Comments 1 to 19:

1. dana1981 at 11:28 AM on 30 July, 2012
   I presume people are going to be asking us what we think about Watts' new preliminary paper on the accuracy of the surface temp record. After a first read of the paper, I think it has some serious problems. First off I don't think the conclusion ("that reported 1979-2008 U.S. temperature trends are spuriously doubled") is supported by the analysis in the paper. I think the methodology was rather poorly chosen. They compared raw data to fully adjusted data, for example not first correcting for time of day of the observations, so there's no way for us to know how much of any discrepancy they find is due to urban heat effects and how much is due to time of day and other biases that are corrected for in the final product. They found a factor of 2 difference between raw class 1 and class 2 station trends and the final adjusted data set, then they simply assume that whole factor of 2 difference is due to urban heat effects without actually checking to see how much is due to other factors. It's an apples and oranges comparison, so they conclusion is not supported. Their conclusion also doesn't pass the 'sniff test'. Over land, surface and satellite warming trends should be roughly equal, yet they find a factor of 1.6 greater warming trend in the UAH TLT data than in their raw class 1 and 2 surface temp data. So basically if their analysis is right, it means UAH is biased high, which is simply an implausible result. More likely the Watts results are low because they haven't corrected for various other biases like time of day. Note that these are just my first impressions after reading over their paper and it's possible I've missed something, but it appears to me like they have some major problems there.
   0 0
2. skywatcher at 14:31 PM on 30 July, 2012
   To paraphrase Darwin, "A house burned down by fire did not tell its story more plainly than did this atmosphere (or planet!). On Watts, he seems to have forgotten some sentiments of his about press releases before peer review
   0 0
3. Rob Painting at 16:21 PM on 30 July, 2012
   The cartoon is spot-on..........sadly.
   0 0
4. Steve Bloom at 16:39 PM on 30 July, 2012
   Don't forget to have a look at Leroy's paper (peer-reviewed? and where is the data supporting the standards?). Consider the Class 3 temp standard, a key break-point for Watts: "Class 3 (additional estimated uncertainty added by siting up to 1 °C) • Ground covered with natural and low vegetation (< 25 cm) representative of the region. • Measurement point situated: o at more than 10 m from artificial heat sources and reflective surfaces (buildings, concrete surfaces, car parks etc.) o at more than 10 m from an expanse of water (unless significant of the region) o away from all projected shade when the Sun is higher than 7°. A source of heat (or expanse of water) is considered to have an impact if it occupies more than 10 % of the surface within a circular area of 10 m surrounding the screen or makes up 5% of an annulus of 5m." So if you have, e.g., a station that meets all Class 1 requirements except for having a ~1.5m shrub at a distance of 10m such that the aforementioned 7 degree standard (which note is relative not to the sensor itself but to the point on the ground directly below) is exceeded for even a few days each year, that's enough for a demotion all the way to Class 4. I'm tempted to blame Leroy for facilitating this stupidity, but of course his expectation is that the standards will be used by scientists, not propagandists.
   0 0
5. heijdensejan at 17:21 PM on 30 July, 2012
   Victor Venema also has his first review of the new Watts paper http://variable-variability.blogspot.in/2012/07/blog-review-of-watts-et-al-2012.html
   0 0
6. zinfan94 at 02:01 AM on 31 July, 2012
   Dana, I have a suggestion that you might keep in mind as you read Watts et. al. draft paper (copied from my comments at Stoat): One thing that really puzzles me about the Watts et. al. draft paper. Isn’t it important in scientific papers to address any prior work that clearly contradicts the current findings? Menne et. al. (2010) did a comparison of the best quality CONUS temperature measurement stations (USCRN) with the USHCNv2 stations. ( Menne paper Figure 7). The USCRN stations have been engineered to provide much more robust and error-free temperature recording and monitoring (three duplicate temperature sensors, isolated sites, consistent designs at every station, additional monitor supervision). Menne et. al. found that the USCRN temperature anomalies correlated extremely well with the USHCNv2 anomalies (r2=0.998 for Tmax and r2=0.996 for Tmin). Now Watts et. al. comes along, and claims that the USHCN is significantly in error against the station subset of only 48 Class 1 sites, and 112 Class 2 sites that he personally selected. The USCRN has 107 locations with 114 monitoring stations (seven stations are duplicated nearby). All of the USCRN stations would likely be considered Class 1 or better using the methodology in Watts et. al. And they were sited across the country to capture the CONUS anomalies accurately. So why didn’t Watts et. al. correlate their USHCN subset results (using the methodology applied in their paper) against the USCRN stations? Essentially the USCRN is the “gold standard” for siting, so the lack of the comparison in Watts et. al. sticks out like a sore thumb. If nothing else, this is an extremely important quality control exercise to ensure that obvious mistakes weren’t made processing the data from the USHCN subset stations. (next comment) To say what I meant in the last comment, a bit more succinctly: Isn’t Watts et. al. claiming (indirectly, given the Menne et. al. 2010 results that show excellent correlation between the USCRN and USHCNv2) that the US Climate Reference Networks stations are substantially in error? If so, this is an extraordinary result. NOAA needs to find out what is wrong with their state-of-the-art network ASAP.
