Month: May 2017

Global Average Surface Temperature Measurement Uncertainties

By Joseph D’Aleo

The most significant uncertainties that must be dealt with to properly analyze temperature trends were compiled by Joseph D’Aleo, CCM, AMS Fellow

In this era of ever-improving technology and data systems, one would assume that measurements would be constantly improving. This is not the case with the global observing network. The world’s surface observing network had reached its golden era in the 1960s to 1980s, with more than 6,000 stations providing valuable climate information.



The number of weather stations providing data to GHCN plunged in 1990 and again in 2005 (as stations in the oversampled lower 48 states were thinned out). The sample size has fallen by over 75% from its peak in the early 1970s, and is now smaller than at any time since 1919. The collapse in sample size has increased the relative fraction of data coming from airports to 49 percent (up from about 30 percent in the 1970s). It has also reduced the average latitude of source data and removed relatively more high-altitude monitoring sites (McKitrick 2010).


We could show many regional or country examples but here is one, Canada. NOAA GHCN used only 35 of the 600 Canadian stations in 2009. Verity Jones plotted the stations from the full network rural, semi-rural and urban for Canada and the northern United States both in 1975 and again in 2009. She also marked with diamonds the stations used in the given year. Notice the good coverage in 1975 and very poor, virtually all in the south in 2009. Notice the lack of station coverage in the higher latitude Canadian region and arctic in 2009.GHAST2.png

Canadian stations used in annual analyses in 1975 and 2009 (source: Verity Jones from GHCN).

Just one thermometer remains in the database for Canada for everything north of the 65th parallel. That station is Eureka, which has been described as “The Garden Spot of the Arctic” thanks to the flora and fauna abundant around the Eureka area, more so than anywhere else in the High Arctic. Winters are frigid but summers are slightly warmer than at other places in the Canadian Arctic.

Environment Canada reported in the National Post, that there are 1,400 stations in Canada with 100 north of the Arctic Circle, where GHCN includes just one.


After the 1980s, the network suffered not only from a loss of stations but an increase in missing monthly data. To fill in these large holes, data were extrapolated from greater distances away.

Forty percent of GHCN v2 stations have at least one missing month, It reached 90% in Africa and South America.

            GHAST3.png                      Analysis and graph: Verity Jones


According to the World Meteorological Organization’s own criteria, followed by the NOAA’s National Weather Service, temperature sensors should be located on the instrument tower at 1.5 m (5 feet) above the surface of the ground. The tower should be on flat, horizontal ground surrounded by a clear surface, over grass or low vegetation kept less than 4 inches high. The tower should be at least 100 m (110 yards) from tall trees, or artificial heating or reflecting surfaces, such as buildings, concrete surfaces, and parking lots.

Very few stations meet these criteria. The modernization of weather stations in the United States replaced many human observers with instruments that initially had warm biases” (HO-83) and later cold biases (MMTS) or were designed for aviation and were not suitable for precise climate trend detection [Automates Surface Observing Systems (ASOS) and the Automated Weather Observing System (AWOS). Note the specifications required a RMSE of 0.8F and max error of 1.9F. ASOS was designed to supply key information for aviation such as ceiling visibility, wind, indications of thunder and icing. They were not designed for assessing climate.


Also, the new instrumentation was increasingly installed on unsuitable sites that did not meet the WMO’s criteria. During recent decades there has been a migration away from old instruments read by trained observers. These instruments were generally in shelters that were properly located over grassy surfaces and away from obstacles to ventilation and heat sources.

Today we have many more automated sensors (The MMTS) located on poles cabled to the electronic display in the observer’s home or office or at airports near the runway where the primary mission is aviation safety.

The installers of the MMTS instruments were often equipped with nothing more than a shovel. They were on a tight schedule and with little budget. They often encountered paved driveways or roads between the old sites and the buildings. They were in many cases forced to settle for installing the instruments close to the buildings, violating the government specifications in this or other ways.

Pielke and Davey (2005) found a majority of stations, including climate stations in eastern Colorado, did not meet WMO requirements for proper siting. They extensively documented poor siting and land-use change issues in numerous peer-reviewed papers, many summarized in the landmark paper “Unresolved issues with the assessment of multi-decadal global land surface temperature trends (2007).

In a volunteer survey project, Anthony Watts and his more than 650 volunteers at found that over 900 of the first 1,067 stations surveyed in the 1,221 station U.S. climate network did not come close to the Climate Reference Network (CRN) criteria. 90% were sited in ways that result in errors exceeding 1C according to the CRN handbook.

