The Blackboard

A tad off topic, but Anthony mentions in his most recent post that:
.
“The divergence between the satellite derived global temperature anomalies of UAH and RSS and the GISS land-ocean anomaly is the largest in recent memory.”
.
While this may or may not be true for UAH depending on how one defines recent memory, the current diff between RSS and GISS (both normalized to a 1979-1998 baseline) this month of 0.319 is smaller than that of March 2008, which was 0.351. To find a greater divergence between UAH and GISS, however, you need to go back to the 1998 ENSO. And the divergences during that El Nino were still quite a bit larger than anything seen of late.

That said, there does seem to be some relationship between ENSO events and the divergence between surface and satellite measurements, and we do have a growing El Nino on our hands right now.

Here is a graph of UAH – GISS and RSS – GISS over the past three decades:

http://i81.photobucket.com/alb.....ure6-2.png

Zeke–

depending on how one defines recent memory

I chuckled at this one because one of the recent blog kerfuffles seems related to Stefan writing the uber-vague

“Some aspects of climate change are progressing faster than was expected a few years ago”

Which was either true or false depending on the precise meaning of “a few years ago” and on who expected things to be progressing slower those few years ago.

It looks like part of the divergence this month may also be due to warmth in Antarctica, which is outside the satellite window. We’ll see how that plays out. I’m not too worried about month -to-month divergences, but I’m a big curious about differences in longer term trends.

Zeke – however the 1998 El Nino had RSS and UAH much higher than surface temperature records, not significantly lower as now. I’m not sure I ever posted here my notes on the tropical comparison question, but the way, that we discussed here many months ago – I ran some further analysis that strongly suggests UAH (and to a lesser extent RSS) have a long-term downward drift at least in the tropical channel that has not been accounted for. At least a start on details of that (not that it’s a particularly original conclusion) in my Hot Spot Redux: analysis of tropical tropospheric amplification

Lucia,
.
I think our country went through enough semantical scarring back when we were discussing the meaning of “is” to focus too much on nit-picks :-p
.
As far as the current divergence, it is odd in-as-much as El Nino tends to manifest itself more in the satellite record than in the surface record. Though looking back at the 97-98 ENSO, at the start of the event satellite records were quite a bit lower than surface records, but this changed dramatically by the end.

http://i81.photobucket.com/alb.....ure8-4.png

[Edit: oops, got the colors reversed on the legend ^. Fixed now]

Zeke–
If the lag is real, maybe the reason it happens is related to the progress of the El Nino event. During El Nino, the surface of the central Pacific warms. This is somehow related to cold water no longer rising and/or warm water subsiding as the winds die down and the water “sloshes” a big like water in a bath tub.

Anyway, initially, the warmth really, really is right at the surface. The average surface temperature rises and is detected.

Then, within over a few months, the air warms, rises, mixes etc. So, it take a bit longer for the temperature near channel 5 to warm.

Of course, maybe there is no lag. In which case, my ‘explanation’ is hooey.

Tetris–
I think the GISS method is unusual, in so far as it’s constantly recomputing local anomalies in a rather involved way. But we don’t actually know that they are intentionally introducing a positive bias or even that the fact that corrections tend to increase the trend might be a symptom of confirmation bias.

Since I don’t have time (or inclination) to try to compute my own monthly GMST results, all I can do is a) report what we find using that metric and b) point out that GISS does revise their metric and notice whether or not the corrections tend to affect computed trends.

If we threw out every GMST measurement someone doesn’t like, we’d be left with NO measurements.

“If we threw out every GMST measurement someone doesn’t like, we’d be left with NO measurements.”

That’s the whole point. If we can’t/won’t agree on how to measure things, we have nothing. AGW’s goal is to destroy information and make political divisions.

Andrew

Andrew/Lucia/Tetris,

Its also worth noting that, post-facto adjustments notwithstanding, GISS still tracks HadCRU pretty well (which is to be expected given the overlap in stations used for both).

http://i81.photobucket.com/alb.....re10-3.png

I think you’re missing a word around slightly.

Considering how historical data seems to get modified all the time, a simple two-dimensional plot of temperature vs. time does not give the whole story. How about a 3D surface plot of temperatures, as a function of both putative measurement date and analysis (data set) date?

Zeke [16285]
Problem is, as SteveM has shown at CA, HadleyCRU is not above a good number of question marks either and is currently the subject of several FOIs.

Zeke–
Yes. Hadley, NOAA and GISS aren’t too far off based on anomalies.

I like trends because some of their features are less affected by choice of start data. But the noise in measurements does affect them more.

