2011-10-05 13:01:36New paper with new take on satellite data vs UAH.RSS
John Cook

john@skepticalscience...
130.102.158.12

Got this email re new research on UAH satellite data:

I want to point you to a new paper documenting the NESDIS/STAR 2.0 analysis of MSU/AMSU data.  STAR uses a novel method of intercalibrating satellite data, which is probably the biggest source of uncertainty with these analyses.  The new paper (and previous papers) can be found below.
http://www.star.nesdis.noaa.gov/smcd/emb/mscat/reference.htm
Website and data here:
http://www.star.nesdis.noaa.gov/smcd/emb/mscat/mscatmain.htm
I think it's worth a post to SkepticalScience explaining this analysis, and comparing it to UAH and RSS.  It has significantly higher trends than either of those and I don't think the results will be easy for skeptics to dismiss on technical grounds.  Also, unlike the UMD and UW analyses, STAR is continually updated and refined, so it's going to be with us for the long haul.  I've talked with Cheng-Zhi Zou and he says that they are working on a TLT product.  When that is released, I suspect that they will come under intense, probably not-very-polite, criticism.

I asked if they were interested in writing a guest post, they declined. So if anyone is keen to write about this, follow the links :-)

2011-10-05 13:43:51Zou et al
Glenn Tamblyn

glenn@thefoodgallery.com...
121.220.15.155

John

 

I covered this approach by Zou et al in this post some time ago . I might have a look at their newer paper and perhaps update the post.

2011-10-05 15:01:45
dana1981
Dana Nuccitelli
dana1981@yahoo...
69.230.101.55

Do it Glenn!  Frankly I think too many people believe UAH and RSS are the only game in town.

2011-10-05 15:22:17
Glenn Tamblyn

glenn@thefoodgallery.com...
124.181.72.163

John

This latest paper is quite detailed and technical and not a fantastic basis for a stand-alone post. It is doing even more fine-tooth comb work on biases in the satellite data. What it has produced is a small change in the trends they calculated in their previous papers (in C/decade):

 

Mid Troposphere     - New 0.141, Old 0.137

Upper Troposphere - New 0.048, Old 0.040

Lower Stratosphere - New -0.332, Old -0.329

The key point is that these tropospheric values are not compensated for biases in the signal due to cooling in the stratosphere - that is the point of them doing a TLT product.

In contrast RSS currently have a trend 1979-2011 for TMT of 0.087 that then produces a TLT of 0.142. So Zou without cooling bias compensation is just as high as RSS with cooling compensation. So expect Zou to give a TLT of 0.18 to 0.2. That is when it would be worth a new post.

If your correspondent has contact with Cheng-Zhi Zou then they might like to suggest that also doing a new product using something like the TLT type approach but for the upper troposphere might be a really good thing to do!

Missing Hotspot anyone?

2011-10-05 15:34:13
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

Glenn,

Pardon my ignorance, and sorry if the answer is easily availble and out there, but I ony saw a TMT product for Zou et al.  Why do they not calculate TLT?

2011-10-05 16:08:48
Glenn Tamblyn

glenn@thefoodgallery.com...
124.181.72.163

Albatross.

 

The TMT, TTS & TLS produced by the 3 teams are simply calculated from 3 different MSU channels on the satellites. Each is at a differing frequency that will tend to get its signal from a spread of different altitudes in the atmosphere. This is based on how solutions to the radiative Transfer Eqn for differing frequencies produce different signal spreads with altitude (the weighting function) - and if you need a more technical description than that you will need to a radiative transfer specialist rather than this humble engineer.

The problem with TMT & TTS is that a significant part of the signal is generated in the cooling lower stratosphere, imposing a cool bias on the other part of the signal arising from the troposphere. TTS for example is roughly 50/50 signal from upper troposphere and lower stratosphere. In contrast the signal spread for the TLS channel is around 95% stratosphere. So if TTS is reporting virtually no warming with its 50/50 split but TLS that reads the next level up in the lower stratosphere is giving a strongly negative trend, then the tropospheric part of the TTS signal must be fairly strongly positive.

TLT is a  synthetic channel. Instead of just using data from one reading of one channel (after jumping through a hundred hoops wrt calibration and biasing is concerned - this is where most of the processing goes), the TLT algorithm uses multiple readings of the same channel at different scan angles then applies an algorithm that takes differences between these readings. This has the effect of accentuating the signal from lower altitude and removing most of the stratospheric bias. So TLT has a weighting function that is far more purely tropospheric and in fact is more heavily weighted towards the lower troposphere. Downside of the method is that increases the signal error - taking differences of multiple readings does that. So while the other channels have a weighting function derived just from the Radiative Physics, TLT has a weighting function derived from the Radiative Physics and an algorithm.

