CRTM Aerosol LUTs compared to NASA LUTs

Andy: Well, I’m Andy Tangborn, I have been at NASA for 23 years, I have been moving to NOAA for IMSG and I have been working with Cory Martin on cycling experiments with JEDI and CRTM. This is a workflow that Cory has first developed. I have been learning how to use it and today I will talk about the forward modelling and the differences from the CRTM. All we want to look at is the forward modelling. It doesn’t really matter what kind of obs we have and I have been putting in MODIS Terra Obs. It really needs to be analyzed over a short period of time, less than a month. We are not in a position to say which one is more accurate. We have a good idea what the optical depth is, but also the individual species. There is a source that is AERONET, but also ATom for the species. And so, this really doesn’t tell us much about which one is more accurate, but rather what’s the reason why they’re so different. The first thing is the experiment I did some time ago. This is three weeks of Mean O-F. The NASA LUTs are generally closer to the Obs, but not all the time. The important thing is that they are different. But it’s not changing the forecast accuracy, but the changes to the actual fields …

Sarah: Are you using an older version? The CRTM LUTs are also coming from NASA.

Andy: I don’t know, Mariusz is knowing more.

Mariusz: This is the most recent table.

Sarah: Ok, so I guess we are talking about an older CRTM version.

Andrew: Ok, the important thing is that they are different. The first thing is to create an HofX difference scatter plot. It gives you an idea of the difference in AOD magnitude. You can see that in some places the CRTM is larger. In some values the NASA values are larger. This includes the influence of all 16 species, so It’s kind of hard to analyze, so I was just looking into one 16-hour window and a limited region and separated the species. So, for a model output I zeroed out all species but one. The location is the Indian ocean. It has land and ocean, and a pollution source over India. The top plot is just a scatter plot of the differences and you can see distinct lines which are probably different species. The second column is over the ocean and the third is over the land. This here has the biggest difference. Over the ocean the correlation between the two schemes is the most different. The individual species are here. Black carbon is a good fit. The difference between ocean and land is negligible. A very different one is the hydrophilic black carbon. The differences are not entirely random, but the difference is not getting with smaller amplitude. If you look at hydrophobic black carbon, it’s similar, probably a little bit better. Hydrophilic organic carbon shows no systematic difference and there is no trend. The differences are so small that everything’s random. The dust values show large systematic differences. There’s no difference over the ocean or over land. You can’t even see the random difference because there’s a systematic bias. For dust-2, the radius is a bit larger, but the situation is the same. The last one is the sulfate. It’s somewhere in between with a big random and systematic component, with the ocean being a bit more random. The only thing I don’t have are the sea salts. They don’t contribute much to the AOD. There’s some uncertainty in my mind on how the CRTM handles them. My understanding is that sea salt bin 1 and 2 are put into bin 1 in the CRTM, but checking that didn’t work out. This is my initial conclusion from this part of the study. There are some differences, the black carbons have a little bit larger systematic differences, but it’s not as large as dust. Again, we can’t make a statement which one is better. The thing that we need to do is to make comparison with in-situ data. That’s what I have, questions?

Ben: Thanks, Andy. For the validation we can’t tell which one is correct. Do you think the AERONET data would be beneficial here?

Andy: Yes, I think Mariusz has already integrated AERONET into JEDI.

Ming Chen - NOAA Affiliate

3:31 PM

Ben, I need to leave for about 20min. I need to pick up my son.

Jim Jung - NOAA Affiliate

3:32 PM

JPSS-2 current launch date is 30 September 2022

Andy: I should also ask if this is the analysis form that might be useful for you?

Mariusz: I have an opinion I can express. Rather than comparing the output with the species, the previous step would be to look at the tables themselves. Just by looking at the differences for the hydrophilic species, there are some significant differences on how those models approach aerosol growth and possibly refractive indices.

Cheng: Hi Andrew, I have already done the comparison of the SSP and refractive index between those two models. We can have a separate meeting on that. And I also agree with Mariusz on first looking into the SSP.

Andy: Ok, Mariusz would probably also be interested in that.

Sarah: I must also say this table is probably GOES-4. If we verify the tables we need to look into making sure that we go in the right direction.

Mariusz: The NASA process of tuning the aerosol model has been going on for some time. I don’t know how objective the tuning process is, but in my mind the current NASA LUTs correspond better to observations.

Sarah: That’s a good point. A lot of tuning is done at NASA. [unintelligible].

Cheng: In CRTM v2.4.1 those NASA tables will also be available and you can call either table.

Ben: We just want to verify that when we say NASA table, we are referring to the same tables. We also need to make sure that we have proper versioning.

Mariusz: NASA tables come with tags. It’s not difficult to keep track of their changes.

Ben: Right, there will be a NASA version and a CRTM version.

Sarah: Ben, when you talk about version, is that for aerosol tables?

Cheng: I just want to say that for the NASA tables I combined all tables in one common item.

Quanhua: I think the NASA and JCSDA tables are formally the same, just the format is different. But at the same time, we can call default and NASA table. Once people accept it, we can use it. So, we can make it simple.

Ben: Any other questions for Andy? Thanks, very nice comparison.

Quanhua: In the past we do not use aerosols [for data assimilation]. There are some people doing aerosol assimilation but it’s based on retrieval products. At STAR we complete a relatively simple UV radiance assimilation, just to check whether the model can do it. I think NCAR expressed interest into UV assimilation.

Ben: We are quickly getting to the point where UV radiance assimilation is becoming of operational importance, especially for regional air quality. Other comments?

Igor: I have one comment. Are there any plans to compare the product with CALIPSO data?

Ben: Yes, Isaac Moradi was working on a Forward operator for space-based radar. Once that’s done we are moving forward with that. So that’s definitely on our radar. Whether that is going to be easy to do, that’s another issue.

Igor: CALIPSO is not trivial. They specifically developed that to make a better retrieval.

Mariusz: CALIPSO is also not available in real-time, just 2 or 3 days. If there is something for real-time, that would be useful.

Ben: Does anybody know what the status of the CALIPSO laser is?

Igor: Still operational.

Ben: Patrick, do you want to give an update on the coefficient generation roundtable?