A reality check of temperature for Wondering Willis Eschenbach

Update 2 - see below for another Reality Check with GISTemp

Update - see below for Reality Check 5

Addendum - see below for Reality Checks 3 and 4

---

Wondering Willis Eschenbach has an article at WUWT today (archived here, latest here). It's a lazy article. One of those silly articles claiming that the global surface temperature datasets aren't as reliable as the tropospheric temperature sets. Willis thanks the UAH duo for providing a reality check. He finished his article with this acknowledgement:

Finally, acknowledgement is due to the originators of the method of satellite temperature measurements, Drs. Roy Spencer and John Christie. It is thanks to them that we have a satellite-based atmospheric temperature record to act as a reality check for the oft-adjusted surface temperature record. Very well done, gentlemen.

This article is to provide Willis with not one but two reality checks.

Willis penned his article to feed Anthony Watts' readers some much needed doubt. The reason they need that doubt muchly at the moment is because this year is shaping up to be another hot one. Whether it'll turn out to be the hottest on record so far or not remains to be seen.

Reality check 1:

Willis wanted a reality check. Here's one for him. Willis puts up this chart from NOAA:

Source: NOAA

He wrote (notice his snide "...their idea of..."):

Here is the NOAA graphic in question, showing their idea of the current year to date in black, and the five warmest years in color.

Man, they are squeezing it to claim this September was the warmest, looks like a three-way tie to me … but I digress.

Willis' comment relating to September shows that he didn't read the NOAA chart properly. The point for September isn't just the month of September, the lines represent the year to date figure. 2014 is now neck and neck with the two hottest years on record: 2005 and 2010. As NOAA explains: "Each month along each trace represents the year-to-date average temperature. In other words, the January value is the January average temperature, the February value is the average of both January and February, and so on."

So here's the first reality check for Willis. According to NOAA, September 2014 was 0.04°C hotter than the previous warmest September, which was in 2005:

Data source: NOAA

The surface is not the lower troposphere

Willis then wrote how he "just laughs":

Now, I have read in a lot of places that we currently have good agreement between the satellite temperature data and the ground temperature data. Each time I read that, I just laugh. While the two measurements are closer than they have been in the past, there are still great differences. 

Willis seems to think that up in the lower troposphere, leaving out the poles, the average global temperature should every month be the same as it is on the surface.  What he neglects to show is how close the surface temperature records are to UAH on an annual basis.

Reality check 2:

Willis praised the scientists at UAH for providing a reality check. Here's another reality check.

The chart below shows UAH, GISTemp and HadCRUT4 from 1979 to the present. I've aligned them all to the UAH baseline - 1981-2010. The lower troposphere record has some higher highs and some lower lows. However they are all very close to each other. Click the chart to enlarge it.

Data sources: NASA GISS, Met Office Hadley Centre, UAH

The UAH record shows a slightly lower trend, mainly because the first three years were all higher than GISTemp and HadCRUT. Still, there isn't much between them. If you look over the 32 years since 1982 the trend is even closer, as the chart below demonstrates.

Data sources: NASA GISS, Met Office Hadley Centre, UAH

Before you get stuck into me for not being purist scientific, I'm not suggesting that anyone discard years of data just to give a "better" answer. UAH is higher in some years and lower in others. I just put up that chart to show that three years at the beginning are largely responsible for the difference in the trends.

You can also get stuck into me, if you want to, because I added the average to date for 2014. After all, UAH and GISTemp have nine months data and HadCRUT has eight. The final number for 2014 is probably not going to be too different to the average year to date.

Anyway, look at the difference. From 1982 to the present you have to go to four decimal places to see a difference in the trend/year between any of them. If you take the entire period, from 1979 to the present, the difference in the trend (highest minus lowest) is less than 0.0019°C/year or 0.019°C/decade or 0.19°C/century if the trends stayed the same for another seventy years or so. Which isn't likely.

