Glenn Loury (Watson Institute for International and Public Affairs, Brown University) discusses climate change with Patrick T. Brown (Stanford University). Early on in the conversation Brown makes that point that climate models are simulations of physical processes, not statistical extensions of past activity into the future. Somewhat later in the conversation Glenn asks how we know climate change is implicated in the year to year changes in climate and weather events.
Brown responds (c. 40:27):
Essentially the one degree [Celsius] of warming that we've seen so far has not caused any huge signals to emerge from the noise of natural variability when it comes to a lot of these weather events. So when it comes to global average temperature, like I said before, the signal to noise ratio is much larger. We basically see a hockey-stick. We see a situation where global average temperature is far outside the envelope of natural variability.But when you start getting down to weather events like, say, hurricanes, or landfalling hurricanes, or most intense hurricanes, then you have a situation where there's decade to decade variability and year to year variability. And that variability so far is larger than any climate change signal.So you don't see any long-term huge trends. But the models that project increases in hurricane activity, they tell us that we shouldn't see the trend yet. So they're not yet being proved wrong. They accurately simulate this year to year variability and this decade to decade variability. They say the signal will emerge from the noise later in the century. But it shouldn't be observable yet necessarily.