So, the question was, how much effect does removing the Arctic sea ice make to the rest of the world? In particular, what is the direct effect, ignoring all the feedbacks that would occur on SST and so on? Happily, an atmosphere-only GCM can answer this question, and happily I had one to hand, viz HadAM3.
Happily also it runs fairly fast nowadays – about 20 years/day on 8 opterons – so here are results from 12 years of run, compared to 50 years of control run. The difference between the two is that I’ve wiped out the Arctic sea ice (in all seasons). It gets replaced with sea at (I’d guess) -1.8 oC – considerably warmer than the sea ice there previously.
So what happens? Not all that much. Here are some zonal mean pictures, by season: top annual, bottom june-july-august (ie NH summer) and middle NH winter. Unsurprisingly the winter changes are largest; perhaps surprisingly the summer changes are small. But then the control run ice is probably not far off freezing in summer anyway. There are 50 black lines for each year of the control run, and 12 blue lines for the anomaly.
The only really obvious change is over the Arctic basin itself. And then a map of the mean change, again by season. I’ve plotted the fractional change – ie, (anomaly-control)/control. Don’t get too carried away analysing individual coloured blobs – many of them are probably not stat sig. It may as well run out for a few more years before being done for sig.
Disclaimer: all this was run off fairly quickly and there may be errors. If you see anything obvious, do let me know.
[Update: 20+ years now, and things settle down a bit. See here for another map pic. The US stuff disappears; but the drought in Australia proves robust :-) ditto some of the stuff over Africa -W]