Share
4,188 Posts.
lightbulb Created with Sketch. 3
clock Created with Sketch.
19/03/18
10:16
Share
Originally posted by tristanc
↑
In the tropics, it’s quite clear that the trend is essentially unchanged. In the higher latitudes, it’s less clear. The NH 50 and 30 mbar signals in particular “look” quite flat, I’d agree.
Remember, though, despite the incessant claims of “no significant warming/cooling since...”, that’s not the only statistical question you can ask (and one of the less useful ones at that). I’d argue that a far more important question is, “is the trend since xxx significantly different from the long-term trend?”
These sorts of analyses are difficult to do properly. They’re bedevilled by all sorts of statistical traps: when you have a choice of many starting and/or endpoints, your chances of getting a meaningless “significant” trend from purely random data grows very quickly; people’s tendency to fit “broken” trends with massive instantaneous step changes with no physical rationale, etc.
Without actually crunching the numbers (because I honestly don’t have the time to do it properly), I’d guess that for most of the extratropical RATPAC data above both answers will be negative: the trends since 1995 aren’t significantly different from zero, but nor are they significantly different from the trends over the full timespan.
Expand
Sure but isnt the trend over the full timespan inclusive of the 1990’s change which is due to mt pinatubo (?) so statistical correlation to the trend inclusive of that event isn’t a proof of an agw effect surely?