Only about 3% met the ideal specification for siting. They 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. They found 68 stations located at wastewater treatment plants, where the process of waste digestion causes temperatures to be higher than in surrounding areas. In fact, they found that 90% of the stations fail to meet the National Weather Service’s own siting requirements that stations must be 30 m (about 100 feet) or more away from an artificial heating or reflecting source.

The average warm bias for inappropriately-sited stations exceeded 1C using the National Weather Service’s own criteria, with which the vast majority of stations did not comply.

GHAST 5.png

In 2008, Joe D’Aleo asked NOAA’s Tom Karl about the problems with siting and about the plans for a higher quality Climate Reference Network (CRN at that time called NERON). Karl said he had presented a case for a more complete CRN network to NOAA but NOAA said it was unnecessary because they had invested in the more accurate satellite monitoring. The Climate Reference Network was capped at 114 stations and would not provide meaningful trend assessment for about 10 years.

In monthly press releases no satellite measurements are ever mentioned, although NOAA claimed that was the future of observations.


The biggest issue though to accurate measurement is urbanization. Bad siting usually enhances the warming effect. Weather data from cities as collected by meteorological stations are indisputably contaminated by urban heat-island bias and land-use changes. This contamination has to be removed or adjusted for in order to accurately identify true background climatic changes or trends.

In cities, vertical walls, steel and concrete absorb the sun’s heat and are slow to cool at night. In surrounding suburban areas (often where airports are located), commercialization and increased population densities increase the temperatures at night relative to the surrounding rural areas. More and more of the world is urbanized (population increased from 1.5 B in 1900 to over 7.1 billion today.


The EPA depicts the typical temperature distribution from city center to rural, similar to the observed minimum temperature analysis surrounding London in mid May (about a 10F difference is shown).


Oke (1973) found a village with a population of 10 has a warm bias of 0.73C, a village with 100 has a warm bias of 1.46 C, a town with a population of 1000 people has a warm bias of 2.2 C, and a large city with a million people has a warm bias of 4.4C.

Zhou et al (2005) have shown global data bases (for China) not properly adjusted for urbanization. Block (2004) showed the same problem exists in central Europe. Hinkel et al (2003) showed even the village of Barrow, Alaska with a population of 4600 has shown a warming of 3.4F in winter over surrounding rural areas, These are but a handful of the dozens of studies documenting the UHI contamination.

Most confirm the warming is predominantly at night. During the day when the atmosphere is well mixed, the urban and rural areas are much the same. This analysis by in Critchfield (1983) for urban Vienna and suburban Hohe Warte shows the temperature traces for February and July.

Screen Shot 2017-05-25 at 1.10.26 PM.pngTom Karl whose paper in 1988 defined the UHI adjustment for the first version of USHCN (which was removed in version 2) wrote with Kukla and Gavin in a 1986 paper on Urban Warming:

“MeteoSecular trends of surface air temperature computed predominantly from [urban] station data are likely to have a serious warm bias… The average difference between trends [urban siting vs. rural] amounts to an annual warming rate of 0.34°C/decade.  … The reason why the warming rate is considerably higher [may be] that the rate may have increased after the 1950s, commensurate with the large recent growth in and around airports. …

Our results and those of others show that the urban growth inhomogeneity is serious and must be taken into account when assessing the reliability of temperature records.” 

Inexplicably, the UHI adjustment Karl argued for was removed in USHCNv2.

Doug Hoyt, once chief scientist at Raytheon wrote: “It is not out of the realm of possibility that most of the twentieth century warming was urban heat islands.”



This is needed when a station is missing data for a month or months. It is accomplished using anomalies. For areas where there are adequate close-by surrounding stations, the assumptions that despite the local temperature differences, most sites will have a similar anomaly (departure from normal) is a reasonable one. But for infilling they can go as far as 1200 km (750miles) away to find data. At longer ranges this become problematic. Take for example northern Canada or the arctic where they must extrapolate over vast distances.


This adjustment that blends data for all stations was designed to detect previously undisclosed inhomogeneities (station moves or siting changes) and adjust for urbanization. It may help detect siting discontinuities but is not an adequate substitute for UHI adjustment. The rural stations if properly sited and the Climate Reference network of stations should be reference to adjust the urban stations.