   0 0
7. LarryM at 04:16 AM on 31 July, 2012
   I haven't read the Watts paper, and probably won't at this time because it hasn't even been submitted to a reputable peer-reviewed journal yet, so at present it has little more scientific credibility than a blog post or an editorial opinion. I suppose SkS will need to comment on it because the paper's advocates will attempt to make a mountain out of an unpublished molehill, but let's not forget that it's not a real paper until it has been revised to address the comments of expert reviewers and an editor, and accepted for publication.
   0 0
8. dana1981 at 04:43 AM on 31 July, 2012
   zinfan - I don't know, Watts seems to be claiming that their new classification approach is a significant change from previous approaches, so I don't know if they would consider the USCRN stations class 1 or 2. We're working on a blog post that outlines the various issues with the paper though. Eevn though it hasn't been submitted, we're viewing it as an opportunity for Watts et al. to correct these problems before they submit the paper. Not that I expect them to do so, but at least we're giving them the opportunity.
   0 0
9. zinfan94 at 07:17 AM on 31 July, 2012
   dana: OK. Got that. I am trying the same approach. I did put this comment up on the WUWT thread, suggesting that they should use the USCRN to identify siting issues. Currently the comment is caught in moderation, and its been awhile, so I copied and pasted the comment before it "disappeared". At the very least, the Watts et.al. team should be asked to do similar work to that done by Menne et.al. 2010, that correlated USHCN data with USCRN data. But they could move a long way to actually demonstrating and quantifying siting biases, if they just used the USCRN as the standard for comparison. Paul K2 says: Your comment is awaiting moderation. July 30, 2012 at 1:10 pm I read the paper, and it has some obvious problems. The most obvious is the lack of comparison between the USCRN and the different classes of stations in this paper. The USCRN should be the “gold standard” for station temperature measurements. If there are siting issues, you don’t need decades of data to spot the problem. The siting issues should be detectable, even with only 4-5 years of daily data. Since almost all of the USCRN stations now have over five years of data, correlating the Tmax and Tmin against the data from the various classes of “selected” stations in this report should be the obvious first step in identifying siting issues. Menne et.al. (2010) did this with the homogenized data from different subsets of stations, and found very strong correlations with the entire USHCN dataset (r2 = 0.98 for Tmax, and r2 = 0.96 for Tmin). The failure of Watts et. al. (2012) to complete the same exercise, should be rectified prior to publication. If the findings regarding siting in this paper are accurate, then the Class 1 and Class 2 sites should correlate strongly with the USCRN station data, and Class 3, Class 4, and Class 5 stations should have significantly lower correlations with the USCRN stations. If the correlations for the different station Classes identified in this paper are not substantially different enough to explain the large differences in decadal trends, then some of the other adjustments are likely responsible for the differences. For example, changes in time of observation, adjustment for a move of a station that was previously sited next to a heat source to a better location (that now allows the station to be classed as Class 1 or 2), switch to a different temperature measurement device or system, etcetera, could explain why smaller classes of raw data don’t track well with the overall trend calculated from homogenized station trend data. Not addressing the USCRN data is a serious shortcoming for this paper.
   0 0
10. dana1981 at 07:45 AM on 31 July, 2012
    Good luck zinfan. I asked a pretty straighforward question in the WUWT comments regarding a discrepancy in the paper and got some rather rude responses from the moderator REP. They're not very open to criticism (or skepticism) over there.
    0 0
11. Tristan at 12:26 PM on 31 July, 2012
    From REP "When you read what Anthony posts, take the time to actually read it, think about it, and the implications will become obvious. We will be moderating strictly. Snark, outrage, disappointment, and instant-analysis of how that stupid Watts got it all wrong will, of course, be snipped." Instant analysis of how that amazing Watts got it all right will, of course, be scooped up and bathed in.
