This reasoning has obviously been constructed to reveal the contradiction of focusing on the level of past temperatures, for instance during the Medieval Warm period or during the Holocene, to demonstrate or refute the possible effect of CO2 on the temperatures today. Many bloggers seem to think that if the Medieval Warm Period was warmer than today, it would rule out any effect of anthropogenic CO2. This is, as the example above illustrates, absurd. Yes, our dear sceptics will retort that the 'hockey team' and in general the IPCC reports have tried to portray the relative levels of temperatures in the Medieval Warm Period relative to today as a proof of the anthropogenic greenhouse effect. This is and was incorrect as well. Scepsis etymologically means 'investigation' and thus a true sceptic would be rather interested in being informed and not so much in scoring political points against the adversary. The effect of anthropogenic CO2 can only be ascertained by comparing two situations that only differ in their CO2concentrations, but which are otherwise equal. The Mediaeval Warm period and today are two moments with different CO2 concentrations, but many other factors could have been different as well. For instance, it is not known whether solar irradiance was lower or higher, or whether volcanism was stronger or more frequent. A direct comparison and a conclusion about the effects of CO2 is thus not possible. It would be possible if we could quantify, and then factor out, the effect of all these other forcings. But this is unfortunately not yet possible. This rather long preface is related to the recent paper by Esper and colleagues. It shows that summer temperatures in Northern Scandinavia display a long-term cooling trend at millennial timescales over the last 2000 years. This was derived from the density of tree wood and from simulations with climate models. Before going into the parts of the papers that I personally find more interesting -allow me another detour. Quite a few blogs have too quickly re-interpreted this seasonal and regional temperature reconstruction as a global annual temperature record, erroneously claiming that global temperatures are declining, even during the last decades. It is a wild extrapolation to believe that Scandinavian summer temperatures are indicative of global annual temperatures. Additionally, long-term millennial trends are one thing. Another quite different thing are the shorter warm and cold excursions around this trend, as for instance the 20th century warming, which have completely different causes. A true sceptic, again, would intermediately point to those wild extrapolations or involuntary errors. Fortunately, we have a lively collection of climate blogs devoted to identify all sorts of errors and pitfalls in climate science, and I am convinced that this one will also be duly exposed by the revolutionary guardians of the internet.
Harking back to the Esper et al. paper... Was it warmer in summer in the Northern Hemisphere 2000 years ago than today ? Quite likely.. Was it even warmer in summer 6000 years ago ? I would say yes. Many other proxies indicate quite warm conditions in Northern summer at high latitudes. For instance, the Siberian tree line was closer to the Arctic coast than it is now. This is not really a surprise. Climate simulations driven by the different orbital configuration of that time display these higher summer temperatures. In the mid-Holocene, the perihelion, which is the point of the Earth's orbit that is closest to the sun occurred in July. Now it occurs in January. Thus, it is not very surprising that the shift of the perihelion from July to January caused summer temperatures in the Northern Hemisphere to display a cooling trend. Has this any relation to the anthropogenic greenhouse effect in the 20th century ? Nothing whatsoever.
Linear trends of the solar insolation caused by slow changes in the orbit of the Earth over the last 2000 years. This spatial pattern is very different from the greenhouse gas forcing, and it obviously can cause regional and seasonal temperature trends of both signs that are much larger than the annual global mean. The global mean of this 'orbital forcing' is very close to zero.
However, it does say something interesting about the impacts of warming, irrespective of what caused these higher temperatures. In a warmer Mid-Holocene 6000 years ago, when perhaps the Arctic was ice-free in the summer, polar bears obviously survived.
From my perspective, and I am sure other co-author may have other points of view - an interesting part of the paper lies elsewhere. First, it was possible for the first time to identify the long-term millennial cooling trend in summer temperatures and high latitudes in tree-ring records, with a value of about 0.3 degrees per millennium- compare this value with the observed global warming trend of about 0.12 degrees per decade . Secondly, this trend is similar to that produced by climate models, although it is not equal - and two similar climate models produce also temperature trends that are roughly similar, but again not the same. To require that two climate models perfectly agree with proxy-based reconstructed temperatures would sound as a minor niggle, but in my opinion this is one of the few situations in which we should expect a close agreement if models were perfect and the reconstructed temperatures are correct. Most of the comparison between climate reconstructions and climate simulations in the Holocene have been conducted over periods in which the external forcing is uncertain, e.g. the past millennium. In these periods the calibration of low-resolution proxies in terms of temperatures, like lake sediments, is particularly difficult . In our case the dendroclimatological records were very long, the changes in the external orbital forcing can be accurately known by astronomical calculations, the target variable is a linear trend over a long period - 2000 years. It can be argued that in this situation the agreement among models and between models and reconstructions should be particularly good. Indeed, when one looks at the large scale pattern of temperature trends simulated by two climate models, they look strikingly similar: cooling at high latitudes and warming (yes, warming!) ) over the Tropics. Yet, squinting over some regions, for instance the Mediterranean, the sign of the simulate the trend is opposite, which indicates that regional feedbacks may also influence millennial trends. Also, the magnitude of the trends in some latitude bands is larger in one of the simulations. Over Scandinavia in summer, the magnitude of the reconstructed cooling trend is larger than in the simulations, which would indicate that models may be under-sensitive to changes in the external forcing (none of our internet guardians picked up this possible interpretation). However, I would not go that far, since the nature of the orbital forcing is quite different from greenhouse gas forcing.
Linear trends of the summer temperature over the last 2000 years simulated by two climate models. The panels on the right display the mean trend for each latitudinal band.
All in all, this paper opens a few interesting questions by extending the play field in which models and high-resolution climate reconstructions can be compared, ushering in an external forcing that is accurately known.