Wondering Willis just looked at the month of September this year and wrote:

What is the reason for this huge difference in the surface and tropospheric records? I think it is a result of two things—the endless upwards adjustment of the surface data, along with the always-growing urban heat island effect.

Now look at the charts of annual averages and tell me, where is this "huge difference"? If there were endless adjustments of the surface data, given the trend of the three sets is almost indistinguishable, does that mean that the satellite record has also had "endless upward adjustments"?

Willis fluffs about and writes:

But whatever the reason, it is clear that the satellite record tells a very different answer than the one given by the practitioners of the dark art of post-hoc historical temperature adjustment. Given my choice, I’d say that the satellite record is the better of the two

Yes, he would say that. It's not that there is much difference on an annual basis. If you look again at the charts above, the annual average for UAH for 2012, 2013 and this year to date is almost identical to that of GISTemp and HadCRUT.

If you are wondering why there are differences between temperatures in the lower troposphere and temperatures on the surface, there was a publication back in 2006 that addresses the subject. It was published by the US Global Change Research Program. Chapter one can be downloaded here.

You'd expect there to be differences on a day to day basis between the lower troposphere and the surface. Probably even on a month to month basis. The atmosphere has its own dynamics. The air isn't the same temperature as the surface going all the way up, or going sideways.

Not only that, but the observations and how they are calculated are quite different. The lower troposphere temperatures come from Microwave Sounding Units on polar orbiting satellites. These readings then have to be adjusted to account for satellite drift and probably other things. Then calculations are made to convert the readings into a temperature scale. UAH (and RSS) satellite records don't provide temperature at a point or particular height in the troposphere, it's an average over a thick layer. Also, the satellites aren't too good over the poles, particularly over Antarctica, where the altitude and snow mess with the readings. UAH cuts out 15 degrees of latitude (7.5 top and bottom). RSS is more conservative, only going between 82.5°N and 70°S.

The surface data is from thermometers on land and on the sea. Again, the poles aren't all that well represented. Antarctica is huge and largely uninhabited. The Arctic isn't that much better. What is known is that the Arctic and parts of Antarctica are heating up faster than probably anywhere on Earth. UAH and RSS probably don't pick up all that. The surface data probably doesn't pick up all that either.

Anyway. I didn't intend to get into too much detail. I just wanted to give Wondering Willis Eschenbach a couple of reality checks.

Addendum: Reality Checks 3 & 4

I forgot about this chart I did earlier. It's another reality check for Wondering Willis. The animated chart shows UAH annual average and the UAH September only average going back in time. What's interesting about this (the reality check of sorts) is that the trend for September at 0.187°C/decade at  is quite a bit larger than the trend for the year as a whole (0.138°C/decade).

Data source: UAH

For the sake of completeness, here is NOAA just for September with a linear trend of 0.163°C/decade, somewhat lower than UAH for that month. Another reality check for Willis :)

Data source: NOAA

---

Update: Reality Check 5

Willis has updated a part of his article I didn't comment on. (H/t Brandon R. Gates in the comments.) I will now.  In the comments Willis Eschenbach wrote:

October 25, 2014 at 10:03 am
My thanks to all of those sharp commenters who pointed out that I was showing the monthly anomalies, not the YTD averages as NOAA had shown. It is one of the joys of writing for the web that my mistakes don’t last long without being spotted, which typically saves me weeks of work chasing blind alleys.

However, doing it the way NOAA did it just makes the difference between the two datasets even larger, viz:

He got it wrong. Willis posted this chart in the comments and the main article:

Source: WUWT

Willis is off his game. He might get a lot wrong but he doesn't usually get things like this wrong. He wrote the caption:

Figure 2. My UPDATED graphic showing the UAH MSU T2LT lower temperature data. Sadly, lack of sleep took its toll, and I showed the individual monthly values in my previous graphic, rather than the year-to-date average. My error has no effect on the conclusions of the post. Note that the MSU anomalies have been re-baselined to match the NOAA anomalies. Data Source. 

I'd say that he is still suffering lack of sleep. Below is what he should have posted. It is comparable to the chart in Reality Check 1 above. "Each month along each trace represents the year-to-date average temperature. In other words, the January value is the January average temperature, the February value is the average of both January and February, and so on". Click to enlarge, as always.

Data source: UAH

In the lower troposphere, 1998 will still be the hottest year in the UAH record, followed by 2010. This year is just edging ahead of 2005 on a year to date basis. 2005 is currently the third hottest year in the UAH record. 2014 would need larger anomalies for the rest of the year (October to December) to keep pace with 2005.

Note that I didn't "rebaseline" UAH. I prefer to use a baseline where I can calculate an average using the same set of data, rather than just deduct some arbitrary amount from all the numbers. UAH doesn't have data for the NOAA baseline period. In any case, we are comparing years within a particular set of data so the absolute values between temperature data sets aren't relevant. It's just the order of years that we are considering. (And the lower troposphere isn't the surface anyway.) If you want to compare the satellite and surface records, I've done so in this article higher up.

[Sou: 26 October 2014]

---

Update 2: Reality Check 6

There was a comment in another thread asking why NASA thinks that this year is looking to be another hot one. So I've prepared another year-to-date chart, similar to the others, using GISTemp. Click to enlarge.

Data source: NASA GISS

As you can see from the thin red line, 2014 is already pretty well level with the second hottest year in the GISTemp record. It wouldn't take much for it to reach or exceed the record of 2010.

Sou 29 October 2010

From the WUWT comments

TYoke  might not be aware that it was the UAH team that made a blunder at one stage, and had to go back and correct it.

October 24, 2014 at 10:49 pm
Too right about the value of the Satellite record. Without it, it seems very likely indeed that the surface guys would have ‘adjusted’ the record to fit alarmist predictions.

norah4you - I haven't got a clue what she/he is trying to say. Anyone?

October 24, 2014 at 10:51 pm
Had someone among scientist bothered taking a good look at last year’s vulcano eruptions in water or close to water in region Berings Straight via Aleutian Islands over to Kodiak Islands, they would have a better understanding of why winds, streams and much more are effected by underwater vulcanos CO2-produce. Also why it’s impossible to use reradiation temperature instead of meassuring temperatures 1 resp 3 meter over actual waterlevel AND also under not only in a few places but at least each 100 square kilometer. The reradiation figure is in itself a fake reading. So many factors are involved in which reradiation can be “found” that this is more How to lie with statistics than not.

Bill 2 points out that Willis misread the NOAA chart.

October 25, 2014 at 12:02 am
NOAA graph is showing year to date average, not individual months. 

I expect in 2090, if Earth is four degrees hotter, there will still be tiny clusters of Sceptical lefty's saying stuff like:

October 25, 2014 at 1:13 am
Suppose last month was indeed “the hottest September evah!” — at least, since reliable records began. When one considers the age of the planet and the length of the period for which no accurate record exists … what useful conclusions may be derived from this hottest of Septembers?

In any case, has the anthropogenic signal been plausibly distinguished from the natural one? If so, can it be definitively asserted that, in the absence of the anthropogenic signal, the Earth would not be warming? Is there any point pursuing this exercise any further?

Neville asks if Wondering Willis is accusing NASA of committing fraud.

October 25, 2014 at 1:59 am
So we can take it from Willis’s quote above that he thinks that GISS etc are indeed publishing/promoting fraudulent temp data?

David A says that 1998 is the hottest year in the UAH record. And it is. But look at the charts above again. They are anomalies from the same 30 year baseline. The trend is remarkably similar, despite some greater year to year extremes in the satellite data.

October 25, 2014 at 3:00 am
Nick, the overwhelming fact is that satellites show that every month of the year was cooler than 1998, with the 1998 anomalies double the 2014 anomalies.