This talk will describe
preliminary results from an eddy permitting
ocean and sea-ice configuration of the M.I.T.
General Circulation Model (MITgcm – see
http://mitgcm.org) that is designed to support
dynamically reasonable estimates of the full
three-dimensional, time-varying oceanic state
on decadal time-scales. Undertaken as part of
the Estimating the Circulation and Climate of
the Ocean (ECCO – see http://www.ecco-group.org)
project, the model configuration presented is
suited to constraints by both remotely-sensed
(altimeter, scatterometer, ocean temperature,
ice cover and gravity) data and by in-situ (temperature
and salinity profilers, mooring, drifter and
float) data using advanced, mathematically rigorous,
assimilation methods (adjoint method, Kalman
filter and Rauch-Tung-Striebel smoother). |
Our model configuration employs a novel gridding
approach that avoids singularities anywhere
on the sphere by projecting grid-lines from
an embedded cube onto the sphere surface. Using
this approach we obtain a truly global grid
with no singularities anywhere on the sphere.
The grid has maximal grid spacing of 25.9 km
and minimal grid spacing of 3 km. The global
coverage of the grid allows us to include an
interactive representation of the polar oceans
and of sea-ice into our assimilation system.
The grid spacing is sufficient to allow a rich
mesoscale eddy field to emerge. Using state-of-the-art
large-scale parallel computer resources, committed
to the project by NASA, the system simulates
several years in a day. This throughput is sufficient
to allow a practical, semi-continuous assimilation
system that partially resolves the ocean eddy
field to be envisaged. In this presentation
we will describe steps taken so far to achieve
an optimal system configuration for this work
and towards our goal of performing a planetary
scale “brain-scan” that will globally
map the full three-dimensional time dependent
state of the ocean and sea-ice system at high
resolution over the last ten to thirty years.
|
