GISS vs CRU/RSS/UAH

[nzucker]

There is not "supposed to be" and there definitely does not need to be high resolution to measure global temperature. You can form an accurate index of temperature by randomly selecting a mere 70 stations and then extrapolating the 1000s of miles between them. The result doesn't change if you use 70, 500, 5,000 or 50,000 stations. +.8C/century.

The large discrepancies where GISS runs 2-5C cooler or warmer than UAH are due to the extrapolations. However, mathematically these balance out. The slight extra warming on GISS is due to the surface stations warming more than UAH.

Maybe we should be "checking" the satellites considering they're not very accurate and other satellite data sources say GISS is too cold. I guess GISS extrapolations are causing a cold bias!!! OH NOES!

Higher resolution means better measurements...the less you have to extrapolate means that more areas are being measured exactly for their temperature anomaly. A global temperature index with 10,000 stations is going to be much more accurate than one comprised of 10 stations, bottom line. The more data you have to guess, the more your margin of error increases. GISS is more likely to give an erroneous picture of global temperatures than the satellites because more land area is being guessed rather than measured. One place being warm doesn't necessarily mean another place 100 miles away is the same warmth, but a thermometer there sure does...

And you keep missing the point...we're not talking about global temperature change on long-term scales of 100 years or 30 years. We're talking about the last decade where there is controversy over how much we have warmed due to the fact that GISS is showing a higher decadal trend than other sources. We all agree that over extremely long periods of time, some of the biases of extrapolation will surely be neutralized. But the problem is, we're trying to determine whether the Earth has continued warming at the expected rate in the last 10-15 years, that's the argument here, and the divergence between GISS and UAH/RSS makes it impossible to do this analysis objectively. In the short term, extrapolation doesn't even out at all. Furthermore, we are talking about specific monthly maps...you still haven't answered my question about what would happen if I were doing a research project on Greenland climate during Winer 2010-11...how would I know whether Greenland was above or below average looking at RSS and GISS maps from March? With all the technology we have today, shouldn't I be able to be more precise than saying Greenland was between 3C above average and 3C below average for the month?

Really sad to see such conspiracy theory posts on a science forum. The methodology and data hasn't changed since GISS was first developed. Hansen didn't "get desperate" and make it warmer magically.

You just keep pointing out the monthly maps and ignoring the fact that there are large areas that are extrapolated too cold as well. GISS runs warmer than UAH not because of the extrapolations, but because the surface data has actually warmed more. Which is why HadCRUT 60S-60N has also warmed much more than UAH or RSS. You have yet to address either of these points and instead you just respond with your typical off-point personal attacks. Very very sad.

So much for this "promise." It is very disappointing that you can't abstain from attacking the person and stick to the substance. I am not disregarding your viewpoint. I have given several clear reasons why it is blatantly wrong which you have not responded to.

First of all, you need to lighten up. My comments about Hansen are often tongue-in-cheek. Do I really think he's manipulating the data because the Earth hasn't been warming as much? No, probably not...of course with the corruption in our government, you can never rule that possibility out, but I wouldn't actually swear to that. You take this climate change stuff so seriously that you can't even joke around about it anymore, have some fun bantering with a friend, etc. Live a little, for Christ's sake! And I don't think it's a personal attack to say that you are frequently obstinate in your arguments; I'm just pointing out that it's a waste of time to continue this line of argument with you because you keep resorting to "Extrapolation doesn't matter over the 30-year full period" and "Extrapolations should mathematically even out according to the Law of Large Numbers" even when we're not talking about either of these things. You bring back the same old arguments when we want to know: why is GISS different in the last 10 years? And, why do certain areas show such a large deviance in monthly maps between RSS and GISS? Someone's gotta be right, someone's gotta be wrong. Following models and 500mb patterns, I think I can ascertain that GISS is probably incorrect. If you want to counter this, you should start posting detailed station data, other sources of temperatures for disputed areas, etc.

The problem is, you're not adding anything new to the argument, Andrew. When I showed that there was nearly 6C divergence in some areas of Greenland between GISS and RSS, you should have been the first to realize how much that is and start investigating it. If you want to defend GISS and the accuracy of its extrapolations, then go find the data, go through the temperatures for each station, and post it here. Also, you keep saying the code hasn't changed....so, you should link to that code, journal articles about the accuracy of GISS extrapolation methods, etc. You don't do any of this...you just resort to the same trite old arguments and ignore the questions ElTacoman and I have posed. Like, why do the GISS and satellite maps diverge in places with more extrapolation, and in places where it is harder to disprove inaccurately warm data? This is suspicious, Andrew, no way around it. You have to start challenging yourself to be a real skeptic again as you were on the radio show. You've drifted back towards the mainstream so much that you see everything they do through rose-colored glasses. Show you are a scientist and start investigating, start asking the questions! Have the courage to challenge those in positions of authority!

