Acquisition of a Computational Facility for Numerical Dynamo Theory

点击翻译按钮获取中文摘要

9724568 Bloxham This grant, made through the Major Research Instrumentation (MRI) Program, provides $178,885 as partial support of the costs of acquiring a new computational facility at Harvard University for modeling the Earth's dynamo using their recently developed 3-d, dynamically self-consistent geodynamo code. Specifically, they will acquire a new Silicon Graphics Origin 2000 server with sixteen CPU's, 45 Gb of disk storage and 2 Gb of expanded memory. The current Kuang and Bloxham (Harvard) 3-d geodynamo model and the Glatzmaier (LANL) and Roberts (UCLA) model, whose results have recently been featured on the covers of the international and multi-disciplinary journals Nature and Science, are the only two currently developed and functional models of the geodynamo in the world. The Kuang-Bloxham model simulates a magnetic field that is generated well above the inner core close to the core-mantle boundary, while the Glatzmaier-Roberts model predicts a field generated just above the inner core. Both models show dipole reversals after a few decay times and the Glatzmaier- Roberts model, which was developed first and has been run long enough to simulate other geomagnetic reversals at ca. 100,000 ybp and 200,000 ybp, is supported by the Earth's paleomagnetic reversal records. The difference between these models lies in the setup of boundary conditions and choice of initial parameters for each of the models and the geophysical basis for each model representation can be and is, argued. Given limited computational resources for this research, both groups are forced to make simplifications to the models that result in codes that can be run in a reasonable amount of time (the Kuang-Bloxham model currently requires one month of dedicated time on their CARY computer for simulation of the Earth's dynamo for only one decay time). Given these limitations, unrealistic values of initial parameters are used (i.e. Ekman number, Prandtl number, Rayleigh number; see rev iew by Merrill). However, it is clear that increased parallel processing and memory are required to run this code within reasonable lengths of time to model the Earth's geodynamo over geologically significant time periods and it is important to support continued development and subsequent simulations using both of these competing models. ***