Essentially, if you were challenged to engineer a season for record melting, wouldn't it begin much like this?
I'd start with strong winds in November blowing from the CAA to the Laptev across the pole. Or maybe that's exactly what happened this season?
I'd want a record warm May, not the average one we are having in the central Arctic. Early melt ponds to get CAB melting earlier and longer.
That's a testable hypothesis:
What do you think? Does a warm May correlate well with record setting melts?
Seeing where '07 is - it doesn't, you mean? I can agree.
However this is not proper test, nor proper hypothesis. You test one which sounds like "warm May always causes record melt". While Richard meant, i think, quite different one: "record warm May leads to greater melt during the season".
The difference is, the former assumes that record warm May is _the_ main cause of record season melt (which is obviously not true); thus your proof indeed demonstrates this one wrong. However the latter assumes that record May warmth is important, but not the main nor not the only (by far) major factor, and if other important factors (most of them) are "working" against whatever May situation was, - then the season melt can end up being very "opposite" to what May temperature could suggest for the season. Example being '07 in the graph, - cold May, record melt.
Since the original question was about the start of record-melt season, i dig "record warmth May" being indeed quite appropriate thing to start with, if one would be asked to "engineer" record season melt. As '07 demonstrated, record melt _back then_ could happen without warm May; doesn't mean the same is true now, 9 years later - after 2012 record, i mean. And in any case, to indeed engineer it, record-warmth May is very not enough.