Then why do the maps almost always match up well for the U.S. and other areas with abundant data/stations, but the places they don't tend to be sparse data regions like Africa, the Arctic/Greenland, and parts of Asia? 2+2 = 4, but you are insisting this isn't the case. If what you were claiming were true, the AMSU and GISS maps wouldn't match up well over most of the globe...but they do.

And it is not mathematically impossible to have a warm bias with extrapolation. In a perfect world, extrapolation shouldn't lead to a warm/cold bias, but because global temperatures are not evenly distributed, it is very possible. If localized stations develop greater warming for whatever reason, that can be extrapolated over a huge area, when it may just be a localized effect. There could easily be an imbalance between stations that have warming and those that have cooler temps.

Exactly! Andrew needs to show that the particular stations GISS uses don't have a bias relative to other stations, and he hasn't done this.

And yes, the suspicious part is of course where the data sets diverge....

[skierinvermont]

Here is HadCRUT 60-60 vs UAH 60-60, once again demonstrating that when we use the minimal amount of extrapolation possible for the surface data, it still shows significantly more warming than the satellite data. HadCRUT 60-60 is warming .020C/decade faster than UAH 60-60. This unequivocally shows that the greater warming at the surface is due to the actual data warming more, not extrapolation.

If you think that extrapolation causes a warm bias, then one must necessarily predict that when one removes the extrapolation, the warm divergence will disappear. We can test that easily. Remove the extrapolation and the divergence remains. This is an unambiguous test of your nonsensical hypothesis. And it unambiguously shows that the actual surface data is warming more. Of course if we use STAR, RAOBCORE, RICH, Fu, or V&G we find the troposphere warming as much as or more than the surface. RSS and UAH are the only ones that don't.

Of course this is completely unsurprising to anybody that understands the Law of Large numbers and its application to repetitive extrapolation. It's possible to extrapolate too warm once, even 50 times, but when you are extrapolating 500+ times, 12 months a year, for 30+ years, the law of large numbers makes it statistically impossible that you will consistently extrapolate too warm.

[skierinvermont]

Oh and zucker.. I'm not going to respond yet again to your same tired old arguments and personal insults. You need to have the "courage" to respond to the points I have made instead of ignoring them. You need to try and think critically and rationally instead of ignoring everything that contradicts what you want to believe. You are and always have been incredibly obstinate close-minded and arrogant. ETC. ETC. ETC.

[nzucker]

Here is HadCRUT 60-60 vs UAH 60-60, once again demonstrating that when we use the minimal amount of extrapolation possible for the surface data, it still shows significantly more warming than the satellite data. HadCRUT 60-60 is warming .020C/decade faster than UAH 60-60. This unequivocally shows that the greater warming at the surface is due to the actual data warming more, not extrapolation.

Of course this is completely unsurprising to anybody that understands the Law of Large numbers and its application to repetitive extrapolation. It's possible to extrapolate too warm once, even 50 times, but when you are extrapolating 500+ times, 12 months a year, for 30+ years, the law of large numbers makes it statistically impossible that you will consistently extrapolate too warm.

No one is talking about 30+ years. Do you not fricking understand that we are talking about short-term deviation? How stupid can you be?

And no one is saying it's just extrapolation. It's probably true that GISS is too warm because its stations are too warm, especially in high-latitude areas like Greenland. Whether it's because they're on the coast, in towns, etc I don't know. But that's what you should be exploring if you want to defend GISS.

Oh and zucker.. I'm not going to respond yet again to your same tired old arguments and personal insults. You need to have the "courage" to respond to the points I have made instead of ignoring them. You need to try and think critically and rationally instead of ignoring everything that contradicts what you want to believe. You are and always have been incredibly obstinate close-minded and arrogant. ETC. ETC. ETC.

I am thinking rationally. Tell me why GISS deviates so much in certain areas! We're waiting...Since GISS is making measurements and extrapolations that deviate from observation of the H5 pattern and from satellite analysis, the burden of proof lies on you to prove these are accurate.

[BethesdaWX]

Here is HadCRUT 60-60 vs UAH 60-60, once again demonstrating that when we use the minimal amount of extrapolation possible for the surface data, it still shows significantly more warming than the satellite data. HadCRUT 60-60 is warming .020C/decade faster than UAH 60-60. This unequivocally shows that the greater warming at the surface is due to the actual data warming more, not extrapolation.

