Application of lateral boundary condition perturbations to help restore dispersion in limited-area ensemble forecasts

Abstract

In a companion paper, the authors showed that lateral boundary condition (LBC) constraints on small-scale error variance growth are sufficient to limit dispersion in limited-area-model (LAM) ensemble simulations. The error growth constraints result from the use of coarsely resolved and temporally interpolated LBCs. The effect is present in any modeling system using "one-way" LBC forcing unless the forcing model has the same resolution as the LAM and the LBCs are updated at every time step. This limitation suggests the need to apply statistically consistent, finescale LBC perturbations at every time step during LAM simulations. In this paper, a new method for implementing LBC perturbations is developed to help counter the above effect by creating a statistically consistent source of error growth along the lateral boundaries. The LBC perturbations are designed to amplify with time while coherently propagating into the domain. The procedure is tested in a controlled and efficient manner using a modified barotropic channel model. Ten-member ensemble simulations are produced over many cases on a periodic channel domain and each of four smaller nested domains. Lateral boundary effects are specifically isolated since the simulations are perfect except for initial and lateral boundary condition perturbations and the use of coarsely resolved and/or temporally interpolated one-way LBCs. Statistical results accumulated over 100 independent cases demonstrate that the application of LBC perturbations capably restores ensemble dispersion, especially on smaller domains where LBC effects propagate quickly through the domain. The paper closes with some comments on the relevance of the LBC perturbation procedure in practical settings. © 2004 American Meteorological Society.