Spatiotemporal Interpolation of High Frequency Irradiance Data for Inverter Testing

Abstract

A model for interpolating irradiance data in space and time is developed using cloud motion vectors. It is tested using two networks of photodiodes in Eastern Canada that measure irradiance at time scales of milliseconds and spatial scales of tens of meters. The model captures network-average variability with skill scores of up to 86% compared to nearest neighbor interpolation. We provide case studies showing how to use this model to study inverter response to rapid irradiance fluctuations. We also use AC power measurements from an inverter collocated with one of the photodiode networks to benchmark our approach and the wavelet variability model.