• Haile T. Woldesellasse

Student thesis: Master's Thesis


Wind is one of the crucial renewable energy sources which is expected to bring a sustainable solution to the global issue of climate change. Several linear and nonlinear multivariate techniques have been used in the past to predict the stochastic character of wind speed. A wind forecast with good accuracy has a positive impact on the reduction of electricity system cost and is essential for the effective grid management. Over the past years, not many studies have been conducted on the assessment of teleconnections and possible effects on the variability of wind speed in UAE. In this study, nonlinear-CCA (neural network based) technique is applied to assess the connection between global climate oscillation indices and meteorological variables of United Arab Emirates. A major emphasis is given to the wind speed measurement in Abu Dhabi, UAE. The wind dataset was obtained from six ground stations. The strength of the nonlinear canonical correlation between the two sets of variables varies with the lag time. The performance of the models is assessed using six error indices. These are the mean squared error (MSE), the root mean squared error (RMSE), the mean absolute error (MAE), the relative mean absolute error (MARE), the relative mean squared error (MSRE) and the relative root mean squared error (RMSRE). The results indicated that NLCCA models provide more accurate information about the nonlinear intrinsic behavior of the dataset of variables than linear CCA models. Besides, the combination of the linear and nonlinear models also increased the accuracy of the prediction for a 3-monthly wind speed data.
Date of AwardMay 2017
Original languageAmerican English


  • United Arab Emirates
  • Wind Speed
  • Renewable Energy Sources
  • Climate Change
  • Abu Dhabi
  • Wind Energy.

Cite this