The Medieval Warm Period seems to play the role of climate porn, attracting lots of attention but actually distracting from more interesting periods in the recent past climate, from which we could learn more about the future.
This is course not totally true. The level of temperatures in the Medieval Warm Period (WMP) could be helpful to estimate how robust the Greenland Ice sheet really is, one of the main uncertainties for the projections of sea-level rise for the next centuries. Or to be assured that certain ecosystems, can or cannot, tolerate temperatures warmer than the ones we are experiencing today. However, for these purposes the MWP is not that critical, as possible it was not the warmest period in the Holocene. Most reconstructions and model results indicate that the Midholocene Optimum was warmer than the MWP, and even more so at high latitudes in summer in the Northern Hemisphere. This is not surprising since the orbital configuration of the Earth was also different. But to help us understand the functioning of the Earth's climate and global scales, including the thorny issue of the climate sensitivity, the MWP is not really relevant and even less so to constrain the skill of climate models. To illustrate my point, let us have a look at the following picture displaying the Northern hemisphere mean annual temperature in climate simulation over the past millennium until the end of this century.
The details of the temperature evolution is of course dependent on many factors that are burdened with high uncertainty, and I am not claiming that this temperature evolution represents the truth. For instance, the future scenario of greenhouse trace gas concentrations is of course not known to be realistic. Or the past variations of solar and volcanic activity are still debated. So the temperature level in the MWP could be, say a few tenths of a degree warmer or cooler than in this simulation. But in view of this picture, it is clear that this details are not really very important for the MWP. First, we see that the projected temperature increase much higher than any possible reconstruction of the MWP may indicate. Te question of whether the MPW was warmer or cooler than today is not relevant to tell whether or not the model future projections are wrong. The level of uncertainty in the climate reconstructions for the MPW is too large to discriminate between bad models or good models: all would probably reproduce temperature levels around the 20th century mean, the same as all reconstructions.
The message here is that if one wants to prove model projections wrong, one has to … ehem prove model projections wrong and not prove that the MWP was warmer than present temperatures. The MWP may be interesting stuff for the blogosphere, but does not say much in either direction about the skill of climate models or about the theory of anthropogenic greenhouse effect.
If one is willing to prove models wrong, it is much more reasonable to focus in the climate periods that are as dissimilar as possible to today's climate and there compare models and reconstructions. Within the Holocene, this period is quite probably the Little Ice Age., when on top of that the proxy data sets are much richer and even long instrumental series may be available. In this period is when the signal-to-noise ration, i..e the amplitude of climate change relative to the uncertainties in reconstructions and modelling should be larger. The possible climate feedbacks and their spatial fingerprints would be also stronger and perhaps detectable in the proxy records. This does not mean that the task of comparing reconstructing with model simulations is easy for the LIA. The uncertainties in the external forcing - volcanoes, solar irradiance, land use.. are not that small. I do not know if the variations of solar irradiance, for instance, really oscillate with a 80 year period, but it seems that our estimation of this amplitude do oscillate with a 5-year period. Starting with the first estimations by Lean et al (1995) of a change of about 0.25% between present and the peak of the Little Ice Age, towards the much smaller changes presented more recently by Krivova et al (around 0.15%) , and ending with even more recent numbers that I have got to know from private communications, in which we would get back again to the Lean et al values, it seems that these estimations display some type of periodic boundary conditions.
Nevertheless, a climate period as recent as the LIA with a flurry of information available should help us et lead to ascertain some of the robust predictions that may not be that dependent on the exact amplitude of the changes in the external forcing. They are not even predictions that are very much depend on a particular climate model. For instance, climate models predict, on large scales, an increase of precipitation in areas that are humid now and a decrease of precipitation in areas that are presently dry. This prediction is mostly based on relatively simple reasoning involving the fact that specific humidity near the surface should increase or decrease with global temperature following the Clausius–Clapeyron equation, whereas global precipitation would change at a slower pace (Held and Soden, 2006). At first sight, there is no reason why this should not be valid for colder periods as well. Thus we would have a clear target to test climate models and reconstructions regarding the global hydrological cycle: mean high-latitude precipitation in the LIA should be lower than present, subtropical precipitation should be higher than present.