Percent snow

What percentage of winter precipitation falls as snow in Australian snowfields? One thing that’s obviously different so far about season 2013 (vs say the miserable 2006) is the proportion of rain vs snow. By the end of July there’d been nearly 500 mm of winter precipitation at Thredbo Village for just 53 cm of snow on the ground at the nearby Spencers Creek snow course. How unusual is that?

Information on precipitation type in the Australian snowfields is surprisingly scarce. The Hennessy study provided some estimates for Mt Hotham from their snowpack model, generally in the range 80 – 85% snow (15 – 20% rain). Beyond that, you could go to the automatic weather station records, which measure co-incident temperature and precipitation, and apply some arbitrary cutoff like say “all snow below 1°C”. But the records there are pretty short and unreliable. For example, all except Perisher Valley froze up in the 18-22 July event and recorded no precipitation.

Here I’m taking a much simpler and cruder approach. I’m just going to compare the snow depth increase at Spencers Creek from the beginning to the end of winter (from 1 June to 1 September) with the total recorded winter precipitation. Assuming zero winter melt, and with both measures expressed as water-equivalent, that will give the winter average percentage of snow precipitation. Of course, there is winter melt, so there will be some error (melt will appear as if it was rain). Also, Snowy Hydro Limited don’t routinely publish the snowpack densities required to calculate water-equivalent snowpack depth (they do measure them). There are published average figures though (Bormann et al, 2013), which are adequate for my needs (average 1 September Spencers Creek density is ~400 kg/m³).

Rocky Valley winter precipitation gauge

Rocky Valley winter precipitation gauge – why a “keep out” sign may be a good idea…

That leaves the winter precipitation record, a surprisingly challenging thing to get right. I first tried the Long Paddock data drill, the standard way to get nice neat, reliable, interpolated historical rainfalls in Australia. Unfortunately the interpolated record it provides for Spencers Creek is not believable. Next I tried assembling my own historical record from various alpine sites, including Spencers Creek itself (there’s data from the 50s and 60s), Charlotte Pass (some early data there looks OK), Guthega Dam and Thredbo Village.  (The pre-AWS Perisher Valley records mostly look unreliable.) The trouble with that effort was the resulting record is simply not tenable — in most years there’s much more snow accumulated on the ground than precipitation recorded!

Accurate measurement of snow precipitation is difficult. A common approach is to use a standard raingauge with a heated cone, so any snow entering it immediately melts and is recorded as water. It seems to be rather difficult to keep that power hungry heater going at a remote site all through winter. More important still is wind. Snow is obviously much more affected by wind than rain is, and in some conditions very little may enter the cone and get recorded. There are designs that address the wind problem but I’ve yet to see one in use in Australia. [Update: Some have been used for the Snowy Hydro cloud seeding studies (photo, photo).]

In my search for a more reliable record I eventually found that AGL Hydro’s record at Rocky Valley Dam near Falls Creek appears much more satisfactory, for reasons I don’t yet understand (aspect perhaps?).  The winter record there only goes back to 1961, so I padded the earlier part with scaled Spencers Creek precipitation data [x1.8(!!) based on the overlap].





Applying the AGL Hydro winter precipitation record to the Spencers Creek snow depths yields generally believable snow percentages, in line with Hennessy et al:


The two sites are ~100 km apart of course, so this is nothing more than a crude indicator; a rough guide. But you can see straight away that June-July 2013 was unusual but not unprecedented. There was a similar winter way back in 1973, a very wet period in our hydrological history.

4 comments to Percent snow

  • Warrie

    Caught red or is that white handed at the rainguage. LOL. Questions and comments. 1.How can you arrive at over 100% snowfall?? 2. At 103 cm on the 8th August it looks like only 5 or 6 years at Spencers are worse. 3.I thought Snowy Hydro in measuring snow depths also obtained % water each week as well. Thus they can keep their finger on the pulse of likely runoff and so plan ahead for generation etc. You need to get a hold of that data – perhaps a la Kepler? 4. Selwyn stayed at 2.7 cm which seems a lucky coincidence or a glitch. Reading taken on 7th not 8th. Cheers… W

    • Gerg

      1. Density wrong or precip measurement error … or just different weather at Falls and Spencers. It’s meant to be just a rough guide.
      2. There’s 13 less than 103.2 at 8 August on the interpolated figures. Not all that rare. Maybe 170 cm is still possible?
      3. Yep, they measure it every week. Maybe need to chase contacts there, but not really worth it for this exercise.
      4. Yeah, odd … but happens, presumably. Must gather that Three Mile Dam record, a better reason to go chasing the Hydro. (It’s a pain extracting it from the website.)

  • Warrie

    Hmmm, just checked WZ for TBo village – 284.3mm for July which is a bit short of 500!! And they are claiming a min of -18 last Sat the 3rd. WZ have had a dogs breakfast for the NSW alpine with a muddle of C Pass mixed in with TBo need I go on…. W

  • Warrie

    Whoops, missed that word “winter”. So it’s 488.6mm for June and July for the 53cm of snow. But there’s just too many periods of rain and melting/sublimation in 61 days. Nearby PB had 445.6mm and is slightly closer to the snowdepth course. Maybe analyse just a two week period. At least when the TBo AWS gives a pptn reading it
    means rain and you could cross reference it with PB…… W

Leave a Reply

You can use these HTML tags

<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <s> <strike> <strong>