AIM Working Group by

What questions or comments do you have for Peter, Jim and Dan?

dcolegrove1 — 2023-10-31 15:48:51 UTC

What questions or comments do you have for Craig and Hugh?

dcolegrove1 — 2023-10-31 15:49:11 UTC

What questions or comments do you have for Matthias and Nick?

dcolegrove1 — 2023-10-31 15:49:14 UTC

What questions or comments do you have for Tammy?

dcolegrove1 — 2023-10-31 15:49:16 UTC

What questions or comments do you have for Cenlin?

dcolegrove1 — 2023-10-31 15:49:18 UTC

What questions or comments do you have for John?

dcolegrove1 — 2023-10-31 15:49:53 UTC

What questions or comments do you have for Alma?

dcolegrove1 — 2023-10-31 15:50:01 UTC

2023-10-31 17:30:44 UTC

Can CESM be used for weather forecasting, rather than climate forecasting? If yes, would the weather forecasts be credible, or is the system too focused on S2S and climate projections to lead to credible weather forecasts? If no, what would it take to make CESM ready to produce credible weather forecasts?

Thanks Peter! Are there some specific observations that could help to achieve your objectives?

2023-10-31 17:31:16 UTC

Does S2S system include chemistry? And what are the plans for including aerosols/chemistry?

2023-10-31 17:31:20 UTC

SIM

2023-10-31 17:35:50 UTC

What are the MMM concerns with SIMA?

Is there a timeline for a functional and portable DA using jedi?

2023-10-31 17:37:08 UTC

2023-10-31 17:37:22 UTC

What is the way forward with DA. DART or Jedi or both? Is there room for both?

Thanks, Craig. What is the timeline for MPAS replacing WRF?

2023-10-31 17:37:40 UTC

Is chemistry included in your MPAS? Do youhave collaborations with EPA?

2023-10-31 17:38:03 UTC

What is uncertain about DART: Is it possible that DART will be decomissioned?

2023-10-31 17:38:29 UTC

How are the CESM subseasonal prediction system and the S2S prediction w/ MPAS complimentary vs. overlapping? e.g., DA in S2S w/ MPAS, but not CESM subseasonal ESP. How will these efforts move forward together?

2023-10-31 17:39:47 UTC

2023-10-31 17:45:55 UTC

Are there any possible synergies between HAO's modeling needs and other models used at NCAR (e.g., MPAS, CESM)?

How closely do you (HAO) work with the core DART people? Are major modifications needed to DA systems for the solar prediction problem?

2023-10-31 17:46:00 UTC

Thanks Matthias! Is it expected that influx of funds to HAO for new staff would significantly help to achieve the future plans?

2023-10-31 17:46:31 UTC

Does EOL currently work with DART and or Jedi developers on including EOL instruments in DA?

2023-10-31 17:50:13 UTC

2023-10-31 17:51:03 UTC

How do you envision achieving this "shift in culture" you mentioned? Does EOL want to develop more in-house expertise about models and DA? Or is the idea to work more with other labs?

Is there a collaboration with ACOM on the data/model integration strategy for chemistry data?

2023-10-31 17:52:13 UTC

What are some specific steps needed to make obs and instrument development in EOL more synergistic with model development? In an ideal world, how would this happen?

2023-10-31 17:52:33 UTC

2023-10-31 17:52:51 UTC

How important are models for the design of new instruments. Do you use models to compute synthetic observations that then guide instrument development?

2023-11-13 15:52:49 UTC

The uncertainty regards how DART and JEDI will coexist. I think it's unlikely DART will entirely go away, but if it does, that ultimate decision would not come from MMM.

2023-11-13 15:53:29 UTC

No chemistry is currently included in MPAS, but I believe there have been some discussions to get that started. We don't have collaborations with the EPA as far as I know.

2023-11-13 15:54:51 UTC

Who knows? I think some of it will depend on how fast the community moves to MPAS. I feel like we've been talking about this for 5+ years at MMM, and the current strategy seems to be to support both WRF and MPAS, but to gradually support MPAS more and WRF less.

