Richard Betts, late of this diocese, posted a tweet the other day referring to the SkS member, Kevin Cowtan, who has produced yet another paper trying to disprove the pause. The approach he took this time was to look at the uncertainties in the satellite and the terrestial temperature measurement and, surprise, surprise, he came to the conclusion that the satellite data had more uncertainties than the terrestrial data. Because of his overt activism I have every reason to doubt Cowtan's scholarship but accepted the general premise that satellites measurements had more uncertainty than terrestial measurements on face value. I also don't have a problem with the homogenisation methodology and assume that the scientists are doing their best to get to an accurate assessment of the global temperature anomalies.

My problem is this. As far as I know there are about 4800 weather stations in the HadCrut4 data set, I'm assuming 3000 argo buoys and a miscellaneous number of data extracted from different forms of measuring sea surface temperatures, and nothing above, or below 80N and 80S respectively. In terms of uncertainty I've assumed that in real terms measuring the temperature of a bucket of seawater is little short of useless, so I'm assuming that the HadCrut4 measurements are obtained from 4800 weather stations and 3000 argo buoys. Each of which represents a pin-prick of space on the Earth's surface. Or put it another way we have 1 measuring station for every 65,400km^2 of the Earth's surface area.

I tweeted back to Richard and asked was this taken into consideration when calculating the likely uncertainties in the HadCrut4 data set, but, not for the first time, I don't believe he understood what I was getting at.(If there's fault it's me not him)

What are the chances of getting a robust view of the Earth's average global temperature analogy out of a network of 7800 measuring stations over a land and sea mass of 510 million kilometres^2 and how to the Met Office and CRU deal with the uncertainty of a signal extracted from 7800 measuring devices over 510 million Km^2? Anyone know?

Thanks Mike, I have problems with much that the clisci community puts out by way of numbers, but when they start taking seriously a paper from the University of SkS purporting to show that there are more uncertainty in satellite measurements of global temperature than there are in a measurement system with around 8000 measurement devices mostly in the USA and Europe purporting to be able to measure the global temperature with less uncertainty it piques my curiosity.

Is it my ignorance? or have they somehow forgotten that the temperature measuring instruments are imbalanced spatially, and that there are few of them? I think I know what Cowtan purports to have proved, and that is that individual temperature measurement stations have less uncertainty than satellite measurements, which may be true, but there is a bigger picture, one which puts the terrestrial measurement stations in the context of their numbers and distribution. In that context the resultant temperature measurements cannot have less uncertainties than satellite temperature measurements. Surely?

Anyway I've asked maybe someone will enlighten me.

To be fair, there are big issues with satellites, such as orbital decay, stitching records from different sources (where have we heard that before?), permanent and intermittent power failures, instrumental drift (who checks the sensors once they are in orbit?), etc. They are not unimpeachable sources of fact. See climate audit for Steve Mc 's reservations.

http://climateaudit.org/2016/01/05/update-of-model-observation-comparisons/#comment-765688

diogenes: I accept there are big issues with satellite records, I think I said that above, but it's quite another story to imply that the thermometer/Argo records have fewer uncertainties given the geography and coverage of these measurement stations and argo buoys (1/200,000q miles).

Thanks for the link though it may educate me.

Does anyone really care what the temperature is over most of the ocean? Surely the land temperatures where people live are the most relevant and maybe where glaciers lurk if you have been worried about the last few thousand years general increase in sea level.

Rob Burton, I think that depends on what point is being discussed. In the context of weather and the science behind global-warming climate claims, it does matter. Or ought to. Just as the temperature measured by satellites up in the troposphere matters for the same reasons.

But to the man/woman on the street, no it probably doesn't matter. Which is why the global-warmers will always be struggling to get people interested, never mind continuously frightened, about such things. And this also why they desperately need an on going supply of claimed disasters and "records": People can see the real weather and climate every day when they walk out of the front door, and they are not alarmed. And most of them are still not alarmed when the BBC tells them to be alarmed.

michael hart

People can see the real weather and climate every day when they walk out of the front door, and they are not alarmed. And most of them are still not alarmed when the BBC tells them to be alarmed.

