Season 2013 snow depth

Discussion in 'Weather' started by Gerg, May 17, 2013.

  1.  
    Gerg

    Gerg Well-Known Member

    [​IMG]

    <ul style="list-style-type: disc"><li> Spencers Creek snow depth data is from Snowy Hydro Limited.</li><li> Depths are recorded on a measuring course north of the Kosciusko Road between the Spencers Creek crossing and Charlotte Pass, at about 1830m elevation. This is a high-altitude, natural snow site (no snowmaking or grooming).</li><li> The measurement method is described <a href="http://web.archive.org/web/20120519133020/http://www.thredbo.com.au/weather-and-conditions/snow-depth-chart/measuring-snow-depth.asp" target="_blank">here</a> (archive link). Measurements are generally made every Thursday, but the series is not uniformly sampled. The first year is poorly sampled (1954), while others are heavily sampled (daily in parts of 1955-1960).</li><li> To produce a uniformly sampled series for analysis the data plotted here have been resampled every Thursday -- by zero-filling at the start and end of each season's record and gap filling by linear interpolation.</li><li> The 7-day moving averages and moving medians are plotted centred, and are lightly smoothed (Gaussian; 5-day FWHM). (The median is the middle value - where half the seasons are deeper, half thinner.)</li><li> The average and median of the annual peak depths exceed the peaks of the moving average and moving median depths because the annual peak depths do not all occur in the same week of the year.</li></ul>
    Last edited by a moderator: Dec 29, 2013
  2.  
    Gerg

    Gerg Well-Known Member

    And all this stuff:

    Full record:

    [​IMG]

    And anomalies from the mean curve:

    [​IMG]

    And the trends:

    Peak depth:

    [​IMG]

    Season integral depth:

    [​IMG]

    Season integral depth is my preferred index of season quality, because it incorporates the effects of season length as well as depth. It's the area beneath the annual snow depth curve, calculated by adding together the depths for each week of the year. Think of it as the average season depth times the season length (from zero to zero); hence that unit &quot;metre.weeks&quot;.
    Last edited by a moderator: Dec 29, 2013
  3.  
    Gerg

    Gerg Well-Known Member

    And more:

    Decades (unchanged):

    [​IMG]

    And cycles, by Fourier analysis (updated):

    [​IMG]

    For context, I add background points from Fourier analysis of 100 randomly generated depth series with similar stats to ours, but with no cycles. The real data plots well within the &quot;random&quot; range except right at 2 years. So, to the extent that there is any cycle at all, it's a very weak 2 year one.
    Last edited by a moderator: Dec 29, 2013
  4.  
    Gerg

    Gerg Well-Known Member

    And all those correlations (2012 in bold):

    [​IMG]

    Southern Oscillation Index (SOI) is the difference between Tahiti and Darwin surface atmospheric pressures expressed as monthly standard deviations x10. It is an indicator of the El Niño Southern Oscillation (ENSO), an east-west quasicycle in Pacific Ocean surface temperature and wind patterns which correlates with precipitation across much of Australia, including with alpine snow. A positive SOI is associated with more (and some say wetter) Australian snow.

    Pacific Decadal Oscillation (PDO) is a long-cycle, mostly north-south variation in the western Pacific Ocean, closely related (but not equivalent) to yet another claimed mode called the Inter-decadal Pacific Oscillation (IPO). Negative long-average PDO is weakly correlated with more snow. (2012 was the second &quot;best&quot; 2-year PDO on record...)

    Indian Ocean Dipole (IOD) is an ENSO-like variation in the smaller Indian Ocean, which correlates with precipitation across southern Australia, including with alpine snow.

