
	UCAR \| NCAR \| CISL \| SCD > CSS Home Computational Science ReSET

Computational Science

HOMME on the IBM BlueGene/L - NCAR researchers, funded in part by the Department of Energy's Climate Change Prediction Program, have built a scalable and efficient spectral-element-based atmospheric dynamical core using the University of Colorado's Computational Science Section's High Order Method Modeling Environment (HOMME). In order for this to be a useful tool for atmospheric scientists it is necessary to couple this core to physics packages employed by the community.

Inverse Ocean Modeling System - This modular system, known as the "Inverse Ocean Modeling System" or "IOM" (Chua and Bennett, 2001) is based on a twenty-year effort. From a control theory perspective, IOM effects fixed-interval smoothing of nonlinear dynamics and nonlinear observing processes. The dynamics may be weak constraints; as a result, the number of computational degrees may be very large. From a statistical perspective, IOM minimizes the estimator of maximum likelihood for random errors in the dynamics and data. IOM makes such large, nonlinear but functionally smooth optimization problems feasible by iterating on the nonlinearities, yielding linear smoothing problems, and by making preconditioned searches in the "data subspace" at each iterate. The system is in the process of being optimized for several computer architectures.

SciDAC-CCCP - The goal of this project is to develop a scalable conservative dynamical core for CCSM that addresses atmospheric transport issues such as mass conservation and monotonicity preservation. Our starting point is a high-order inherently conservative shallow water model based on the discontinuous Galerkin (DG) method (Nair et al. 2004, Mon. Wea. Rev.) for the cubed-sphere. To reduce development costs we will integrate this model into the NCAR high-order multiscale modeling environment (HOMME) as this contains a 3D dynamical core based on the continuous Galerkin method which efficiently scales to thousands of processors. However, there are several issues remaining. Extending the model to a conservative 3D dynamical core requires that the low-order finite-difference approach employed in HOMME for the vertical discretization be replaced. One promising approach is high-order Lagrangian vertical advection methods based on conservative remapping. Another issue is achieving high integration rates. For this we consider implicit/semi-implicit and OIF schemes (St-Cyr and Thomas, (2004), Appl. Num. Math.). Finally, we need to determine whether this core is suitable for climate simulations and its computational efficiency. We will leverage work currently underway with IBM and Sandia National Laboratories to integrate CAM and CRCP physics into HOMME for AMIP and Aquaplanet simulations on the IBM BlueGene/L and Cray Red Storm platforms.

BOB "Built On Beowulf" Dynamical Core - BOB is an efficient spectral dynamical core, written in F77, and primarily designed to run efficiently on small clusters composed of distributed memory computers using MPI.