Multi-year ground-based irradiance dataset in a northern urban climate

A multi-year irradiance dataset has been collected using secondary standard pyrheliometer and pyranometers at a Canadian urban test site. Detailed screening algorithms, instrument bias corrections, and uncertainty analyses were undertaken to produce a high accuracy, validated dataset for global horizontal irradiance (GHI), direct normal irradiance (DNI) and diffuse horizontal irradiance (DHI). The inter-comparison of the three instruments is included within a study of the DHI shadowband correction model, and confirms agreement with mean bias errors less than 4%. Comparative studies of measurements with satellite data were undertaken, and the agreement for GHI daily values with NASA satellite data exhibited a relative mean bias error (rMBE) of -1.7% and a relative root mean squared error (rRMSE) of 24.3%. The agreement for DNI hourly values in comparison to SolarAnywhere v3 was -1.2% with an rRMSE of 8.8%. In all three statistical analyses, the agreements are within the instrumental uncertainty, providing very strong validations of the ground-based dataset and of the accuracy of the satellite-derived data. Frequency distribution versus DNI for ground and satellite data are found to have some differences in-line with the nature of their measurement characteristics.