In fact most people over about 18 think what's the fuss I seen much worse, those under 18 think if this is as bad as it gets what's the fuss?

Phil Clarke posted some interesting links on the Adjustocene cartoon thread and I for one am now better educated than I was regarding the nature, extent and reasoning behind the land-based temperature adjustments than I was.

Having read the articles he linked to, I was left the impression that the people concerned are almost certainly sincere in their attempts to produce meaningful temperature runs with some claim to accuracy, but I was also left wondering if the whole exercise isn't ultimately futile. For all the effort put into making the adjustments sensible, accurate and realistic, I still couldn't help feeling there was a "finger in the air" element to it all. Adjusting for changing the time of day when measurements were recorded (in the past, mostly in the afternoon, today, mostly in the morning) sees today's temperatures adjusted up by about 0.5C to compensate, while adjusting for the UHI sees temperatures go down by c 0.2C. The 0.5C I can accept, though the figure seems a little arbitrary; the 0.2C for UHI strikes me as inadequate, but who knows? And that's a large part of the problem. Nobody really knows, for all their fancy algorithms. Much the same can be said for their (smaller) adjustments the other way for sea-based temperatures. Are they sincere? In my view, almost certainly. Have they got it right? In my view, almost certainly not. What would they have to do to get it right? I haven't a clue, but I doubt if anyone else truthfully has, either.

Which leads into your point about satellites. I accept without reservation that there are problems with temperatures suggested by satellites too. They are also adjusted, and no doubt the adjustments are made with equal sincerity. As a layman, it seems to me that the need for the adjustments to the satellite temperatures is less, as they have been around for a much shorter time, and the issues requiring adjustments aren't so dramatic as things like relocating sites, and changing the time of the day when the measurement is recorded. So, on balance, I trust the satellite records more than the land and sea-based ones, but if I'm honest, that's more a gut feeling than anything else.

I think the point you make is a very important one, however. Satellites offer much wider coverage of the planet, whatever their limitations. Land-based measurements are highly biased towards the USA, so yet more adjustments are required to compensate for that. Land-based measurements are (outside of the USA and parts of Eurpose) sparse, so that introduces the concept of in-fill. Ocean-based measurements strike me as even more sparse, for such a huge area. And, as you point out, there's nothing if anything beyond 80 north and south. Which ultimately makes me wonder, what's the point? At least until we are able to find a much more comprehensive and accurate and reliable way of measuring the temperature world-wide. Isn't that where satellites were supposed to come in?

Jan 31, 2016 at 9:28 PM | Unregistered Commentermichael hart

Michael, my main point was that in order to get a "global" temperature using thermometers then we have to guess a huge area of the Earth that doesnt have them and largely we really do not care about. At least the satellites should be pretty consistent at getting the temperatures truely globally. But if you were concerned about climate change in Northern Ireland say I would just look at the very long record in Armagh and say that basically covers climate change for the last few hundred years and leave it at that.

Mark, for an idea of how difficult it is to get a satellite record of atmospheric temperatures, see http://www.drroyspencer.com/2015/04/version-6-0-of-the-uah-temperature-dataset-released-new-lt-trend-0-11-cdecade/

If sufficiently interested, a numerate person can download the thermometer data from around the world and, armed with no more than Excel, compute her own temperature index. It is not too difficult - at least one skeptical blogger has done this. The result will be a graph showing rising temperatures. If she wanted to do the same for satellite data... she probably couldn't. It is too difficult.

It would also probably be too difficult for her to produce the planet-wide grid of surface temperatures by the opaque krigging procedures from the thermometers. No one, or few people, pretends the process is trivial. (Least of all the 'guardians' of the surface data: Otherwise they might get asked if what that are doing really needs all that money. Roy Spencer, however, produces the satellite data on a shoestring budget.)

