by Judith Curry
A discussion of Section 6.1 of Alan Longhurst’s book Doubt and Certainty in Climate Science.
In the original thread, numerous commenters were critical of Chapter 4 Can a global mean temperature be measured? Personally, I find Chapter 6 to be much more insightful: Chapter 6 Regional patterns of temperature change over land surfaces.
Here are excerpts from Section 6.1 Regional anomalies in the evolution of SAT during the 20th century. My excerpts focus on the U.S.; read the entire chapter for a broader perspective (and also the relevant figures).
A really extraordinary press release issued by NASA GISS in 2006 suggested that: “it must now be recognised that the USA is the sole region that did not warm progressively during the 20th century “.294 This was based on a paper authored by James Hansen who noted that “The U.S. mean temperature has now reached a level comparable to that of the 1930s, while the global temperature is now far above the levels earlier in the century. The successive periods of global warming (1900-‐1940), cooling (1940-‐1965), and warming (1965-‐2000) in the 20th century show distinctive patterns of temperature change suggestive of roles for both climate forcings and dynamical variability”. This statement had its origins in a reworking of the Goddard SAT data, corrected by satellite night lights for the urban effect) that came to conclusion that has since disappeared from view: during the 20th century the surface air temperature history of the contiguous United States differed very significantly from the rest of the world. The US data suggest that the end of the 20th century was no warmer than the warm 1940s, while the data representing the ‘rest of the world’ exhibit a pattern that closely resembles the iconic global pattern.
This analysis faces us with an absurd choice: either (i) to suppose that the radiative effect of CO2 in the atmosphere somehow selectively excludes the United States, or (ii) to suppose that the USA is the sole region for which a reasonable correction for the urban and other anthropogenic effects can be made. Of course, the choice is not hard to make.
Regional analysis with rural data alone is easily done in the conterminous USA which is a unique region in the high proportion of rural compared with urban locations: at a randomly-‐chosen position in Kansas, within a radius of 255 km there are 22 rural stations (with data mostly starting around 1905-‐1910 and continuing to the present), 5 small towns and one city of 550,000 inhabitants. Elsewhere, data for many regions include very few rural stations indeed: within 350 kms of Lyons in eastern France there are 3 cities of a million or more, and only 11 rural sites, of which 6 are at high altitude (Pic du Midi, and so on) while the other 4 are in Switzerland or on the north Italian plain. One ‘rural’ site is the little town in which the international airport of Milan is sited, directly adjacent to an aggregation of 1.3 million people! Data from these two groups of stations would match the difference between US and global data patterns.
Despite such problems, I have assembled several regional groups of rural 120 stations which show that there was no unique global pattern of temperature change during the 20th century over all land surfaces: rather, there are characteristic regional patterns that may be tentatively associated with characteristics of regional circulation patterns in the atmosphere and ocean; these patterns are not seen in the standard regional archives that are dominated by observations made in urban sites, or in regions where farming practices have changed during the 20th century, and where the natural vegetation has been progressively destroyed.
Progressively towards the northeastern USA, this symmetrical pattern changes so that stronger warming occurs at the end of the century, and this dominates rural data from regions north and northeast of the Great Lakes, as seen (below) in Canadian rural stations from the Quebec-‐Ontario borders.
This pattern, with stronger warming in the final decades of the 20th century than rural stations in the western and southern USA, is “very largely attributable to unforced natural causes”, according to a recent study.295 This conclusion is based on analyses that suggest that this pattern is closely associated with strong negative trend exhibited by the NAO during the last 30 years or so. Experiments using CMIP296 techniques with prescribed anthropogenic forcing alone fail to reproduce the circulation changes associated with NAO, nor yet the associated tropospheric warming.
Observations show that annual mean geopotential heights in the upper troposphere have increased in the arctic zone since about 1980, especially in the NE Canada-‐Greenland sector, an increase unlikely to be due to changes in surface temperature, but perhaps rather associated with negative values of the NAO, itself associated with the PDO signature in the North Pacific and with the wave-‐train pattern that links this region with positive trends in Canada and Greenland and also with circulation anomalies over the North Atlantic. This mechanism, it is suggested, accounts for half of the recent warming in this region – but, it should be noted, the ‘recent warming’ referred to here is what is indicated by instrumental data that have not been filtered for the urban effect.
The regional pattern that is characteristic of eastern Canada shown in the previous plot may usefully be termed the North Atlantic pattern, since it matches the pattern of the NAO and also is characteristic of stations around the coasts of that ocean; it also matches the pattern of the Atlantic Meridional Oscillation (AMO) that codes the relative temperatures of the North and the South Atlantic Ocean basins (p. NN).
Clearly, the regional pattern of warming at rural locations during the 20th century is complex and it is not easy to discern a single global pattern for this change. Of course, a single model is not what one should expect, given the shifting but repetitive pattern of atmospheric pressure systems (and hence of wind direction and the transport of heat) at global scale. But this informal survey does suggest that the mid-‐century warming, and the subsequent period of cooling, was prominent across much of the northern hemisphere from North America, east of the Sierra, to western Russia. It also suggests that the prime influence on climate pattern of this region is centered in the Atlantic Ocean -‐ as you would expect to be the case from the pattern of ocean currents in the North Atlantic. Alternation between the two extreme states of the NAO, with a characteristic frequency that matches the solar 60-‐80 year cycle, appears to control climate modes over his vast region.
It is also clear from these plots that the iconic temperature curve for the entire globe discussed at the beginning of this chapter conceals the existence of regional differences that are very significant. It should also be noted that the progression of warming in the surface temperature data in the GHCN departs significantly from the change in the troposphere, obtained from satellite MSU sensors and discussed in Chapter 4. Even though these data for the troposphere are available only since the 1970s, they must be the best indicator of the real radiative effect of CO2 in the atmosphere but, despite this, they are seldom discussed in analyses of changing global temperatures.
Despite the regional patterns discussed above, we have to deal with the fact that some version of the global plot of SAT data, with or without the inclusion of SST data for the oceans, has become the iconic pattern that defines anthropogenic global warming, and has been widely disseminated in simplified form; this has the very serious consequence that this is now the pattern to which modelling results must conform if they are intended to simulate the radiative effects of carbon dioxide in the atmosphere.
This brief and informal survey of rural station data appears to demonstrate conclusively that the global plot of SAT offered by Hansen and shown above cannot be considered as representative of the reaction of the lower atmosphere globally to the release of anthropogenic CO2 -‐ although that is how it is presented. This is a critical issue, because it is implausible that somehow the USA should be exempt from anthropogenic global warming; the only reasonable interpretation of the plots is that the difference is driven rather by the quality of the data than by the state of the planet: it will not be forgotten that the USA has far and away the best coverage of station data, predominantly from rural stations, and that it is better processed and curated than in most of the remainder of the globe.
For reference, I point to these previous posts:
I have long been a proponent of a regional focus on climate change, both to support scientific understanding and support policy making (see my talk from the climate adaptation workshop).
Read the rest of Chapter 6 also; there are some really good insights throughout the chapter.
Moderation note: I will treat this as a guest post: keep your comments relevant and civil. Alan is reading the comments, and will participate if the discussion is productive. Of course criticisms are expected and welcome, but please avoid ‘piling on’ and try to keep the discussion productive.