When I posted a graph showing total snowfall at Mt. Rainier from 1921 to the present, I got a lot of guesses as to why I was posting the graph. Almost all of the guessers thought that I posted it in order to make an argument on one side of the great climate change debate. They were wrong.
The best guess was that I posted the graph to show a trend. And, in fact, there is a trend, which you can see estimated in the dashed line in the plot. If you believe the line, then you would say that the total snowfall at Rainier has been increasing by a little more than 2 inches a year since 1921.
(The equation was estimated with a simple linear model, in which the total snowfall(y) is a linear function of the year(x). Those who like to do mindless extrapolation will notice that the average snowfall at Paradise will be close to 1,000 inches in 150 years — and that it was minus 3500 inches in 1 AD. Mindless extrapolations can be entertaining.)
Having shown you that estimated trend, I will now say that I don't take that estimated line seriously. I don't think a linear trend is a good description of the data. Rather than give you my own description of the data now, I will leave that as a problem for the readers, for another week or two.
I am doing that for a nefarious purpose. I hope to persuade at least a few readers to look at some data analytically, rather than as a source of ammunition in an ongoing political quarrel.
Cross posted at Jim Miller on Politics.
(You can find the sources of the data at my original post. I made one minor change in the data set, restoring the fractions to four of the years.
Those who know even a little statistics will want to know that the estimated trend is significant at the 0.01 level, but that the R2 is a lousy 0.1188.
You can see some pictures of the snow at Rainier, taken in March, 2008, here.)
Posted by Jim Miller at July 16, 2009 10:38 AM | Email ThisFor example, within the ~100 years shown, the snowfall appears to peak (roughly) about every eleven years, which might indicate a correlation with the sun spot cycle (I'm using my imagination as to what the data might be for the missing years).
Hypothetically, combining the eleven-year cycle with a wet trend with a period of more than a hundred years could explain the data, while also preventing the sort of nonsensical results you mentioned that accompany "mindless extrapolation".
Posted by: ewaggin on July 16, 2009 12:32 PMThanks.
Posted by: Gary on July 16, 2009 12:53 PM@4: It didn't go over my head... I'm stating that it had no point unless he wanted to impress his fans with his Excel skills.
Posted by: demo kid on July 16, 2009 02:35 PMThanks.
Posted by: Gary on July 16, 2009 02:46 PMPosted by: Gary on July 16, 2009 03:01 PM
Is the govt going to control the annual snowfall on Mt Rainier? Just wondering....
Posted by: scottd on July 16, 2009 03:12 PMIt's common knowledge that it'll get colder (hence more snow) before is gets warmer.
We know this because Al Gore told us so it MUST be true.
".....How long can you tread water?.........."
Posted by: Sam Adams on July 16, 2009 03:17 PM
Now that's just plain hurtful, after I have spent this much time learning the basics of R. (I experimented with the business graphs in Open Office, but didn't find them adequate for much of what I want to do.)
More seriously: Demo kid, I really do hope that you will try to come up with an explanation for the pattern(s) seen in that graph. And that other readers will do the same -- just as I said in the post. (If you want a hint, send me an email.)
Incidentally, I am using R for other graphs; here's one that should interest both Republicans and Democrats. (I plan to put up the next one in that series at the beginning of August.)
Posted by: Jim Miller on July 16, 2009 03:29 PMPosted by: scottd on July 16, 2009 03:36 PM
Posted by: scottd on July 16, 2009 03:39 PM
I don't know what the target is. That's why I asked.
Posted by: Gary on July 16, 2009 03:39 PMGiven the 10+ year break between measurements, my guess is that either the snow gauge was moved or the Park Service put up a building next to it sometime after WWII.
The really scary linear extrapolation is the median daily temperature in Seattle from 1/1/09 to 7/17/09. At the current rate of change, Seattle will be too hot to support mammals by the summer of 2011. Your pets will die. By the winter of 2014, the climate will be too hot for water to remain a liquid, and Puget Sound will boil, and by 2015 the climate will be so nasty that I'll finally be able to afford a house with a view on Queen Anne.
This can be prevented: if the Seattle City Council passes a $.07 per kilowatt surcharge on electricity, payable to me, I can guarantee that the expected environmental apocalypse will not occur. However, you must act now. If we wait until October, that will be too late for anyone to believe me.
Ah, who am I kidding. Everyone who believed me now will still believe next December.
Posted by: David D on July 16, 2009 09:36 PMWith respect to your graph, I'll say it again: it means nothing. There might be some point you're trying to make later, but you're not making one right now.
I would, however, reformat the figure. The lines don't mean anything, but suggest that you're interpolating between the points. It's a bad habit that a lot of folks get into (and yes, I know that there are plenty of examples of folks that do this!), but since this isn't a continuous function, it doesn't really make sense. In fact, a bar chart would probably be more valid.
And in terms of any of the piddling comments on climate change made by the intellectually challenged: this is a record of snowfall, which is precipitation and NOT temperature. Under climate change scenarios, precipitation is expected to change in EITHER direction depending on the location. The issue for the northwest is more the speed and timing of snowmelt -- especially with respect to their impacts on water supplies and hydroelectric power.
Posted by: demo kid on July 17, 2009 10:04 AMWith that caveat, I imagine if you you'll want some other category to regress, such as snowfall or precipitation from another point, average temperature, something. If you possess knowledge of weather patterns, you should be able to recognize what variables could further isolate the year factor (or offer exculpatory evidence of time's effect).
But after saying all that, my instinct is that snowfall is essentially a stochastic process (read A Random Walk Down Wall Street).
Posted by: SlipperyPete on July 17, 2009 01:37 PMPosted by: G Jiggy on July 17, 2009 02:44 PM
For those interested in actual facts, see this article wherein a lead IPCC author admits that we still don't know enough to implicate CO2, even as we watch our Congress debate plunging us further in to debt and other damaging economic schemes over CO2.
Or read this article, wherein NCAR finds that the Sun is more responsible for climate than humans. To most rational people, this is obvious. But for those interested in the political and economic plays of the Alarmists, no such obvious facts such as the heat content of the oceans as compared with tiny human influences need be examined. It's simply their play for control, alarm and greed. That's why Al Gore says the science is settled even as papers are produced daily that are still trying to understand earth's climate.
Jim Miller is acting in a much more scientific manner than his weak critics. Let's see Demo Kid post his own blog with better analysis. That's not going to happen, because that takes a lot more work than sitting back and just writing anonymous and cowardly ad hominem swipes.
The bottom line: "There appears to be something fundamentally wrong with the way temperature and carbon are linked in the climate models." This inconvenient truth, is one that drives those in the political sycophancy class like demo kid, to the desperation of ad hominem.
Posted by: Jeff B. on July 19, 2009 08:51 AM