Instead through homogenization the rural areas are contaminated by urban stations, Dr. Edward Long from NASA examined a set of rural and urban stations in the lower 48 states both raw and adjusted. After adjustment, the rural warming rates increased 5 fold while urban warming rates were only slightly reduced. This augmented not eliminated UHI contamination.



The other data set that presents a challenge for a precise assessment of global average surface temperature (GAST) is world’s oceans, which cover 71% of the globe.

Major questions persist about how much and when to adjust for changing coverage and measurement techniques from buckets to ship intake, to moored and drifting buoys, satellite skin temperature measurements and now ARGO diving buoys.


ARGO network of 3341 diving buoys and floats introduced in 2003 should improve the assessment going forward though NOAA chose instead to adjust them warmer based on the fact that ship intake data was typically warmer than the buoys.


For more see here and here.

See Research Report here.

Invalidating the EPA’s Endangerment Finding

Authors’ Comments on their Two Research Reports:

On the Existence of a “Tropical Hot Spot” & The Validity of EPA’s CO2 Endangerment Finding, Abridged Research Report, August 2016, and

On the Existence of a “Tropical Hot Spot” & The Validity of EPA’s CO2 Endangerment Finding, Abridged Research Report, April 2017, Second Edition

On April 29, 2017, TWTW stated the following:

“Revised Paper by Wallace, Christy, and D’Aleo: In his testimony, Christy discusses the simple statistical model used in the August {2016} paper by Wallace, Christy, and D’Aleo. At the time of Christy’s testimony, the paper was undergoing revision and made stronger {Emphasis Added}. The paper has been reviewed by several experts in relevant sciences and statistics.”

The authors would like to clarify the situation; two separate and distinct research activities were carried out, each culminating in a separately peer reviewed research report. They were each published simultaneously on many different web sites, but 7 months apart. Importantly, there have been no revisions to either research report.

Both research efforts set out to test for the Existence of a “Tropical Hot Spot” and the Validity of EPA’s CO2 Endangerment Finding. Both dealt carefully and properly with econometric simultaneous equation parameter estimation issues in the two separate structural analyses that were carried out. And, both efforts involved the same three authors. Each analyzed the same Tropical, Contiguous U.S. and Global Temperature data sets.

“The objective of this research was to determine whether or not a straightforward application of the “proper mathematical methods” would support EPA’s basic claim that CO2 is a pollutant. Stated simply, their claim is that GAST is primarily a function of four explanatory variables: Atmospheric CO2 Levels (CO2), Solar Activity (SA), Volcanic Activity (VA), and a coupled ocean-atmosphere phenomenon called the El Nino-Southern Oscillation (ENSO.)”

However, the model explanatory variables used in the two separate research activities were very different. Readers should recall frequent debates among climate scientists as to which Natural Factor explanatory variable was most important – solar or oceanic/ENSO activity. The first research effort focused on testing the explanatory power of using just ENSO variables (i.e., specifically MEI related variables) and volcanic activity and was publicly released as the August 2016 Peer Reviewed report.

The Peer Reviewed Second Edition, publicly released in April 2017, explicitly included all three Explanatory variables, that is, solar, volcanic and oceanic/ENSO activity. From a purely statistical analysis standpoint, the results were invariably excellent in both modeling exercises.

The temperature data measurements that were analyzed were taken by many different entities using balloons, satellites, buoys and various land based techniques. Needless to say, if regardless of data source, the structural analysis results are the same, the analysis findings should be considered highly credible. The fact that two separate research efforts came to the same conclusions implies that the findings should be considered quite robust.


Press Release and Research Report “On the Existence of a “Tropical Hot Spot” & The Validity of EPA’s CO2 Endangerment Finding: April 24, 2017

As a prelude to the Press release and Research Report, I have chosen to post the introduction on the press release and significance of this research report by Dr. Alan Carlin, a report reviewer and Retired Senior Analyst and manager EPA from his web site.

Second Edition of path breaking Research Report Further Shows the Scientific Invalidity of Climate Alarmism

Despite Saturday’s so-called “March for Science,” the almost simultaneous release of a second edition of a Research Report showing the exact opposite of what some of the marchers claim to be the conclusions of climate science has brought home the Orwellian reality that the marchers have gotten their claims concerning what the science says exactly backwards.  The Climate March website says their forces of “The Resistance” won’t tolerate “institutions that try to “skew, ignore, misuse or interfere with science.” If the marchers really support science, they should be supporting climate skeptics, not the climate alarmists. How Orwellian can you get? The science is clear.

