A simple theoretical model demonstrates that some amount of randomness in s
now and ice mass balance is sufficient to reproduce empirically observed po
wer law and exponential distributions of snow patch and glacier sizes. No o
ther assumptions about the underlying topography or snow accumulation and a
blation processes are necessary to extract this important spatial property.
The inclusion of additional geometrical and physical processes can alter t
he specific scaling constants of the size distributions, but the fundamenta
l behavior remains unchanged. Specifically, for snow patch and glacier size
s less than some correlation length the size distribution is a decreasing p
ower law, and for sizes larger than the correlation length the distribution
decreases rapidly as an exponential. The solution is based on a mapping to
a relatively well explored class of problems in percolation theory.