Apr 22 2008
Revisiting Tucson USHCN: Regional Correction Factor
In July 2007, the USHCN surface station at the University of Arizona was profiled for the first time. Being in a parking lot, it is a less than an ideal location. The pictures have been shown repeatedly on Watts Up and Climate Audit, as well as in presentations by Dr. Roger Pielke Sr. The temperature record was defended by both me and Herman and Jones using a similar methodology.
In a recent post at Watts Up, he profiles a station that is near a concrete walkway. I commented, basically saying that just because it is near the walkway doesn’t necessarily mean it will introduce a warming bias. Which lead him to reply:
Though given your defense last year of your UofA USHCN station in the parking lot, I suppose I can see why you don’t see much issue here. Your defense of the UofA station suggested you don’t think siting is important.
In the intervening months, I have used (with the suggestion of someone much smarter than me) a method that looks at the temperature difference between a station and the regional average. I define the regional average as the 10 closest stations within 1000km. The results from this analysis have been commented at Climate Audit, and have generally recieved praise from the ‘coolers’ group, and not so much praise from the ‘warmers’ group.
Getting back to the Tucson site, the reply by Herman and Jones at Dr. Pielke’s Climate Science Weblog included a figure of the yearly temperature record at UofA according to GISS, after corrections. [GISTemp at Tucson UofA]
One thing that should be immediately obvious is the large jump that occurred around 1998. This does not look natural, and should be investigated. Herman and Jones attribute the large decrease in temperatures in 1998 to the El Nino. Looking at their figure 2, the 1998 jump for the UofA temperature is around 3C, while at the airport it is only 1.5C. Also, the airport data quickly returns to values 1.5C higher than 1998 in subsequent years. The UofA temperature also returns, but only by 1.5C and not 3C.
The temperature sensors at the Tucson airport and UofA are only separated by 7 miles, and such inhomogeneous readings are not expected. This suggests that there are instrument or microsite issue with either the airport temperature or the Tucson temperatures. I applied the same method as before for looking at microsite issue.
 |
The black line represents the difference in temperatures between the regional average (the stations in the legend) and UofA. The red line represents a smoothing of the black, and can be thought of as a regional correction factor to the UofA data. Where the red line is positive means that the UofA parking lot stations is warmer relative to surrounding stations, and where the red line is negative means that the UofA station is cooler relative to surrounding stations.
The red dashed vertical lines represent entries in the UofA stations history file. According to Herman and Jones, the physical location of the station has moved twice since 1940; in 1948 and 1997. Note: the USHCN station history file ends in 1995. I can’t do anything about that. I’ve highlighted 1998 in cyan because Herman and Jones say:
The temperature sensor that was installed in 1997 is an aspirated device built by R.M. Young that is part of an electronic station package. Data from this device entered the record sometime in 1998. Prior to that time, the devices used were non-aspirated.
From the above figure, it appears that since the station last moved in 1997, there has been a cool bias at the Tucson UofA surface station. The timing of the abrubt change in the figure is difficult to ascertain. I am unsure whether it is because of the station move or because of the change in thermometer. Because the change represents a negative bias, it appears that the change in thermometers has not been corrected for in the temperature record.
In addition to showing that the Tucson UofA surface station has a negative bias with respect to surrounding stations since 1998, a correction can be applied to the original non-corrected data to account for whatever microsite or equipment changes that took place.
 |
Because the recent temperature anomaly at Tucson UofA is biased towards low temperature compared to the regional average, it is not surprising that the correction increases the trend. Even without the correction, the UofA station has a trend of 0.303 F/decade (0.168 C/decade) since the beginning of the record. After the correction is applied, the trend increased to 0.360 F/decade (0.200 C/decade).
The Tucson region has experienced rapid growth since this COOP station was founded. And with that brings a large urban heat island. From this analysis, it is impossible to determine how much of the warming trend in the Tucson region is due to global warming and how much is due to the urban heat island.
