aperezhortal/pyvet

Python implementation of the Variational Echo Tracking algorithm

IMPORTANT!: Project moved to the pySTEPS nowcasting initiative

This project is now part of the pySTEPS_ nowcasting initiative. The latest version of pyVET can be found in https://github.com/pySTEPS/pysteps

This repository is deprecated and it is left here to maintain a registry of the pyVET development history, but be aware that is not up to date.

pyVET - Variational Echo Tracking Algorithm

The pyVET package provides a python implementation of the Variational Echo Tracking presented in Laroche and Zawadzki (1995).

This algorithm is used by the McGill Algorithm for Prediction by Lagrangian Extrapolation (MAPLE) described in Germann and Zawadzki (2002).

Installing PyVET

Dependencies

The pyVET package need the following dependencies

• future
• numpy
• cython
• scipy
• skimage
• matplolib (for examples)

Install from source

The latest version can be installed manually by downloading the sources from https://github.com/aperezhortal/pyVET

Then, for a global installation run:

python setup.py install

For user installation:

python setup.py install --user

If you want to put it somewhere different than your system files, you can do:

python setup.py install --prefix=/path/to/local/dir

IMPORTANT: All the dependencies need to be already installed!

pyvet's Issues

Minimization procedure not finding absolute minimum

In some cases the minimization procedure do not converges to the absolute minimum (or one close to it). It converges to a local minima that is far away from the actual absolute minima.

For example, a rotor displacement field cannot be retrieved but the VET. Run simpleTest.py script in examples for more details.

Add support for real observations

Add a weighting factor to the Cost function to take in to account the quality and the availability of observations.

Adjust naming convention according to google standards

Update code to follow google standard.

https://github.com/google/styleguide/blob/gh-pages/pyguide.md

Replace morph function by skimage.transform.warp

Instead of using the cython morph function, use the warp transform from skimage.
Compare performance of each function.

Add Sphinx documentation and installation instructions

Add Sphinx documentation and installation instructions

Add control of minimization options in VET subroutine

Add necessary keywords to VET function be passes to the scipy minimization function.
**kwargs seems a good way of implement this.