Breakout Summary Report


ARM/ASR User and PI Meeting

Earth System Model Evaluation and Development Using ARM Data Sets: Community Modeling Activities
25 October 2022
2:00 PM - 4:00 PM
Ann Fridlind, Christina McCluskey, Shaocheng Xie, Hugh Morrison

Breakout Description

Recommendations from the 2021 report of ARM's Cloud Processes and Measurements Science Group highlighted the potential value of additional emphasis on community model evaluation exercises. Cloud-focused community modeling exercises have been historically undertaken with international participation through the World Climate Research Program's Global Energy and Water Exchanges (GEWEX) project panels, and many of those were founded on ARM data sets. Such exercises often serve as a foundation for Earth system model development and improvement, whether at the case study or statistical level (e.g., most climate models can be configured to run past single-column model case studies). Within the context of ARM aerosol and cloud data sets and how they are now being used for Earth system model evaluation, development, and improvement, this session invites wide-ranging discussion of past, current, and potential future GEWEX Atmospheric System Studies (GASS) community modeling activities, as well as WMO and other collaborative groups using long-term data sets. We also invite general discussion of the aerosol-cloud-precipitation-radiation-dynamics uncertainties that are most limiting confidence in estimates of climate sensitivity, and how these could potentially be better constrained in models using ARM observations via future community activities. One desired outcome from this breakout session is to support incubation of aerosol-cloud-related modeling activities that could be proposed to GASS for broader international community participation.

Main Discussion

The discussion was divided into three sections. The first talks provided introduction, background, and summary of two existing community model evaluation projects:

2:00-2:05 Background and agenda/Ann Fridlind
2:05-2:15 Successes and barriers/Hugh Morrison
2:15-2:25 The GASS Diurnal Cycle of Precipitation project/Shaocheng Xie
2:25-2:35 COMBLE project/Tim Juliano and Florian Tornow

Brief discussion included emphasis on the two current projects listed above. Tim was asked about the model resolution selected for the planned COMBLE intercomparison. He reviewed his slide showing the current proposed resolution, but reported that this is not currently a final choice and adherence to the suggested specification will not be required within reason owing to the project goal of comparing all simulations with observations (rather than solely to one another in a highly constrained format). That approach will also allow use of microphysics schemes that may require a smaller domain owing to computational expense. Discussion closed with Shaocheng emphasizing the need for large-eddy simulation and cloud-resolving model contributions at this phase of the Diurnal Cycle of Precipitation project.

The second set of talks addressed ESM evaluation using the Earth Model Column Collaboratory (EMC2) radar-lidar simulator package to compare model cloud properties in large-eddy simulation and large-scale models to ARM remote-sensing measurements:

2:40-2:50 EMC2  for LES, ModelE3, and CESM/Israel Silber
2:50-2:55 EMC2  for E3SM/Yuying Zhang and Jingjing Tian

Peter van Leeuwen suggested that data assimilation techniques used to move from the space of observations to model quantities could provide useful context for instrument simulators. Ann commented that the use of simulators arose at GISS owing to a lack of capability to robustly define uncertainty in observed cloud phase in any other way, especially in cases of loss of lidar signal—a lesson that others had learned earlier, leading to COSP2, which provided the subcolumn foundation also used in EMC2. Yuying was asked about the current choices available in the COSP2 package, with interest in the differences between COSP2 and EMC2 shown. The subcolumn distribution of clouds and precipitation was thought to be one of the largest uncertainties in current radar and lidar simulators for low-resolution ESM evaluation. Toshi Matsui suggested that we not use subgrid sampling, but average observations in time and space consistent to ESM grid boxes. The importance of agreement between simulator parameters and model physics was agreed as a critical component.

The third set of talks raised the question of incubating new activities from the perspective of leading ESM uncertainties:

3:00-3:10 Aerosol physics most limiting ESM skill/Susannah Burrows
3:10-3:20 Cloud and precipitation physics most limiting ESM skill/Andrew Gettelman
— Contributed lightning slides/Xiaohong Liu on cloud phase in CESM/E3SM

Open discussion following these slides ranged quite widely in less than about 30 minutes remaining in the session. Bill Collins raised the question of SO2 biases across climate models, which Susannah Burrows commented upon. Steve Klein advocated for advancing new model tests of cloud physics, and agreed that proceeding on a cloud regime basis is a promising approach. Thijs Heus asked whether there is a role for LASSO case studies as a driver.

Discussion was also initiated regarding logistical needs from the community to participate in model evaluation projects such as the DCP and COMBLE projects. This included discussion in the context of broader international efforts such as the pan-GASS and the International Cloud Modeling Workshop (ICMW, taking place every 4 years in conjunction with the International Commission on Clouds and Precipitation). Lulin Xue described activities with the ICMW and how they could inform future model evaluation projects.

Key Findings

There was a general sense that this is a new discussion to be having, from the perspective of using ESM biases to drive decisions regarding model-observation evaluation. There was broad interest to continue the discussion, as indicated by a raise of hands at the close of the discussion.


How to have the required discussions to develop ideas and pathways forward? How to support sustainable exercises in the community?


Continue the discussion to incubate ideas and approaches. A dedicated workshop would be a logical next step.


The approach for coordinating model evaluation projects that is currently being advanced has promise but requires further development with more dedicated time for discussion, ideally with more structure to break down and develop approaches.

Future Plans

A first workshop proposal could be prepared by the leads of this breakout. Two leading outcomes of the workshop could be (1) re-envisioning the original Randall diagram for model improvement shown by Hugh Morrison, and (2) establishing a living list of ESM uncertainty targets. This foundation would precede discussion of specific use of observations by design. Sustaining the group and inviting all relevant contributors would be a key aspect, in order to support fully open participation—and sustainability. In other words, the objective is to communally incubate community exercises that are sustainable (e.g., potentially related with modular components supported by infrastructure where that makes sense) rather than imposing such exercises.

Action Items

Seek program management guidance regarding a virtual or in-person workshop proposal.