(No Ratings Yet)
Loading ...Related Posts:
30 Responses to “Revisiting Tucson USHCN: Regional Correction Factor”
To reduce spam, comments are automatically closed 30 days after the last comment. If you would like to comment on any closed thread, please use the contact form at the top of this page.
Now that surfacestations.org has built up a fairly substantial database, might not a better approach to a regional average be a group of stations selected for being high quality based on neutral criteria?
My concern is this. Arizona is both a rapid population growth state and one with a lot of hot, sunny days. The rapid growth makes it more likely that site issues will be created with time — paving an area, or a new building. The hot, sunny climate might suggest that such influences might have a more significant effect than in a cloudier state.
So why not just use stations free of apparent site issues?
Atmoz, just to mention that deep in the documentation at the CRN site is the interesting information that the original CRN design involved pairs of stations sited quite near each other, the idea being that one could use the paired data to eliminate such things as microsite bias. This design was abandoned when it was discovered (IIRC by testing an initial pair built at the NOAA Asheville site) that there were substantial differences in the data that weren’t amenable to being pinned to any particular site-related bias.
john, the premise of the surfacestations project is that the scientists who have been working with this data for many years are either fools or conspirators. Watts et al have indeed identified numerous problems with the USHCN sites (as has Atmoz), but I would point out that the existence of the CRN is proof that such problems have been known to the relevant scientists for many years. The key point is that the USHCN is so massively over-sampled (i.e., has vastly more stations than are needed) that a gross result like the national temperature trend is nonetheless obtainable with reasonable accuracy. Where the USHCN turns into a data train wreck is at the regional level, and it was to deal with that problem that the CRN was built. BTW, a first pass was done on the Watts “high quality” stations and the results were more or less consistent with NOAA and GISS. The disappointment was palpable, but Watts is able to keep morale up by running lots of posts of the “it’s the sun!” and “Gore is fat!” varieties.
Which reminds me: There’s plenty of CRN data available now if all one is interested in is the national trend, but Watts and minions have expressed no interest in it. Interesting, that.
It is worth noting that the paper by Micheals and MiKitrick suggusts that the US network is pretty good despite the issues. The real problem is if these kinds of microsite issues exist in the US network then why should have any confidence in the stations in places like Russia, China, India or Africa?
Atmoz, your latest exchange with Watts certainly illustrates the problem. When I first looked at the post, I figured I must be missing something since I couldn’t see how the photo illustrated a microsite bias. In particular, there was no apparent T difference between the side of the MMTS facing the sidewalk and the one not (although a slight difference between the lit and unlit portions can be seen). Is it incorrect to conclude that the photo actually makes a point opposite to the one Watts wants?
[Reply: It probably is incorrect to conclude that the photo shows the opposite of what his point. But it is probably also incorrect to conclude that the photo validates his point. I doubt that one photo taken on one day is going to tell us much of anything about the temperatures recorded at that station.]
Also, it’s interesting how his response to you resorts first to a rigid application of the rule. Among other things, he ignores the bleedin’ obvious fact that the CRN *siting* rule would have no need to differentiate between a small chunk of thin concrete (the sidewalk) versus a large, thick piece that would actually have a chance of affecting air temperatures more than a foot or two away. Also, since a uniform surface is specified for the inside of the CRN enclosures, there’s no reason to be concerned about the thermal characteristics of the ground immediately surrounding the sensors, which in the real world would be important but Watts ignores because it’s not in the CRN rules. My guess is that in this particular case the frequency with which the grass gets cut is a far more significant factor than that sidewalk. Having established (to his own satisfaction) that the CRN rules are appropriate for his purposes, it would not be useful to then examine the ways in which they aren’t.
I think he protests a bit too much about the impracticality of taking those temp readings, BTW. Watts makes much of his living selling little weather stations, so I think he knows exactly what sort of results he’d get.
