[pct-l] Sierra snow plot

[jeffmoorehead1 at cox.net (Jeff Moorehead)]

I like this method. If you have the actual dates (or extrapolated in case of 
the outage years), what is the mean and standard deviation of the 0 snow day 
and how does the latter change as the actual date approaches it? You could 
'jacknife' the estimates for past years and ask how accurate is the running 
mean (which would be the best predicted 0 snow day) on any given date 
compared to the actual observed 0 snow day. You could translate this into a 
statement that says something like "on this date, there is a X% chance of 
error in predicting the following date for 0 snow"? People would be 
interested in these dates.
Interesting idea.

> I've got a sneaky new plot that "predicts" Sierra snow levels.
>
>    http://theory.lcs.mit.edu/~e_lehman/utyplot.png
>
> This is not your typical snow graph, so let me explain...
>
> * The Y-axis indicates the snow level measured by a sensor at Upper
> Tyndall Creek, which is at 11400' and pretty close to Forester Pass.
>
> * The X-axis measures days relative to "Ray Day", June 15.  So the plot
> runs from about four months before to one month after Ray Day.
>
> * On the left side of the graph, there is a single red line.  This is the
> actual snow level at Upper Tyndall Creek this year-- 2005.  Of course, the
> red line ends today (April 2).
>
> The remainder of the graph is a big mass of multicolored lines.  I know
> some are difficult to distinguish, but this is the sneaky part.
>
> Today, the snow water equivalent at Upper Tyndall Creek is 38.7 inches.
> Suppose that in each previous year on this date-- April 2--- someone had
> hiked up to the snow sensor with a shovel and either piled on more snow or
> shoveled off some snow so that there was exactly 38.7 inches-- just like
> today.  The graph shows how the meltoff would have proceeded from that
> "corrected" snow level.
>
> Looked at another way, the graph show what will happen THIS YEAR if our
> meltoff pattern matches a previous year.  So that mass of colored lines
> represents the "envelope of uncertainty".  The snow level for the
> remainder of the year is almost surely going to run somewhere through that
> big tangle.
>
> For example, one dark blue line on the graph plunges to zero before Ray
> Day.  So if the melt pattern this year follows the melt pattern that year,
> Upper Tyndall Creek will be bare of snow well before June 15.  (This area
> actually was snow-free by Ray Day in about 2/3 of the years from
> 1970-2004-- so that's something to hope for!)
>
> Below the dark blue line, there is a pink line which drops very fast, but
> never makes it down to zero.  In that year (1981), the meltoff was
> super-fast.  However, on April 2, 1981 there were only 16.0 inches of snow
> at Upper Tyndall Creek.  So the pink line ends after only 16.0 inches of
> meltoff.
>
> On the other hand, there have been several years where Upper Tyndall Creek
> didn't drop 38.7 inches below the April 2 level until a *month* after Ray
> Day.  The corresponding lines run into the lower right corner of the
> graph.
>
> I'll update this graph from time to time, if I remember.  (If you're
> curious, send me a reminder and I'll do it.)  With each passing day, the
> "envelope of uncertainly" will grow smaller...
>
> /Eric
>
> P.S.  Actual snow levels for 1970-2005 are plotted here:
>
>  http://theory.lcs.mit.edu/~e_lehman/utyplot_abs.png
>
> The flat lines followed by jumps that appear in some years are due to
> sensor outages.
> _______________________________________________
> pct-l mailing list
> pct-l@mailman.backcountry.net
> unsubscribe or change options:
> http://mailman.hack.net/mailman/listinfo/pct-l
>