Hennessy et al, 10 years on

The definitive report on the effect of climate change on Australian snow is undoubtedly the CSIRO/ANU study of Hennessy et al (The impact of climate change on snow conditions in mainland Australia, K Hennessy, P Whetton, I Smith, J Bathols, M Hutchinson and J Sharples, CSIRO Atmospheric Research, August 2003). It’s much-cited, and formed the basis of the snow section (5.2.6) in the authoritative Climate Change in Australia report of 2007.

Amazingly the Hennessy report is 10 years old next month. Time to check progress…


Hennessy has always seemed pretty pessimistic, most obviously in this often reproduced set of maps (slightly edited from Hennessy et al Figure 9):

Hennessy Figure 9

Wow, hardly any natural snow by 2050!  Actually that’s a complete misreading of the report’s findings.  We need to understand what the study team did, and said.

The Hennessy study used a high-resolution spatially distributed snowpack model — representing snow accumulation and melt — to estimate time-varying snowpack fields for historical and future climate scenarios.  The model operated on 2.5 km² grid and used a sophisticated interpolation scheme to estimate climatic inputs for each grid square through the full time history.  For future scenarios, the challenge then was to do a meaningful regional climate change prediction, something that even today, 10 years on, remains problematic.  A common approach is to use some sort of downscaling from global climate model output, a purpose for which most GCMs were never intended.  The Hennessy study used a 2001 regionalisation done by CSIRO Atmospheric Research based on multiple GCM results (including their own) for the IPCC SRES emission scenarios.

So far, so good; logical and sensible.  The challenge with interpreting Hennessy begins with the way they presented the results.  For whatever reason they chose to provide just two estimates, intended as lower and upper bounds to probable outcomes (at claimed 95% confidence):

  1. A “low” climate change scenario, using a combination of the lowest predicted warming and the smallest predicted precipitation ‘decrease’ (actually a small increase)
  2. A “high” climate change scenario, using the highest predicted warming combined with the greatest predicted precipitation decrease.

Unfortunately for interpreting the results, neither of those is any sort of likely outcome — they’re meant to be upper and lower bounds — so neither is that “2050 high” snow depth map that everyone gasps at.  The connection of the “low” and “high” scenarios with the IPCC scenarios is also somewhat opaque in the report.  It’s apparent from Appendix 2 that the “low” and “high” temperature scenarios are based on the IPCC TAR (2001) range of projections (and so far we appear to be tracking near the high end of those).  But how the precipitation projections fit with IPCC is less clear.

Before we get to prediction vs performance, there’s a couple more quirks.  The CSIRO/ANU team of course compared their snowpack model with historical observations and it performed pretty well (their Appendix 1).  Unfortunately they found that about the worst performance was at our most interesting snow depth record, Spencers Creek.  The reason is unclear. (It might be that the Spencers Creek snow course over-reports the area’s snow depth due to some local effect.)  To adjust for the snowpack model miss-fit, I’ve multiplied the observed Spencers Creek depths here by 0.9.  The second quirk is that Hennessy et al were primarily interested in impacts on resorts, so they didn’t actually report future depths for Spencers Creek. Instead they provided results for Mt Perisher.  That is no big issue because Mt Perisher is almost the same elevation (1835 vs 1830 m) and only 4 km away to the north-east.  The depths there are likely to be similar.

So here’s how it looks, at Mt Perisher / Spencers Creek:

The background is Hennessy et al’s Figure 12.  It shows their snowpack model results averaged for three 20-year intervals centred on about 1990, 2020 and 2050; the last two for the two change scenarios, “low” and “high”.  I’ve superimposed the real Spencers Creek data for their 20 years centred near 1990 (scaled by 0.9) in pink dots.  You can see that the model doesn’t do too badly with the adjustment (the melt leg is too fast).

I’ve also plotted the Spencers Creek average for the most recent 10-and-a-bit years, which centre on about 2008 (blue dots; also scaled by 0.9).

It’s clear that by 2008 we were already past the “low” climate change prediction for 2020 and approaching the “low” change prediction for 2050.  We’re obviously not tracking anywhere near that scenario, which is hardly news given the absence of any effective emission controls. (It’s based on a low emission IPCC scenario.)  On the other hand, nearly two-thirds of the way through the first change interval (at 2008 in 1990 – 2020), we’re still nowhere near the “high” climate change prediction for 2020.  Substantial acceleration would be needed to get close to that, and at 2012 there was no sign of that in the data.

So what to make of Hennessy? I think we’re probably tracking close to their best estimate expectation, somewhere midway between “low” and “high”. The trouble is they weren’t game to actually tell us what that was.


More:  It’s worth noting that Hennessy and colleagues published further on this work in a 2008 paper*; not so much an update as further analysis of results focusing on predicted snowmaking demands. They still provided only the low and high change predictions for 2020 and 2050 — without any “best estimate” — but they did give more explanation of the basis for choosing the lower and upper bound cases.

(* Hennessy KJ, Whetton PH, Walsh K, Smith IN, Bathols JM, Hutchinson M, Sharples J (2008) “Climate change effects on snow conditions in mainland Australia and adaptation at ski resorts through snowmaking“, Clim Res 35:255-270)


2 comments to Hennessy et al, 10 years on

  • theflyingfordanglia

    Very interesting look at the dire future facing Australian ski resorts. Is there any evidence to suggest NZ will face a similar fate? What about Canadian resorts? Or the famous Niseko?
    Interested in hearing your thoughts.