That meeting tomorrow should be interesting.
It seems to me that in order to use surrounding regional sites to help correct for problems, you’ve got to make sure none of the surrounding regional sites was adjusted near the same time. Is that correct?
[Reply: It depends. For the sake of this blog, I've made the assumption that the temperature changes due to station moves / equipment changes / etc. will be about the same from station to station. That means when I take the average of the 10 surrounding stations, any changes at one individual station will be 1/10 the magnitude of at the station of interest.
In the example above, there is a station move (or something) that occurs at Mesa at around the same time as the change in Tucson. It is quite obvious when looking at the station data, but it does not appear in the regional averaged data. So when the Tucson data is subtracted from it, we're left with the changes at Tucson representing the majority of the changes seen.
If this were to be used as an actual correction, then yes. Much more care would be needed to looking at the station histories of the surrounding stations.]
Steve, isn’t it obvious? That big white mass next to the temp sensor is like sticking the sensor in a toaster.
When the temperature is rising it would tend to impart a cooling effect and when the temperature is falling it would tend to impart a warning effect.
The number of sites that have serious issues should make any person seeking the truth to question our ability to see past the noise.
MikeEE
Raven, of course the problems are worse in many places. But even with the low-quality stations and all the interpolations that have to be made, the temperature trend at the global scale just doesn’t take that much data to establish reasonably.
But this raises a potentially interesting question, which is what happens if we take the U.S. and other reliable long-term segments of the global T record (Europe and Australia, e.g.), compare them with the satellite data and then see how consistent that result is with the presumably less reliable regions of the global land T record. I haven’t seen that done, but I’ll bet the the answer is… actually not very interesting.
FYI the plan is to extend CRN-like stations globally. Interestingly there’s at least one USCRN station being sited in Siberia, IIRC for purposes of getting a consistent Arctic-wide calibration.
MikeEE, which big white mass? Do you mean the sidewalk? If that were noticeably heating the housing, we should see some sort of differential heating effect at the lower right of the housing. It isn’t there. Similarly, if the sidewalk were putting off that much heat we would expect to see the effect in the bordering grass. We don’t.
Steve L, I don’t know if there’s a conscious effort to avoid e.g. simultaneous sensor upgrades but based on the records I’ve seen it’s a non-issue.
“Where the USHCN turns into a data train wreck is at the regional level, and it was to deal with that problem that the CRN was built.”
But isn’t that what was exactly done for this post; use ‘regional’ information to modify measured data?
For engineering and scientific experimental investigations, measured data are changed only by use of a calibrated instrument that can be traced back to a reference standard. Temperature data collected in a flow field, for example, are never modified by reference to other data measured nearby.
“My guess is that in this particular case the frequency with which the grass gets cut is a far more significant factor than that sidewalk. Having established (to his own satisfaction) that the CRN rules are appropriate for his purposes, it would not be useful to then examine the ways in which they aren’t.”
The rules cannot be discussed, applied, or modified on the basis of guesses. The CRN rules are not simply ‘appropriate for his purposes’, they are the CRN rules.
“The key point is that the USHCN is so massively over-sampled (i.e., has vastly more stations than are needed) that a gross result like the national temperature trend is nonetheless obtainable with reasonable accuracy.”
How does the concept of over-sampling apply when different instruments in different spatial locations are taking data of different mico-effects at different times? I guess we can always fall back on The Law of Large Numbers.
Finally, we are talking about modifications to measured data. They are not adjustments or corrections. They are modifications, pure and simply. “Having established (to their own satisfaction) that the modifications are appropriate for their purposes, it would not be useful to then examine the ways in which they aren’t.”
Dan, however much you might wish otherwise, the rules in question are for *siting* the CRN stations *only*. They are of at best very limited value in grading USHCN stations.
This passage –
“For engineering and scientific experimental investigations, measured data are changed only by use of a calibrated instrument that can be traced back to a reference standard. Temperature data collected in a flow field, for example, are never modified by reference to other data measured nearby.”
