You need initial conditions and boundary conditions for up to the end of your run time. Your boundary condition data can extend beyond your forecast length, but it can't be shorter, as this is a fatal error.
A model without boundary condition data will do nothing useful at the boundaries. Eventually, this could contaminate the forecast.
Let me give you an example. Suppose your initial conditions are entirely in a warm sector with a strong arctic front just to the north of the boundary at t=0. Given a winter time scenerio, you would expect that cold air to plunge south in your domain. Without the use of the boundary data from another model, this would never happen, as your run wouldn't know there is a strong front near the domain.
Kevin W. Thomas
Center for Analysis and Prediction of Storms
University of Oklahoma