    Antarctic Oscillation (AAO; also called &quot;Southern Annular Mode&quot; or SAM) is a measure of how tightly the circumpolar winds (&quot;polar vortex&quot; in one usage) blow around the pole. A loose pattern (negative AAO) leads to more polar storms reaching southern Australia, and more snow.
    Last edited by a moderator: Dec 15, 2013
  5.  
    Claude Cat

    Claude Cat Gone Fishing Moderator

    [​IMG]
    Last edited by a moderator: Nov 12, 2013
  6.  
    Doonks

    Doonks Go ahead, make my day! Moderator

    this proves global warming exists
  7.  
    Mrstr_Chief

    Mrstr_Chief Dedicated Member

    With the data you have, its quite clear the SOI, PDO, IOD, AAO by themselves don't correlate very well with snow depth, don't know why some people use it for their long range forecasting...
  8.  
    filski

    filski Part of the Furniture

    Nice work Gerg.
  9.  
    Gerg

    Gerg Well-Known Member

    True enough. You prefer, perhaps, tea leaves?
    Last edited by a moderator: Dec 13, 2013
  10.  
    Vermillion

    Vermillion Pool Room

    Moon phases.
  11.  
    Claude Cat

    Claude Cat Gone Fishing Moderator

    You been talking to Ken Ring again?
    Last edited by a moderator: Dec 13, 2013
  12.  
    Vermillion

    Vermillion Pool Room

    Me and him are like GFS and the Frog.
  13.  
    Falls expat

    Falls expat New Member

    A couple of things I would add about the teleconnections (AAO, SOI, MJO etc). If you look a pure correlation stats on these they are fairly poorly correlated. However, that is the nature of random aspect of weather. The reason they correlate poorly is often they cancel each other out. If they all line up supporting a similar pattern that is when the extreme/strong events occur. The tricky bit is determining which one is the dominant pattern driver at the time if at all.

    Take for example the MJO. I use this for forecasting weather in Europe and globally to add value to existing NWP forecasts. However, for example, if you look at pure stats on how the MJO affects W Europe the correlation is very poor.

    Take this last NH winter. If you look at the MJO in late Autumn and early winter it correlated well with known correlations (see Cassou). This enabled me to predict changes in the temperature well in advance of the NWP. However, in early to mid January, a very strong Sudden Stratospheric Warming event (SSW) occurred over the North Polar atmosphere. This suddenly became the main pattern driver which swamped the MJO signal. If you average the MJO over many years, there would be many times that the MJO would have been overridden by something else and hence the correlation stats are poor.

    This does not mean they are useless for predicting, but you have to take the teleconnections as a group and see if they favour one outcome or not.

    One last thing. If you haven't already, some of you may want to look at Atmospheric Angular Momentum as a forecasting tool as well. See here AAM . Yellow reds translate to W wind anomalies, greens/purple to E wind anomalies. This can be useful to see how the tropics interact with the mid latitudes and can be a useful tool. I think if you look at this you may be able to see why this early season trough was allowed to drive so far north to dump significant snow.

    Falls
    Last edited by a moderator: Dec 15, 2013
  14.  
    filski

    filski Part of the Furniture

    Bloody hell, good to see you post Falls!

    For anyone not aware, Falls is a long term poster and meteorologist working around the globe.
  15.  
    Falls expat

    Falls expat New Member

    Yeah I moved a bit closer to home. Now in Singapore. Doing same job. Just the wife got transferred here. Sorry for diversion away from topic.

    Has anyone looked at sunspot research yet? Some interesting finding from Scaife at the UK Met Office. Initially they thought that the increase in energy during high sunspots was negligible and therefore could not influence the global climate. However, there is a fairly good correlation between low sunspots and global temperatures in past history and his work goes some way to explaining why.

    Turns out that during high sunspot periods the sun emits up to 8 times the normal UV radiation incident on the outer atmosphere. UV is fundamental in Ozone production and warming of the upper atmosphere.

    If this correlation is true then most experts think the next sunspot cycle will be almost non-existent similar to the late 1800s which corresponded to the N Hemisphere &quot;little Ice age&quot;.

    I suspect this is good news for a ski industry that sits right on the borderline of viability especially in the background of human induced global warming. (Please no discussion on CO2 here, I am bored of it).
  16.  
    Sandy

    Sandy Dark Sith Lord of the Pool Room Moderator

    Hi Falls, good to hear from you!!!!!! [​IMG]


    With the sunspots, have you got a link for that data?

    i .e. Are they implying that one/some high sunspot activity cycle will be followed by a low activity cycle? If so, why the next cycle?
    Last edited by a moderator: Nov 12, 2013
  17.  
    Falls expat

    Falls expat New Member

    Last edited by a moderator: Dec 15, 2013
  18.  
    Sandy

    Sandy Dark Sith Lord of the Pool Room Moderator

    Excellent!!!!!!! [​IMG]
    Last edited by a moderator: Nov 12, 2013
  19.  
    Falls expat

    Falls expat New Member

    Forgot to answer your question Sandy.

