Spatio-Temporal Downscaling of Hourly Solar Irradiance Data Using Gaussian Copulas

摘要

This paper presents a novel method for downscaling hourly solar irradiance data to higher resolution in both space and time. The method is based on transforming any point in two-dimensional space and time to a position in a propagating cloud field, the internal spatial variability of which is modelled with a Gaussian copula. By relating the mean hourly clear-sky index to probability distributions for 15-s irradiance and exponential decorrelation rate, the required inputs to the copula model are reduced to cloud field velocity and hourly average clear-sky index. The model is applied to irradiance data from a sensor network and is shown to accurately downscale hourly data from one sensor to 15-s resolution and 17 dispersed locations on individual days, reproducing key statistical features of the empirical network data.