Teleconnections and Analysis of Long-term Wind Speed Variability in the UAE

  • Mussie S. Naizghi

Student thesis: Master's Thesis


Wind energy accounts for a small share of the global energy consumption in spite of its widespread availability. One of the obstacles hindering wind energy use is the lack of proper wind speed assessment models. The objective of this study is to examine the longterm variability of wind speed in the United Arab Emirates (UAE) and its teleconnections with various global climate oscillations by using wind speed collected from six ground stations and gridded reanalysis dataset. Linear correlation analysis and wavelet analysis have been applied to characterize the interaction. Correlation analysis of average wind speed in a three-month moving window with the climate indices did not reveal strong correlation except for North Pacific and Mediterranean Oscillation Index. For these indices, however, strong negative (positive) correlation is observed at lag-4 (lag-10) months. Further correlation analysis based on annual average wind speed in a three-month moving window over 30 years to identify seasonality exhibited moderate correlation with few climate indices during some months. Furthermore, continuous wavelet transform (CWT) and wavelet coherence (WTC) are employed to model the wind speed variability and coherence with the climate indices to identify the dominant driving process for the wind in the UAE. CWT of wind speed in all stations demonstrated annual periodicity while few showed a six-month periodicity as well. WTC analysis indicates that the wind speed in the UAE is mainly influenced by the North Atlantic Oscillation, East Atlantic Oscillation, Southern Oscillation Index and Indian Ocean Dipole primarily in the time scales of 2-3 and 5-7 years period. The first two indices simultaneously modulate wind speed in the summer while the last two influences winter and autumn wind speeds. In terms of long-term change, annual wind speed in three of the six stations showed significant trend at the 5% level. Additionally, five of the stations but one presented change points. Attempts to use multiple linear regression to formulate forecasting models using climate indices as the predictors following stepwise regression for predictor selection gave poor results indicating that non-linear models have to be studies in the future.
Date of AwardMay 2015
Original languageAmerican English
SupervisorTaha B. M. J. Ouarda (Supervisor)


  • Wind Speed
  • Wind Energy
  • Energy Consumption
  • Assessment Models
  • Wind Energy in the UAE
  • Forecasting Methods.

Cite this