Currently graphics tests in iris use matplotlib (average pixel difference) for graphics testing:
https://github.com/matplotlib/matplotlib/blob/master/lib/matplotlib/testing/compare.py#L264

Since problem specific tolerances are apparent and this becomes ever more apparent when running tests on other platforms for which these tests were not designed (OS X), I propose an investigation of an alternative graphics testing method.

Assigned: @cpelley