    We are at or near the peak of the latest sunspot cycle (24) which is only about half that of the previous one (23). In between these two was one of the longest sunspot free periods in recent times which corresponded to a series of cold winters across Europe and Asia, but not so over the USA.

    This low sunspot period is one of the reasons people think the rapid warming associated with Greenhouse gases has leveled off in recent years although this is speculation.

    There are a lot of readings coming from the sun that are similar to the late 1800s low sunspot period and hence why they think the next cycle will be very low.
  20.  
    Sandy

    Sandy Dark Sith Lord of the Pool Room Moderator

    Great that's what I was after
  21.  
    filski

    filski Part of the Furniture

    Falls, congrats on the move, especially after the run of winters recently!

    If I may, I might challenge that assertion about recent NH winters and Solar Spots/ UV. Sudden Stratospheric Warming seems to have a good coorelation with cold outbreaks in the UK at least. It also goes some way to explaining the difference in effects noticed in northern and Southern Hemispheres. Surely being solar origin, UV would have a similar effect in the SH (even if it were lessened) to match the NH?

    I'm less clear on the causes of SSW though. Maybe you know of Judah Cohens work?
  22.  
    Falls expat

    Falls expat New Member

    Yes true about UV being global, but most ozone is produced over tropical regions anyway then migrates towards the poles. I am speculating that the southern hemisphere ozone hole may influence the UV interactions when the sun returns in spring. I also think that the the main changes due to this are over the polar vortex region which excludes large parts of the southern hemisphere land masses and therefore their observation network.

    As far as SSWs are concerned, they are are assumed to be just wave breaking of the polar jet which eventually injects larges amounts of subtropical warmth into the polar region. This then causes polar vortex to break down and even split up in the N Hemisphere. This can happen in the SH polar vortex, but it is much stronger and more stable than the NH so it doesn't happen very often or to the same extent. The NH also has more meridional transports of air due to land mass/oceanic boundaries and large mountain ranges which act to keep the polar vortex less stable and prone to breaking up.
  23.  
    BlueMountains

    BlueMountains New Member

    Great to have you around Falls expat, have been reading your posts for years and learning heaps.

    I'm predicting a slow start to the season with a nice finish. Snow depth of 1.8M.

    I don't like the look of the general patterning at present but we all know that things can change quickly. The SAM is heading north again and the LWT looks very weak/flat to me. The polar front is well south and that isn't good. Its been a virtual frontless autumn here in the Blue Mountains and our current rain total for May is just over 3mm!

    I also don't like the SST anoms in the Bight and further south and also in the Tasman. ECL development looks the go to me, but not really cold setups, more rainy at this stage.

    I'm always hopeful though and it could be going nuts by mid or late June in the ski fields, so fingers crossed.

    I've got nearly 400 followers up this way on Twitter that love their snow.

    http://www.blackheathweather.com/twitter.html
    Last edited by a moderator: Dec 15, 2013
  24.  
    Snorkler

    Snorkler Part of the Furniture

    So if the AO is -ve but the PNA is +ve, it leads to not much snow in the Sierras? [​IMG] That is the PNA overides the -ve AO Teleconnection, when the two turn -ve though we're in for a good Sierra Dump.
    Last edited by a moderator: Dec 15, 2013
  25.  
    Oldie

    Oldie Well-Known Member

    http://www.landscheidt.info/?q=node/50
    This site has a lot of info on sunspot activity in relation to the earths weather and &quot;grand minimums&quot; etc causing little ice ages. I was linked to it by a site called Ice age now, so I don't know if it is biased or not. Worth a look though.
    Last edited by a moderator: Dec 15, 2013
  26.  
    Falls expat

    Falls expat New Member

    I look at the AO and PNA virtually every day in my job and I have found that probably more than 75% of the time the PNA and AO have the opposite sign. Similar with the NAO as well having the opposite sign.
  27.  
    FourSquare04

