Exceptions that disprove the AGW 'rule'

McShane &Wyner tried to create the same graph from the same data but could not. They conclude:

"Using our model, we calculate that there is a 36% posterior probability that 1998 was the warmest year over the past thousand. If we consider rolling decades, 1997-2006 is the warmest on record; our model gives an 80% chance that it was the warmest in the past thousand years. Finally, if we look at rolling thirty-year blocks, the posterior probability that the last thirty years (again, the warmest on record) were the warmest over the past thousand is 38%."[page 37]

So, even using Mann's dubious data and employing a variety of statistical methods, McShane & Wyner's model suggests that there is only an 80% chance that one recent decade was the warmest of the last 1000 years, and 1998 is most likely not the warmest year [64% against] and the last 30 year period is also unlikely to have been the warmest [62% against]. In other words, the type of weather we have now has all occurred before, and in the not too distant past when CO₂ was supposedly low.

6 McKitrick, McIntyre, Herman 2010

If the IPCC models are right about the feedbacks, we would see a hot spot 10km above the tropics. AGW theory says this should happen because more water will have been evaporated to this part of the atmosphere and would have caused rapid warming.

McKitrick et al found that the model predictions are about 4 times higher and outside the error bars of the weather balloons and satellites measurements.

McKitrick et al's findings have been replicated by Fu et al who also find a discrepancy between the models and observations about Troposphere warming, although not to the same extent as McKitrick et al do. However, in a follow-up paper, McKitrick and Vogelsang not only confirm that the predictions of warming by the models have been exaggerated but also show the small amount of recent warming was due to a natural climate shift in 1977. This climate shift has been noted by many other researchers and means global warming is playing an even smaller role then predicted by the models.

7 Anagnostopoulis, G.G., Koutsoyiannis, D., Christofides, A., Efstratiadis, A., and Mamassis, N.

If McKitrick et al shows that the IPCC global computer models can't model the present and therefore the future, Professor Demetrius Koutsoyiannis and his team show those models can't even model the past

In his 2008 paper Koutsoyiannis compared the model predictions from 1990 to 2008 and whether those models could retrospectively match the actual temperature over the past 100 years. This test of retrospectivity is called hindcasting. If a model has valid assumptions about the climatic effect of variables such as greenhouse gases, particularly CO₂, then the model should be able to match past known data.

Koutsoyiannis's 2008 paper has not had a peer reviewed rebuttal but was subject to a critique at Real Climate by Gavin Schmidt. Schmidt's criticism was 4-fold; that Koutsoyiannis uses a regional comparison, few models, real temperatures not anomalies and too short a time period.

Each of Schmidt's criticisms was either wrong or anticipated by Koutsoyiannis. The period from 1990-2008 was the period in which IPCC modeling had occurred; the IPCC had argued that regional effects from global warming would occur; model ensembles were used by Koutsoyiannis; and since the full 100 year temperature and rainfall data was used in intra-annual and 30 year periods by Koutsoyiannis anomalies were irrelevant.

In 2008 Koutsoyiannis found that while the models had some success with the monthly data all the models were "irrelevant with reality" at the 30 year climate scale.

Koutsoyiannis's 2010 paper "is a continuation and expansion of Koutsoyiannis 2008". The differences are that (a) Koutsoyiannis 2008 had tested only eight points, whereas 2010 tests 55 points for each variable; (b) 2010 examines more variables in addition to mean temperature and precipitation; and (c) 2010 compares at a large scale in addition to point scale. The large, continental scale in this case is the contiguous US.

Again Koutsoyiannis 2010 found that the models did not hindcast successfully with real data from all the 55 world regions not matching what the models produced. The models were even worse in hindcasting against the real data for the US continent.