In this paper, we create a Markov chain model to generate random wind profiles up to an altitude of 40 km based on actual wind data over Luc Ngan dist., Bac Giang, Vietnam. The wind speed and direction are assumed statistically uncorrelated; therefore, two independent Markov chains are built for these quantities. For the wind speed, the probability of transition between discrete states is approximated by the Weibull distribution whereas the von Mises probability density function is applied to the transition of the wind direction. The relationships between the coefficients of the probability density functions and the altitude at each state of the wind speed and direction are obtained. The statistical properties of the simulated wind profiles are compared with those of the actual data to validate the present Markov-chain model.