Testbed 1 archive ADCIRC datasets have errors that prevent loading using Iris

During the SciPy 2014 code sprint today we discovered that at least some of the ADCIRC datasets in the testbed 1 archive thredds catalog have errors that prevent them from loading using Iris. I modified the NcML for one ADCIRC dataset until it loaded, discovering two problems:

there was an erroneous FillValue global attribute which needed to be removed
several variables that contained maximum values, and therefore didn't have a time dimension, nonetheless had a coordinate attribute with the value time y x.

Here's what I added to this NcML file: /data/comt_1_archive/inundation_tropical/UND_ADCIRC/Hurricane_Ike_2D_final_run_with_waves/00_dir.ncml

 <remove type="attribute" name="_FillValue"/>

     <variable name="depth">
      <attribute name="coordinates" value="y x" />
     </variable>

     <variable name="swan_HS_max">
      <attribute name="coordinates" value="y x" />
     </variable>

which then allows this to work:

import iris
url = 'http://comt.sura.org/thredds/dodsC/data/comt_1_archive/inundation_tropical/UND_ADCIRC/Hurricane_Ike_2D_final_run_with_waves'
var = 'sea_surface_height_above_geoid'
cube = iris.load_cube(url,var)

We need to check all the adcirc datasets and correct as necessary.

ioos / comt

Testbed 1 archive ADCIRC datasets have errors that prevent loading using Iris #15