I don’t have it either. The NOAA raw data (with or without TOBS) is what we want. GISS “raw” data is just NOAA adjusted data. (And I wouldn’t trust their reverse-adjusted stuff even if I could find it–which I can’t.)

OTOH, I am sure the NOAA raw data exists because St. Mac made a set of color maps of it to compare USHCN1 adjustments over the 20th century. So I’m guessing it’s available–somewhere–over on CA.

I suspect that adjusted CRN1&2 stations match adjusted CRN4&5 stations quite well. But not for the reason one might think! What I bet has happened is that INSTEAD of CRN 4s & 5s being adjusted down to the rates of CRN 1s & 2s, CRN 1s & 2s have been adjusted UPWARDS to match the rates of CRN 4s & 5s.

We’ll find out when Anthony et al. do the hammer-and-tongs analysis.

Until station move effects are quantified on each station record, it’s still not the true, unbiased temp history — unless the station had no moves.

JohnV’s analysis (posted above) used 62 CRN123 rural stations with a reasonably comprehensive geographic distribution (http://www.opentemp.org/_results/20071011_CRN123R/map_crn123r.png). That said, there is certainly the possibility that the analysis may change now that more data is available. My hunch is that it will not, but part of the fun of science is being proven wrong.

I do (and have) apologized in hindsight for not emailing you when writing up that story in October ’07. My only concern with the project early on was statements by you and others involved that seemed to imply that siting problems in individual stations implied that the overall GISS reconstruction for the U.S. was seriously flawed, with no analysis at the time to back it up. JohnV presented the first thorough analysis of the data collected, and his preliminary conclusions were interesting (though, admittedly, they only relied on 34 percent of all stations).

OpenTemp itself seemed to be created out of a legitimate interest in creating a transparent easy to use analysis tool for the Climate Audit community, and a number of other folks (mosh among them) used it for various things. It is a shame that it was discontinued, though real life does have a habit of getting in the way of our hobbies. It might be interesting to revive it or create something similar for your analysis, so that others can use the tool to perform various tests. For example, if will be interesting to look into why CRN45 stations trend higher than CRN123, and if that is confined to urban areas or extends to rural as well (which will hopefully tell us a bit about UHI!). An open source independent reconstruction of surface temperature will be a great resource (given the mess that is GISS code) for use by everyone, and now that the data is finally to the point at which you feel that it is ready to analyze, it would be a shame for a useful tool to go to waste.

I agree that CRN45 had a higher warming rate than 123 (as I posted over at Lucia’s place the other day). Its yet another interesting result that may come out of the data.

Since you followed JohnVs work on Crn123, you should have followed my work on CRN45? So what did I show about CRN45? and why did JohnV agree with it? It strikes me as Odd that the early work done my JohnV and me has receieved such uneven coverage. Basically I found and JohnV confirmed that CRN45 stations had warming trends higher than CRN123.

I have been round and round trying to locate “RAW” data.
I checked out WUWT & CA Resources links.
I checked out Craig Ipso’s CO2Science climate data.
There is a data set that is “raw + USHCN adjusted”.
Every data set was raw + something.
And none matched the B91 handwritten observer forms.

Anthony, what data set(s) are you using for your study?

And a lot of those CRN1s and 2s are ASOS setups in airports. So you have the HO-83 sensor warm bias plus the various airport issues.

Then there is the issue of raw vs. adjusted data. And whether TOBS should be included with raw. TOBS is a legit thing to adjust for — but is the method used by the NOAA legit? I’m sure I don’t know. Maybe that’s the ONE and ONLY thing they just happened to get right. (Or not.)

It’s not as easy as just totting up the various ratings and comparing the trends.

You make an interesting suggestion. I’m curious what you see that would be beneficial about incorporating an untested code base with a first look analysis of the CRN ratings of the US surface stations.

