Introduction

This is a Python implementation of the Gibbs SeaWater (GSW) Oceanographic Toolbox of TEOS-10. Extensive documentation is available from http://www.teos-10.org/; users of this Python package are strongly encouraged to study the documents posted there.

This implementation is based on GSW-C for core functions, with additional functions written in Python. GSW-C is the work of Frank Delahoyde and Glenn Hyland (author of GSW-Fortran, on which GSW-C is based), who translated and re-implemented the algorithms originally written in Matlab by David Jackett, Trevor McDougall, and Paul Barker.

The present Python library has an interface that is similar to the original Matlab code, but with a few important differences:

• Many functions in the GSW-Matlab toolbox are not yet available here.

• Taking advantage of Python namespaces, we omit the “gsw” prefix from the function names.

• Missing values may be handled using numpy.ma masked arrays, or using nan values.

• All functions follow numpy broadcasting rules; function arguments must be broadcastable to the dimensions of the highest-dimensioned argument. Recall that with numpy broadcasting, extra dimensions are automatically added as needed on the left, but must be added explicitly as needed on the right.

• Functions such as Nsquared that operate on profiles rather than scalars have an axis keyword argument to specify the index that is incremented along the pressure (depth) axis.

A wrapper around gsw called gsw-xarray exists for xarray. It adds CF compliant attributes when possible, units, and name.