Ticker for June 12, 2007

                
MESONET TICKER ... MESONET TICKER ... MESONET TICKER ... MESONET TICKER ...
June 12, 2007      June 12, 2007      June 12, 2007      June 12, 2007



More Correlations

Yesterday, the Ticker meandered into the nefarious world of correlations,
using statistics to bring the earth-shattering news that Oklahoma's
wetter summer months tend to also be cooler. That's neat, but we admit
its old news.

So, we'll play around with other correlations this week that might
illustrate more useful climate signals. Today we'll look at how one
month's temps feed into the next month's precipitation. A specific
example: what can you predict about May's precip using April's
temperature?

The answer is "not much" (tomorrow's correlation is more fun):



In fact, for most of the year, for most of the state, correlations
between previous month's temp and current month's precip are laughingly
negligible. July precip might hold some marginally-useful information
about August temps, but not enough to make confident, bold statements.
Something on the order of "well, gee, you can kinda-sorta see that, in
a limited way, warmer (cooler) Julys predict drier (wetter) Augusts".
And we'll see why (in a kinda-sorta sense) on Thursday. However, the
correlations of about 0.2 are still enormously small, explaining less
than 5% of the variance seen in August temps.

And, interestingly, you can make the same awkward, hesitant statement
about November temps feeding into December precip. And we really don't
know why, to be honest.

Finally, we're dealing with very small correlations, which is not
a statistically mature thing to do ... but ... we can't help notice the
profound difference in the April Precip lines for the panhandle (light
blue) and the southeast (dark red). These truly are two different
climate regimes, and nothing draws out that difference like the active
spring season.