Sensitivity of limited area model data assimilation to lateral boundary condition fields

Abstract

Experiments with a limited area model forecasting system have been carried out to show the importance of accurate lateral boundary conditions, also during data assimilation forecast cycles. Advection of observed information from data-dense areas to data-sparse areas is an important process in any data assimilation, and for limited area assimilation, this process should also work properly across the lateral boundaries. A case study including a rapid cyclogenesis in the vicinity of Iceland is used to illustrate the deterioration of forecast quality as a result of too old forecast boundaries during data assimilation.