Installing Iris on OSX 10.9.2

Installation

Following the instructions here.  I use the Canopy virtualenv which is set up for you when you install Canopy, and I make sure this is activated so that the correct modules path is used.  Then install

  • Cartopy is in the Canopy package manager, so is easy to install using the GUI
  • Pyke – can be installed using pip as in ‘pip install pyke’ which should take care of dependencies
  • udunits – install via homebrew

Finally, clone the github repository (I use the github app, which I find works really well, so just click on ‘clone in desktop’ and choose a location for cloning) and run ‘python setup.py install’ from the command line using, I dunno, iterm.  After this, ready to go.

 Getting started with a pp file

  • May need to generate a pp file first using um2pp: um2pp /path/to/um_output_file /path/to/pp_file
  • May need to convert your pp to bigendian using bigend:  bigend -32 /path/to/pp_file path/to/32_bit_bigendian_pp_file

If you need any, my sample pp files are in ~ptg21/iris_test_data on Hector

Initial script

import iris
import iris.plot as iplt
import iris.quickplot as qplt
filename='test_32.pp'
cubes=iris.load(filename)
print (cubes)
13: unknown / (unknown) (model_level_number: 60; latitude: 73; longitude: 96)
14: unknown / (unknown) (model_level_number: 60; latitude: 73; longitude: 96)
15: unknown / (unknown) (model_level_number: 60; latitude: 73; longitude: 96)

so some work still to do here, but the data are still available:

wind_speed=cubes[65]

contour=qplt.contour(wind_speed)

plt.gca().coastlines()

plt.clabel(contour, fontsize=6)

plt.show()

figure_1

EDIT 18/02/2015  – Conda now works well (http://www.scitools.org.uk/iris/docs/latest/installing.html)

Advertisements