A few things. By all means, please read #2 and make the according Maps, either that or teach me how to make the maps

1) Yes you are correct, you can get a reasonably accurate temperature measurement By Extrapolating thousands of miles...this is true.......but this doesn't change the fact that the Higher Resolution modules (HADCRUT is the best in this case) will be MORE accurate than the Low Resolution Modules (GISS) where they measure/cover.

2) The GISS deviation began this decade (In the late 1990's), so that is when we'll start the analysis. You don't need to include unrelated timeframes before the late 1990's when GISS began to deviate, because that is Unrelalated to the current GISS deviation! If GISS will eventually come back in line with other sources, then fine, but that doesn't change the fact that it has deviated since the +AMO took over the Arctic.

So you'll start from ENSO peak in 1998 to ENSO peak in 2010 Comparing HADCRUT to GISS, HADCRUT to UAH, then GISS to UAH... if you don't mind.

3) STAR is being held out of Mainstream Operation due to errors regarding infrared channels, which I've posted links on maybe 15 times in the past 6 months, go back and look if you wish This may not be fixable, but if can eventually be figured out, then STAR will eventually go Mainstream when the problems are worked out. If there were no problems with STAR, then it would be mainstream right now.

There is no reason to even mention STAR on this Forum unless it becomes Mainstream.

4) Yes, the Surface has been warming faster than the LT overall, and that is basic proof that Lessening LL Global Cloud Cover has Caused a significant portion of the warming since the beginning of the Satellite Era (before then is uncertain). More SW (visible) light reaching the Surface radiated from first contact, warming from the Surface Up, since LW radiation (Heat) would result from the transformation os Visible light onto the Earth's Surface, and would warm the System most Significantly right at the Surface, and less so the higher you go. A 2% Change in GCC = 1.2W/m^2 of increased heat overall, the IPCC sites 1.4W/m^2 for all Drivers including the release of GHGes Since 1790. A 4% change in GCC equates to 2.4W/m^2 of increased heat...so....yeah, this is a Major Problem in Climate Science.

Since GCC Changes overall are Long Term (evidenced by BE^10 Proxies (Isotopes found in Ice Cores created with Solar correlation to increased Visible light lagging GCR's), this makes it equally Important to find a way to Accurately Measure GCC long term...its key to Climate Science.

GCC Generally trends long term based on BE^10 proxies, that is proven fact, when comparing BE^10 Proxies to Ice Core Temperatures, it is clear GCC changes as a response to Solar Magnetic Activity are a Major Climate Driver.

[skierinvermont]

No one is talking about 30+ years. Do you not fricking understand that we are talking about short-term deviation? How stupid can you be?

If you are comparing monthly maps of GISS and AMSU, as you have been, then the 32 year divergence is what makes GISS overall warmer than UAH/RSS. How can you "be so stupid?"

As I have shown, and as the math dictates, the actual surface data is warming more even with minimal extrapolation (IE HadCRUT 60-60 vs UAH 60-60) and the 30+ year divergence is not due to extrapolation.

[nzucker]

If you are comparing monthly maps of GISS and AMSU, as you have been, then the 32 year divergence is what makes GISS overall warmer than UAH/RSS. How can you "be so stupid?"

As I have shown, and as the math dictates, the actual surface data is warming more even with minimal extrapolation (IE HadCRUT 60-60 vs UAH 60-60) and the 30+ year divergence is not due to extrapolation.

But we're talking about maps for this year, and the divergence in the last 10 years. Yes, historically the satellites have shown less warming than the surface, which may lend more credibility to the global cloud cover argument as Bethesda claims, but recently GISS has been warmer than all the sources, even land-based. No one is claiming that the surface wouldn't warm more without extrapolation....ElTacoman and I are the first to admit that extrapolation is NOT the only reason why GISS/Hadcrut are warmer. But recently, GISS shows considerably more warming than Hadley, and shows warm anomalies in places that the satellites detected to be cold. All I want to know is that accuracy is being preserved. Why is GISS so out of line sometimes with satellite measurements of the same exact places? Why does this always happen in data sparse regions where it's hard to find the answers?

I mean Andrew, if you said NYC was -3C this month, and I said, no Central Park was +2C, wouldn't that seem a little weird? In this day and age, we shouldn't have such massive discrepancies. And it's not just for the purposes of global temperature trends...when I study winter patterns, I want to know what 500mb patterns cause what anomaly distribution, and that's hard to do if you have 5C discrepancies showing up in Greenland or the Arctic.

