The integration of wind power into the power system causes difficulties in system operations due to its associated uncertainties. Spatial correlation is one aspect of uncertainties and derives from geographically dispersed distribution. However, spatial correlation is ignored by most studies, which decreases the modeling accuracy. In this paper, a high-dimensional t-copula model is first proposed to preserve spatial correlation characteristics. Through this model, adequate simulated data are sampled, which forms scenario generation. K-means clustering method is further used to reduce the number of scenarios generated. A case of 4 wind farms selected from AEMO dataset has been implemented to illustrate the establishment of the spatial correlation model. Then, a modified IEEE 24-bus test system is utilized to conduct optimal power flow (OPF) with and without the consideration of spatial dependence. The comparison results show that with spatial correlation accessing, the generation cost would have a certain increase. This implies, if the spatial correlation is ignored, the overall cost of OPF is underestimated. Furthermore, with the penetration level of wind power increasing, a lager cost gap is obtained, thus proving there exist benefits with the inclusion of spatial correlation during system modeling.