This paper investigates the capability of the Multimodel Ensemble approach to improve the deterministic forecast of wake vortex behavior and to produce reliable vortex habition areas. Therefore the models D2P, APA 3.2, APA 3.4, APA 3.8 and TDP 2.1 were provided within the framework of a NASA-DLR cooperation. In a previous study the Bayesian Model Averaging (BMA) approach, that computes the ensemble forecast as a weighted sum of PDFs, turned out to be promising. For this reason the focus of this paper lies on the further development and assessment of this method. While the previously presented methods did not take into account that the error increases temporally, the new approach considers growing uncertainties. In addition, combined confidence areas for the vertical and lateral vortex position are derived from bivariate probability density distributions. For training and evaluation wake vortex campaigns accomplished by NASA (MEM95, DFW97, DEN03, MEM13) and DLR (WakeMUC, WakeFRA, WakeOP) are employed.