    FourSquare04 Dedicated Member

    Falls expat, good to see you're still around [​IMG]
    Any predictions for the upcoming 2013 winter?
    Last edited by a moderator: Nov 12, 2013
  28.  
    Falls expat

    Falls expat New Member

    Not really. All depends on how the Antarctic polar vortex sets up for the start of winter and probably more importantly what happens in the tropics. I have not really looked at it to be honest. However, you need combined subtropical and polar jets to dump snow over SE Australia. Most of the time they are separated allowing high pressure to dominate.

    The following image shows the Global Atmospheric Angular Momentum plot up until the 18th May.

    [​IMG]

    Note the lines I have drawn on it showing how there has been a major shift in AAM over both the N and S Hemisphere. There is probably a seasonal change aspect to the easterly anomalies moving north but also a westerly wind burst probably induced by a recent MJO that sent westerly anomalies into southern hemisphere and this probably induced the large trough and snow event in recent days.

    Worth noting that the atmosphere right now is adopting a more El Nino like pattern with anomalous W flow over the tropics. SSTs obs right now suggest the opposite so one wonders if there is an abrupt change on either the SSTs or the global circulation coming soon to address his imbalance.
    Last edited by a moderator: Dec 15, 2013
  29.  
    BlueMountains

    BlueMountains New Member

    This abrubt change of the global circulation could be the SAM, that is forecast to drop dramatically into the negatives in June?

    This is where Falls has a lot more insight than most of us, being able to apply a broader awareness of the subtleties, beyond the usual indicators that most of us observe.

    The BoM's seasonal outlook has been released and they are going for wetter than average across good portions of Oz. Not surprising I guess with the SAM progged to drop and the IOD looking helpful as well.

    This may not be great for the ski season though, with some warmer days and nights quite possible. I do think it could usher in some big fronts by maybe June or July though, in an attempt to counter the lack of fronts in autumn.
  30.  
    PG

    PG Old And Crusty

    'Outlook confidence' for rainfall over the Alps is rubbish however.
  31.  
    Gerg

    Gerg Well-Known Member

    Revised statistical prediction model:

    Spencer's peak (cm) = 1438 - 0.61 x year - 24.8 x AAO + 0.86 x SOI - 13.2 x IOD - 10.4 x PDO - 0.17 x last year's depth


    AAO, SOI &amp; IOD are June-July-August means as before, and PDO is the 2-year mean to August. I capitulate and add the apparent correlation with last year's depth, for those enamored with cycles (implements negative lag-1 auto-correlation, a bit like a weak 2-year cycle).

    The new model explains 31% of the variance (~27% detrended), which is about 3% better than the old one. The estimate is still only good for about +/-50cm at 1-sigma (a range covering about two-thirds of years).

    The 2013 prediction for AAO 1.0, SOI 0.0, IOD -0.2 and PDO -1.4 increases to 170cm

    (Note that I'm not reducing the AAO/SAM ... unconvinced.)


    Model goes like this:

    [​IMG]

    My previous years plotted there are hindcasts, not a priori predictions like Peterson's, so not really a fair comparison.
    Last edited by a moderator: Dec 29, 2013
  32.  
    Gerg

    Gerg Well-Known Member

    Of course, if AAO were to suddenly drop to zero (for the JJA mean), that would add a healthy 25cm to the prediction [​IMG]
    Last edited by a moderator: Nov 12, 2013
  33.  
    davidg

    davidg Dedicated Member

    Last edited by a moderator: Dec 15, 2013
  34.  
    Gerg

    Gerg Well-Known Member

    Yep, but it jumps around*. Need to look at the long trend (<a href="http://www.cpc.ncep.noaa.gov/products/precip/CWlink/daily_ao_index/aao/month.aao.gif" target="_blank">3-month running mean</a>).

