Synthetic generation of high temporal resolution solar radiation data using Markov models

In this paper, a model for the synthetic generation of 1-min global solar radiation data starting from the daily clearness index is presented. The model is constructed by treating the process generating a normalized form of the 1-min clearness index sequence as a Markov process. Two sets of three-year global solar radiation data taken at 1-min intervals from two locations in Japan are used to construct the Markov transition matrices. Because different days have different statistical characteristics due to different weather conditions, the days in the data set are first clustered into groups based on the daily clearness index values. Transition matrices are then formed for each group and consequently used to synthetically generate 1-min global solar radiation data. Second-order Markov models are selected based on the partial autocorrelation functions of the measured data. The statistical characteristics of the measured and synthetic data sets are found to be in close agreement thus confirming the validity of the model.