Temporal discretization

Several different time integration schemes were tested during the course of model development. The original model used the Leap-Frog (centered in time and space) scheme together with the Asselin Asselin (1972) temporal filter (to control computational modes). The Leap-Frog time integration scheme has really been the "workhorse" of modern meteorology, and has been used extensively in GCM, NWP and mesoscale research models. It has been used extensively in numerical modeling as it is a one step method, however, there are computational mode problems so that it is used in conjunction with a time filter. An additional drawback of the Leap-Frog scheme is in terms of computer memory, as prognostic variables at two time levels must be stored (present and 1 past time step). In addition, it is not self starting, so another method is generally used for the first time step, such as a simple predictor-corrector method. In summary, the Leap-Frog method is being increasingly abandoned in the favor of other schemes, such as the rather popular RK method. In ASP, options to use the Leap-frog scheme have been removed in favor of the Runge-Kutta time integration scheme.