Improving a solar irradiance model using statistical methods
LE3 .A278 2011
Master of Science
Mathematics and Statistics
Mathematics & Statistics
The sun is a very powerful source of energy and has been used since the beginning of the existence of man to survive. Clouds reduce the amount of solar irradiance that travels to the Earth. The Turquoisat model developed by Turquoise Technology Solutions Inc., Montreal, uses satellite imagery to determine the solar irradiance at a given location on the Earth. The total irradiance on a horizontal surface can be divided into two parts: direct irradiance and di use irradiance. These two parts are a ected by clouds di erently. Direct irradiance decreases as there are more clouds while di use irradiance increases. It turns out that the current relationship between cloud cover and solar irradiance depends on the total irradiance. In this thesis, we propose a new methodology using the di erent components of solar irradiance. Fur- thermore, by comparing the Turquoisat model output to solar irradiance values gath- ered from ground stations from across Canada it can be seen that during the morning the Turquoisat model underestimates the solar irradiance that reaches the Earth, and during the evening the model overestimates the solar irradiance. In an attempt to account for the bias in the Turquoisat model, we use a Gaussian process modelling approach to model the discrepancy between the observations from ground stations and the Turquoisat model output. These two adjustments improve the accuracy of the Turquoisat model.
The author grants permission to the University Librarian at Acadia University to reproduce, loan or distribute copies of my thesis in microform, paper or electronic formats on a non-profit basis. The author retains the copyright of the thesis.