The assessment of wind resources must be carried out to choose wind farm sites adequately. Additionally, input data on surface roughness maps and topographic maps are required to evaluate wind resources, where input data accuracy determines the accuracy of their overall analysis. To estimate this accuracy, we used met-mast data in Jeju and produced the ground roughness value for the Jeju region. To determine these values, an unsupervised classification method using SPOT-5 images was carried out for image classification. The wind resources of the northeastern part of Jeju were predicted, and the ground roughness map of the region was calculated by the WindPRO software. The wind speed of the Pyeongdae region of Jeju from the ground roughness map was calculated using WindPRO as 8.51 m/s. The wind speed calculated using the remote sensing technology presented in this study was 8.69 m/s. To assess the accuracy of the measured WindPro and the remote sensing technology values, we compared these results to the observed values in the Pyeongdae region using met-mast. This comparison shows that remote sensing data are more accurate than the WindPro data. We also found that the ground roughness map calculated in this study is useful for generating an accurate wind resource map of Jeju Island.