    (* Link is just AAO calc'd off the 14-day GFS model. Veracity... I dunno, maybe it is falling. 2m would be nice...)
    Last edited by a moderator: Dec 15, 2013
  35.  
    BlueMountains

    BlueMountains New Member

    2008 had a notably positive SAM and we had around twelve snow days in the upper Blue Mountains. It doesn't always translate to better fronts with a lower SAM. As Falls suggested, other drivers can influence outcomes. With the changing climate, I sometimes wonder how strong these traditional signals will be for this season and the next.

    Interesting times. I sense we will learn plenty this year.
  36.  
    Gerg

    Gerg Well-Known Member

    No depth? Should still be something to measure:

    [​IMG]
    Last edited by a moderator: Nov 12, 2013
  37.  
    Gerg

    Gerg Well-Known Member

    Updated... Lads been out on S&amp;R?
  38.  
    Gerg

    Gerg Well-Known Member

    I've updated the Fourier analysis (above too):

    [​IMG]

    This is done with the Excel FFT. That algorithm only works for series lengths that are integer powers of 2, so I padded out our 59 years of record to 64 with trend values.

    The background points are from Fourier analysis of 100 randomly generated depth series (no small task...). I just used the linear trend plus random Gaussian residuals (not quite right, but close).

    As before, the real data plots well within the &quot;random&quot; range except right at 2 years. So, if there's any cycle at all, it's a very weak 2 year one (plus harmonic at 4).
    Last edited by a moderator: Dec 29, 2013
  39.  
    filski

    filski Part of the Furniture

    Interesting, but not sure what to make of it. Bump at 11yrs isn't strong enough?
  40.  
    Gerg

    Gerg Well-Known Member

    13 out of 100 randoms are higher. If you look, you will find. It's called confirmation bias.
  41.  
    Gerg

    Gerg Well-Known Member

    Only 30% of recorded years have no snow at Spencers Creek at 7 June (though that percentage would be higher now with the downtrend). It's 0% by 1 July:

    [​IMG]
    Last edited by a moderator: Dec 29, 2013
  42.  
    Gerg

    Gerg Well-Known Member

    Nope. There are plenty of warming indicators in our snow depth record, but lack of early season snow does not appear to be one of them:

    [​IMG]
    Last edited by a moderator: Dec 15, 2013
  43.  
    filski

    filski Part of the Furniture

    A week movement, in a season lasting around 17 weeks is somewhat significant.
  44.  
    Gerg

    Gerg Well-Known Member

    But in the wrong direction! I plotted 10 years as well ... line jumps around a bit, so I doubt a week means much here. The decades plot above suggests most change so far is in the peak and tail of the season. Start-up looks much the same as always.
  45.  
    JustinH

    JustinH Well-Known Member

    Curious... If its not a bother, could you post a 30 to 20, 20 to 10 and 10 to 1?
    Thanks for sharing what you you've done, interesting and unexpected
    Edit - Please
  46.  
    Gerg

    Gerg Well-Known Member

    This is not a very powerful method (in the statistical sense), so 10 years isn't really enough data. Maybe 20?

    (( What I'm doing is running a 7-day window over the weekly resampled record and fitting a percentile to that. The curve is then smoothed with the same Gaussian used for the depth plot. Ten years gets you ~10 data points in each 7-day window ... not good. Twenty is better, but still weak. I could increase the window size, but that would reduce the resolution (making it harder to hit well known obs, like that there's never been a recorded July zero). ))
    Last edited by a moderator: Dec 29, 2013
  47.  
    JustinH

    JustinH Well-Known Member

    I was more curious with the comparison to earlier measurements in the 50's 60's 70's? Three 19's maybe?

    Do you know if the method of measurement always been the same since the 1950's?
  48.  
    Gerg

    Gerg Well-Known Member

    Above ... might need a page refresh.
    One for Vego. I suspect the method is probably consistent, but the effort may not be. For example, did they motor out to measure every early season fall in past years the way they do today, now that everyone is determinedly watching?
    Last edited by a moderator: Dec 13, 2013
  49.  
    JustinH

    JustinH Well-Known Member

    Sorry I'm meaning something like 1956-1974,1975-1993 and 1994-2012 (18 year windows?)
    Its probably a waste of your time as they are all probably too similar and nothing significantly stands out
  50.  
    Gerg

    Gerg Well-Known Member

    We've done enough. Within likely error bars, they're all the same.