In this paper, an unscented Kalman filter (UKF) coupled with a nonlinear model-predictive controller (NMPC) for a hydraulic wellbore model with multi-variable control and tracking is presented. In a wellbore, high drill string velocities in operational sequences such as tripping might result in surge and swab pressures in the annular section of the wellbore. To overcome these challenges, a controller incorporating safety and actuator limits should be used. A second-order model is used to predict axial drill string velocity downhole. With a NMPC specifying the block position trajectory, choke flow reference, desired backpressure pump flowrate and stand-pipe pressure, we can automatically supervise and control the pressure in the wellbore. To compensate for unmeasured states, an estimator is designed to predict the frictional pressure forces in the wellbore and filter noisy measurements. A stochastic approach for the hydraulic model is taken, including variance of the average fluctuations for the flow and pressure states. Comparing three NMPC configurations, the result of using an integration of the tracking error in the prediction model gave best offset-free tracking of the bottom-hole pressure. The controller compensates for the unknown fluctuations, and is shown to be robust towards model mismatch. Including the mechanical system in the NMPC prediction model, we can effectively constrain the predicted axial drill string velocity to reduce the pressure oscillations and achieve tracking of bottom hole pressure and choke differential pressure. The outcome is shown through extensive simulations to be an effective control strategy, reducing the pressure spikes while tripping.