In this paper, we present a parameter estimation procedure for a 3D Navier-Stokes problem coupled with a 0D system. The idea is to get a reduced order (0D) model from the 3D model to do the parameter estimation. The estimation is performed on this reduced order model by using the Unscented Kalman Filter (UKF). Clinically measured data are taken into account during the parameter estimation and after a first 3D simulation with these parameters, a new iteration of the estimation procedure can be performed by updating the reduced order model according to the 3D results. This process can be repeated and leads to an iterative algorithm. Such a method is applied on a haemodynamics problem: the simulation of the blood flow in the aorta using Windkessel boundary conditions. The 3D results are shown to be in good agreement with the clinical data, showing that the parameters are well estimated.