    0 0
12. warm at 19:26 PM on 31 July, 2012
    I suggest a complementary approach to contradict Watts' paper: provide other evidences of recent warming in USA that are not compatible with low warming trends. For instance this study: http://passthrough.fw-notify.net/download/112635/http://www.wwww.aslo.org/lo/toc/vol_52/issue_5/2013.pdf Spatial analysis of ice phenology trends across the Laurentian Great Lakes region during a recent warming period The results clearly show ice retreat in the great lake area. The Watts results show that this region experienced low trend (0.135C/decade vs 0.37 "official" trend): how ice could melt without significant warming ? I found another article river temperature trend (unfortunately, paywalled ) http://www.esajournals.org/doi/abs/10.1890/090037?journalCode=fron Also a recent article on forest phenology (end of growing season) (paywalled) Trends in fall phenology across the deciduous forests of the Eastern USA http://www.sciencedirect.com/science/article/pii/S0168192312000500 This article use USHCN data Other approach, glacier change, such as this study: http://www.glaciers.pdx.edu/fountain/MyPapers/SittsEtAl2010_GlaciersMt.Adams.pdf Twentieth Century Glacier Change on Mount Adams, Washington, USA "The main driver of glacier recession appears to be summer air temperature, as little change in precipitation has occurred over the past century. All three temperature data sets show a significant increase in summer temperature beginning around the 1980s (e.g., Nylen 2004, Lilliquist and Walker 2006) corresponding to the rapid retreat in glacier area during the latter part of the 20th" In the northeast US, Watts' result shows very little warming (0.078) as compared to "official" trend (0.247). Glaciers are melting without warming ??
    0 0
13. Eric (skeptic) at 19:37 PM on 31 July, 2012
    zinfan94, your question and an answer to it were posted at WUWT. To briefly follow up here, I would not expect a USCRN comparison as part of a reanalysis of 1979-2008 trends for the same reasons that one was not done here: http://www.agu.org/pubs/crossref/2012/2012GL051387.shtml (I don't have that full paper however).
    0 0
14. warm at 21:14 PM on 31 July, 2012
    "Their conclusion also doesn't pass the 'sniff test'. Over land, surface and satellite warming trends should be roughly equal, yet they find a factor of 1.6 greater warming trend in the UAH TLT data than in their raw class 1 and 2 surface temp data. So basically if their analysis is right, it means UAH is biased high, which is simply an implausible result. More likely the Watts results are low because they haven't corrected for various other biases like time of day." I have the "definitive" answer to that issue: Temperature Trends at the Surface and in the Troposphere http://climate.envsci.rutgers.edu/pdf/TrendsJGRrevised3InPress.pdf This study used GFDL model to simulate surface and TLT (such as UAH) temperature trends. At the extra-tropical northern lattitude, a more rapid warming of surface as compared to lower troposphere is expected, in line with the results of "official" trends (0.22 in the lower troposphere, 0.3 at the surface...). The amplification factor used by Watts to match TLT and surface is a pure "fudge factor" that have no scientific basis...
    0 0
15. dana1981 at 03:53 AM on 1 August, 2012
    Thanks warm, that's a useful reference. I've used it in our response post to the Watts paper, which we'll probably publish on Friday.
    0 0
16. zinfan94 at 06:43 AM on 1 August, 2012
    Eric (skeptic): Wrong Menne paper. Try this one: JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 115, D11108, doi:10.1029/2009JD013094, 2010 On the reliability of the U.S. surface temperature record Matthew J. Menne,1 Claude N. Williams Jr.,1 and Michael A. Palecki1 Received 27 August 2009; revised 24 December 2009; accepted 7 January 2010; published 8 June 2010. In this paper Menne et. al. compared USHCN stations of different siting quality with the USCRN stations.
    0 0
17. John Hartz at 06:53 AM on 1 August, 2012
    To the surprise of no one, Muller’s Op-ed and Watts’s press release have generated a goodly number of blog articles over the past few days. One that merits a careful read is: “More evidence attention-grabbing climate studies prematurely rushed and potentially flawed” by Jason Samenow of the Capital Weather Gang, Washington Post, July 31, 2012 To access this blog post, click here.
    0 0
18. Bernard J. at 14:21 PM on 1 August, 2012
    John. I know that you'll be aware that Watts is intending to release a second draft soon - if it hasn't already been done it might be useful to anticipate where any future revisions will end up taking any final paper that might actually make submission.
    0 0
19. zinfan94 at 00:53 AM on 2 August, 2012
    OH NO!!! Another DRAFT??? Please spare us the Pontificating Press Releases, the Grandstanding in the press (NYTimes etc.), and the enormous waste of effort spent investigating a non-issue. Why doesn't AW get some real peer review? Go to the High School in Chico, find the science teacher, and use some of his haul from Heartland to pay the teacher to review his work before subjecting us all to this again. The process of releasing and publicizing not-even-half-baked propaganda, doesn't work. In the end, the process has even damaged (or destroyed) his own reputation. I hope you are wrong.
    0 0

You need to be logged in to post a comment. Login via the left margin or if you're new, register here.