A Comparative Study on Solar Radiation Datasets for Photovoltaic Energy Prediction at High Latitudes
Abstract
This study presents a comparative analysis on solar radiation datasets for solar energy forecast. Such datasets differ among the sources (i.e., satellite observations, reanalysis, weather stations), the radiative parameters (i.e., global horizontal irradiance, diffuse horizontal irradiance and direct normal irradiance), and the time resolutions (i.e., one-minute, one-hour). The plane of array irradiance amounts impinging on a photovoltaic panel (PV) are determined through the decomposition and transposition models implemented in the pvlib-python tool and the latest methods found in literature. Then, the PV energy production is quantified and validated against experimental data. Such data are collected from the PV integrated in the rooftop of the Test Cell Lab research facility, in Trondheim (Norway). The main goal is to determine which dataset can better predict the PV energy production at high latitude locations. Datasets providing measured direct normal and diffuse horizontal irradiance amounts allow to reduce the length of the model chain; therefore, they are expected to be preferable.