The notion that authors should seek to collaborate with critics of their work is a potent idea. It has the potential downside of tripping on human foibles that we all posess, but the legitimizing of the objectivity sounds like it would have significant benefits. Perhaps an article by you on the Yale Forum on Climate Change & The Media (http://www.yaleclimatemediaforum.org/index.php) would go a long way in explaining those benefits to all.

Regardless of disagreements we have had in the past, you’ve accomplished quite a feat crowdsourcing surface station ratings and gathering everything together. It will be interesting to see how the analysis pans out, though I suspect that unless there was an obvious bias in early station survey choice the results will be similar to the earlier effort: http://www.opentemp.org/_results/20071011_CRN123R/crn123r_gistemp_5yr.png.

I’d also suggest potentially bringing on JohnV as a coauthor. Your differing opinions on climate science would help legitimize the objectivity of the end-result (though, peer review and transparency of methods helps as well).

REPLY: Zeke, thank you for the recognition. While you do raise an interesting suggestion, I’ll point out that JohnV never so much as bothered to contact me before writing his code and releasing his very first results using only a handful of stations. Later I did some discussion with him, via Climate Audit, and the result you reference above is a result of that but uses only 17 CRN1/2 stations with poor spatial distribution. That you see his very initial effort, sans any early collaboration in the design or even courtesy notice to me to say that he was doing the project, as somehow legitimizing my own effort now speaks to your own bias. If you visit opentemp.org and type in my name in the search box in various ways, you won’t find it mentioned there, which seems a poor way of providing attribution for the work upon which his was predicated.

As I’ve pointed out many times, if I had done an analysis at that stage, in late summer 2007, with so little coverage and poor distribution, and found a significant difference that went the opposite way of JohnV’s finding’s, you and many others would be all over me crying “foul” and pointing out why my effort was flawed and premature.

Yet JohnV did exactly that, and not only is his work held up as an example of why the surfacestations effort itself was flawed in many of the forums you frequent, including your own (and I’ll point out that you also didn’t bother to even contact me before writing your own opinion on the project, an objective reporter gets both sides first), but JohnV gets a “free pass” for publishing an analysis prematurely.

But instead of doing a premature analysis, I’ve kept at the task of data gathering, and only published updates of the census data and distribution, while highlighting interesting issues with stations. JohnV’s opentemp effort appears to have no updates since late 2007. At this point I have no idea what the analysis will show on national or regional scales. Does siting matter on a national scale? I think so, but I’m not excluding the possibility that it does not once the data is appropriately filtered. We’ll find out.

Had this been a funded project from the outset, few people would even know about it, but making this an all volunteer effort, the only way to make it happen is to keep it visible. I’ve taken a fair amount of flak for that, but given that NOAA, NWS, NCDC, and GISS have not taken on this task of quality control evaluation of siting, I think the fact that we’ve been able to get as far as we have means that I made the right choice.

Besides, as Even points out, the issue is more complex than just comparing stations due to regional trends. Opentemp may have some useful tools, but the way it was presented at outset certainly didn’t lend itself to collaboration later. I won’t rule out the possibility of using opentemp for some portion of the analysis, but collaboration probably isn’t in the cards without some public fence mending.

I will say though that your suggestion legitimizes the effort, having gone from early vocal complaints of “pictures don’t matter” to now suggestions of collaboration with an early critic, points to the recognition that maybe pictures matter after all.

Anthony Watts

No, nooo. That was me, not Scott. Guess I’d better hurry.

On my side, I wouldn’t be surprised if you actually obtain a larger warming trend doing that. Remember that since the beginning of the 20th century, many American cities “de-industrialized” in the sense that many “hot” industrial plants left the cities themselves to go the countryside (because of pollution concerns). Including chemical plants, steel industry, and the like.

I have to disagree. This isn’t a study of urban heat island effect. This is about siting issues. In fact, your explanation would probably increase the trend. If you use rural stations to calculate UHI of the non-rural stations, and the the rural stations are warmer because of industrialization, then the rural staion isn’t going to allow a good calcuation of the UHI.