[skierinvermont]

But we're talking about maps for this year, and the divergence in the last 10 years. Yes, historically the satellites have shown less warming than the surface, which may lend more credibility to the global cloud cover argument as Bethesda claims, but recently GISS has been warmer than all the sources, even land-based. No one is claiming that the surface wouldn't warm more without extrapolation....ElTacoman and I are the first to admit that extrapolation is NOT the only reason why GISS/Hadcrut are warmer. But recently, GISS shows considerably more warming than Hadley, and shows warm anomalies in places that the satellites detected to be cold. All I want to know is that accuracy is being preserved. Why is GISS so out of line sometimes with satellite measurements of the same exact places? Why does this always happen in data sparse regions where it's hard to find the answers?

I mean Andrew, if you said NYC was -3C this month, and I said, no Central Park was +2C, wouldn't that seem a little weird? In this day and age, we shouldn't have such massive discrepancies. And it's not just for the purposes of global temperature trends...when I study winter patterns, I want to know what 500mb patterns cause what anomaly distribution, and that's hard to do if you have 5C discrepancies showing up in Greenland or the Arctic.

The reason the GISS monthly maps look warmer than the UAH maps overall is that GISS has diverged over the last 30 years. This is true regardless of the extrapolations or if we use minimal extrapolation. Changing the extrapolation doesn't remove the divergence. Therefore, the extrapolations are not why GISS monthly maps look slightly warmer overall.

GISS is not intended to accurately represent 500mb patterns on a monthly basis. There are lots of other data sources available for this, and I believe you know that, and are just trying to obfuscate the issue.

[tacoman25]

No.. AMSU and GISS shouldn't match up across the rest of the globe. That is absurd. Because of the extrapolations GISS will be much cooler and much warmer in large areas. There is a tendency towards warmth, because the surface stations have warmed more than the satellites (which is an undeniable fact because HadCRUT 60-60 and GISS 60-60, which have much less extrapolation, have both warmed more than UAH or RSS). This unambiguously proves the reason for the greater warmth on GISS (or HadCRUT+UAH poles) is not the extrapolation. Even when you take away the extrapolation, and look only at HadCRUT 60-60, there is greater warming than on UAH or RSS.

It's pretty funny watching everybody go ape-**** over extrapolation when the actual reasons for the divergence between GISS and HadCRUT can be found if you take the time to examine their methodology in detail.

I don't care what "should" happen. I've looked through and compared a lot of GISS and AMSU maps, and there is a strong tendency for them to agree fairly closely in places that have dense ground data like the U.S. There is much less agreement in the Arctic, parts of Africa, and parts of Asia.

[skierinvermont]

I don't care what "should" happen. I've looked through and compared a lot of GISS and AMSU maps, and there is a strong tendency for them to agree fairly closely in places that have dense ground data like the U.S. There is much less agreement in the Arctic, parts of Africa, and parts of Asia.

This is sort of like "duh" ... no surprise there. I'm sure Hansen or any other scientist would tell you GISS is intentionally designed that way. In fact, I'd rather GISS use fewer stations and have these big funny looking extrapolations than use more stations which might be of lower quality.

I'm just waiting for somebody anybody to answer one question:

If the extrapolations are what cause the warm divergence, why, when I remove the majority of extrapolations and use HadCRUT 60N-60S, does the divergence with satellite data remain?

The obvious answer is that the extrapolation do NOT cause the divergence, since removing them does not also remove the divergence. The divergence is due to the actual surface thermometers warming more than the satellite data for the troposphere above (at least according UAH/RSS).The extrapolations have NO effect on the divergence between surface and satellite data from UAH.

[BethesdaWX]

This is sort of like "duh" ... no surprise there. I'm sure Hansen or any other scientist would tell you GISS is intentionally designed that way. In fact, I'd rather GISS use fewer stations and have these big funny looking extrapolations than use more stations which might be of lower quality.

I'm just waiting for somebody anybody to answer one question:

If the extrapolations are what cause the warm divergence, why, when I remove the majority of extrapolations and use HadCRUT 60N-60S, does the divergence with satellite data remain?

The obvious answer is that the extrapolation do NOT cause the divergence, since removing them does not also remove the divergence. The divergence is due to the actual surface thermometers warming more than the satellite data for the troposphere above (at least according UAH/RSS).The extrapolations have NO effect on the divergence between surface and satellite data from UAH.

GISS 60/60 does have a warmer trend than HADCRUT 60/60, but that isn't enough to explain the divergence. Since you're a pro-AGWer, you should hope that the extrapolations ARE the cause... and that the surface has not been warming faster than the LT. The LT has to warm faster if our understanding if the GHE is accurate, this is a requirement, (AKA: Warming of the Surface and Enhanced warming of the LT through source of re-emission).....

Unfortunately, the case here, the surface has been warming faster than the LT as you stated, which is more represenative of Decreased LLGCC, which very slight devations have large effects of several W/m^2. Its a huge factor that could make or break our thoughts on the causation of the warming seen.....which is why I'm dumbfounded that we have not tried with more effort to find a reliable method of GCC measurement.

Once I can get my freakin PHD... dammit I'll gather a group of scientists to build a reliable GCC measurement System if we don't have one by then.

[skierinvermont]

GISS 60/60 does have a warmer trend than HADCRUT 60/60, but that isn't enough to explain the divergence. Since you're a pro-AGWer, you should hope that the extrapolations ARE the cause... and that the surface has not been warming faster than the LT. The LT has to warm faster if our understanding if the GHE is accurate, this is a requirement, (AKA: Warming of the Surface and Enhanced warming of the LT through source of re-emission).....

Unfortunately, the case here, the surface has been warming faster than the LT as you stated, which is more represenative of Decreased LLGCC, which very slight devations have large effects of several W/m^2. Its a huge factor that could make or break our thoughts on the causation of the warming seen.....which is why I'm dumbfounded that we have not tried with more effort to find a reliable method of GCC measurement.

Once I can get my freakin PHD... dammit I'll gather a group of scientists to build a reliable GCC measurement System if we don't have one by then.

Or more likely, the satellite data is simply wrong, as argued by Fu, V&G and Zou. As well as the radiosonde data running warmer.

[nzucker]

Or more likely, the satellite data is simply wrong, as argued by Fu, V&G and Zou. As well as the radiosonde data running warmer.

LOL, the satellite data is way less likely to be wrong than GISS.

[skierinvermont]

LOL, the satellite data is way less likely to be wrong than GISS.

If you say so. Actual scientists familiar with satellite technology disagree.

Moreover, some of the satellite data must be wrong by definition since the differences between the various satellite sources are much larger than the differences between the surface and UAH. Given the large discrepancies between various satellite sources it is logically impossible for the satellite data not to be wrong.

[nzucker]

If you say so. Actual scientists familiar with satellite technology disagree.

Moreover, some of the satellite data must be wrong by definition since the differences between the various satellite sources are much larger than the differences between the surface and UAH. Given the large discrepancies between various satellite sources it is logically impossible for the satellite data not to be wrong.

Well, Roy Spencer has said that the margin for error is less than .05C/decade, so apparently he thinks it's reasonably accurate.

You can say the same about HadCRUT and GISS...one of them must be wrong because Hadley is .02C/decade cooler.

Also, the RSS maps have a better appeal to the eye than GISS, which looks pretty phony.

[skierinvermont]

Well, Roy Spencer has said that the margin for error is less than .05C/decade, so apparently he thinks it's reasonably accurate.

You can say the same about HadCRUT and GISS...one of them must be wrong because Hadley is .02C/decade cooler.

Also, the RSS maps have a better appeal to the eye than GISS, which looks pretty phony.

Not all error is quantifiable, and even +/-.05C/decade is higher than that for GISS or HadCRUT. Moreover, the error estimates for Zou are smaller than that for UAH, and Zou shows over 2X as much mid-tropospheric warming as UAH, so one or both of them must be wrong by definition. The radiosonde data also is closer to Zou for the MT. UAH is a complete outlier for the MT, and is still the lowest for the LT. UAH's mid-tropospheric trends really discredit it.. there is simply no way the mid-troposphere has warmed that little. RSS and the radiosondes show 50% more mid-tropospheric warming, and Zou shows 2X as much. If it can't even get remotely close for the MT, how can we expect it to be at all accurate for the LT?

Also the divergence between GISS and HadCRUT over the last 30 years is smaller than .02C/decade as you stated, more like .005 or .01C/decade. Which is much smaller than the divergence between surface and satellite or between UAH vs RSS vs V&G vs Fu vs Zou vs RICH vs RAOBCORE. And once we infill HadCRUT with UAH, there is no divergence whatsoever. We have very very strong agreement at the surface for medium term trends. The same cannot be said for the troposphere using satellite or radiosonde data where there is wide disagreement between sources.

"looks phony" is not a scientific argument. It is not intended to look spatially accurate month to month.

