This is the most important graph I’ve produced for this site. It shows the future of natural snow cover in Australia, which is that by around the end of this century there won’t be any. That’s none at all.
A series of posts describing this graph and its derivation culminates here.
What is the fit like for a higher order polynomial trend?
Well you could do this (third order fit to just the extended moving average), but that would be pretty silly:
I argued at length here why a quadratic fit is the appropriate choice for a case with clear acceleration. I think our peak snow depth series are that. Then, of course, I broke that rule here by using a fourth order fit; but with the caveat, “certainly not a great choice technically”. It’s used there because it’s a nice visual fit to the early near stasis followed by rapidly increasing change.
Don’t you think it is about time to check the data – and discard the trends?
http://www.snowyhydro.com.au/our-energy/water/inflows/snow-depths-calculator/
Anyone who thinks that climate is not cyclical should look at the historic data IMHO
Your page and graphs are great, thanks. This ‘analysis’ is a bit silly though. Why wouldn’t you do some tests with a null that the peak depth is a stationary timeseries with no structural breaks? If you can reject that at a reasonable confidence level (?) then start to think about alternatives. ref Don Andrews work on detecting structural breaks at unknown breakpoints or anything similar.