Season 2016 roundup

The 2016 Australian snow season was a good one by modern standards, even though the peak snowpack depth of 170.5 cm¹ was well below the long term (62-year) average of 197 cm. The season was late to start — the first measured depth was on 23 June — and very late to finish (last depth 22 November). Ongoing snowfalls up high through the second half of the season kept the cover and surface in better than usual shape … but one of those managed to screw my nearly perfect pre-season peak depth prediction. The eventual peak on 6 October was the second latest ever recorded, just behind 1962 at 8 October.

Spencers Creek snow depth, from Snowy Hydro

It’s worth noting that at lower elevations the 2016 season was far less impressive. Here is the record from Deep Creek, south of Cabramurra, NSW, at ~1620 m elevation.

Deep Creek snow depth

Deep Creek snow depth, from Snowy Hydro


The prediction

I predict the season peak snow depth at Spencers Creek using a simple seven-parameter multiple regression model based on well-known climatic modes and influences. My pre-season prediction for 2016, made in late May, was for a peak depth of 159 ± 44 cm. For a while that looked like it might hit to within a couple of centimetres … until that pesky near half-metre fall in early October. It pushed the peak up to 170.5 cm, still well within the prediction error margin.

The pre-season prediction has to use projections (guesses!) for many of the model parameters. In assessing how the model went, it’s fairer to look at what it would have predicted if it had the actual parameter outcomes for the season. Done that way (called ‘hindcasting’), the model peak depth prediction would have jumped to 209.6 cm, largely on the strength of lower than expected local sea surface temperatures (0.26°C lower — a lot!) and a slightly lower winter average Antarctic Oscillation Index. The long term hindcasted ‘prediction’ vs performance looks like this:

Spencers Creek season peak snow depth – hindcast ‘prediction’ vs performance



For me the stand-out influence on 2016 — apart from those unexpectedly ‘low’ sea surface temperatures — was the negative Indian Ocean Dipole; nIOD in the jargon. Qualified commentators were calling the 2016 nIOD event the strongest on record, except they didn’t always mention that they were referring to a rather short Bureau of Meteorology official record. Here is the Japan Marine-Earth Science agency’s IOD chart; you can see that there have been as large (or larger, on some measures) nIOD events in the fairly recent past.

JAMSTEC Indian Ocean ‘Dipole Mode Index’ (DMI) record

The Indian Ocean Dipole is an east-west, ENSO-like variation in the tropical Indian Ocean, which correlates with winter precipitation across southern Australia, including with alpine snow. The negative phase positions warmer than normal tropical waters north-west of Australia, from which streams of tropical moisture can feed south into the circumpolar train of winter weather systems. From about July 2016, the nIOD seems to have provoked a series of what I call north-westerly feed events (more correctly ‘north-west cloud bands’). Here is the impressive Andrew Miskelly’s view of just one of them (click for video link):

A 2016 ‘north-westerly feed’ event.

These tropical feed events seem often to be marginal on temperature, bringing snow to higher elevations but only rain to mid and lower levels. That may partly explain the poor lower level snow season in 2016.


The charts

The updated full records:

Snowy Hydro Australian snow depth records

Snowy Hydro Australian snow depth records

Snow depth record

Spencers Creek snow depth record

Spencers Creek snow depth anomalies

Spencers Creek snow depth anomalies


The season peak and integral depth:

Peak depth trend

Spencers Creek peak snow depth trend

Spencers Creek peak snow depth timing

Spencers Creek peak snow depth timing

Spencers Creek integral depth trend

Spencers Creek integral snow depth trend


The decades, and the future:

Spencers Creek snow depths by decade

Spencers Creek snow depths by decade

Future trend of Australian snow depth

Spencers Creek snow depth future


The correlations (2016 highlighted):

Spencers Creek sea surface temperature correlation

Spencers Creek sea surface temperature correlation


The dumps:

Spencers Creek "dumps"

When the dumps happen



1. Snow depth at ~1830 m elevation at Spencers Creek near Charlotte Pass, NSW, midway between Perisher Valley and Thredbo; data from Snowy Hydro Limited.