9:00 AM-9:45 AM
Chair: Kenneth R. Jackson, University of Toronto, Canada
Room: Convocation Hall
Global atmospheric models used for climate simulation and weather prediction are formulated in the same way, the difference between the two applications being spatial resolution and length of run. These atmospheric models describe three dimensional fluid flow above the spherical earth with state-dependent forcing. They consist of numerical approximations to the fluid dynamical component of the atmosphere, generally referred to as the dynamical core, and approximations for sub-grid scale processes, referred to as physical parameterizations, which provide the forcing and damping of the resolved scale fluid flow. The speaker will briefly review the current numerical methods of choice for the dynamical cores which have become accepted because they treat the fluid flow problem accurately and provide natural, economical solutions to the difficulties associated with spherical geometry. He will point out the advantages of these methods, illustrate their deficiencies, and discuss why improved methods are needed to eliminate these deficiencies which are becoming more troublesome as the physical parameterizations become more complex and interactive. One important question is how to evaluate new schemes for global atmospheric models, and how to isolate errors due to the numerical aspects from those due to the highly interactive physical parameterizations. The speaker will illustrate the difficulty of dealing with the complete model, pointing out the need for idealized test cases. He will describe current test cases and discuss their strengths and weaknesses.
David L. Williamson
Climate Modeling Section
National Center for Atmospheric Research