The accurate prediction of wind speed (WS) can facilitate the effective utilization of wind energy. However, the complex nonlinear relationships between WS and meteorology variables make WS prediction a challengeable work. Furthermore, considering significant wake effects between adjacent wind turbines, the wind regimes subjected by different turbines tend to exhibit obvious spatio-temporal coupling characteristics. This article proposes a layered-vine copula-based WS prediction framework that considers both spatial correlations between turbines and meteorological information. The first layer extracts the spatial correlations of WS and uses a D-vine structure to describe the multidimensional WS dependence of wind turbines and develop a conditional quantile regression model. The second layer determines the impact of meteorological variables on WS, whereby the key variables are selected and the C-vine regression model is established for prediction correction. Finally, the proposed method is verified using real data measured from a wind farm that the correlations between WS and its influencing factors can be well modeled and thus increase the prediction accuracy.