Home Page

Friday, November 19, 2010

Fast And Loose With The Data

As mentioned before, I've been looking at this latest study come out regarding Lake Tahoe. I have trouble with the most-newsworthy projections of the study regarding the climate changes at Lake Tahoe over the next century:

About 55 percent, on average, of the precipitation at lake level in Tahoe now falls as snow. By 2055, the study predicts snow will drop to about 45 percent – and to just 30 percent by the end of the century.

The result: a shorter ski season, and perhaps a spring without snow-capped peaks ringing the lake, said Robert Coats, lead author of the study and a visiting scholar at UC Davis.

"We're looking at a shift from snowfall to rainfall, increased melt rate, and earlier melt," Coats said. "Once you lose the snowpack, then you lose the late-spring water supply. So drought could begin earlier in the year."
That's a powerful metric there, the percentage of precipitation falling as snow. Here is a graph from the 2009 TERC report showing the percentage of precipitation falling as snow for Tahoe City over the last century:

















In the 2010 report, simulations carry the projection into the future (GFDL A2; see chart below). The projection is supposed to be a basinwide average (even though it looks like it's congruent with the Tahoe City data):

But I found this all rather strange. Do observers with the National Weather Service collect data regarding the percentage of precipitation falling as snow? One might - might - be able to glean that statistic from standard airport observation stations (provided one decides how to pigeonhole ambiguous ice/water mixtures like sleet), but the long record here, back to 1910 for Tahoe City, suggests COOP station data instead was used. They don't gather that kind of data at COOP stations. So, what kind of data did they use?

According to the 2009 TERC report:
Snow has declined as a fraction of total precipitation, from an average of 52 percent in 1910 to 34 percent in present times. In Tahoe City, snow represented 37 percent of 2009 total precipitation, consistent with the long-term decline.

These data assume precipitation falls as snow whenever the average daily air temperature is below freezing. (Precipitation is summed over the Water Year, which extends from October 1 through September 30.)
Oh! These aren't measured data at all! They are a synthetic construct. We don't know how well the assumption of temperatures-below-freezing-means-snow holds. It is always possible, of course, for precipitation to fall as snow even if the average daily temperature is above freezing, provided temperatures are below freezing at the instant precipitation is falling.

So, what these charts are actually saying isn't that the snow precipitation fraction is going down, but that the average daily temperature is going up. The scattered data points, which give the illusion of measured data, or data simulations, actually represent nothing of the sort. It's pretty-much all sleight-of-hand!

Tahoe City is at lake level, a relatively-warm place, so it is quite possible for rain to fall there even as snow is falling on the nearby, colder peaks. So, it's too early to host a wake for the snow-capped peaks of the Sierra Nevada.

Warmer daily average temperatures can be consistent with an earlier snowmelt, of course, but if that's the metric one is measuring, that's the one that should be used, and not some other, superficially-appealing construct.

I am not among global warming skeptics, but it's easy to appreciate their frustration when confronted with synthetic data that purport to show one thing, when they actually show another.

Bait and switch; bait and switch. This is about the right mind set, right here....

No comments:

Post a Comment