An adaptive time-step for increased model efficiency

Typically, in numerical weather prediction models, the integration time-step is a user-configurable constant that is chosen at the beginning of a simulation. The minimum necessary time-step is driven by the most extreme vertical and horizontal motions expected throughout a model simulation. Caution is required in choosing the time-step because a time-step that is too long will cause model instability and simulation failure, while a time-step that is too short will require unnecessary computing power. Further, since the time-step remains constant, unnecessary computing power is used when the most extreme vertical and horizontal motions subside.

Here, a method is presented for automatically setting and adapting the time-step during a simulation based upon the underlying weather conditions. The time step is increased when the maximum vertical and horizontal motions within the modeling domain decrease, resulting in an increased average time-step. With the increased average time-step, fewer steps are required to complete the simulation, yielding significant improvements in model efficiency.

In operational 12-km simulations, using the WRF-ARW core, for the continental United States, the implementation of the adaptive time-step has yielded a mean model run-time decrease of 36%, with a maximum decrease of 43% and a minimum decrease of 27% over 70 simulations. The adaptive time-step has been tested in simulations with resolutions ranging from 4 to 36 km and with simulations that utilize nests.