Advertisement

Watts not to love: New study finds the poor weather stations tend to have a slight COOL bias, not a warm one

Analysis of actual U.S. data disagrees with Anthony Watts’ primary conclusion.

My guest blogger today is one of the best meteorologists around, Dr. Jeff Masters, former Hurricane Hunter and now Director of Meteorology for the Weather Underground. There’s so much damn stuff to blog on, I didn’t get around to the amazing new study that, as DotEarth’s Andy Revkin put it, “throws cold water on the allegation that bad weather stations have amplified America’s warming trend” — allegations made by former TV weatherman Anthony Watts who runs the anti-science website WattsUpWithThat.We knew that the “good or best” weather stations provide data that matches the overall U.S. temperature record (see Must-read NOAA paper — Q: “Is there any question that surface temperatures in the United States have been rising rapidly during the last 50 years?” A: “None at all.”). But as Revkin explains, “In essence, the paper, On the Reliability of the U.S. Surface Temperature Record (pdf), concludes that the instrument issues, as long acknowledged, are real, but the poor stations tend to have a slight cool bias, not a warm one.” Like Revkin, I first saw this on Masters’ Wunderblog, and he gave me permission to excerpt it at length here.Former TV weatherman Anthony Watts, who runs the popular global warming contrarian website, “Watts Up With That”, was convinced that many of the U.S. network of surface weather stations had serious flaws in their siting that was causing an artificial warm bias in the observed increase in U.S. temperatures of 1.1°F over the past century. To address this concern, Watts established the website surfacestations.org in 2007, which enlisted an army of volunteers to travel the U.S. to obtain photographic evidence of poor siting of weather stations. The goal was to document cases where “microclimate” influence was important, and could be contaminating temperature measurements. (Note that this is a separate issue from the Urban Heat Island, the phenomenon where a metropolitan area in general is warmer than surrounding rural areas). Watts’ volunteers — 650 strong — documented the siting of 865 of the 1,218 stations used in the National Climatic Data Center’s U.S. Historical Climatology Network (USHCN) for tracking climate change. As reported in Watt’s 2009 publication put out by the Heartland Institute, the volunteers “found stations located next to the exhaust fans of air conditioning units, surrounded by asphalt parking lots and roads, on blistering-hot rooftops, and near sidewalks and buildings that absorb and radiate heat.” Watts surmised that these poorly-sited stations were responsible for much of the increase in U.S. temperatures over the past century, due to “a bias trend that likely results from the thermometers being closer to buildings, asphalt, etc.” Watts concluded, “the U.S. temperature record is unreliable. And since the U.S. record is thought to be the best in the world, it follows that the global database is likely similarly compromised and unreliable”.

Figure 1. A poorly sited temperature sensor in Marysville, California, used for the USHCN. The sensor is situation right next to an asphalt parking lot, instead in the middle of a grassy field, as it is supposed to be. The sensor is also adjacent to several several air conditioners that blow their exhaust into the air nearby. Image credit: surfacestation.org.Analysis of the data disagrees with Watts’ conclusionWhile Watts’ publication by the Heartland Institute is a valuable source of information on siting problems of the U.S. network of weather stations, the publication did not undergo peer-review — the process whereby three anonymous scientists who are experts in the field review a manuscript submitted for publication, and offer criticisms on the scientific validity of the results, resulting in revisions to the original paper or outright rejection. The Heartland Institute is an advocacy organization that accepts money from corporate benefactors such as the tobacco industry and fossil fuel industry, and publishes non-peer reviewed science that inevitably supports the interests of the groups paying for the studies. Watts did not actually analyze the data to see if taking out the poorly sited surface stations would have a significant impact on the observed 1.1°F increase in U.S. temperatures over the past century. His study would never have been publishable in a peer-reviewed scientific journal.

Figure 2. Annual average maximum and minimum unadjusted temperature change calculated using (c) maximum and (d) minimum temperatures from good and poor exposure sites (Menne 2010). Poor sites showed a cooler maximum temperature compared to good sites. For minimum temperature, the poor sites were slightly warmer. The net effect was a cool bias in poorly sited stations. The dashed lines are for stations ranked by NOAA, while the solid lines are for the stations ranked by surfacestations.org.

