Atmospheric Solvers for a New Earth System Model (ESM)

Bill Skamarock

NCAR/MMM

Pressing scientific problems in weather prediction and climate science are driving the need for global nonhydrostatic atmospheric models, and at NCAR the development of a new Earth System Model (ESM) has been proposed to fulfill this need. The solvers for the Navier-Stokes equations lie at the center of these models, and a number of issues associated with spatial discretizations of the sphere, the numerical methods used with these discretizations, and the continued growth of massively parallel processing (MPP) computer architectures, has resulted in large number of very different candidate solvers. In contrast to the last decades of the previous millenium, when spherical-transform-based global models were predominant, there appears to be no obviously-superior solver architecture among the candidates at present, and the existing latitude-longitude grid models have drawbacks that are increasingly difficult to overlook. I will present a brief overview of the needs for the new solvers, requirements and desirable features for them, and some of the recent development efforts and their advances and outstanding problems.