A new algorithm to estimate sky condition based on 5 minutes-averaged values of clearness index and relative optical air mass

This work describes a new algorithm to characterize sky condition in intervals of 5 min using four categories of sun exposition: apparent sun with cloud reflection effects; apparent sun without cloud effects; sun partially concealed by clouds; and sun totally concealed by clouds. The algorithm can also be applied to estimate hourly and daily sky condition in terms of the traditional three categories: clear, partially cloudy and cloudy day. It identifies sky conditions within a confidence interval of 95% by minimizing local climate and measurement effects. This is accomplished by using a logistic cumulative probability function to characterize clear sky and Weibull cumulative probability function to represent cloudy sky. Both probability functions are derived from frequency distributions of clearness index, based on 5 minutes-averaged values of global solar irradiance observed at the surface during a period of 6 years in Botucatu, Southeastern of Brazil. The relative sunshine estimated from the new algorithm is statistically comparable to the one derived from Campbell-Stocks sunshine recorder for both daily and monthly values. The new method indicates that the highest frequency of clear sky days occurs in Botucatu during winter (66%) and the lowest during the summer (38%). Partially cloudy condition is the dominant feature during all months of the year.