Someone told my mother-in-law that the day of the first snow would predict the number of snows.  For instance, this year the first snowfall in these parts was on November 21 so this model would predict 21 distinct snow dumps this year.

To test this, I tried to keep track of the number of snows during the 2008/2009 season.  This generated some controversy within the Smucker household surrounding the definition of a snowfall.  I argued that if it snowed and then stopped and then snowed again, this constituted two distinct snowfalls even if these events happened during the same afternoon.  Amy thought I was liberal at times in my snowfall quantity assignments.

At any rate, I counted 27 snowfalls this season, the last one being on April 7. It is obvious, then, that the proposed model is faulty.  Instead,  we can predict the number of snowfalls by taking the day of the first snowfall and adding to it one less than the day of the final snowfall.

This revised model fits the data perfectly.