Downscaling of spatial irradiance based on cloud advection using transfer functions

Abstract

Spatiotemporal aggregation of solar irradiance occurs when a spatially distributed receiver (e.g. a PV generation facility) collects variable geographically distributed irradiance and reduces it to a single electrical generation output. Models of this phenomenon exist, and are designed to take variability from a single point irradiance monitor and predict how that variability will be reduced by aggregation. We have applied these models in a revered manner to assess whether the models can be used to predict the variability of a single point measurement given an aggregate irradiance time series as an input. Results for an advection-based model show that this approach leads to overprediction of the high frequency variability due to overprediction of the site-to-site correlation. Incorporating predictions of site pair decorrelation from the wavelet variability model can temper the degree of overprediction and produces more realistic point time series. Further work may be warranted to further improve upon these efforts and enable reliable, transfer function-based downscaling of irradiance data.