble simulations, machine-learning-aid tuning strategies, and the adoption of instrument simulators. The seven short talks covered: 1) new modeling tools to help global storm-resolving models (GSRM) zoom into ARM sites, such as Regionally Refined Model of SCREAM (RRM-SCREAM) and Doubly Periodic SCREAM (DP-SCREAM, analogous to a standalone CRM); 2) new analysis approaches including the development and application of cell tracking algorithms on convective processes, e.g., TOBAC and PyFLEXTRKR; 3) new observational data analyses for GoAmazon, CACTI, and TRACER; 4) observational needs for quantifying aerosol-cloud interactions, and reduction of persistent dynamical and microphysical biases. Specifically, extracting signals related to key processes requires analyses of both large comprehensive data sets and detailed case studies with sufficient spatiotemporal environmental context, sometimes available from models, for sampling-limited measurements.       

Issues

N/A

Needs

N/A

Decisions

N/A

Future Plans

This is our first time to organize a session focusing on high-resolution modeling and application of ARM data for such model diagnosis and improvement. Many of our talks and discussions were about mesoscale convective systems (MCS) and the associated processes and observations. MCSs are significantly better simulated in kilometer-scale models compared to traditional lower-resolution climate models. Subgrid processes remain a challenge in km-scale models, especially the simulation of boundary-layer clouds and their transition to deep convection and upscale growth into MCSs, which is at the interface of convective scale and mesoscale processes including the representation of cloud entrainment and detrainment.  We hope to continue this session and discussion next year.

Action Items

1) Potentially propose for an AGU session in 2023.

2) After the session, continuous discussion focused on potential collaborations to assess E3SM and SCREAM using GoAmazon, SGP, CACTI, TRACER and the upcoming SEUS data.