... such as the median?

We're sure you know the median is often the best measure when the data have extreme outliers so why not use it instead of the average which is much less representative of a season-total's snowfall?

Shouldn't you at least evaluate the difference?