Zou and co seem to be focussed on the bias and calibration issues so adding the TLT stage is an extra step they haven't taken yet. Also the rawer TMT data is still of use to climate modellers for example. Thier model will produce vertical temperature profiles/trends from which thay can use to calculate what reading TMT should be seeing and compare it to the actual data.

In contrast TLT is probably of more relevence as a 'headline' warming signal rather than as detailed data. Ind it is exactly this that makes it important in the broader AGW debate. And why I suggested another synthetic channel for the upper troposphere (TUT say), isolating out the layer of interest, might be really useful in the broader debate.

If a synthetic TUT were calculated, from the numbers given for TTS and TLS above, a TUT trend of 0.3 wouldn't surprise me. And with Zou implying a TLT of 0.18-0.2, bingo, there is your upper troposphere hotspot. I have never understood why the likes of Ben Santer haven't collaborated with the satellite temp guys to and get this done. It may not have high scientific utility but it sure as hell would have PR utility!

2011-10-06 04:25:14
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

Hi Glenn.

Many thanks for that well thought out and detailed explanation.  I was familiar with the stratospheric contamination issue, but thought that applied only to the TMT data, but until today I (embarrassingly) did not know that the TLT temperature data are based on synthetic channel data, I thought that there were drived using a different channel (channel 4); nor did I realise that contamination from the stratsophere was still an issue for the TLT data.  So they use channel 5 data to generate the synthetic data for TLT estimates...cool.

Good point about Santer working with Zou et al.  I am also curious why non of the model validation studies have incorporated the STAR data, well not as far as I am aware of at least.

2011-10-06 05:58:48comment
Robert Way

robert_way19@hotmail...
134.153.162.53

wow

2011-10-06 07:13:52
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

Hi Robert,

Wow ?  Is that a good "wow" or abad "wow"? ;)

2011-10-07 09:21:52comment
Robert Way

robert_way19@hotmail...
134.153.162.53

It's a wow at the change in TLT that we might be looking at haha

2011-10-07 09:46:53
Albatross
Julian Brimelow
stomatalaperture@gmail...
199.126.232.206

Tks Robert.  Yup, if what Glenn expects is true, that will really throw a cat amongst the deniers and skeptics ;)

2011-10-07 14:41:36Spectral separation
KR

k-ryan@comcast...
69.138.165.234

So - there are several channels (4? More?) available from the satellite data. how many atmospheric levels are we really interested in from those?

I'm asking because I work (from time to time) with spectral separation issues, and am curious - can anyone point me at appropriate references for the inverse functions used to more accurately separate the data on particular levels, removing crosstalk?

2011-10-16 17:24:45
Glenn Tamblyn

glenn@thefoodgallery.com...
124.177.56.131

KR

Sorry for the delay responding. I wrote a post on this general subject back in March that covers a lot of this ground. The earlier satellites had 4 channels that were used for Troposphere/Lower Stratosphere measurements. Later satellites had more channels with 3 of them roughly corresponding to the channels from the earlier satellites that were used for Climate work - these are basically weather satellites and the climatological use piggy backs off this.

A lot of this is about the weighting functions for each channel. RSS have a good graphic of the weighting functions here, about 1/2 way down the page.

Essentially there are 4 channels on the older sat's that have relevence. Chan 1 would be ideal for near surface values but too much of its signal originates from the surface rather than the atmosphere so in practice it isn't usable. Chan 2 (Chan 5 on the newer sats) is weighted so most of its signal originates roughly in the middle of the troposphere, but with some signal from above and below, including about 15-20% from the lower stratosphere - this is published as the TMT date series. Channel 3/7 is nominally upper troposphere but nearly 1/2 its signal originates from the stratosphere so it will report the sum of a warming and a cooling zone. Channel 4/9 is nominally lower stratosphere. Some of its WF stretches higher into the stratosphere but only about 5% originates in the troposphere.

There are other channels tuned to higher altitudes and these are used by the community that is interested in the Ionosphere etc. But they aren't involved in the mainstream satellite temperature record products. The newer sats also have channels 1-4 which I believe use signals from water vapour rather than oxygen as their basis but these aren't used for the more public products either.