One of the ultimate goal of ensemble forecasting is to reliably estimate the time-evolution of probabilistic density functions (PDFs) of meteorological fields, given existing of intrinsic uncertainties in both initial condition (IC) and model physics. However, due to model error and imperfect IC perturbations used in real-world ensemble forecasting, it is believed that this task is extremely difficult if not impossible. Currently, little is known about the impact of model error or imperfect perturbations on the evolution of ensemble-based PDFs in operational numerical weather prediction (NWP) models. Since it's impossible to study the PDF evolution for full fields in full space, this preliminary study uses only a small subset of fields at fixed locations to demonstrate how their PDFs evolve when model error or imperfect IC perturbation or both of them present in an ensemble system. The NCEP short-range ensemble forecasting (SREF) system which based on two regional models (Eta and RSM) and bred IC perturbations was used in this study.