The authors of a path breaking August 2016 research report released today a Second Edition of their Research Report.  The conclusions disproving the validity of USEPA’s three lines of evidence for their 2009 Endangerment Finding for Greenhouse Gases and very clearly demonstrating the lack of a statistically significant impact of increasing atmospheric levels of CO2 on global and tropical temperatures remain the same. However, the analysis process utilized is both more elegant and easier to understand. It demonstrates that Natural Factors involving solar, volcanic and oceanic activity fully explain the Earth’s tropospheric and surface temperatures. And, that CO2 plays no significant role.

Skeptics have long argued that fluctuations in global temperatures are not primarily due to human-caused emissions of CO2 from using fossil fuels to improve their lives, and have generally attributed these fluctuations to changes in the sun, our source of heat and light.  The importance of solar, and other natural factor fluctuations has now been shown to be the case despite many tens of billions of taxpayer dollars spent by the US and other governments to try to disprove the obvious and mislead the public on this central scientific issue in the climate debate.

So the new Edition does not contradict any of the conclusions reached last fall, but now provides a more understandable and common sense explanation for fluctuations in global and tropical temperatures. Nothing that U.S. EPA, the UN, or even President Obama have done, or even could have done, could have had significant effect on the Earth’s temperature.  The effect of their attempts to do so will be to line the pockets of “renewable” energy sources at the expense primarily of the less well-off the both in the US and the rest of the world and of decreasing the productivity of green plants and humans by discouraging the use of fossil fuel energy and thus CO2 emissions.

Previously climate skeptics have raised myriad reasons why reducing human CO2 emissions would have little effect on global temperatures despite arguments based on elaborate climate models that had never been proper validated. These Climate Models invariably predict that higher CO2 levels will lead to higher temperatures. The Research Report 12 separate times invalidates this assumption. It robustly invalidates the argument that reductions in CO2 emissions as advocated by the UN and the Obama Administration will have statistically significant (i.e., different from zero) effect on global temperatures. So they are a total waste of taxpayer and ratepayer dollars. And, very harmful to job creation, economic growth and the poor.



Abridged Research Report Second Edition, April 2017

A just released peer reviewed Climate Science Research Report has proven that it is all but certain that EPA’s basic claim that CO2 is a pollutant is totally false.

All research was done pro bono ICECAP NOTE: as has been the case with the numerous comment filings, letters, editorials, research reports and Amici briefs to the DC Circuit Court and SCOTUS. The team of scientists, economists and climatologists are not in for profit but we truly care about our environment and its inhabitants. We apply the seemingly forgotten scientific method and utilize rigorous statistical analysis techniques to determine the validity of politically driven claims, frivolously accepted and used to justify policies that force those who can least afford it to ride the Green Express Train to Energy Poverty.

This research failed to find that the steadily rising Atmospheric CO2 Concentrations have had a statistically significant impact on any of the 14 temperature data sets that were analyzed. The tropospheric and surface temperature data measurements that were analyzed were taken by many different entities using balloons, satellites, buoys and various land based techniques. Needless to say, if regardless of data source, the analysis results are the same, the analysis findings should be considered highly credible.

The analysis results invalidate EPA’s CO2 Endangerment Finding, including the climate models that EPA has claimed can be relied upon for policy analysis purposes. Moreover, these research results clearly demonstrate that once the solar, volcanic and oceanic activity, that is, natural factor, impacts on temperature data are accounted for, there is no “record setting” warming to be concerned about. In fact, there is no Natural Factor Adjusted Warming at all. The authors of this report claim that there is no published, peer reviewed, statistically valid proof that past increases in atmospheric CO2 concentrations have caused the officially reported rising, even claimed record setting temperatures. And, EPA’s Climate Models fail to meet this test.


From the Second Edition Report:


The objective of this research was to determine whether or not a straightforward application of the “proper mathematical methods” would support EPA’s basic claim that CO2 is a pollutant. These analysis results would appear to leave very, very little doubt but that EPA’s claim of a Tropical Hot Spot (THS), caused by rising atmospheric CO2 levels, simply does not exist in the real world. Also critically important, this analysis failed to find that the steadily rising Atmospheric CO2 Concentrations have had a statistically significant impact on any of the 14 temperature data sets that were analyzed.

