In traffic flow modelling with macroscopic approach, the velocity variable is depend on density. The higher the density, the lower the velocity and vice versa. In this research, Finite Volume Methods (FVM) with the Flux-Corrected Transport (FCT) for numerical approximation is used. FCT also compared with the other numerical approximation such as Upwind Scheme and Lax-Wendroff. The traffic data that consist of the surpassing vehicles through the domain control were collected from merging lane phenomenon in JL. Perintis Kemerdekaan, Bandung. Each lane will be representated as a flux. Then, polynomial least square is used to approximate the velocity function. The results are v(rho) = 29.03rho^2 - 56.11rho + 30.26 for polynomial second order lane 1 with 2.167 RMSE and 0.5744 R-square, v(rho) = 64.3rho^2 - 97.64rho+39.51 for polynomial second order lane 2 with 5.865 RMSE and 0.6904 R-square, and v(rho) = 19rho^2 - 47.67rho + 29.96 for polynomial second order lane 3 with 2.934 RMSE and 0.7574 R-square. Lastly, the velocity function is substituted in Flux-Corrected Transport approximation to simulate the position of a leading vehicle. The results are, the lower the density the emptier the queue and vice versa.