Correlating the Weather

"You can prove anything with statistics." … Humphrey Appleby

"There are lies, damned lies, and statistics." …Mark Twain… or, perhaps, Benjamin Disraeli

"Rain is grace; rain is the sky condescending to the earth; without rain there would be no life." …John Updike .

Dog Days started in the first week of July, and will end around the time this issue of the Dispatch is published. Traditionally this is a period when the weather gets hot and humid; it used to be a topic of conversation, but since blogs, ipods and air conditioning became commonplace most people take the attitude of, "Well, it's summer… what would you expect?" But this year, the onset of Dog Days coincided with a prolonged dry spell that got everyone's attention. Lawns turned brown, cornfields dried up, suburbanites fretted about whether they would be allowed to water their flowerbeds, and the few farmers who haven't yet conceded their land to developers began to worry if they could make it through another year. And people of my generation remembered old times.

My parents built a house in 1923, and my father proceeded to dig a well near the back door, using dynamite to penetrate the hard West Virginia shale. He was lucky enough to find water at a depth of 16 feet without blowing up either himself or the house. The well supplied water for drinking and cooking via a pump at the kitchen sink; water for laundry came from a cistern filled by rainwater drained off the roof. Since there was no bathroom in the house, these two sources were sufficient to supply the family in

normal times. However, we were always aware of the system's limitations, and from my earliest memories I was cautioned to conserve water during Dog Days. My grandmother was a weather watcher of long and superstitious experience, and she believed if Dog Days started in a certain phase of the moon we would have a drought. I never was able to pin her down about the exact connection with the moon, but she was right occasionally, and no one seemed to remember it when she was wrong.

My early childhood years coincided with the drought that resulted in the Dust Bowl, and there were several summers when our well went dry. Luckily, grandma's well was on a better aquifer; I recall my father carrying water from it each day in our 3-gallon milk bucket, while I toddled behind him with a pail improvised from a coffee can, spilling most of it before we got home. This aspect of the "Good Old Days" ended just after the war, when "city water" was piped into our area from Fairmont, about three miles away. But although that was 70 years ago, even now when Dog Days arrive the memory comes back, like a salmon returning to the stream of its hatching after years at sea.

As I write this, we are nearly 10 inches below the normal rainfall total for the year. Official records have been kept locally for the past 140 years, starting in 1867 at Mount St. Mary's and then, for the past 50 years, by the late Lucille Beale; they show that our average annual rainfall is just under 44 inches. Recently, just for curiosity, I made a graph of these records; the main thing it revealed is why it is so hard to prove anything about what causes our weather. The data points from one year to the next vary in a wildly erratic pattern that looks completely random at first glance; the driest years were 1888 and 2001, when we had about 27 inches, and the wettest was 1996 with over 75 inches. The over-all trend looks as if the average may have dropped a couple of inches since 1867, but the procedure for calculating the regression line on the graph that would prove or disprove such a conclusion is tedious, and I haven't got around to doing it yet. However, after staring at the graph for a while you can begin to see a general pattern of alternating wetter and drier periods occurring at roughly 20-year intervals.

Making sense out of weather patterns is fiendishly difficult work; information piles up at rates that can swamp the memory capacity of our biggest computers. Efforts to deal with it before the computer age sometimes produced confusion that reached comic opera proportions, yet some of our basic theories about weather arose from those efforts. Beginning in the late1930s, Dr. C. G. Abbott of the U. S. Weather Bureau did a Herculean study of over 20 different components of weather in an attempt to find what had caused the Dust Bowl drought. He made a graph of these factors as they had varied over many years; since he had no computer, he plotted each factor by hand in a different color on a sheet of graph paper that was over 30 feet long. The graph was so unwieldy that he had to keep it rolled up around a Quaker Oats box like a Biblical scroll, and he worked on it until he was past 90. He claimed his graph showed a correlation between periods of drought in the midwestern U. S. and sunspot cycles. Most people regarded him as somewhere between mildly eccentric and outright loony; however, years after his death meteorologists found that his method predicted a drought that occurred in the late 1960s.

The problem with Abbott's theory was that sunspots occur in maximum numbers every 11 years, and at that time there was no known mechanism by which they could influence rainfall. Since 1960, satellite measurements have shown that the sun's radiation varies with the sunspot cycle, and the cycles themselves occur in pairs, so they could actually cause weather to change at 22-year intervals. Present-day measurements seem to confirm a correlation between sunspot cycles and wet or dry periods, but correlation is not proof of cause and effect. When teaching statistics to my ecology classes, I used to remind them that the growth of the stock market in the 1960s correlated very well with the length of women's skirts over the same period of time.

At the point where we are now, we can only hope that history will repeat itself. In the past 140 years there were three years in which we had less than 30 inches of rain, and each of those was followed by wetter than normal seasons. But the dots and lines on my graph sound a cautionary note; since 1960, the fluctuation in rainfall from year to year and cycle to cycle seems to have gotten more extreme. Whether this correlates with global warming is an interesting question; proof of such a correlation will be extremely difficult, and the nature of science is such that the question will probably not be settled in my lifetime. In the meanwhile, I have to be satisfied with the wisdom of a farmer who was quoted years ago in the Saturday Evening Post: Dog days will end. Rain will come; it always has, and it always will.