Wind speed is one the most important variables of the wind phenomena. Thus, it is important to investigate the main temporal and spatial characteristics of the wind speed to benefit greatly from its power. in this study a systematic analysis used in detecting long-term correlations, known as detrended fluctuation analysis (DFA), is applied to wind speed data from 20 wind generation stations located at the northwestern part of Turkey. The analysis revealed that all of the wind speed records have a crossover in the scaling exponent (a). At long time scales, the data exhibit long-range correlation around alpha similar to 0.87. No meaningful relationships between the scaling exponents and some topographic properties (elevation and distance from the sea) of the observation stations are observed, i.e. all of the wind records gave almost similar scaling exponents regardless of their topographic properties. (C) 2009 Elsevier Ltd. All rights reserved.