Guest post by Samuel I. Outcalt, Professor Emeritus of Physical Geography, The University of Michigan, Ann Arbor
Abstract: After noticing the strange time dependent behavior of the difference between the NOAA USHCN Average Annual Temperature data in the unadjusted RAW data. The RAW data was plotted for the 1965-2011 time period. The RAW data plot revealed the start and end of the Modern Warm Regime, which lasted from 1976 (the base of the “hockey stick”) until the onset of the 21st Century. The footprint of the regime in the area is validated by mean annual geothermal data from 3 boreholes along Trail Ridge Road in Rocky Mountain National Park.
Background: The strange behavior of the NOAA difference between the NOAA USHCN average annual temperature [AAT] and the raw data [AATRAW] at Boulder, Colorado is presented as Figure 1.
Figure 1. The trace of the RAW and adjusted Average Annual Temperatures at Boulder, Colorado display sectors of high amplitude variation interlaced with long runs of uniform adjustments.
Previous investigations of the adjusted Boulder average annual temperature data indicated that the Modern Warm Regime was not viable on the integral trace of that record. This was considered unusual as inflections of major regime events are known to produce inflections and extreme values of the integral trace in serial data (Outcalt et.al.(1997)). The integral trace of the USHCN Average Annual temperature data is presented as Figure 2.
Figure 2. The Integral Trace of the Average Annual Temperature does not indicate the onset of the Modern Warming Regime.
In summary, the RAW and Adjusted Average Temperature [ATA] data over the record length 1898-2012 displays interlaced sequences of large amplitude and uniform adjustments rather than a pattern of smooth transitions and the Adjusted Data does not even hint at the wide spread 1976 climate regime change which is a major feature of global climate data (see Figure 3).
Figure 3. The integral traces of three global climate series all show major inflections near the 1976.
The missing 1976 onset of the Modern Warming might be present in the RAW data integral trace for the 1965-2011 time period if rather heavy handed adjustments to the AAT data had masked or attenuated the inflection.
The 1965-2011 ATTRAW Data: The 1965-2011 RAW data is displayed as Figure 4.
Figure 4. The RAW data for 1965-2011.
In Figure 4. The integral trace displays a minimum in 1979 near the 1976 global climate transition and a minimum a 1999 near the end of the Global Warming Regime. Further, the 5 year running mean and integral trace display downturns in 2005 and 2008. These results appear to support the hypothesis that heavy handed adjustment had masked or attenuated the signature of the beginning and end of the Period of Modern Warming in the USHCN ATT data.
It should be noted that the method if Hurst Rescaling introduced in an paper by Outcalt et.al.(1997) is exceedingly robust and has been used by Runnals and Oke (2006) to detect weather station site moves.
The method unfortunately also detects transitions introduced by data adjustment which seems be true in the case of the Boulder Annual Average Temperature [ AAT ] data.
A further indication of the footprint of the modern warming data can be found in the thermal profiles of mean annual temperatures derived from hourly observations at probes placed in at 1 m intervals in three 6 m boreholes along Trail Ridge Road in Rocky Mountain National Park. The boreholes are approximately 39 miles northwest of the Boulder Weather Station at the Boulder Municipal Airport. These mean annual temperatures are extremely robust as they were derived from probes recording at an hourly interval for a calendar year (Janke (1911)). In addition, the annual mean temperatures were calculated at all the measurement levels by the author and dated using the thermal disturbance methods published by Terzaghi(1970). These data are presented as Figure 5.
Figure 5. The Trail Ridge Mean Annual Temperature Geothermal Profiles.
Using the inflection produced by the 1976 global regime transition at 6 meters in BH2 the overlying inflections dated at 9.3, 2.2 and 0.3 years BP. As the data were taken during a calendar year from the summer of 2010 to 2011 the inflections date from the turn of the century to 2011. The range of inflection dates may be artifacts introduced by local site effects (slope, exposure, winter snow cover, ground water migration etc..) as well as the assumption of the same apparent thermal diffusivity at all the sites.
Conclusions: The footprint of the Modern Warming Regime that was recovered in the RAW data from the Boulder Weather Station is the probable result of the heavy handed uncritical adjustment of the AAT data set. Hurst Rescaling is a powerful method for detecting regime transitions in climatic data. However, the method is also sensitive to uncorrected site moves and adjustment artifacts. The alteration of the Boulder RAW data set was so severe that it masked and attenuated both the onset and end of the Modern Warming Regime.
Another aspect of Hurst Rescaling is the difficulty of introducing false regime inflections into serial climate data. One must first alter the data integral and differentiate the integral to a synthetic data set. However, the differential would probably not remotely resemble the initial data. The perpetrator would then be faced with nearly infinite iterations to fine tune the data.