[nzucker]

Not all error is quantifiable, and even +/-.05C/decade is higher than that for GISS or HadCRUT. Moreover, the error estimates for Zou are smaller than that for UAH, and Zou shows over 2X as much mid-tropospheric warming as UAH, so one or both of them must be wrong by definition. The radiosonde data also is closer to Zou for the MT. UAH is a complete outlier for the MT, and is still the lowest for the LT. UAH's mid-tropospheric trends really discredit it.. there is simply no way the mid-troposphere has warmed that little. RSS and the radiosondes show 50% more mid-tropospheric warming, and Zou shows 2X as much. If it can't even get remotely close for the MT, how can we expect it to be at all accurate for the LT?

Also the divergence between GISS and HadCRUT over the last 30 years is smaller than .02C/decade as you stated, more like .005 or .01C/decade. Which is much smaller than the divergence between surface and satellite or between UAH vs RSS vs V&G vs Fu vs Zou vs RICH vs RAOBCORE. And once we infill HadCRUT with UAH, there is no divergence whatsoever. We have very very strong agreement at the surface for medium term trends. The same cannot be said for the troposphere using satellite or radiosonde data where there is wide disagreement between sources.

"looks phony" is not a scientific argument. It is not intended to look spatially accurate month to month.

I don't think radiosondes can be considered a source of global temperatures because they aren't even distributed globally...Most of them are in North America with very few in the upper-mid latitudes of Asia.

Also, UAH was right with the radiosondes until recently, so unless that's coincidence it must have some handle on things.

GISS may not be spatially accurate each month, but we shouldn't be seeing 5C discrepancy. Saying the extrapolations are wrong is another way of saying the surface stations are too warm, so it is in fact related to the original difference between LT temps and surface stations.

[nzucker]

And once we infill HadCRUT with UAH, there is no divergence whatsoever. We have very very strong agreement at the surface for medium term trends. The same cannot be said for the troposphere using satellite or radiosonde data where there is wide disagreement between sources.

"looks phony" is not a scientific argument. It is not intended to look spatially accurate month to month.

I believe UAH+HadCRUT mid-latitudes still yields a slightly cooler result than GISS when you look at the last 10 years.

I don't expect 100% spatial accuracy but I'd like to think that we would have agreement on which places were above/below average.

[skierinvermont]

I don't think radiosondes can be considered a source of global temperatures because they aren't even distributed globally...Most of them are in North America with very few in the upper-mid latitudes of Asia.

Also, UAH was right with the radiosondes until recently, so unless that's coincidence it must have some handle on things.

GISS may not be spatially accurate each month, but we shouldn't be seeing 5C discrepancy. Saying the extrapolations are wrong is another way of saying the surface stations are too warm, so it is in fact related to the original difference between LT temps and surface stations.

Nobody said radiosonde data represents global temperatures. But the fact that over the areas covered by BOTH, radiosonde data runs much warmer than UAH for the the mid-troposphere and to some extent the lower troposphere lends evidence that UAH is not very accurate.

Indeed I agree the divergence it is related to the difference between LT temps and the surface stations. NOT anything inherent to the process of extrapolation. Which is what I have been saying all along, while you have been going on and on about how extrapolations cause a bias. Glad to see you have come around finally.

[skierinvermont]

I believe UAH+HadCRUT mid-latitudes still yields a slightly cooler result than GISS when you look at the last 10 years.

I don't expect 100% spatial accuracy but I'd like to think that we would have agreement on which places were above/below average.

Only looking at the last 10 years... but if you want to talk about the last 10 years then UAH and RSS have been diverging wildly from each other over that short period. That's what happens when you look at short periods.

Over the 30 year period HadCRUT+UAH poles matches GISS perfectly, while UAH, RSS, Zou, Fu, V&G, RAOBCORE, RICH all show very different 30 year trends.

Over the 10 year period, HadCRUT+UAH poles matches GISS very closely with a very small difference, while UAH, RSS, Zou, Fu, V&G, RAOBCORE, RICH all show very drastically different trends.

So no matter what time period we look at, the agreement is much closer for the surface than for the troposphere.

[nzucker]

Nobody said radiosonde data represents global temperatures. But the fact that over the areas covered by BOTH, radiosonde data runs much warmer than UAH for the the mid-troposphere and to some extent the lower troposphere lends evidence that UAH is not very accurate.

Indeed I agree the divergence it is related to the difference between LT temps and the surface stations. NOT anything inherent to the process of extrapolation. Which is what I have been saying all along, while you have been going on and on about how extrapolations cause a bias. Glad to see you have come around finally.

