An Overview of High-Performance Computing and Challenges for the Future. http://www.netlib.org/utk/ people/JackDongarra/talks.htm

not available at time of publication. Eugenia Kalnay Department of Meteorology University of Maryland [email protected] MS28 POD Model Reduction of Large Scale Geophysical Models 4DVAR is a powerful tool for data assimilation in meteorology and oceanography. A major hurdle in use of 4DVAR for realistic models is dimension of control space ,generally equal to size of model state variable and typically of order 10 − 10 and high computational cost in computing cost function and gradient requiring integration of model and adjoint. This led to introduction of a reduced model approach of POD type by projecting the full model dynamics into reduced space. Experience with a large-scale PODbased reduced model and adjoint for unstructured ocean model, 3-D finite element Imperial College Ocean Model (ICOM) including adaptive mesh capability is presented along with work on a ocean reduced gravity model. An approach using 4-D VAR data assimilation with reduced order POD model will be outlined. Ionel M. Navon Florida State University School of Computational Science [email protected] MS28 Adaptive Observation of Contaminant Plume, Hurricane, and Ocean Systems Two new algorithms for Adaptive Observation are pre-