Hello, you might want to have a closer look at the International Standard IEC 61724 "Photovoltaic system performance monitoring - Guidelines for measurement, data exchange and analysis". You will find there useful information on monitoring the irradiance. The sampling interval for the irradiance is specified to be 1 min. or less, so I would say you sampling interval of 15 min. is too large. The standard also explains how to integrate or average your data according to the sensor and value you are measuring. If you want to measure hourly data that you want to import in PVsyst, the values should be representative for the hour that follows the time label. If for example you measure a value for 15:00, it should reflect the data from 15:00 to 16:00. For minimizimg your statistical errors, you need to keep the sampling interval small. To keep systematic errors small you should carefully calibrate your sensors and measure at a place that is representative for your whole array. If you can afford a redundant measurement this will also reduce the uncertainties. The standard also gives guidelines on how to check the quality of your measured data.