2023-11-13 15:56:07 UTC

JEDI is probably the way forward for MPAS-based data assimilation because it has more capabilities than DART. But, DART could remain a very useful research tool. Personally, I think there's room for both, but we'll have to see how things evolve. It's unclear if JEDI will ever really become a good research tool or if it will be too operationally focused.

2023-11-13 15:58:25 UTC

JEDI definitely already works in many capacities, especially when interfaced with MPAS. There's no real timeline for further improvements as far as I know, but the software continues to rapidly evolve.

2023-11-13 15:59:22 UTC

Hugh might add more, but my (Craig's) concerns are that I simply don't have a good understanding of how the project will evolve and how useful it will ultimately be.

2023-11-28 22:54:38 UTC

We have worked closely with the DART group for a while (10+ years) on implementation of DA in WACCM and WACCM-X. There is also work related to DA in TIE-GCM and solar dynamics. The collaboration with DART has been especially helpful for implementing DA in these models.

The DA problem for solar/space weather is slightly different compared to atmosphere DA, so some modifications are needed. This is due to it being partly a forced/boundary-driven system as well as the significantly fewer number of observations

Are there any plans to add MUSICA capabilities into MPAS-A, CM1, or FastEddy?

2023-11-29 20:25:12 UTC

Reply: Chemistry with MPAS should be added through SIMA. Limited efforts to add GOCART chemistry/aerosols into MPAS-A.

Currently CESM only has hydrostatic dynamics and so cannot be used for high resolution weather forecasting. Development is proceeding to add mpas and se non-hydrostatic dynamic cores. This will also require development of new physics parameterizations appropriate at these scales - SIMA.

2023-11-29 20:27:21 UTC

We are currently running S2S with both a lowtop model as well as a high top model which includes chemistry.

2023-11-29 20:28:58 UTC

You mentioned AI emulation for aerosols/chemistry, is there much effort underway in ACOM on this now? Or is it more a future direction?

2023-11-29 20:32:07 UTC

Our next step in the ongoing incremental improvement of the S2S system is to add DA capability.

2023-11-29 20:34:46 UTC

2023-11-29 20:34:55 UTC

Could you use an (updated) CESM-WACCM configuration for predictability and DA capabilities?

Is there an understanding of how much DA improves chemical forecasts? If this is substantial, is it an area that should be a significant priority going forward

2023-11-29 20:35:18 UTC

What would be the CISL resources allocated to support AI versus physics-based models?

2023-11-29 20:38:33 UTC

In which areas can we develop stronger collaborations between acom/cisl?

2023-11-29 20:39:20 UTC

2023-11-29 20:40:31 UTC

Will changing hardware and computing restrict our ability to continue current scientific / DA activities? How?

I assume that usint Noah-MP in the UFS modeling system means that it will be coupled through CMEPS? If so that should make it fairly trivial to incorporate into CESM.

2023-11-29 20:44:51 UTC

Does the RAL work on S2S predictions include collaboration with the work in CGD on S2S?

2023-11-29 20:49:34 UTC

Are chemistry/aerosols integrated in WRF-Urban?

2023-11-29 20:52:01 UTC

2023-11-29 20:52:38 UTC

Does coupling of models with FastEddy imply refactoring those models for GPUs?

What primarily sets the direction for these various model-development efforts in RAL? Is it mainly individual externally funded projects? If yes, how are these efforts sustained over the long term?

2023-11-29 20:52:44 UTC

2023-12-15 15:51:25 UTC

For solar and magnetospheric models it is absolutely critical to have dynamical cores that do include the evolution of magnetic field and are capable with dealing with highly compressible fluids and supersonic flows (development of shocks). Dynamical cores used for the Earth's atmosphere do not include these features and implementing them would have little benefit for applications to the Earth system. There may be some connections we could explore with regard to radiation transport and atmospheric chemistry. In the solar chromosphere we need to solve rate equations for atomic level populations coupled with the radiation field, which do have some similarity to chemical reaction frameworks.

