Breakout Summary Report

Breakout Description

In 2020, the DOE ARM COMBLE field campaign set out to study cold-air outbreaks (CAOs) over the Norwegian Sea. CAOs globally exchange large quantities of energy between both atmosphere and ocean and the high and mid-latitudes, and thus play an important role in weather and climate. In late 2021, a model intercomparison project (MIP) was initiated to focus on shallow mixed-phase clouds that occur under convective CAO conditions observed during COMBLE. In mid-2022, a COMBLE MIP white paper was accepted by the GEWEX Atmospheric System Study (GASS) panel, describing a two-phase project for LES and SCM models. In phase one, droplet and ice concentrations are specified. In phase two, the specification adds an observation-informed initial aerosol size distribution profile with modally specified composition and hygroscopicity, adequate for prognosing droplet activation and heterogeneous ice formation. The COMBLE MIP effort has been an early adopter of the DOE ARM Data Workbench to support each stage of participation, in addition to providing access to the DOE Cumulus HPC for computationally intensive forward-simulation for model evaluation against ground-based lidar and radar measurements at Andenes. We anticipate that this MIP effort will lead to developing a sustainable framework to efficiently facilitate model evaluations against AMF observations in COMBLE. This session is intended to highlight first model submissions and evaluation with ARM data, in addition to summarizing the supporting role of the ARM Data Workbench in enhancing participation and facilitating the model-observation evaluation. Both modeling and observational participants are invited to help guide the MIP analyses and observational evaluations going forward, as well as to provide feedback and suggestions regarding the new ARM Data Workbench capabilities.

Main Discussion

This breakout session focused on recent developments with respect to the COMBLE LES/SCM Model-Observation Intercomparison Project (MIP). Ann Fridlind kicked off the session by providing a brief overview of past MIPs that have focused on arctic clouds, including M-PACE, SHEBA, ISDAC, and CONSTRAIN, while highlighting similarities and differences between those MIPs and the COMBLE MIP. Ann finished by motivating the need for the COMBLE MIP, and why, more generally, we should focus on cold-sector mixed-phase cloud studies. The second block of the session highlighted recent efforts by our team to advance COMBLE MIP, with particular emphasis on aerosols, observations for model evaluation, and integration of the MIP activities with the DOE ARM Data Workbench. Presentations in this part of the breakout were given by Abbey Williams, Daniel Knopf, Florian Tornow, and Maxwell Grover. Lastly, Timothy Juliano gave an overview of the COMBLE MIP activities to date, including preliminary results using the ARM Data Workbench to conduct analysis.

A discussion period of ~40 minutes followed the presentations. Overall, the COMBLE MIP efforts were well received, with the group particularly interested in the ARM Data Workbench activities. One main discussion topic was regarding the ability to extend the MIP to E3SM and other global models in nudged mode. One participant pointed out that hierarchical modeling is a good approach, and that the COMBLE LES/SCM represents a “pilot project” that can be expanded. Maxwell noted that the CESM tutorial used JupyterBook, and so many users in the climate modeling community are likely familiar with the Python-related infrastructure. Maxwell also mentioned that the Workbench could be extended to other ARM/ASR projects, since there are templates available for COMBLE, OPEN ENG focused on ARM notebooks, and Open Science tutorials. Additional discussion centered on developing more focused science questions and hypotheses to be asked/investigated for the COMBLE MIP.

Key Findings

Abbey Williams demonstrated excellent CCN-droplet closure at Andenes, and consistent indications that the accumulation mode appears to be roughly 50% sea salt. Daniel Knopf showed remarkably good aerosol-INP closure at both Zeppelin and Andenes sites, comparing three parameterizations. Models also consistently produced remarkably realistic cellular structures at the main Andenes ground site. However, LES at 100-m grid spacing were unable to reproduce early roll formation as strongly as observed. We note that the CONSTRAIN CAO intercomparison also did not produce rolls, for reasons that were not understood. LES studies in the literature are not generally well linked to specific observed cases and appear to exhibit very weak rolls relative to observations. The Ovchinnikov group also found difficulties in this area, with notable sensitivity related to large-scale forcing. In short, this is a long-standing problem that is not well understood. Ability to better constrain the case for COMBLE MIP specifically was highly limited by complete lack of direct observational data regarding (thermo)-dynamic structure at the ice edge and upwind. Furthermore, requesting 100-m grid spacing LES (a reasonable minimum) for domains large enough to contain the cells at the main observing site (cells of order 30-50 km in diameter and 4 km in depth) made higher resolution difficult to justify. We note that nudged winds (rather than geostrophic, as now specified) were not preferred by the international climate modeling community for SCM or the cloud modeling community for LES.

Issues

Matt Shupe expressed concern that model surface fluxes were compared with the ERA5 model, without any supporting observations. He asked if observations were available. We debated the potential use of Bear Island and Andenes profiles + satellite SST data as an input to flux-profile estimates of surface fluxes. Additionally, there was a lot of interest to consider somehow including nudged global models, whether climate models or storm-resolving with or without nested grids. Ann thought that comparison with LES and SCM could be quite straightforward to do if Lagrangian trajectories were extracted from such models. Time series information could also be extracted at the Andenes and Zeppelin sites.

Needs

Based on the general sentiment of modelers in the ASR community, one key result from the breakout session was the need to more effectively and efficiently evaluate numerical models across a range of scales (i.e., large-eddy simulation to global climate) using ARM measurements. It is also important to conduct such comparisons in a unified manner to strengthen the validity of findings in studies focused on model-observation evaluation. Conducting deliberate MIPs is one approach to addressing the above challenge. An additional point related to MIPs focused on cloud transitions/evolutions from a Lagrangian perspective is the critical need for traditional fixed measurement platforms to be complemented by non-fixed measurement approaches (e.g., aircraft or unmanned aerial systems). Such quasi-Lagrangian measurements are important for modelers to not only constrain forcings but also conduct robust evaluations.

Decisions

It was agreed that the current COMBLE MIP setup is an adequate starting point and that the Workbench structure is a really excellent framework for moving forward in modular additional projects. Regarding the “rolls problem”, in the early phases, it was agreed that it is now a good approach (after extensive tests that basically failed) that the leadership team will engage the wider community in whether any participants can produce rolls that are objectively similar to satellite observations in translating LES. If adjustments to the forcing are reasonably suggested, they could be incorporated before the final, largest-domain simulations are requested following leadership team tests with several models.

Future Plans

A clear path forward to strengthen the value of MIPs using ARM measurements includes the expansion of the ARM Data Workbench. Considering the strong collaboration between the COMBLE MIP and ARM Data Workbench teams, we believe that the COMBLE MIP can act as a template for future MIPs with plenty of opportunities to enhance the workflow. In the future, this may also allow for a MIP effort to explore multiple case studies within a given field project, which to our knowledge has not yet happened. Such an effort would be very powerful since one limitation of a MIP is that it examines only a single case, which may not represent the broader conditions characterizing a given geographic region.

Action Items

We note several key action items stemming from this breakout session:

• Test substantially higher-resolution simulations for the early COMBLE-MIP period, which were not yet attempted owing to lack of upwind data and a focus on the large-domain results most relevant at the main Andenes site. It is quite unclear whether this will help.


• Discuss the potential for additional surface flux constraints based on COMBLE measurements with campaign PI Bart Geerts.


• Release the MIP to the community via the Workbench within several weeks. Then meet with the E3SM, ModelE3, and potentially CESM groups to discuss a possible extension to global models in nudged mode, within several months.