Record heat despite a cold sun – Discussion

By Sebastian Luening, Frank Bosse and Fritz Vahrenholt


On 14th November 2016 Stefan Rahmstorf (“stefan”) of the Potsdam Institute for Climate Impact Research (PIK) published on the climate blog Realclimate an article entitled „Record heat despite a cold sun”. In this article he discusses a temperature prognosis which we first published 2012 in the book “Die kalte Sonne”. An English translation of the book came out 2013 under the title “The Neglected Sun”. In his blog post, Stefan Rahmstorf attempts to demonstrate that the solar development does not match with the temperature evolution and hence has only a negligible effect on climate. Furthermore, he argues that our temperature prognosis has essentially failed.

First of all, it is good to see that our work is being considered by a prominent climate scientist and by this has re-entered the public climate debate. Nevertheless, we disagree with the conclusions drawn by Stefan Rahmstorf and would like to take the opportunity to comment on the issues raised in his article. To this end, we address the following points:

  • Is solar development really incompatible with temperature development?
  • Does it make sense to evaluate a prognosis only a few years after it was published?
  • How did we arrive at our prognosis and why do we think it will still be successful?
  • How likely are high climate sensitivity scenarios?


1) Is solar development really incompatible with temperature development?

In his starting figure, Stefan Rahmstorf shows the time evolution of global temperature, CO2 concentration and solar activity from 1950 onwards. Unfortunately, the first part of the 20th century is not shown which would have offered interesting insights into possible climate driving mechanisms. In Figure 1 we have extended the graph to 1900 and illustrate solar activity based on a total solar irradiance (TSI) reconstruction by Steinhilber et al. (2009) based on cosmogenic radionuclide 10Be measured in ice cores. The rapid warming of the first half of the 20th century coincides well with a steady increase in solar activity. Attribution of this warming therefore is not trivial as also CO2 increased contemporaneously.

In the 1960s and 70s temperatures dropped, corresponding with a fall in solar activity while CO2 continued to climb upwards. Recent research suggests that the negative phase of 60 year ocean cycles may have been the main reason for this colder interval (Gervais, 2016; Meehl et al., 2016; Tung and Zhou, 2013). Solar activity picked up again in the 1980s/90s reaching some of the highest values, making the second half of the 20th century one of the most active solar periods of the past 10,000 years (Solanki et al., 2004).

Solar activity began to gradually decline in subsequent 11-year solar cycles in the 2000s and 2010s, as marked by the downward trend in the TSI curve by Steinhilber et al. (2009) (Fig. 1). Notably, the reduced solar activity roughly coincides with the so-called warming hiatus or slowdown that commenced around 1998. Again, ocean cycles may have played a major role in initially boosting and eventually terminating the phase of rapid warming that took place 1977-1998 (Meehl et al., 2016).

Due to the inertia of the climate system, time lags of a few decades with regards to external triggers have to be expected. The drop in solar activity during the early 21st century may therefore be only fully implemented in global temperatures in the coming years to decades, if solar activity plays a more important role than currently assumed by the IPCC. Stefan Rahmstorf’s solar representation misses the important ramp up to the exceptionally high solar plateau in the second half of the 20th century. Looking at the interval 1898 to 1997, solar activity (sensu Steinhilber et al. 2009) shows an even better (R=0.78) correlation with temperature than CO2 (R=0.75).


Figure 1: Time evolution of global temperature (GISS), CO2 concentration and solar activity (Steinhilber et al. 2009).


2) Does it make sense to evaluate a prognosis only a few years after it was published?

Climate change temperature prognoses refer to mid- and long-term developments, and do not intend to cover effects related to fast-paced El Nino, La Nina or volcanic events. It therefore does not make sense to evaluate a prognosis only four years after it was published, especially not during an El Nino year such as 2015/16. Apart from this, the temperature dataset chosen by Stefan Rahmstorf and its way of smoothing are debatable.

The GISS data have experienced repeated large administrative changes and therefore are considered by part of the climate research community as unstable (Fig. 2). Furthermore, Rahmstorf chooses averages over a 12 months period which unfortunately further emphasizes the El Nino peak. A better choice might have been monthly temperature data which by the time when Rahmstorf’s blog article went online in mid November 2014 had already returned down to values of the pre-El Nino temperature plateau. In terms of smoothing, a longer-term moving average would make better sense, e.g. 37 months, which helps to reduce some of the El Nino and La Nina short-term temperature effects. In the case of a 37 month smooth, the last smoothed data point is from mid 2015, shortening the reality-test interval of the prognosis down to merely three years.

Continue reading this article at WUWT (where it has been first published and where it can be commented on)