STA 290: Joshua Patrick

Statistics Seminar Series

Thursday, November 6, 4:10pm, MSB 1147 (Colloquium Room)
Refreshments at 3:30pm in MSB 4110 (Statistics Lounge)

Speaker: Joshua Patrick (Post-doctoral Scholar, Dept of Statistics, UC Davis)

Title: Spatio-temporal modeling of solar irradiance data”

Abstract: Recent advances in photovoltaic research has begun to make solar energy a more reliable and significant contributor to electricity grids. Resource assessment, system design, and energy planning have become some of the challenges that utility companies must address. These challenges are all based on the amount of irradiance measured on a grid of solar panels which is the basis for the amount of electricity generated. The modeling and forecasting of irradiance is not only important in determining amount of energy produced by an installed solar energy system but also in the planning of a future system. A suitable spatio-temporal modeling framework is needed to meet these challenges. Recent work has shown that spatio-temporal models can be useful for systems with sparse locations and limited cloud cover conditions. Models that can take into account the complex space-time interactions and that explain the variability caused by various cloud cover conditions is what is needed for solar energy systems to have a bigger part of the utility-scale electricity grid. We will discuss the challenges of detrending the so-called “clear sky model” which is needed for the process to be stationary. From this clear sky model, we will examine some ways in which a reliable and accessible cloud cover covariate can be used in explaining the variability caused by different weather conditions. The nature of the spatio-temporal covariance structure will also be discussed and some preliminary ways in which it can be modeled. These topics will be explored through irradiance data obtained from the University of Oregon Solar Radiation Monitoring Laboratory.