|Date: Friday, November 07, 2014
Location: 1084 East Hall (3:00 PM to 4:00 PM)
Title: Adaptive Mesh Refinement (AMR) and Variable-Resolution Techniques for Weather and Climate Models
Abstract: The atmosphere is characterized by motions that cover a wide range of spatial and temporal scales. They range from planetary-wave scales of order O(10,000 km) to mesoscales of order O(10-100 km) down to microscales of order O(1 m) and finer. Processes like rainfall even happen at the sub mm-scale and always need to be approximated via so-called physical parameterizations. Resolving all atmospheric processes on a computational grid is not possible in today's weather and climate models. Typically, climate models employ horizontal grid spacings of around 100 km, and thereby even have difficulty representing important mesoscale phenomena such as tropical cyclones which have a typical diameter between 500-1000 km.
One solution to this problem is the use of Adaptive Mesh Refinement (AMR) and variable-resolution techniques in atmospheric models. AMR grids can follow a feature of interests with a high-resolution nested grid as the flow evolves. Variable-resolution techniques describe static grid refinements in selected predetermined regions (e.g. an ocean basin for the study of tropical cyclones) and do not move during a forecast. The talk will first give an overview of the partial differential equations in weather and climate models. It then surveys the many mathematical design decisions that determine the accuracy of the numerical schemes and the resulting physical flow fields. Particular attention is paid to the numerical design of the AMR and variable-resolution techniques which are now emerging approaches for future-generation climate and weather models.
Speaker: Christiane Jablonowski