Another Problem With The Use Of Multi-Decadal Surface Temperature Trends To Diagnose Global Warming and Cooling – An Effect Due To The Elevation Where The Temperature Measurements Are Made

There have been a series of research papers which document major unresolved problems with the use of multi-decadal trends in surface temperature to diagnose global warming and cooling (e.g. see and see and references therein).

There is an early paper of mine which documents another effect in mountainous terrain, and, more generally, wherever there are significant elevation changes over relatively short distances.

Our paper that documents a variation of mean monthly temperatures as a function of terrain height is

Pielke, R.A. and P. Mehring, 1977: Use of mesoscale climatology in mountainous terrain to improve the spatial representation of mean monthly temperatures. Mon. Wea. Rev., 105, 108-112.

As shown in Table 1 of this paper, the variation within the year is estimated as ranging from -5.23C per kilometer in January to -6.61C per kilometer in July; a difference of 1.38C per kilometer of elevation change. If a station were moved an elevation change of just 100m, , for example, a monthly mean temperature change of 0.138C between January and July would be found even if there were no other effects. This effect  has nothing to do with long term climate change but is due the existence of an elevation dependence on anomalies due the change of location of the observation site.

In terms of constructing multi-decadal assessments of  temperature trends,  even relatively small elevation changes in station locations will introduce a change in temperature trends and anomalies which is just due to the variation of the elevation of the site. The Global Historical Climate Network (and see for a post on the new USHCN Version 2 analyses)  may be able to correct to some extent for this effect in their homogenization methodology if the change in the data sharply defined, however, it is a quite complicated issue since the effect varies within the year (due to the variation of the average change of temperature with elevation during the year) and due to the variability of individual months of this change of temperature with height (see Table 1 in Pielke and Mehring, 1977 for the statistics). This will mask the magnitude of this effect.

There is a second issue. The construction of an annual average of temperature trends and anomalies includes this seasonal variability. Thus for a temperature measurement of trends and anomalies at a single height to be representative of a deeper layer of the troposphere, it must be assumed that the trends and anomalies at all elevations in complex terrain must just be shifted by an equal amount (i.e. it does not matter what elevation the measurements are made at).  That is there needs to be an elevation invariance to the trends. This requires that the  monthly mean change of temperature with elevation that is presented in Figure 1 does not change over time under larger scale climate variability and change.  This stringent assumption needs to be tested but it does not seem likely.

As the issues with the use of the surface temperature to diagnose global warming and cooling continue to mount, it is imperative that policymakers who are using the global average surface temperature anomaly (e. g. the “+2C threshold; e.g. see) recognize the data has serious issues with its quantitative robustness.