ECML/PKDD 2015, Workshop on Data Analytics for Renewable Energy,
Porto, Portugal, 11th September, 2015

Invited Talk

Taha B.M.J. Ouarda
Institute Center for Water and Environment (iWATER),
Masdar Institute of Science and Technology,
P.O. Box 54224, Abu Dhabi, UAE.

The accurate temporal mapping of renewable energy potential at different resolutions (from the sub-hourly to the multi-decadal level) is vital for the effective harvesting and integration of renewables. The integration of renewable energy has been affected by the deficiency in the development of suitable long term potential assessment models. In a number of instances, the solar and wind energy fields have suffered from projects that were developed based on relatively short records and without consideration of the inherent inter-annual variability in such variables as solar irradiance and wind speed. The projects did not deliver the expected short term outcomes affecting hence the credibility of the field. In the present work we examine and model the long-term variability of diffuse horizontal irradiance (DHI), direct normal irradiance (DNI), the resulting global horizontal irradiance (GHI) and wind speed based on the UAE case study. The teleconnections of these variables with various global climate indices are studied. It is shown that a number of low-frequency climate oscillation indices represent efficient long-term predictors of these variables. Wavelet analysis were used to characterize the interaction between these renewable energy variables and climate indices. Continuous wavelet transform is used to analyze the periodicity in these variables. Wavelet coherence analysis demonstrated that these variables are mainly associated with the North Atlantic Oscillation, East Atlantic Oscillation, El NiƱo Southern Oscillation and the Indian Ocean Dipole indices in a number of time bands. These indices modulate the variables of interest in different seasons of the year. Trends in these variables are also studied and integrated in the long term estimates of renewable potential. A stochastic model that reproduces non-stationary oscillation (NSO) processes by utilizing ensemble empirical mode decomposition (EEMD) and non-parametric techniques is used to predict the evolution of these variables into the future.

Speaker Bio
Professor Taha Ouarda is Professor and Head of the Institute Center for Water and Environment at Masdar Institute. His specialization is in statistical hydrometeorology and risk analysis. Dr. Ouarda was Chairman of the Canada Research Chair on the Estimation of Hydro-meteorological Variables, and Chairman of the Industrial Chair in Statistical Hydrology at the National Institute for Scientific Research. He has also served as President of the National Canadian Committee on Statistical Hydrology. Dr. Ouarda holds a PhD in Civil Engineering from Colorado State University. He has developed several approaches and computer softwaresthat deal with a range of problems in hydrometeorology. He has also led several international projects dealing with hydro-meteorological modeling. Dr. Ouarda is the author of more than 400 articles in his field.