Reconstruction of the time-mean absolute velocity field of the ocean circulation (GWI.5798)

In the near future, satellite observations of sea-surface height will become available, which will enable to determine the absolute velocity field over key areas in the global ocean. This project is intended to develop the necessary computational techniques to be able to derive the full time-mean three-dimensional velocity field in key regions of the global ocean from these observations.
To determine this velocity field, an inverse problem has to be solved within a chosen ocean model, given the observations of sea-surface height and those of the surface forcing of the flow (momentum and buoyancy fluxes). Because of the large number of degrees of freedom in the ocean model, a critical aspect in the solution of this inversion problem are efficient linear systems solvers.

In this project, such a method will be constructed through improvement of a recently developed iterative solver (MRILU) by using ideas from direct solvers and domain decomposition techniques. The MRILU method yields a multi-level incomplete factorization of the matrix associated with the linear system. This factorization is then applied as a preconditioning in a conjugate-gradient type method used to solve the linear system. The project will result in two deliverables:

1. an efficient linear systems solver, incorporated into a robust solution method for optimization of large-dimensional systems on massively parallel computers; and
2. an estimate of the absolute three-dimensional ocean velocity field in certain key areas on the globe, in particular in the North Atlantic Ocean.

Resultaten over het impliciete globale oceaanmodel zijn recentelijk gepubliceerd (J. Comp. Physics, 192, 452-470, 2003).

The results of the absolute velocity calculations of the ocean circulation using the GOCE and altimetry measurements will have direct interest from the agencies which have developed the particular instruments. The results will provide feedback for the improvement of existing Earth observation instruments and development of future ones. In the Netherlands, SRON (Utrecht) is involved in the GOCE mission (dr. R. Koop).

The project results will be of high interest, and will certainly be applied by, the climate modelling community. An equilibrium three-dimensional ocean circulation will lead to a better initialization of continental-scale as well as regional-scale climate models. In the Netherlands, regional as well as global climate modelling is performed by KNMI (Komen, KNMI).

Projectleider

Status van het project