The Construction of 3-d Neutral Density for Arbitrary Data Sets

Abstract:

The Neutral Density variable allows inference of water pathways from
thermodynamic properties in the global ocean, and is therefore an
essential component of global ocean circulation analysis. The widely
used algorithm for the computation of Neutral Density yields accurate
results for data sets which are close to the observed climatological
ocean. Long-term numerical climate simulations, however, often generate
a significant drift from present-day climate, which renders the existing
algorithm inaccurate. To remedy this problem, new algorithms which
operate on arbitrary data have been developed, which may potentially
be used to compute Neutral Density during runtime of a numerical model.
We review existing approaches for the construction of Neutral Density
in arbitrary data sets, detail their algorithmic structure, and
present an analysis of the computational cost for implementations on
a single-CPU computer. We discuss possible strategies for the
implementation in state-of-the-art numerical models, with a focus on
distributed computing environments.