The core of scenario-based automated driving simulation testing is the generation of high-coverage concrete test scenarios. Therefore, the German PEGASUS project proposed a framework for scenario development termed as functional-logical-concrete scenarios. However, there is currently a lack of systematic implementation methods from logic to concrete scenarios. To solve the problem, this paper proposed a concrete scenario generation method based on the combination of Gaussian Mixture Model (GMM) and Markov Chain Monte Carlo (MCMC) sampling, and then applied it on the high-coverage cut-in scenarios generation on freeways. Firstly, we extracted 2,422 real freeway cut-in segments from a naturalistic driving dataset in China, and calibrated a cut-in vehicle (CV) kinematic trajectory model. Subsequently, we selected seven parameters from the trajectory model and the test vehicle (TV) initial state for the cut-in logical scenario description, and constructed a joint probability density function with GMM for the parameters. Furthermore, we generated concrete cut-in scenarios for TV and CV using MCMC sampling. By calculating the Jensen-Shannon divergence (JSD) between generated and actual trajectories, we found that 5 × 105 concrete cut-in scenarios achieved high-coverage for the logical cut-in scenarios. The result demonstrated the effectiveness of the method for high-coverage concrete scenarios generation from logical scenario in this study.