Agenda Item 1:

Introduction

Discussion:

Introduction by Benjamin Johnson:

Ben: I will show you some preliminary cloud coefficients and we can discuss the AOP planning and any deliverables that still need to happen.

Result:

Tasks:

N/A

Responsible People:

Deadline:

Agenda Item 2:

Ben’s Talk

Discussion:

Ben: [shares slides] I created a new MW cloud coefficient file. What you can see highlighted are the things that I changed in the CloudCoeff code. I expanded the frequencies and effective radii. I used Mie theory here to get the normalization on the Legendre coefficients right. This box shows the default CRTM frequencies. In the expanded files I changed the stepping and added effective radii above 2mm. The reason we want to do that is to capture some precipitation better. For the expanded frequency file, I also added frequencies above 2 GHz to capture ICI frequencies. The GEMS frequencies are missing.

This plot shows the mass extinction coefficient. There are a couple of things to note:

-       The mass extinction for low frequencies is lower. In the higher region is consistency. I only plotted this for 190 GHz and if we look at the pattern the extinction peaks at a certain point. We didn’t see this too much in the original CloudCoeffs because we didn’t go to such high frequencies.

-       There are some differences but the behavior is consistent overall.

-       For liquid water there are similar issues.

-       Solid SSA looks really consistent.

-       Liquid SSA and solid asym. parameter also looks good.

My goal here was just to create a placeholder for the table. For the phase function coefficients things are getting weird. You can see a discrepancy in the P_11 coefficients, which is probably a normalization issue. Whenever I normalized the coefficients I created more than the number of streams, so this might be the cause.

In this graph you can see CRTM brightness temperature as a function of channels.

For ATMS you can see a difference between default CloudCoeffs and mine by about 5 Kelvin. What doesn’t look good is when you go out to the maximum effective radius there is a big difference.

Mark: One question, which color is for the default CRTM?

Ben: Default is solid and dashed is new.

Yingtao: That goes so low to the 180K? Your table there is probably a normalization problem.

Mark: For the dashed red curve there is probably a problem at Ch. 16.

Ben: This is about 87 K there.

Ben: This is for ice, the SNOW type. I am more concerned about the lower channels. The higher channels are oxygen channels.

The last slide shows the brightness temperature difference. This should be the solid line here. You can see a BT difference of about 60K at the lower ATMS channel. The next thing probably is fixing the phase function normalization issue and including non-spherical particles.

Mark: The reference in the paper states that it already includes the Guozhen Liu database with non-spherical particles.

Yingtao: In that you also added the polarization?

Mark: Yes.

Ben: Where is the file?

Mark: It was in the file that I sent.

Yingtao: In the table you sent with version 3 for the non-spherical part is from Guozhen Liu?

Mark: Guozhen Liu sent me his polarized table.

Yingtao: You said you sent the Mie coefficients?

Mark: For the liquid particles it’s Mie but for nonspherical particles it’s Guozhen Liu’s table. I contacted him and asked if he has a polarized table.

Yingtao: I remember you said there’s a placeholder?

Mark: That’s in the 2019 version.

Ben: The name of the file is CloudCoeff_pol.bin

Mark: I shared the code for the alpha version in 2019 and last year.

Ben: We have a lot of work to do here. The next step we have to do here is get the normalization right and look at the beta version.

Mark: What you did is also very good. The Europeans will have channels up to 600 GHz for the ICI channels.

Ben: I know that some of the guys on the European side have already created their coefficients for ICI.

Mark: Even for the DA in the operational centers they’re using RTTOV.

Ben: Their lookup tables are sensor-specific and kind of a mess.

Yingtao: I have a question, now that we have multiple lookup tables from Ben and Mark. What will be the LUT for version 3.0?

Ben: We haven’t made a decision yet. We need to make sure the polarization is correct.

Mark: We can discuss which table to use when version 3.0 is ready. In the past we didn’t consider certain frequencies. You can never extract the table from someone and not change.

Ben: We are also considering specific microphysics with Penn State. I haven’t included it in 2.4 yet because I thought it wasn’t ready yet.

Mark: JCSDA and STAR probably need to have what we call “Golden Data”.

Ben: Right. This will be a substantial part of what I am proposing for AOP2021. Now we are relying on O-B, but we also want to have field campaign results.