I feel like it's sort of the same thing...we're saying the extrapolations create a bias because the stations are too warm to represent all of the area that is extrapolated. So if you're saying the extrapolations are wrong, you're saying the stations are running warm. I don't really see much difference here. Extrapolating a warm station to cooler areas would be a source of bias, and that's what the GISS maps show...a broadbrushing of warm station data without the nuance to discern areas that didn't come in as warm.

Also, UAH was paralleling radiosondes very closely until recently..I believe around 2000. So we'd need to find out why a divergence started.

But it's severe weather time, not global temperature time...maybe we finally get an EML in here Wednesday and rock a little bit?

[skierinvermont]

I feel like it's sort of the same thing...we're saying the extrapolations create a bias because the stations are too warm to represent all of the area that is extrapolated. So if you're saying the extrapolations are wrong, you're saying the stations are running warm. I don't really see much difference here. Extrapolating a warm station to cooler areas would be a source of bias, and that's what the GISS maps show...a broadbrushing of warm station data without the nuance to discern areas that didn't come in as warm.

Also, UAH was paralleling radiosondes very closely until recently..I believe around 2000. So we'd need to find out why a divergence started.

But it's severe weather time, not global temperature time...maybe we finally get an EML in here Wednesday and rock a little bit?

No they are too very different arguments. You have argued repeatedly that extrapolation inherently creates a bias. I have argued that it does not, and the reason GISS runs warm has nothing to do with extrapolation but rather the fact that the surface data itself is simply warming more. And my argument is definitely demonstrated by the fact that even when we remove nearly all the extrapolation, and look only at HadCRUT 60S-60N, the divergence with the 60S-60N UAH data remains intact. Removing the extrapolation does not remove the divergence.

There is a huge difference between saying the methodology of "extrapolation creates the bias" and the reality that the actual physical thermometers are warming more.

[skierinvermont]

bump

I believe UAH+HadCRUT mid-latitudes still yields a slightly cooler result than GISS when you look at the last 10 years.

I don't expect 100% spatial accuracy but I'd like to think that we would have agreement on which places were above/below average.

Only looking at the last 10 years... but if you want to talk about the last 10 years then UAH and RSS have been diverging wildly from each other over that short period. That's what happens when you look at short periods.

Over the 30 year period HadCRUT+UAH poles matches GISS perfectly, while UAH, RSS, Zou, Fu, V&G, RAOBCORE, RICH all show very different 30 year trends.

Over the 10 year period, HadCRUT+UAH poles matches GISS very closely with a very small difference, while UAH, RSS, Zou, Fu, V&G, RAOBCORE, RICH all show very drastically different trends.

So no matter what time period we look at, the agreement is much closer for the surface than for the troposphere.

[nzucker]

No they are too very different arguments. You have argued repeatedly that extrapolation inherently creates a bias. I have argued that it does not, and the reason GISS runs warm has nothing to do with extrapolation but rather the fact that the surface data itself is simply warming more. And my argument is definitely demonstrated by the fact that even when we remove nearly all the extrapolation, and look only at HadCRUT 60S-60N, the divergence with the 60S-60N UAH data remains intact. Removing the extrapolation does not remove the divergence.

What I'm saying is that extrapolation from a small number of stations seems to be missing cooler areas within a large area of above normal temperatures. So I guess you can say it two ways: the extrapolation is wrong (in missing the cooler areas due to low resolution) and the stations are warm (relative to the cooler areas that they are also responsible for measuring by extrapolation).

This is also being suggested as a source of short-term divergence, and why GISS has been warmer recently. I think we all admit that the long-term divergence is simply because the LT has warmed less as per satellite analysis; you can either explain this using global cloud cover theories, or by trying to prove UAH/RSS wrong.

[skierinvermont]

What I'm saying is that extrapolation from a small number of stations seems to be missing cooler areas within a large area of above normal temperatures. So I guess you can say it two ways: the extrapolation is wrong (in missing the cooler areas due to low resolution) and the stations are warm (relative to the cooler areas that they are also responsible for measuring by extrapolation).

This is also being suggested as a source of short-term divergence, and why GISS has been warmer recently. I think we all admit that the long-term divergence is simply because the LT has warmed less as per satellite analysis; you can either explain this using global cloud cover theories, or by trying to prove UAH/RSS wrong.

Well then you are just still wrong. You are still suggesting that somehow the process of extrapolation inherently creates a warm bias in the global temperatures.