– proves that you don’t get it at all. Even the CRN data will require that sort of treatment. Interestingly there are hundreds of scientists working on this stuff who seem to think that the techniques used are perfectly suitable for “scientific experimental investigations” (not sure why you added the “experimental”), and then there’s… you. Who to believe?
As a “lukewarmer” I always appreciate your dedication to getting things right. keep up the good work
last time I looked at the UofA data Atmoz, I recall senora being an
intresting comparsion. I might be misremembering, but did you
have a look at that?
[Reply: I haven't. I can't figure out a good way to download the CRN data.]
Anthony, is visiting Karl et al, so I’ve asked him to ask NCDC if we can get a friendlier access to the CRN data being collected. You can join that request if you like
[Reply: I sent them an email about updating the HCN station history file a while back. Never heard back from them.]
I think their budget/staff situation is not good, thanks to the yahoo-in-chief. This will tend to result in less-than-ideal service on “extras.”
steve bloom, let me explin the crn ratings. when anthony was part
way through the project, i found the document and suggested he use it to rank the stations. Why? because in the end if you want to do any kind of analysis you need to be able to rank the sites. So some form of repeatable criteria needs to be applied. The thing I find troubling is that people take the 5C ‘error’ for a class5 to
be a fact. I do not. It’s something to test. using opentemp
I found the bias to be on the order of .15C
without having history files, however, this result is interesting but
not conclusive.
WRT to oversampling, yes ushcn is oversampled. The siting study
of CRN suggests that 138 ( something like that) good stations
are all that is required. So, one could theoretically, start the process if weeding out non compliant stations and suffer no loss of
information, and it would of course take the whole issue off the table
[...] I did a post called Revisiting Tucson USHCN: Regional Correction Factor. I thought it would be my last post on the UofA surface station in a while, but I guess not. The US [...]
Steve Bloom,
No, you wouldn’t necessarily expect to see a thermal gradient. There is radiative and convective heat transfer from the sidewalk.
The convective part wouldn’t show up as a gradient because at that range it should be a uniform warming effect.
The radiative heat transfer would only show up if the housing is a good insulator, and if you look closely it does appear to have a bit of a temperature gradient – but that’s just eyeballing it. Also, what is the material? If metal, there shouldn’t be any gradient because it is such a good conductor. If it is plastic a gradient would probably still would be indiscernible in the photo.
MikeEE
“The thing I find troubling is that people take the 5C ‘error’ for a class5 to be a fact.”
Indeed! My French is very poor, but I strongly suspect that the reason AW hasn’t posted that Leroy translation is because the seat-of-the-pantsness of the whole exercise will then become clear (seat-of the-pants being just fine for *siting* criteria, of course).
“So, one could theoretically, start the process if weeding out non compliant stations and suffer no loss of information, and it would of course take the whole issue off the table.”
So OK, it won’t be long now before John V.’s exercise can be repeated with a larger sampling and that will be that. NCDC will do no such thing because there’s useful data to be had from even the crappiest station. Once the CRN-USHCN calibration has been done (IIRC not less than 5 years from now), the RCMers are going to need the data from those crappy stations (suitably de-crapped, of course). The irony (for some) is that the surfacestations project, if complete, will be a big help when that time comes.
Steve Bloom and others:
Surely CRN siting rules exist for a reason?
Surely one would expect NCDC/NOAA/whoever to follow their own rules?
Certainly I would hope that the answer to these questions would be “yes”. Yet it appears that the rules have NOT been followed, and that until surfacestations.org started cataloguing these stations, no-one had bothered to check. This seems to be a systemic failure - one that, hopefully, can now be corrected.
