A semiparametric spatio-temporal model for solar irradiance data

TL;DR

This paper evaluates semiparametric spatio-temporal models for estimating aggregate solar irradiance in PV plants, demonstrating improved accuracy over simple interpolation and analyzing covariance structures using data from a Hawaiian PV plant.

Contribution

It introduces and assesses semiparametric models for solar irradiance estimation, showing their effectiveness over traditional interpolation methods and examining covariance assumptions.

Findings

Semiparametric models outperform simple interpolation in accuracy.

No evidence supports assuming a separable covariance structure.

Models are validated using real data from a 1.2 MW PV plant in Hawaii.

Abstract

Design and operation of a utility scale photovoltaic (PV) power plant depends on accurate modeling of the power generated, which is highly correlated with aggregate solar irradiance on the plant's PV modules. At present, aggregate solar irradiance over the area of a typical PV power plant cannot be measured directly. Rather, irradiance measurements are typically available from a few, relatively small sensors and thus aggregate solar irradiance must be estimated from these data. As a step towards finding more accurate methods for estimating aggregate irradiance from avaialble measurements, we evaluate semiparametric spatio-temporal models for global horizontal irradiance. Using data from a 1.2 MW PV plant located in Lanai, Hawaii, we show that a semiparametric model can be more accurate than simple intepolation between sensor locations. We investigate spatio-temporal models with separable and nonseparable covariance structures and find no evidence to support assuming a separable covariance structure.