Under the background of the large-scale increase in wind power utilization,the application of typical scenarios to deal with the uncertainty of wind output is of great significance. Aiming at the spatial-temporal correlation among the output of multiple wind farms,an improved scenario generation and reduction method is proposed,and an evaluation method is introduced to test the quality of the generated scenarios. The exponential function is used to construct a multivariate covariance matrix that reflects the temporal correlation of wind power,and the Copula function is used to build a multi-wind farm spatial correlation model. A large number of initial scenes are generated by performing spatio-temporal correlation non-linear transformation and equal probability inverse transformation on the cumulative probability distribution function of random numbers and historical data. The K-means clustering method is improved,and the optimal number of clusters is determined by the elbow method and the clustering effectiveness index,and then the representative spatial-temporal correlation wind scenarios are reduced. Finally,four evaluation indicators are proposed to test the quality of the scenarios. The calculation results show that the volatility,climbing situation and spatial-temporal correlation of the scenarios generated by the proposed method are more consistent with historical data. The proposed method has a higher coverage of actual measured wind power values than another method does.