The effects of this failure are many and varied, but for me it boils down to this: what many assumed (and a reasonable assumption it was) to be pristine data uncontaminated as much as possible by microsite affects is, in reality, contaminated. As far as I can see, no-one can say how much or for how long, and the best we can do is to apply the uncertainty apparent in the current conditions to the full record. That may mean adding up to 5C of uncertainty to a *lot* of stations, I know, but what else can we do? We cannot correct these errors because we do not have any information on their magnitude through time, and so any corrections would, at best, be guesses and I would think that, if anything, such corrections would actually INCREASE uncertainty, not reduce it.
Given that the trend in temperatures seen seems to be on the order of 1C, and microsite contamination up to 5C, what does this mean for AGW? I don’t know the answer, but surely “significant” and “wide ranging” are part of that answer!
[Reply: The units for your trend are in the wrong units, suggesting that you may not know the difference between the temperature and a temperature trend.]
“Surely CRN siting rules exist for a reason?”
Yes, to *site* the *CRN* stations, for which purpose they worked very nicely. They were not intended to be used for the completely different purpose of *grading the data quality* of the completely different *USHCN* network.
The site surveys themselves are indeed useful, just not for the reason Watts had in mind.
Atmoz: 1C/century. And yes, I know it’s actually *less* than this, but it’s in your favour. And I did say “…order of…”.
Steve Bloom: So when Watts and his volunteers find that more than 1/2 the USHCN stations they have surveyed fall into CRN class 4 or 5, this does not concern you? This information does not cause you to pause and ponder the implications? You don’t think that it “matters” that we have documentary evidence that the likely errors are up to 5 times the century scale trend at more than 1/2 the stations of the best network in the world? Your faith is touching, but I’m afraid I don’t share it!
Both: you missed the point and nit-picked. Do you dispute the CRN rating that Watts has given the stations? Do you dispute the error margins the CRN ratings imply? If so, please explain (cites would be nice for the second). If not, what do you think it means?
I want to be clear that I am trying to learn - if there are good reasons why I’m wrong, please tell me what they are.
[Reply: See here and links therein.]
Neil, you say “likely errors,” but based on what? The original paper by Leroy includes no calculations from which the possible errors might be derived, and indeed has question marks behind all of the error quantities. IOW, the possible errors are completely seat of the pants. If one is grading new sites, that’s fine since we are interested in eliminating even faintly plausible biasing factors and have no reason to care about their magnitude. If on the other hand one is interested in using identified microsite influences to screen or adjust data, a more rigorous approach is needed. Doing that correctly makes a lot less sense than just building a new system followed by an upgrade of the present one, thus the CRN and the HCN-M.
Take for example the comparison Atmoz just did between the UA USHCN station and the nearby new CRN station, bearing in mind the the former is pretty much #1 in the Watts parade of horribles. Note the neat alignment and consider how that might be possible. As mosh pointed out, this comparison wouldn’t have caught a step-change resulting in a bias (which is why the CRN-USHCN “burn-in” period is for ten years rather than four), but it settles the question of whether there is still useful data to be had even from the “worst” of USHCN stations.
Steve Bloom said:
“The original paper by Leroy includes no calculations from which the possible errors might be derived, and indeed has question marks behind all of the error quantities. IOW, the possible errors are completely seat of the pants.”
Yep. It’s a best guess, no doubt. But no-one has bothered to pin it down any better - at least that I’m aware of. I’m guessing you don’t have a cite to someone who has done so, or you’d have provided it. So that’s where we are - no-one has pinned it down any better than the one peer reviewed “guestimate”. What else should I use? Should I cite a blog? Would you take it seriously if I did? With all due respect to our host, why should I accept *his* blog posts over, say, Climate Audit, or CO2Science or some other “skeptical” blog? Why should I accept *any* blog post over a peer reviewed paper? I’m pretty sure I’d get pasted if I did and you didn’t like the conclusions!
“If one is grading new sites, that’s fine since we are interested in eliminating even faintly plausible biasing factors and have no reason to care about their magnitude. If on the other hand one is interested in using identified microsite influences to screen or adjust data, a more rigorous approach is needed.”