If this were the case, then removing the extrapolations would remove the warm bias relative to the satellite data. It does not. The difference between surface and satellite data remains the same whether we use extrapolation or not. Extrapolation is 100% irrelevant to the divergence.

The stations do not "miss' the cool areas. As I have shown, GISS repeatedly extrapolates far too cool over vast areas when a cool bubble lands on a station. If I were to compare GISS to STAR I would say that GISS looks like it is extrapolating too cold. In reality, it has nothing to do with the extrapolation. It relates to the physical data itself.

[nzucker]

Well then you are just still wrong. You are still suggesting that somehow the process of extrapolation inherently creates a warm bias in the global temperatures.

If this were the case, then removing the extrapolations would remove the warm bias relative to the satellite data. It does not. The difference between surface and satellite data remains the same whether we use extrapolation or not. Extrapolation is 100% irrelevant to the divergence.

The stations do not "miss' the cool areas. As I have shown, GISS repeatedly extrapolates far too cool over vast areas when a cool bubble lands on a station. If I were to compare GISS to STAR I would say that GISS looks like it is extrapolating too cold. In reality, it has nothing to do with the extrapolation. It relates to the physical data itself.

Of course extrapolation should not create a bias in global temperatures. But when we are analyzing one year, or ten years, on one source (GISS)...it's possible that there's a bias. There are not enough trials to rely on the Law of Large Numbers.

Here is an example...note how much warmer GISS is over South America. Where is that bubble of very warm anomalies coming from?

[skierinvermont]

Well that is exactly what I have said all along. Previously you have said the law of large numbers is not applicable, but I am glad to see you now understand it is applicable and that extrapolation does not inherently create a bias in the long run.

The reason GISS runs warm is that the surface thermometers are warming faster than the air above them. Or the satellite data is just wrong. It has absolutely nothing to do with extrapolation.

I like how you focused in on the tiny bubble of warmth GISS has in S America but ignored the huge bubble of >+4C anomalies in Europe which GISS is completely lacking. RSS has all of Great Britain, France, Germany, Denmark, southern Sweden and Finland and northern Spain in 4-6C anomalies, while GISS has them in 2-4C anomalies.

RSS also has +4C anomalies in Mongolia and In Mexico and Texas which GISS is lacking. Or what about the 0C anomalies north of Europe where RSS has large +anomalies?

[tacoman25]

This is sort of like "duh" ... no surprise there. I'm sure Hansen or any other scientist would tell you GISS is intentionally designed that way. In fact, I'd rather GISS use fewer stations and have these big funny looking extrapolations than use more stations which might be of lower quality.

I'm just waiting for somebody anybody to answer one question:

If the extrapolations are what cause the warm divergence, why, when I remove the majority of extrapolations and use HadCRUT 60N-60S, does the divergence with satellite data remain?

The obvious answer is that the extrapolation do NOT cause the divergence, since removing them does not also remove the divergence. The divergence is due to the actual surface thermometers warming more than the satellite data for the troposphere above (at least according UAH/RSS).The extrapolations have NO effect on the divergence between surface and satellite data from UAH.

The divergence we have been talking about all along is the GISS divergence over the past 7-10 years, in which they have been running warmer than all other sources in both the Arctic and the rest of the globe. As you have admitted before, this pretty much has to be due to their extrapolations. I don't understand why you acknowledge this and yet always come back to the same, irrelevant points that are not even referring to the same divergence.

[tacoman25]

Well that is exactly what I have said all along. Previously you have said the law of large numbers is not applicable, but I am glad to see you now understand it is applicable and that extrapolation does not inherently create a bias in the long run.

The reason GISS runs warm is that the surface thermometers are warming faster than the air above them. Or the satellite data is just wrong. It has absolutely nothing to do with extrapolation.

Wrong. That doesn't explains why GISS has been running warmer than HadCRU 60/60 since the early/mid 2000s.

[tacoman25]

I like how you focused in on the tiny bubble of warmth GISS has in S America but ignored the huge bubble of >+4C anomalies in Europe which GISS is completely lacking. RSS has all of Great Britain, France, Germany, Denmark, southern Sweden and Finland and northern Spain in 4-6C anomalies, while GISS has them in 2-4C anomalies.

RSS also has +4C anomalies in Mongolia and In Mexico which GISS is lacking.

The anomalies don't match exactly, but the fact is that as is often the case, GISS has an area of major warmth (in South America this time) that is non-existant in the AMSU map. Both maps have very similar anomaly patterns over North America, Europe, Australia, and most of Asia and Africa. But for some reason, GISS had a heatwave in South America where the satellites had nothing of the kind.