Modeling Solar Irradiance and Solar PV Power Output to Create a Resource Assessment using Linear Multiple Multivariate Regression

The increased use of solar photovoltaic (PV) cells as energy sources on electric grids has created the need for accurate solar irradiance and power production estimations. In the present paper, we establish a novel technique for solar irradiance estimations and creating a resource assessment dataset by utilizing numerical weather model variables, satellite data, and the surface radiation budget (SURFRAD) network and the Integrated Surface Irradiance Study (ISIS) network measurements. The solar irradiance outputs are global horizontal (GHI), direct normal (DNI), and diffuse horizontal (DHI). The technique is developed over the United States by training a linear multiple multivariate regression scheme at ten locations and then applying the coefficients to independent variables over the whole geographic domain of study.

The irradiance estimations are used as inputs for a solar PV power-modeling algorithm to compute power production estimations at every grid cell across the domain. The dataset is analyzed to predict the capacity factors for solar resource sites around the USA for the three years of 2006-2008. Statistics are shown to validate the skill of the scheme at geographic sites independent of the training set. In addition, we show that more high quality (geographically dispersed) observation sites increase the skill of the scheme.