Spencers Creek peak snow depth cycles

Spencers Creek peak snow depth cycles

Snow depths have been recorded weekly by Snowy Hydro Limited at Spencers Creek near Charlotte Pass, Australia since 1954. This shows the Fourier transform of the season peak depth record, to look for the cycles. Fourier transforms mathematically decompose a signal into its component frequencies or cycles. The relative signal strength at each cycle length indicates the extent to which a cycle of that length is present in the data.

Most of the apparent signal is, of course, just from random variation. For some context, I add background points from Fourier analysis of 100 randomly generated depth series with similar statistics to the Spencers Creek record, but with no cycles. The real data plots well within the “random” range except right at 2 years; so, to the extent there is any cycle at all, it’s a very weak 2 year one (perhaps with a weak harmonic at 4 years).

I don’t know of any physical reason why a 2-year cycle should exist, so I suspect that it, too, is just the result of chance. Probably the data contain no observable cycles.

2 comments to Spencers Creek peak snow depth cycles

  • Dave12

    Hey Gerg, just been readong on the effect of sunspots on the Rhein freezing and skating races in the Netherlands. Also noticed that these sunpots seem to coincide somewhat with good snow seasons here but the opposite. That is, low sunspot activity leads to cold winters in Europe, but here it seems to be high sunspot activity coincides with good snow seasons (2000-2004, 1990-1992, 1981 (but nothing either side), the early 70’s and early 60’s). There is also Chalet history that the late 40’s had some amazing deep snow depths. It’s an 11 year cycle approx and your Fourier thing above (of which I know nothing at all) shows a peak at 9-11 years. Have you seen any study done on this correlation?

    • Gerg

      As I said to another who pointed out that weak peak at 11 years — 13 of 100 random sequences plot higher. Which of course doesn’t prove that the peak is the result of chance, but it does mean that it’s unlikely to be statistically significant at any meaningful level.

      I come from a place with a long history of nutty sunspot-based weather prediction (google up Lennox Walker), so I guess I’ve always been an automatic sceptic. The difference in sun output across the 11 year solar cycle is small, and the inertia of our climate systems is large. I notice, though, that Ken Green’s latest paper includes some interesting snow depth vs sunspot analysis for Australia (http://instaar.colorado.edu/AAAR/journal_issues/abstract.php?id=2896 – paywalled, unfortunately).

Leave a Reply