As to adjustments, yes, I certainly agree - there is no point in making adjustments if you are unsure of what you are trying to adjust for. As to screening, I beg to differ - I doubt you would seriously accept a record from a thermometer if you new nothing about it’s siting and history. So I ask you to look at the USHCN documentation that Watts has provided - there are some *very* good sites, and some *very* poor ones. That and the original siting report (if any) are all we have to go on as far as I know. Certainly you couldn’t adjust these records without a great deal more information, but there’s no reason you couldn’t select only sites that show both as “good” and now (2008) as “good” too, is there?
It seems to me that you (and Atmoz) have accepted the CRN rating Watts placed on the sites, but dispute the magnitude of the uncertainty. That’s fine. I’m sure many would. But I have no other peer reviewed paper suggesting otherwise, so what am I to do? Shouldn’t I accept the peer reviewed paper until and unless someone else publishes something to refute it or refine it?
Do you see my problem? If I rely only on peer reviewed papers, I must conclude that there are errors up to five times larger than the “signal”. If I accept what Atmoz and RealClimate blog about, then I would i) conclude it doesn’t matter or is wrong and ii) have to accept other blog entries over peer reviewed papers. Which leaves me in an impossible position because there are more blogs than I could hope to keep up with and they don’t always agree!
Atmoz: Thanks for the links. Sorry, but I don’t see the relevance to my questions and observations. As far as I can see (and I’ll admit I didn’t look *that* closely), there is nothing on microsite bias contamination over time or any reference to what appropriate error margins might be. Can you be more specific? I know it sounds “bitchy” or “snarky” or whatever, but really - you can do better than that sort of arm-waving!
There’s extensive documentation over at the NCDC site, Neil. FYI much of this is are technical papers that aren’t peer-reviewed. The Leroy paper isn’t, for example, even though one might think so based on the way Watts treats it. The reason no-one has tried to do better is because what Leroy did was fine for *siting* purposes and to do it for purposes of adjusting data is probably impossible. One could do it for an individual site, but such an exercise is harder (and far less useful) than just setting up a new station.
If you need a pointer to specific documentation, note that all of those NCDC sites have email addresses for the technical folks. In my experience they’re very helpful.
Of course one can go through an exercise of rating the USHCN stations (by whatever standard) and derive a national temperature trend using only the “good” stations, but the reason NOAA hasn’t done it is because it’s not a scientifically useful exercise.
BTW, Neil: One reason to accept Atmoz’s views on issues such as this over those of those of the other sites you mention is because Atmoz is rather more qualified than they are.
Steve, this is looking more and more like the same fiasco as UHI - initial estimate is quite high, some spot checks and non-rigorous analysis indicate it’s a non-problem, and because it’s what you want to find, no-one looks any more closely at it. I guess it will take someone like Ross McKitrick to take an interest and do a full and proper analysis before we have an answer that reflects reality - as opposed ignoring it and hoping it all goes away, all the while denigrating those who, like Watts, are at least attempting to amass some data on the (non)problem. I need to find Bender - I think I have another item for his list…
“BTW, Neil: One reason to accept Atmoz’s views on issues such as this over those of those of the other sites you mention is because Atmoz is rather more qualified than they are.”
The last refuge of the scoundrel - argument from authority. I remain highly amused that many climate scientists appear to propose this as a reason to trust their results, but do not seem to accept that same argument about, say, statistics - where even Wegman is doubted!
Be that as it may, perhaps you can explain to me why I should trust this particular “authority” when it remains anonymous?
Well, Neil, it’s pretty obvious you’re hearing what you want to hear from the usual suspects. By all means continue to not independently examine the source materials I pointed you toward. Just consider that in doing so you end up with an awful lot of shaky assumptions, e.g. thinking that the Leroy paper was peer-reviewed. Ross McKitrick is an economist, BTW. Do you go to climate scientists for your economics? And if you like Wegman, don’t forget that he testified that he’s convinced that AGW is a serious problem notwithstanding any niggles about MBH 98.