Acknowledgments:
The author is indebted to Dr. Jason Janke of a Metropolitan State College in Denver for giving him access to the rather unique borehole data from Trail Ridge Road.
References:
Janke, J (2011) personal communication.
Outcalt,S.I.,Hinkel,K.M.,Meyer,E . and Brazel,A.J.(1997) The application of Hurst rescaling to serial geophysical data. Geographical Analysis 29, 72-87.
Runnalls,K.E. and Oke,T.R.(2006) A technique to detect micro-climatic inhomogeneities in historical records of screen-level air temperature. Journal of Climate 19: 959-978.
Terzaghi,K (1970) Permafrost, J. Boston. Soc. Civil Eng. 39(1): 319-368
Section 2c, p.961 = concise, efficient explanation of Hurst Rescaling:
Runnalls, K.E.; & Oke, T.R.(2006). A technique to detect micro-climatic inhomogeneities in historical records of screen-level air temperature. Journal of Climate 19, 959-978.
http://journals.ametsoc.org/doi/pdf/10.1175/JCLI3663.1
As Paul Dennis says (February 27, 2013 at 9:51 am), center the data and do a running sum.
This method probably goes back a lot further than recorded human history. It’s useful. That’s why it has endured and why people can reinvent it intuitively without ever having seen or heard of it.
Sensible serial data exploration always looks at integrals & derivatives. There’s a good reason why differential equations usually have x, x’, & x” terms. It’s called orthogonality.
It’s not always enough to look at unsorted integrals. For example terrestrial circulation in January has a totally different configuration than in July. Why ignore something so fundamental? Just because of bad conventions? That’s not sensible. Have a look at some seasonal integrals in the appendices here:
Solar-Terrestrial Volatility Weaves
There are plenty of other examples where sorting matters. The explorer who ignores fundamentally important conditional dependencies — particularly if they are nonlinear — is guaranteed to fall victim to paradoxical misinterpretation.
common sense about paradox = rare & valuable
Whew! Thought it was me.
RE: Hurst Rescaling
What would a saw-tooth time series look like after Hurst Rescaling?
Let say that it is a drift towards warmer (Stevenson Screen paint getting old) over a couple of years, then it gets repainted.
The Hurst transform (Q) would create a scalloped shape with sharp points upward and rounded U shapes between repaintings. The bottom of the U is NOT a transition!
The first dirivative (Q’) is the just time series translated by the mean, the saw tooth.
Philip Bradley says:
February 27, 2013 at 5:12 am
1976 was when catalytic converters were mandated in the USA in all new vehicles. Other factors contributed to the decline in aerosols, and hence decreased clouds and increased solar insolation around this time, but catalytic converters caused the ‘regime shift’.
//////////////////////////////////////////////////
The introduction of catalytic converters may have been a factor, but also power generation was being cleaned up as from the 1970s and that may well have been more of a factor. Generally, because of a combination of factors, there were clearer skies and less solar dimming as from the 1970s onwards and these clear skies may have played a part in the observed late 1970s to late 1990s warming.
But what now of the pollution put out by China, India and Brazil? What evidence is there of a return to solar dimming?
ferdberple says:
February 27, 2013 at 6:07 pm
What the author is saying is that heavy handed adjustments to try and improve climate records have the opposite effect. They mask what is really happening with climate.
Sort of like average temperature paints a false picture of the climate. A desert and a rain forest may have near identical average temperatures, but completely different climates. By processing the data, climate science has hidden the climate, not revealed it.
///////////////////////////////////////////////////////////////////////////////
The concluding sentence is a very perceptive comment.
I frequently hark on about the inappriateness of using averages and how this hides and distorts the real position. One thing that you can be fairly sure about is that in the real world, the average condition is rarely encountered.
It is often said that models to not do well when making predictions/projections on a regional basis, but , it is further claimed, that they are quite OK for an assessment of the global position. One reason why models do not perform well on a regional basis is that far too often averaged data is fed in. Madness, since climate is local not global (and that is why there are different climatic belts throughout the globe not just one). The only global climate is that planet Earth is presently in an inter glacial period.
This maybe easier to understand if it was translated into pidgen english and re written by Willis into a commentary of the vagarities of scientific english. I feel a little lost like Alice in wonderland.
This is a very important study. It points to key features of recent climate history – e.g. a recent termination of a warming phase, and establishes a useful method for detecting adjustment of climate records, an issue in which many here affect to have a strong interest. So folks should not attack it too much for using dense scientific language. Some scientists find it natural to convey scientific information as it is, rather than dressed up in pink lace.
Hurst is emerging as quite an important dude in climate – he gets his name on fractal analysis of climate, and now on a method to sniff out fraudulent tampering with the temperature record. Respect.