I’ll be showing things baselined with anomalies when the Hadley data arrives (which should be soon.)

I’m also looking at something puzzling….. (Puzzling because I thought a particular tests was not sensitive, and I didn’t expect to get a rejection yet. I’m checking for mistakes.)

Tetris: UAH and RSS aren’t flawless either. All temperature records have their issues, but it is somewhat comforting that they all produce generally comparable results.

http://i81.photobucket.com/alb.....ture11.png

Zeke [16295]
Sure. By definition no data source is flawless. But at UHA you have folks like Christie and Spencer [and their counterparts at RSS] who will go to great lengths to explain what they are doing in terms of data adjustments and re-calibration, why and how, and publish it so you and I can figure out whether they’re hoodwinking us or not. That is certainly not the case at either GISS or Hadley, who simply refuse to share raw data and algorithms, thus the need to go through the whole FOI process. As I have argued many times before, while Mann, Steig, Amman, Jones, etc., etc., etc. appear to think that this is totally normal, refusing to make one’s raw data and calculations available to others so they can verify how you got your results is quite simply not science. It’s pseudoscience on par with homeopathy, astrology and fish gut reading.
Last but not least, whatever flaws there may be with the satellite data, that data is by its very nature more reliable because it wasn’t collected by sensors placed on blacktop parking lots or in front of air conditioner exhausts or airport runways [ref: Watt's surfacestations project].
In my line of work, GISS would not be accepted as a credible data source/reference.

Tetris,

Oddly enough, a few years back it was various “team” members (Sherwood, Santer, and Trenberth among others) who uncovered considerably issues in the impact of orbital drift on satellite measurements. Temps on the satellite side have been considerably more contentious than temps on the surface over the past decade, at least in the scientific literature.

Zeke,

The satellites are contention in the literature because they don’t show enough warming to keep the alarmists happy. If the ‘drift’ problem resulted in warmer trends you could safely bet that no one would have cared and the anyone who questioned the trends would be told that ‘drift does not matter’.

The bottom line is all of the data available is contaminated by confirmation bias. However, in the case of satellites, we have two different groups using the same raw data but with different biases. This competition, IMO, keeps both groups honest and that makes the satellite data the least biased of the available data.

Zeke,
As I said, there may very well be issues with the satellite data, but at least these are discussed and debated with hard facts at hand. Not so with GISS and HAD data, which we know beyond any doubt have been and continue to be “adjusted” without the outside world being provided an opportunity to verify what is going on, and we continue to be asked to accept this data at face value as been credible. An “adjusted” June GISS anomaly that is in excess of one order of magnitude out compared to consistent RSS and UHA data? What am I not getting this twisted “trust me” story?

It’s very much like the mechanism put in play at RC with the recent Kyle Swanson piece. Yes, the argument goes, there has now been an undeniable 9-10 “lull” in the warming [remarkable admission at RC, be said in passing] caused by some yet to be fully understood momentous climatic event, possibly the very 1998 super el nino used until recently by the same RC team to explain away the ongoing absence of warming! But says the RC sanctioned Swanson piece, the underlying man-made CO2 driven warming is still there and what is more, the GCMs show that the warming will pick up again after 2020. What I find mind boggling arrogant in all of this is that we are asked to trust the very same GCMs that spectacularly failed to predict the present “lull”. The same circular “trust me” argument as with the GISS data.
Everything else being equal, I’ll trust the satellite data over the GISS and HAD hodgepodge any time of the day.

Zeke (comment #16275) – I’m not seeing the surface temperature data in your plot, just the ENSO index. My understanding was that global surface and satellite numbers actually tracked one another very well through the 1998 peak, with satellite just being a multiple of surface – as your figure in #16295 seems to show also. I.e. I believe both surface and satellite temps (on the global scale) took a while to respond to the 1998 ENSO.

Arthur,

The red line is (UAH + RSS) / 2 – (GISS + HadCRU) / 2 all normalized to a 1979-1998 baseline. When the ENSO index started rising rapidly surface temperatures rose faster than satellite temps (e.g. the red line was negative). Satellite temps only shot up much higher than surface temps right as the ENSO event was ending. That said, during the whole ENSO event the satellite and surface temps did largely move in the same direction.

Here are the temperatures (as opposed to the difference between the temp series):

http://i81.photobucket.com/alb.....ture14.png

Zeke-the “comfort” that comes from”similar” behavior fades when you realize that the satellites and surface should NOT give similar results. Just look at the variability-the bulk lower atmosphere should, on a global basis, have trends about 1.2 that observed at the surface.

Some recommended reading on this topic since people seem to be very confused about it:
http://icecap.us/images/upload.....onse_2.pdf
http://www.climatesci.org/publ...../R-345.pdf
http://www.agu.org/pubs/crossr.....8864.shtml

Incidentally, that satellites have uncorrected cool drift in the tropics, as Arthur proposes, is decidedly implausible. I recommend he review the extensive literature on the issue of satellites before making such pronouncements-and rather than rely on crap like the “stratospheric influence” BS or RC type author brigades (Santers and other biased “scientists”), he should really read more of Christy et al.’s papers since they have by far the most knowledge of satellite issues.

Andrew_FL,

As far as I’m aware, the TLT band is supposed to track close to the surface, while the middle troposphere band is supposed to be warmer. That said, I’m not by any means an expert in tropospheric warming, and I know that satellites have a number of issues measuring temperatures in different parts of the troposphere due to other parts of the troposphere and lower stratosphere bleeding in.

Radiosonde measurements, for example, show faster TMT warming than surface warming: http://www.ncdc.noaa.gov/img/c.....dec-pg.gif

I looked at the Channel5 AMSU. In the past two weeks the troposphere warmed dramatically. So, this isn’t either/or. There may be a time delay between warming at the ocean surface and warming in the lower troposphere.

Zeke-I suggest you bone up on it-Kotzblatch et al. above is a good start. As for ’sondes, that issue is a BIG mess. but-which data are you looking at? There are a bunch of different data sets, some better than others. Ar you looking at HadAt, RATPAC, ROABCORE, what?

BTW, UAH long ago inplement a method to account for the “bleeding” issue.

Spencer, R.W., and J.R. Christy, 1990. Precise monitoring of global temperature trends from satellites. Science, 247, 1558-1562.

Made use of satellites different viewing angles to get around stratospheric influences. The issue, in fact, was dealt with at the very beginning of satellite research!

http://tcsdaily.com/article.aspx?id=120304F

A little more:
http://tcsdaily.com/article.aspx?id=050504H

The changes shown in my graph do not look anywhere near as dramatic as in the blinking comparator graphs Anthony posted. The reasons?

Typical WUWT. The changes to global Gistemp Nylo noted are slight, as this post says, and would be too small to show in a blink plot. So WUWT posts the much-remarked cumulative effect of adjustments to US temperatures over ten years. And, sure enough, a chorus of indignation over this “new” atrocity.

Speaking of WUWT. Watt’s up with the post on airports? I don’t get it. Is it half finished? Are airports the new air conditioners and barbecue grills? Please advise.

Boris–
The problem with thermometer being concentrated at airports is they are more likely to be affected by UHI. If, hypothetically, every measurement station in South America was at an airport, I suspect no amount of data manipulation could diagnose this. So, yes, airports are, metaphorically speaking, barbecue grills.

Zeke (#16313) – right, that was pretty much my recollection of the temperature curves, sorry I misinterpreted your difference plot. I would say from that plot that it looks like surface temperatures took about 6 months to show significant (out of normal variation) warming after the 1998 ENSO began, and the satellite record perhaps 8 months. Interesting that in the present case there doesn’t seem to be that time-lag – but we’re very early into it so who knows.

Andrew_FL – my suggestion of a long-term negative bias in satellite tropical temperatures is based solely on some statistical analysis of the UAH and RSS data in comparison with surface records, and doesn’t depend on the source of that bias. It could be calibration problems between satellites, it could be stratospheric contamination, it could be something completely different. Given the history of corrections to the satellite record so far, I think your confidence is slightly misplaced.

In particular, over a 10-year time-span, tropical amplification is clear and large (a factor of 1.8 or more) in those records, but it disappears over a 20-30-year timespan. There are also significant issues in the standard deviation patterns of temperature differences that need some further investigation. What I find implausible is a physical mechanism that could account for such a change in the character of Earth’s response processes between 1 decade and 2 to 3.

Nick Stokes or any other science boffin

If need be, please excuse my naivete…I confess to a rather high degree of scientific illiteracy! In regards to Gisstemp, Anthony Watts seems to have demonstrated that rather strong positive forcing agents exist at many of the US temperature monitoring sites (bbq’s, airports, AC systems etc.) used by GISS to calculate temperature anomolies. Nick mentions the “much-remarked cumulative effect of adjustments to US temperatures over ten years”, as opposed to the apparent slight changes to global Gisstemp. Why is this so? I am presuming that there would be a relatively high degree of monitoring stations outside the US that would be effected by similar forcings particularly in countries that haven’t the sufficient hardware or have not developed robust monitoring systems.
best wishes, ian

Ian, there’s at least one reason applying mainly to US. About half the nett adjustment there is the time of observation (TOBS) adjustment. The time at which temperatures are recorded can cause deviations for various reasons (eg intro of daylight saving), but in the US, it mainly relates to the max min thermometers used. After daily reading, these are reset, and record the min and max for the next 24hr.