Atmoz isn’t really anonymous, although as he will readily admit he’s “only” an atmospheric sciences grad student. For my part, I paid attention to the comments here by known experts in the field, which have satisfied me that Atmoz knows whereof he speaks. Sure he knows less than such experts, but it’s also clear from past posts that he avoids claiming knowledge that he lacks. FYI, for an example of someone who’s truly anonymous, try bender.
“Well, Neil, it’s pretty obvious you’re hearing what you want to hear from the usual suspects. ”
I could easily make the same claim against you, Steve.
“By all means continue to not independently examine the source materials I pointed you toward. Just consider that in doing so you end up with an awful lot of shaky assumptions, e.g. thinking that the Leroy paper was peer-reviewed. Ross McKitrick is an economist, BTW. Do you go to climate scientists for your economics? And if you like Wegman, don’t forget that he testified that he’s convinced that AGW is a serious problem notwithstanding any niggles about MBH 98.”
Yes, McKitrick is an economist - well trained in time series statistics and their analysis. Should I ignore his time series statistical analysis of climate data because he’s not a climatologist? Or should I ignore Mann’s analysis because he, by his own admission, is not a statistician?
“Atmoz isn’t really anonymous, although as he will readily admit he’s “only” an atmospheric sciences grad student. For my part, I paid attention to the comments here by known experts in the field, which have satisfied me that Atmoz knows whereof he speaks. Sure he knows less than such experts, but it’s also clear from past posts that he avoids claiming knowledge that he lacks. FYI, for an example of someone who’s truly anonymous, try bender.”
I suspect, from reading lucias blog amongst others, that he (Atmoz) sometimes presents data in such a way as to support the consensus position. There’s nothing wrong with that PROVIDING that people recognise this for what it is - advocacy, not science. That might sound harsh, but in my view (and I’m not alone in this) it’s important in science to look for faults in your pet theory, not look for reasons to support it.
And yeah, I know Atmoz is not completely anonymous - but if it’s so well known, why isn’t it mentioned here on his blog?
I’d just like to add somewhat to the second last paragraph of my previous post - on re-reading it, it does seem somewhat harsher than I perhaps intended! (sorry if it does!)
I want to make it clear what I’m trying to get across, so perhaps I can explain it this way. The example I will use is the “refutation” of Lucia’s falsification of IPCC projections. This is not, and never was, a refutation - it was a refinement of the falsification. In order to refute the falsification, it would be required to show one of the following: errors in the data set; errors in the methodology; errors in the math. None of these were provided. Instead, what we got was that using a subset of the data Lucia used, and using a variation of the methodology she used, we can come to a different conclusion - we went from “a therefore x” to “a AND b therefore x”. You will notice that I am not refusing to believe the analysis of either side here, and I’m not preferring one over the other - I am saying that they are both right, and that the fact they are both right can possibly provide us with some indication of the underlying mechanisms involved that each analysis on it’s own cannot. That they reach different conclusions from the same data does not mean we have to choose one over the other, because they are using different methods and therefore illuminate subtly different aspects of the issue.
In my view, there are too many highly polarised “experts” in climate change who - it seems to me - fall into the trap exposed above. They apparently feel that, in the example above for instance, one or the other must be right and fail to consider that they may both be correct and attempt to understand what that seeming paradox can tell us. Further, I would suggest that once this is pointed out, someone who can not or will not see that paradox (and attempt to understand and/or explain it) is no longer being objective - in short, they are then advocates. It was perhaps somewhat rash of me to assume that Atmoz is an advocate - I have no evidence that he has considered what I have outlined above, after all. Of course, he no longer has that excuse…
I hope that makes it somewhat clearer.