Previous snow studies have relied on continuous clearing
of a set of control modules [12] or on the comparison of
actual system output to modelled system output over a given
time period [6], which is the methodology adopted by this
work. It has, however been shown that many accepted
modelling techniques can have a large portion of seasonal
and absolute bias due to spectral effects and module power
tolerances, in addition to not properly accounting for albedo
irradiation. Therefore, this work uses the adaptive modelling
algorithm proposed in [15] to reduce these biases and