... If the models show a slowdown in the rate of decline as zero ice is approached and also the data is tending to show this recently...
Which models are you referring to? I've never seen a physics-based model with good predictive ability on this trend. True, there are models that don't work well, e.g. for the IPCC studies. If the model can't find the recent trend then it's not much use for predicting future trends.
Why do you say the data is showing a slowdown recently? If I look back at the linear fit of #97 then the residuals are all over the place and I don't see that trend. Based on the second plot, I suspect a quadratic fit would also curve downwards rather than upwards.
Lots of models. There are lots of graphs like attached below, some with many more model runs on them.
>" good predictive ability on this trend."
Clearly not they are all over the place on level of ice and also the trend. While some don't have enough data to see, all the model runs where you can see the change in trend where ice approaches zero is for the trend to get less steep as zero ice is approached.
Do we throw out all evidence because they are all over the place wrt level and trend? Or, do we say yes not much good for level or trend, but it looks like they all agree on trend in slope as zero ice is approached? So don't use them where they are bad but do use them for what they are good at, ie suggesting the change in trend as zero ice is approached declines.
>"Why do you say the data is showing a slowdown recently?"
The 4 parameter gompertz fit has a single inflection point. Whether I use Sept or April, that inflection point occurs in 2005. We have 12 years data since that inflection point. If there was only ~6 or fewer years data since the inflection point, I would be inclined to the opinion that 4 parameters might be too many parameters and I was overfitting. So just random residuals in last few years was allowing a better fit by using too many parameters. However with 12 years data past the infection point, that is too much data led and seems to me to be indicative that the data is showing a decline in the rate of decline.
See 4 parameter gompertz fit at top of this page.
>"If we're looking for a physical reason for a slowdown then I would point to the existence of a 'sanctuary region' against the North side of the CAA and Greenland where the ice hasn't melted out in any year."
Certainly wouldn't disagree with that being one physical reason, but I think there are lots of others.
I would tend to add in albedo feedback to this explanation (as well as deep water). In areas where ice moves out of the area, melting and ice movement allows albedo to drop and more sunlight energy be absorbed and this obviously helps additional melting. In contrast where ice piles up against Greenland & CAA, ice tend to move into area. So rather than melting and movement of ice causing extra area to open up, movement tend to close up areas opened by melting so it is much harder to get albedo drops to assist the melting.
If the ice retreats to a smaller area, that relevant area receives less sunlight energy so less volume melts seems quite possible. I guess this is complicated by winds bringing warmer temperature air so it isn't clear whether this accelerates or decelerates the volume of melt so maybe we need to look to the data and/or models?
Then there is oft discussed failure of MYI to make it around Beaufort gyre leading to rapid collapse of MYI to much lower proportion of the ice over a few years, but once we are down to these lower levels there is more FYI which almost completely recoveres itself each winter.
Are the people on these forums thinking through these reasonings and rejecting them because they don't believe they are significant compared to positive feedbacks they believe in? Or are they just rejecting the reasonings because it is just not exciting or they want to see catastrophic decline in sea ice or .... ?