Agenda Item 1:

Agenda presentation by Ben

Discussion:

Agenda Presentation by Ben:

• Overview of the Zenhub issues
• Core team updates
• JCSDA Quarterly Review upcoming on July 23^rd
• Slides for the review need to be prepared

Any questions about the quarterly review? -No

Result:

N/A

Tasks:

N/A

Responsible People:

Benjamin Johnson

Deadline:

N/A

Agenda Item 2:

Overview of Zenhub Issues

Discussion:

Ben:

CRTM Roadmap on Zenhub is shown. As an example, the CRTM Surface codesprint hasn’t started yet because of the pandemic.

As a further example, Yingtao’s NLTE epic hasn’t been started yet, so please contact Ben to set up issues.

Looking at an individual Epic, it shouldn’t have an individual point estimate. The points should always come from the underlying issues. Epic #61 estimate is set to none and Hamideh should create some issues attached to it.

Result:

N/A

Tasks:

-       Create Zenhub issues and Epics for your work

Responsible People:

Everybody

Deadline:

AOP2020

Agenda Item 3:

Core Team Updates

Discussion:

Patrick’s Update:

Existing AVHRR3 coefficients will be submitted to Emily Liu for model testing.

A visible LBLRTM run has been completed and it was found that it doesn’t include extinction from Rayleigh scattering while it is included for the Thermal IR. Mark Liu has confirmed that this is an issue since the CRTM already has a separate Rayleigh scattering module. An impact study on this issue would be necessary.

Yingtao: Are you using default molecule cross-sections?

Patrick: - Yes.

Cheng’s Update:

Cheng: CRTM is now able to read in NetCDF aerosol properties for the default aerosol properties, so the goal is to update CRTM I/O to accommodate for more aerosol types, such as GOCART and CMAQ. Meanwhile it was found that the aerosol size definition was done for the original properties. Also, even though the original scattering properties were almost certainly computed using a Mie code, the original model is not available anymore.

Yingtao: What’s the ultimate goal of Cheng’s work?

Cheng: A more flexible aerosol model, including the default, GOCART and CMAQ.

Yingtao: We want CRTM to include a more flexible aerosol scheme. I have already completed modifications for GOCART and CMAQ in the CRTM which can be used as the basis for future modifications. Will you compute new scattering tables?

Cheng: Yes, that is what I am doing right now, mainly for dust.

Yingtao: So basically you are going to update the aerosol table for CMAQ and GOCART.

Ben: Any more comments? – No.

Ben’s Update:

Ben: CRTM_dev is essentially the old trunk. On Github I have created release branches that are essentially equivalent to the old release builds. There are two JEDI Release builds that include fixes specific to JEDI. There will also be a CRTM-3.0 alpha release that requires communication with Mark Liu.

The other issue I am working on is unit- and regression testing for the CRTM. The CRTM trunk will be merged to CRTM_dev.

We are in contact with the maintainers of the HITRAN database and they are eager to cooperate. The discussion will revolve around how HITRAN can support CRTM and RTTOV development.

The other issue is Spectral Response Function updates and antenna corrections. A workflow for this needs to be developed. If you have any experience on that please let us know.

There have been a lot of recent changes on the CRTM that might be included in a new REL-2.4.0, such as bug fixes, cloud fraction issues, and OpenMP. This is a somewhat preliminary discussion.

There is a polarized GPS-RO development underway mostly in cooperation with JPL and a team in Spain. Oriented hydrometeors over the GPS-RO ray can induce polarization, even though it’s a forward-scattered L-band signal.

Mark Liu: Is this for the (unintelligible)-Module?

Ben: No, we lack L-band support for the hydrometeors right now however.

Mark Liu: So this is independent from the CRTM?

Ben: Yes, but we do the scattering calculations and the JPL is interested in other instruments, such as radar, which would require my radar CRTM code.

Mark Liu: CRTM requires this additional functionality for looking from one side of the earth to the other.

Ben: Yes, including slant-path RT in the CRTM has been a long-standing discussion.

Yingtao: So for this one you only need the scattering properties and vertical profiles from the CRTM. Are you including the spherical geometry in the ray tracing of the RO?

Ben: There is a 2D operator outside of the CRTM developed at the Obs team. They assume a fixed ray path between observer and sender. Our only contribution will be if the ray path intersects a cloud layer. A Jacobian for the cloud layer would be important however.

Cloudy radiance support needs to be kick-started again for polarization. The effective radius will be extended from 2500 to 10000 microns to avoid the observed truncation issue.

Result:

-       NIR, Visible and UV band transmittance computations have been performed with LBLRTM.

-       A NetCDF interface for aerosol scattering properties in the CRTM is available.