The temperature data measurements that were analyzed were taken by many different entities using balloons, satellites, buoys and various land based techniques. Needless to say, if regardless of data source, the structural analysis results are the same, the analysis findings should be considered highly credible.

Thus, the analysis results invalidate each of the Three Lines of Evidence in its CO2 Endangerment Finding. Once EPA’s THS assumption is invalidated, it is obvious why the climate models EPA claims can be relied upon for policy analysis purposes, are also invalid. And, these results clearly demonstrate – 14 separate and distinct times in fact— that once just the Natural Factor impacts on temperature data are accounted for, there is no “record setting” warming to be concerned about. In fact, there is no Natural Factor Adjusted Warming at all. Moreover, over the time period analyzed, these natural factors have involved historically quite normal solar, volcanic and ENSO activity. At this point, there is no statistically valid proof that past increases in atmospheric CO2 concentrations have caused the officially reported rising, even claimed record setting temperatures.


Icecap Note:

To not be able to invalidate the hypothesis (commonly called prove) that CO2 was a major (i.e., statistically significant) driver of changes in global temperatures, what are called simultaneous equation parameter estimation techniques must be applied. If CO2 in fact does not have a significant impact, it is rather straightforward to test how much of the earth’s temperature variation can be explained by natural factors (e.g., solar, oceanic and volcanic activity.) It turns out that once these natural factor impacts are removed, the NF Adjusted Temperatures have a flat trend and bear no statistical significant relationship to CO2. These analysis results were found for 14 separate topical and global temperature data sets. This research finding leaves no room for CO2 to have any measurable impact on global atmospheric and surface temperatures. The scientific method requires that such analyses findings be fully reproducible.  And, these are.

Unlike most peer (pal) reviewed papers, the authors made available to our peer reviewers and anyone (including you) access to the data sets. Also a blueprint to the methods was explicitly described in the Preface and a full set of summary statistical results in the report. Anyone with knowledge of the proper use of regression analysis involving simultaneous equation systems can fully understand and replicate this work. However reading the Preface would lend those without simultaneous modeling experience, the ability to fully understand the findings of this work.

The Undersigned Agree with the Conclusions of this Report:

Dr. Alan Carlin
Retired Senior Analyst and manager, US Environmental Protection Agency, Washington, DC.
Author, Environmentalism Gone Mad, Stairway Press, 2015.
Ph.D., Economics, Massachusetts Institute of Technology, Cambridge, MA.
BS, Physics, California Institute of Technology, Pasadena, CA.

Dr. Theodore R. Eck
Ph.D., Economics, Michigan State University
M.A, Economics, University of Michigan
Fulbright Professor of International Economics
Former Chief Economist of Amoco Corp. and Exxon Venezuela
Advisory Board of the Gas Technology Institute and Energy Intelligence Group

Dr. Craig D. Idso
Chairman, Center for the Study of Carbon Dioxide and Global Change
Ph.D., Geography, Arizona State University
M.S., Agronomy, University of Nebraska, Lincoln
B.S., Geography, Arizona State University

Dr. Richard A. Keen
Instructor Emeritus of Atmospheric and Oceanic Sciences, University of Colorado
Ph.D., Geography/Climatology, University of Colorado
M.S., Astro-Geophysics, University of Colorado
B.A., Astronomy, Northwestern University

Dr. Anthony R. Lupo
IPCC Expert Reviewer
Professor, Atmospheric Science, University of Missouri
Ph.D., Atmospheric Science, Purdue University
M.S., Atmospheric Science, Purdue University

Dr. Thomas P. Sheahen
Ph.D., Physics, M.I.T.
B.S., Physics, M.I.T.

Dr. George T. Wolff
Former Chair EPA’s Clean Air Scientific Advisory Committee
Ph.D., Environmental Sciences, Rutgers University
M.S., Meteorology, New York University
B.S., Chemical Engineering, New Jersey Institute of Technology

This Pro Bono Research Is Dedicated to the Memory of Dr. William M. Gray (Emeritus) Professor of Atmospheric Science, Colorado State University

The authors of this research are very much interested in knowing the names and credentials of individuals who would like to add their names to the list of scientists whose names may appear in the report under the following statement:  “The Undersigned Agree with the Conclusions of this Report.”

After reading and thinking about this research report, if you would like to have your name added to the list, please send your name and credentials in a fashion similar to those listed in the August 2016 Research Report.

Please send this information to the following dedicated email address: