High-resolution limited-area models have been successfully used for both the simulation and prediction of regional and local mesoscale weather events (Anthes et al. 1989). This success can depend on accurately specifying the initial state of the atmosphere and its variation in time along the lateral boundaries of the model domain. Synoptic-scale global or regional analyses produced at operational weather centers do not provide sufficient resolution for the initial conditions required by a mesoscale model. The observations themselves must be used to enhance a regional analysis or to correct an initial-guess field produced by the mesoscale model itself (e.g., by integrating the model from an earlier operational regional analysis). Often the data themselves are of insufficient resolution to effectively enhance a global- or regional-scale analysis alone, and the data must be used in combination with the high-resolution mesoscale model (Anthes et al. 1989). To provide the lateral boundary conditions for the mesoscale model, forecasts from an operational global or regional model are usually used. For such an operational forecasting system, the forecasts are often too infrequent to provide sufficiently accurate lateral boundary values for the high-resolution mesoscale model. In this case it is useful to nest the high-resolution mesoscale model inside a limited-area model. This provides a link between the mesoscale model and the sparse data and infrequent boundary values provided by forecasts from a global or regional model.