Mike Jackson

I think I did keep my eye on the pea - my ultimate conclusion was: "Which ultimately makes me wonder, what's the point?"

Raff

Thanks for the link - very interesting; I'm adding to my knowledge all the time. Having read about all the adjustments required for satellites, land-based measurements, and ocean-based measurements, however, I think my conclusion remains unchanged: "Which ultimately makes me wonder, what's the point?"

michael hart, yes, the better an answer you want the more work you have to do. But you can get a long way with the basic temperature data and Excel, which is not true of satellite data.

Mark, in a way you are right. We know CO2 will cause rising temperatures and that we need to reduce emissions. So measuring what is happening is just an academic exercise. Is that what you meant?

Raff

Not really, no! I accept that we have GW (not really a surprise, as we emerge from the Little Ice Age). I am sceptical about AGW, and disbelieve entirely in CAGW. Thus I don't accept that we need to reduce CO2 emissions.

However, even if I did, then I never cease to be amazed at the illogical and counter-productive policy prescriptions of the "green" activists. I always put "green" in inverted comments, by the way, as I see nothing green about the rush to blight our beautiful landscape with wind turbines and solar panels. I'm in Athelstan's camp on the other thread, where he said he can put up with the ugliness of nuclear power stations, reluctantly, because at least they achieve something positive, or words to that effect. Living in Cumbria, I think Windscale/Sellafield is a hideous eyesore, but it's as nought to the ring of steel destroying the beauty of this wonderful county.

However, I had better stop there, as I'm going O/T and am in danger of derailing this thread, which is not my intention.

Mike Jackson has it right. What we are looking at is a set of temperature measuring devices over the globe, say around 4000 on land and 3000 at sea (ignoring SST made with buckets). The 4000 are heavily skewed to the USA and Europe.

If over the last 120 years, they had all been measured in the same way, at the same time of day, by the same people then there may, just may, be a signal in the time series that is telling us how the Earth's temperature is performing. If we aren't using the same weather station measured in the same way at the same time of day and have to make adjustments to bring them together, it is my view you could not produce a time-series with any meaningful signal in it and to pretend to be able to do so is disingenuous.

I'll leave to you each to take on board that the Met Office and it's fellow temperature series supporters claim that the system they're using can measure temperatures past and present to 100ths of one degree Celsius.

I wonder if anyone else feels that allowing the people who forecast the temperatures to check whether the forecasts are right or not is a quality control issue?

Feb 2, 2016 at 6:00 PM | geronimo
Yes, testing of software, in this case models, should be done by people not involved in creating the software and who are paid to break it. Dr Jones of UEA had it right of about breaking things, and he should have had faith in the basic "rightness" of his data the fact he didn't speaks volumes but doesn't mean he knew it was wrong.

Geronimo

I'll leave to you each to take on board that the Met Office and it's fellow temperature series supporters claim that the system they're using can measure temperatures past and present to 100ths of one degree Celsius.

I discussed this earlier with Radical Rodent and Micky H Corbett.

The annual average is based, if your figures are correct, on 7000*365=2,555,000 measurements.

Where d is the precision of the mean and n is the sample size

d~1/√n

A thermometer can be read to the nearest 1C. n=1, d=1C

The mean of 100 measurements (n=100) can be expressed to d=0.1C

For n=10000 d=0.01C.

For n=1 million d=0.001C.

For n=2555000 d=0.0006C

The Met Office monthly and annual figures are expressed to 0.01C, which is statistically quite conservative.

EM, they are not all measuring the same the "same" temperature in the same place at the same time. Averaging them is not statistically meaningful or appropriate in the way you think they are.

If I rapidly took a million digital measurements of a cow's fundament, and then one from the South Pole, and averaged them obtain a number I expressed to several decimal places, what would that tell me?

Nothing.

Try reading Essex, McKitrick Andresen, again: "Does a Global Temperature Exist?"