GeoffBeacon: You may want to read the comments on this post: http://neven1.typepad.com/blog/2012/09/arctic-ice-melt-20-years-of-co2-emissions.html#more and also this from Tamino: http://tamino.wordpress.com/2012/10/01/sea-ice-insolation/ While many here, including myself, think that the Arctic will be largely ice free in summer by 2020 as a result of AGW, that does not necessarily mean we accept Wadhams or the Arctic Methane group's reasoning. I for one am very dubious of a number of their claims - they are one of the few scientific groups to whom I would attach 'alarmist' as a perjorative.

Nice work Wipneus. Here's an interesting exercise: Using the uncertainty intervals, you can calculate the probablility of seeing a new record for any year in the future. By eye, for 2013 the probability of seeing a new PIOMAS record is approximately: Exponential: 90% Gompertz: 55% Linear: <1% For 2014 it is approximately: Exponential: 98% Gompertz: 80% Linear: <1%

Karl: Models are mere predictions, fiction only until later facts show how well they relate to the real world. I think that definition doesn't really capture the role of models in science. Models (in any field and of any complexity, from the motion of the planets to the X-ray crystallographic models of DNA to climate science) are attempts to map physical laws onto observable systems. Failure to model the observable system may arise from either the laws being wrong or the model being inadequate. When it comes to climate science, models exist on a huge range of scales and complexities, from zero-dimensional models relating solar irradiance, albedo and greenhouse forcings to temperature, to detailed three dimensional models of the time evolution of climate system. The former have survived many severe tests in explaining most of the past billion years of the earth's climate, including the glacial cycle and 20thC climate (and temperature projections into the future). See http://www.agu.org/meetings/fm09/lectures/lecture_videos/A23A.shtml>this lecture for more. The latter have strengths and weaknesses - for example they can reproduce the glacial cycle and recent models can deduce the existence of the El Nino cycle from first principles, but as we know they are not so good at explaining the loss of Arctic ice.

Neven - your question of whether the models can catch up is actually about the mindset of the modellers. A simple extrapolation of observed ASI extent loss by 2000 would have given an incomparably more accurate forecast than the median of the AR4 projections - and that failure was the modellers' choice, not the available maths. Putting orthodox theory before observation seems to me a pretty poor sort of science, and given the continuing kant about summer ice loss 'perhaps by 2030 to 2050', it would appear that doctrine is pretty impervious to blatant failure. Lewis: I don't think that's fair. Not training the models on 20thC climate is a deliberate choice by the modellers, because if they were to train the models on 20thC climate, then the ability to use 20thC climate as a benchmark of the models would be lost. To illustrate, a GCM could probably be persuaded to give earlier loss of sea ice by tweaking it to increase the ocean currents. However, if the real reason the model was failing was because of a different unmodelled feedback, by introducing a wrong correction you would have destroyed the ability of the data to detect and correct the error. The approach of the modellers seems to me to be both rigorous and correct, and thus the data is able to do what it is supposed to - critique the models.

If Lennartvdl's link is actually Wadham's calculation, then I'm afraid this may be an own goal. From the Arctic News article: Professor Wadhams estimates the present summer area of sea ice at 4 million square km, with a summer albedo of about 0.60 (surface covered with melt pools). When the sea ice disappears, this is replaced by open water with an albedo of about 0.10. From Atmosphere, Weather and Climate Written By: Chorley, Roger G. Barry and Richard J, chapter 3: A large proportion of the incoming solar radiation is reflected back into the atmosphere without heating the earth’s surface. The proportion depends upon the type of surface (see Table 3.2). A sea surface reflects very little unless the angle of incidence of the sun’s rays is large. The albedo for a calm water surface is only 2 to 3 percent for a solar elevation angle exceeding 60° , but is more than 50 percent when the angle is 15° .