QUANTIFYING THE VARIABILITY OF SOLAR PV PRODUCTION FORECASTS

The actual power produced by a solar photovoltaic power system varies according to the refraction, reflection, and absorption of radiation by the atmosphere. Standard production forecasts, however, do not address production uncertainty probabilistically. Using thirty years of historical data from the National Solar Radiation Database (“NSRDB”), we use stochastic simulation to evaluate the production uncertainty that is otherwise ignored by traditional production forecas ts. In the case documented herein, we find significant diff erences between the forecasts made by conventional production forecasting models and those designed explicitly to reflect the uncertainty found in actual historical experience. Having more accurate information about production uncertainty should facilitate more project-appropriate financial structures, reducing risk and increasing investor returns.