Fortunately, a proper analysis of the impact of these poorly-sited surface stations on the U.S. historical temperature record has now been done by Dr. Matthew Menne and co-authors at NOAA’s National Climatic Data Center (NCDC). In a talk at last week’s 90th Annual Meeting of the American Meteorological Society, Dr. Menne reported the results of their new paper just accepted for publication in the Journal of Geophysical Research titled, On the reliability of the U.S. Surface Temperature Record. Dr. Menne’s study split the U.S. surface stations into two categories: good (rating 1 or 2) and bad (ratings 3, 4 or 5). They performed the analysis using both the rating provided by surfacestations.org, and from an independent rating provided by NOAA personnel. In general, the NOAA-provided ratings coincided with the ratings given by surfacestations.org. Of the NOAA-rated stations, only 71 stations fell into the “good” siting category, while 454 fell into the “bad” category. According to the authors, though, “the sites with good exposure, though small in number, are reasonably well distributed across the country and, as shown by Vose and Menne [2004], are of sufficient density to obtain a robust estimate of the CONUS average”. Dr. Menne’s study computed the average daily minimum and maximum temperatures from the good sites and poor sites. The results were surprising. While the poor sites had a slightly warmer average minimum temperature than the good sites (by 0.03°C), the average maximum temperature measured at the poor sites was significantly cooler (by 0.14°C) than the good sites. As a result, overall average temperatures measured at the poor sites were cooler than the good sites. This is the opposite of the conclusion reached by Anthony Watts in his 2009 Heartland Institute publication.

Advertisement

Why did the poorly sited stations measure cooler temperatures?The reason why the poorly-sites stations measured cooler temperatures lies in the predominant types of thermometers used at the two types of sites. An electronic Maximum/Minimum Temperature System (MMTS) is used at 75% of the poor sites. These MMTS sensors are attached by cable to an indoor readout device, and are consequently limited by cable length as to how far they can be sited from the building housing the indoor readout device. As a result, they are often located close to heated buildings, paved surfaces, air conditioner exhausts, etc. It turns out that these MMTS thermometers have a flaw that causes them to measure minimum temperatures that are slightly too warm, and maximum temperatures that are considerably too cool, leading to an overall cool bias in measured average temperatures. In contrast, only 30% of the “good” sites used the MMTS sensors. The “good” sites predominantly used Liquid in Glass (LiG) thermometers housed in wooden shelters that were more easily located further from the buildings where the observers worked. Since the poorly-sites stations were dominantly equipped with MMTS thermometers, they tended to measure temperatures that were too cool, despite their poor siting.

Figure 3. Comparison of U.S. average annual (a) maximum and (b) minimum temperatures calculated using USHCN version 2 temperatures. Temperatures were adjusted to correct for changes in instrumentation, station relocations, and changes in the time of observation, making the trend from good sites show close agreement with poor sites. Good and poor site ratings are based on surfacestations.org. For comparison, the data between 2004–2008 taken by the new high-quality U.S. Climate Reference Network (USCRN, black dashed line) is shown, and displays excellent agreement for that time period. Image credit: Menne 2010.Independent verification of recent USHCN annual temperaturesClearly, the siting of many of the surface stations used to track climate change in the U.S. is not good. To address this issue, in 2004 NOAA created the U.S. Climate Reference Network, a collection of 114 stations in the continental United States for the express purpose of detecting the national signal of climate change. The stations were sited and instrumented with climate studies in mind, and can provide an extremely high-quality independent check on the old USHCN network. Each of 114 stations at 107 locations (some stations were installed as nearby pairs) is equipped with very accurate instruments in a triplicate configuration so that each measurement can be checked for internal consistency. As shown in Figure 3, the USCRN air temperature departures for 2004–2008 are extremely well aligned with those derived from the USHCN version 2 temperature data. For these five years, the the difference between the mean annual temperatures measured by the old USHCN compared to the new USCRN was just 0.03°C, with a mathematical correlation coefficient (r-squared) of 0.997. Menne et al. concluded, “This finding provides independent verification that the USHCN version 2 data are consistent with research-quality measurements taken at pristine locations and do not contain spurious trends during the recent past even if sampled exclusively at poorly sited stations. While admittedly this period of coincident observations between the networks is rather brief, the value of the USCRN as a benchmark for reducing the uncertainty of historic observations from the USHCN and other networks will only increase with time”. The authors finally concluded, “we find no evidence that the CONUS temperature trends are inflated due to poor siting”.

Crediting Anthony WattsThe surfacestations.org effort coordinated by Anthony Watts has made a valuable contribution to science, helping us better understand the nature of the errors in the U.S. historical temperature data set. In his talk last week at the AMS conference, and in the credits of his paper, Dr. Menne had some genuinely grateful comments on the efforts of Anthony Watts and the volunteers of surfacestations.org. However, as of this writing, Watts has made no mention on surfacestations.org or on wattsupwiththat.com of Dr. Menne’s study.JR: Masters is much more generous to Watts than Watts has been the entire scientific community, which he has repeatedly accused of fraud and bad faith. Watts has responded on his website, but most of the response is non-substantive, process-related whining that appears to be contradicted by reporting by Andy Revkin at DotEarth.UPDATE: Brian Angliss has a good post on the paper here, and he points out others who have written on it:

Related Posts: