Limited-area models (LAMs) are presently used for a wide variety of research and operational forecasting applications, and such use will likely expand greatly as the rapid increase in the performance/price ratio of computers and workstations makes LAMs more accessible to novice users. The robustness of these well-tested and documented models will make it tempting for many to consider them as turn-key systems that can be used without any experience or formal training in numerical weather prediction. This paper is intended as a tutorial and caution for such prospective model users, with the specific purpose of illustrating that, in spite of advanced physical-process parameterizations and high resolutions permitted by faster computers, and modern mesoscale data for initial conditions, there is still a basic limitation to predictability with a LAM-lateral boundary conditions (LBC). Illustrations are provided of previous work that show the serious negative effects of LBCs, and guidelines are provided for helping to minimize their negative impact on forecast quality.

No abstract

No abstract