Now if you read early in the morning, you get a cool bias. The reason is that on a cold morning, you’ll record a cold min for the previous day, reset to the current (cold) temp, which may well be the min for the next 24 hr. The same cold morning gets recorded twice. If you read the temp in the afternoon, you’d risk recording a warm day twice, so that’s a warm bias.

The original NWS advice was that max-min thermo’s should be read in the late afternoon. But generally, operators became required to read the rain gauges in the morning, so over the years they shifted the time. When you allow for this, you get an upward trending TOBS adjustment.

You can read about the US adjustments here.

Comparison with the surface record? You gotta be kidding me!

Your lack of confidence borders on denial of reality.

The reason amplification vanishes over long time scales is almost certainly for the reasons elucidated by Kotzblach et al. Maybe you should read up on my references before you conclude that you have any clue what you are talking about.

You might also want to read:
http://www.climatesci.org/publ...../R-321.pdf

By now the evidence for the accuracy of the satellites is piling on-and the evidence against the surface data is unbelievably so. You are either behind the curve or in denial…

Ta Nick…perhaps i should think about taking meteorology 101 if I ever grace the halls of a tertiary establishment again.

Anthony’s graph is about USA. GISS made an important update in all of USA’s data on August 7th 2007, as reported in their updates to analysis page. Of course, the change was towards warming. But that change hardly affected the Global temperatures.

However a similarly big methodology change happened in April 2006 which did affect global temperatures a lot. This change is described in their updates to analysis web page as having a “negligible” effect on global temperatures. However it was by far the biggest change that I am aware of sinde December 2005 and raised the Global Temperature Trend by a 10%. In the following graph (see link below) you can compare GISTEMP data as it existed in February and June 2006, together with the corresponding trends for the period 1880-2005. Notice how they rise the trends by strongly cooling the past (and some warming of the present). This was achieved by removing data from gridcells near Anctartica which didn’t show the “right” ammount of warming, and replacing them with nothing at all.

http://www.elsideron.com/Trend.....change.PNG

I forgot, Lucia, in this link you can find several versions of the GISTEMP GLB.Ts+dSST data as they were published in different dates. The available dates start in December 2005.

http://web.archive.org/web/*/http://data.giss.nasa.gov/gistemp/tabledata/GLB.Ts+dSST.txt

Regards.

Andrew_FL – Klotzbach’s paper (not yet published) seems to be looking at global trends, not the tropical ones I was referring to – the discrepancy they are highlighting is primarily the warming trend over land, which is considerably greater in the surface temperature series (and somewhat greater in RSS than UAH – in fact RSS and Hadley almost entirely agree on the land trend). Their argument says almost nothing about possible errors in the satellite data, and instead focuses on a large number of Watts-style complaints about the surface data (which is the same issue the Pielke paper you most recently pointed to is on about too). I would think the recent NOAA report on surface station sampling should put those claims to rest.

And of course they also downplay RSS data against UAH for some reason. So of all the temperature series out there, only UAH is reliable according to those folks. Right – I think Occam’s razor might come to a different conclusion.

Note how, rather than addressing the points of the paper, Arthur simply dismisses it. And he ignores the evidence supporting UAH and claims to apply “Occam’s Razor”-really he is making the argument from numbers-Whatever, Arthur.

So far I have not seen the correct answer for why GISS constantly changes past temperatures. The correct answer is laid out in Hansen et al (2001) found at http://pubs.giss.nasa.gov/docs.....n_etal.pdf
The methodology used by GISS is to infill missing data by refiguring the trend as new data becomes availible. If you look at page 18 of the paper it shows that this methodology lowers past temperatures while raising temperatures current temperatures.

Another problem is that the changes used for USA climate are not used for the Rest of the World, but the temperature trends are presented as if they were.

Finally, while this may not be a problem, when I see that the total adjustments match the long term trends and that the raw data shows no such trend.

Arthur Smith,

The idea that the tropical surface temperature record is some sort of gold standard to which satellite data must conform is ludicrous. The coverage is sparse and QA/QC is questionable at best. The simplest explanation of an increasing difference between satellite and surface data in the tropics is that the surface record is biased warm rather than a cool bias in the satellite data.