Mark: We worked with Ben Hao now and he found that in the past we didn’t have reference data above 30km. Even good radiosondes stop at 25km. The COSMIC-2 data has a much better height. They can provide a profile up to 60km. When I compare with the CRTM computation and those profiles than this structure changes. The model always needs to be developed.

Ben: You have non-LTE as well.

Mark: Not for MW. For IR you have non-LTE but not for MW.

Ben: I read a paper on non-LTE effects on SSMIS.

Mark: I had a paper on the Zeeman split and Yong Chen had a paper on non-LTE.

Yingtao: When you said you did a calculation with COSMIC did you use the CRTM?

Mark: What we found is when we use CRTM for ch. 15 of ATMS we have a high peak for the weighting function. The standard deviation is small but there is a bias sheet.

Ben: Is it a problem that the area COSMIC covers is very broad?

Mark: That is probably not a problem since we are looking for statistical mean values. We had a high vertical resolution but low spatial resolution.

Yingtao: Yong Han developed the regression coefficients.

Mark: At the time I mainly developed the ADA solver and Yong Han and Yong Chen were in charge of the transmittance and Paul was overall in charge of the code. People were working on different parts but were discussing together, so I know what Yong Chen did on the transmittance part. Yong Han’s data, I didn’t ask him about the Zeeman split but I think I can find the data he used. So far, we just guess a relation to the magnetic field. We added Zeeman splitting for ATMS but we cannot definitely say. Yong Han used actual measurement data. I don’t know which field data he used.

Ben: The observation data might be better now because they have these electron density observations.

Mark: There’s no data.

Yingtao: We still have access to his code and his data, right?

Mark: I don’t know, before he left he told me something but I never looked at his directory. Is Tong Zhu working for the CRTM team? Did he get a copy? Probably Yingtao you can ask Tong Zhu.

Kevin: I’m sorry Mark, what is this for?

Mark: We were talking about the Zeeman splitting. This work was done by Yong Han in 2003. By 2017 Tong Zhu was working on the CRTM. At least what I know Tong Zhu did ask Yong Han where the data for the Zeeman split is.

Kevin: I guess that would be worth to talk out with Tong. Previously it was really just developed for SSMIS.

Mark: In the team they are interested to save the code and the data.

Ben: It looks like RTTOV is applying Zeeman to other MW sensors as well.

Result:

Differences between a Mie reference cloud coefficient table and the CRTM default tables have been observed.

Tasks:

-       Investigate correct phase function normalization

Responsible People:

Benjamin Johnson

Deadline:

Agenda Item 3:

Short Updates from Ming

Discussion:

Ben: Looks like Ming has been trying to push some code to the JCSDA repository. Ming you are pushing the interface codes right now?

Ming: Yes, I will also share the source code.

Mark: Ming, is it just the interface?

Ming: [unintelligable] I was reading the plan with Paul to use OOP features. That requires a lot of programming. That requires modifying in a lot of places. The best idea is to replace the current top-down design with a bottom-up design. Just as advice. I can make the surface part very flexible. We can take advantage of the CSEM. The current CSEM is for scalar. When we have the vector code we need to modify this part. The other thing is user defined. In the CRTM development plan from Paul there’s a user-defined emissivity.

Mark: For the planning for version 3.0, your part will be ready by February?

Ming: Yes.

Ben: Good. The sooner we can get it the better. Set

The other part I wanted to discuss is Yingtao’s non-LTE work?

Yingtao: Yes, I did some work for CrIS.

Ben: Please coordinate that either with me or Patrick, Patrick is preferred.

Ben: Kevin, have you provided any feedback to the Joint Center on the AOP2021?

Kevin: On the NOAA side we are still working through everything. When do you want to provide a draft AOP?

Ben: The last week before the retreat in April, so we can circulate it among everybody.

Mark: This will be FY2021?

Ben: Yes, UCAR FY April 1^st to March 31^st each year.

In February everything should be almost ready. That needs to go to the MOB to sign.

Mark: By AOP 2021 we should also include the Zeeman splitting. Mark, I didn’t have a chance to discuss with Tom.

Result:

Ming Chen has begun to upload code to Github

Tasks:

-       Upload CSEM interface code to Github

Responsible People:

Ming Chen

Deadline:

February 2021