2023-12-15 16:03:35 UTC

Yes! We have a dire need for software engineering support. On the solar side we have 0 SEs helping with model development. On the geospace side we have a part time SE for model support. We are thin staffed on the scientific side, anybody departing is leaving a gap nobody else could fill.

Yes, UFS uses ESMF/NUOPC.

2023-12-18 05:04:14 UTC

RAL works on various S2S prediction aspects, e.g., fire, drought, precipitation, etc., but not much collaboration with the work in CGD currently likely because they are driven by different projects and sponsors.

2023-12-18 05:06:10 UTC

Yes, WRF-urban can be run with online coupled chemistry (i.e., WRF-Chem-Urban). The urban physics will affect meteorology, which further affects chemistry/aerosols, while chemistry/aerosols will affect meteorology which further feed back to urban surface processes.

2023-12-18 05:07:40 UTC

Yes, it is the goal, but might start with the cpu version first.

2023-12-18 05:12:28 UTC

As RAL heavily relies on external funding, most model-development efforts are driven by external funded projects. RAL always seeks common interests between NCAR/RAL and sponsors to achieve a best strategy so that the model development efforts align with NCAR/RAL strategic plans and at the same time satisfy sponsors' needs.

2023-12-18 05:19:05 UTC

For example, the NCAR community Noah-MP LSM is used in the NOAA UFS system, which provides a good opportunity for RAL to develop and enhance Noah-MP with external funding from NOAA. Similarly for WRF-based modeling systems. This is the strength of NCAR developing community models, which are widely used by different external sponsors.

2023-12-18 05:19:14 UTC

However, it is indeed a challenge for RAL to keep such efforts sustainable in a long term.

2023-12-18 05:20:43 UTC

CISL views AI and physics-based techniques as valuable in advancing our understanding of NCAR science objectives. We will continue to support whichever algorithmic approach our science community needs to advance scientific inquiry.

2023-12-18 15:23:37 UTC

Currently, ACOM and CISL have a strong ongoing collaboration regarding the optimization and GPU enablement of the MUSICA solver. We hope that this collaboration will continue and potentially expand into other software that is important for the atmospheric chemistry community.

2023-12-18 15:24:29 UTC

Currently, 20% of the computing capability of Derecho is provided through GPU-based computing. We expect that in future systems, the percentage of computing capability provided by GPU-based computing will either stay the same or increase. Software that is limited to either CPU-only or GPU-only execution will not be able to easily adapt to future changes in either the NCAR or wider computing environment.

2023-12-18 15:27:10 UTC

2023-12-19 18:12:56 UTC

JEDI is already functional and portable. Questions revolve more around timeliness of getting certain capabilities added and user-friendliness.

2023-12-20 23:05:17 UTC

I agree with Craig about SIMA, and would add that the direction isn't clear to me (or at least, IMO hasn't been communicated clearly, but others may have a different opinion).

2023-12-21 18:25:01 UTC

Not directly but we would be happy to do so.

2023-12-21 18:29:04 UTC

We are gaining more expertise in data assimilation within EOL since hiring Junkyung Kay, for example. Significant progress in modeling and data assimilation is only possible through strong collaborations across NCAR. We have been working most closely with MMM and RAL and are discussing further potential collaborations. We hope to be involved in data assimilation of all EOL datasets, including APAR.

2023-12-21 18:31:57 UTC

Yes, EOL has strong collaborations with ACOM, including with a joint appointments. This could also be further strengthened.

2023-12-21 18:35:57 UTC

This is another excellent question! Allison, our new EOL Director, has stated that she sees instrument development going hand-in-hand with modeling and data assimilation. As a relevant example, we have recently led OSSE (Observing System Simulation Experiment) data assimilation analyses of our water vapor MPD (MicroPulse Differential Absorption Lidar) instrument.

2023-12-21 18:39:14 UTC

I think strong communications and collaborations with MMM, RAL, ACOM and CISL are necessary to make this happen. We first need to identify the people, projects and funding to pull together the diverse expertise to communicate and contribute together toward this effort.