Diagnostic tools for 3D unstructured oceanographic data

TL;DR

This paper demonstrates how open source software and high-level programming languages can be used to efficiently extract diagnostic quantities from 3D unstructured ocean model data, simplifying analysis of complex mesh structures.

Contribution

It introduces a method for constructing diagnostic tools for unstructured ocean models using Python and open source packages, applicable across different unstructured grid codes.

Findings

Tools can be built with Python, NumPy, SciPy, VTK, and MayaVi.

Method applied to flux calculation in unstructured mesh data.

Approach is adaptable to various unstructured ocean model outputs.

Abstract

Most ocean models in current use are built upon structured meshes. It follows that most existing tools for extracting diagnostic quantities (volume and surface integrals, for example) from ocean model output are constructed using techniques and software tools which assume structured meshes. The greater complexity inherent in unstructured meshes (especially fully unstructured grids which are unstructured in the vertical as well as the horizontal direction) has left some oceanographers, accustomed to traditional methods, unclear on how to calculate diagnostics on these meshes. In this paper we show that tools for extracting diagnostic data from the new generation of unstructured ocean models can be constructed with relative ease using open source software. Higher level languages such as Python, in conjunction with packages such as NumPy, SciPy, VTK and MayaVi, provide many of the high-level primitives needed to perform 3D visualisation and evaluate diagnostic quantities, e.g. density fluxes. We demonstrate this in the particular case of calculating flux of vector fields through isosurfaces, using flow data obtained from the unstructured mesh finite element ocean code ICOM, however this tool can be applied to model output from any unstructured grid ocean code.