-       Build Release branches on CRTM_dev are available.

-       A meeting with the HITRAN team has been set up.

Tasks:

-       Sensitivity test for Rayleigh scattering extinction in LBLRTM results.

-       Modify CRTM aerosol scattering I/O to allow more flexibility.

-       Compute L-band hydrometeor scattering properties.

-       HITRAN discussion

-       Polarized cloudy radiance support

Responsible People:

CRTM Core Team

Deadline:

AOP 2020

Agenda Item 4:

In-Kind Issues

Discussion:

Ben:

GPU porting has been discussed for Daniel Abdi.

Has this been started already?

Daniel: I cloned the CRTM repository and was able to compile it with GNU. It seems that there have been a lot of changes in REL-2.4.0 .

Ben: Don’t wait for REL-2.4.0, use 2.3.

Mark Liu: If you do the GPU update for 2.3, do you need to make any changes for 2.4 . Are there any automatic tools for converting the code?

Daniel: OpenMPIs offload instruction can be tested initially.

Ben: Porting to GPU requires extensive rewrites of the code.

Mark Liu: For OpenMP only few modules need to be changed. For GPU everything needs to be changed.

Ben: If this works out it might be possible to maintain a parallel version of the CRTM. It should be assessed which parts need to be changed.

Mark Liu: Is there an estimate how long a modification would take.

Daniel: We don’t have to port everything. We can port hotspots with OpenACC, but copying data will still take time, so we will start with porting only one subroutine and measure the time.

Ben: Daniel, can you create a Zenhub issue for that and add me? – Ok.

For the pyCRTM, Edward Hyar sent some slides on pyCRTM and he is optimizing the package by Bryan Karpowitz. External dependencies should be removed for a “student test”. Issues are e.g. that pyCRTM doesn’t run on 32bit Linux.

Yingtao: Will pyCRTM be part of a future CRTM release?

Ben: -Yes, we have the permission from the owners such as Bryan Karpowitz.

Nick, do you want to give an Update on SSE?

Nick: About the past two weeks?

Ben: Just a general update and if you need testing.

Nick: We are at the testing stage with Jim in the GSI. Jim is doing double-difference calculations for different channels to show the temperature dependence. There were some initial issues. Ming Cheng has provided code for emissivity that is not a LUT, but parametric. Now things are looking like we expect them to, but there are still some things that don’t quite make sense.

I have also worked with Sergio at UMBC to include it in kCarta. I will continue the work with Jim and then I will begin writing manuscripts, one on development and one on impact.

Ben: The SSE temperature Jacobian also needs to be tested.

Nick: Yes, Jim is looking into that, but I have only seen one plot for an atmospheric column.

Ben: Ming, do you want to give an update on CSEM?

Ming: Yes, before I go to the CSEM, I wanted to know because you list the L-band impact on SSE, do you plan to implement CSEM?

Ben: Hamideh is not on the CRTM team.

Ming: I have already seen that Hamideh made some changes to the CRTM code. Can you ask here to discuss future development impacting the CSEM and CRTM? For a new model implementation step-by-step tests are important.

Ben: If you have some slides on your work can you forward that?

Ming: Yes, I can provide a comprehensive evaluation. Future surface development should run through the CSEM.

Ben: Yes, but without the CSEM code it is difficult to be responsive to such developments.

Ming: The CSEM is an external library that should already be established as a repo on the STAR gitlab. We haven’t decided yet on the specifics of the repository maintenance. We need to add the CSEM interface in CRTM_dev though, so we need to discuss the next steps there.

Ben: A longer discussion is necessary here.

Yingtao: NLTE for shortwave IR was already done for CriS at STAR half a year ago with updated temperature profiles from Spain. I am working with Sergio from UMD. I have presented the results at AMS and they look good. But the results are only available for CriS, not IASI. All hyperspectral instruments in the CRTM should be added.

Ben: Are there any new profiles available?

Yingtao: No, the profiles are from Spain and they are already used at ECMWF.

Ben: This is something that is important for us. Can you start a Zenhub issue?

Yingtao: Yes, I’am still learning.

Ben: Is there anything that has been missed in the agenda?

Result:

-       Yingtao: NLTE is available for CriS.

-       Nick: SSE changes are being tested.

-       Ming: CSEM code is available on STAR Gitlab.

Tasks:

-       Daniel: Apply OpenACC to a CRTM subroutine and test it on a GPU.

-       Ming: Provide access to CSEM code.

-       Nick: Provide documentation on the SSE tests.

-       Yingtao: Create Zenhub Epic on NLTE.

Responsible People:

CRTM In-kind contributors

Deadline:

AOP 2020