v1.7.1

CIS 1.7.1 fixes

• Fixed an issue where gridded->ungridded interpolation would unmask masked source arrays
• Support for Pandas 0.24

v1.7.0

CIS 1.7 features

• Extended the definition of the MODIS_L3 plugin to include monthly files.
• Added plugins Aerosol_CCI_L3 and Cloud_CCI_L3 to read Level 3 CCI products.
• Extended the Aeronet plugin to read files from:
  • The Version 3 Direct Sun Algorithm,
  • The Maritime Aerosol Network,
  • All versions of the Spectral Decomposition Algorithm,
  • The All_Sites_Times_*.dat files now distributed under the "Download All Sites" link.

Incompatible changes

• Renamed the plugins Aerosol_CCI and Cloud_CCI to Aerosol_CCI_L2 and Cloud_CCI_L2 to be consistent with MODIS.
• Updated to Iris 2.0.0 and pyHDF 0.9.0 (removing previous workarounds).
• Gridded / gridded collocation of time coordinates is no longer supported since iris no longer allows the
  determination of whether a point lies within a bounded region for datetime-like objects

Bugs fixed

• In a PP file, if a var_name contains spaces the plugin will now attempt to replace them with underscores.

v1.6.0

CIS 1.6 features

• Implemented ungridded - ungridded collocation performance improvements (#9)
• Improved reading of NCAR-RAF style NetCDF files
• Performance improvement when reading many (100s of) NetCDF files
• Improved support for reading CALIOP L2 data

Bugs fixed

• In previous versions of CIS cartographic weights were calculated for all collapse operations regardless of the dimensions being collapsed. This may have given unexpected values in some cases. CIS now only calculates weights when collapsing over latitude.
• Plot colourbars are now associated with the plot axes rather than the figure, which makes it easier to make
  multi-axes plots with CIS.
• Fixed an issue where UngriddedData.copy() didn't copy metadata
• [JASCIS-373] get_variable_names now gets passed the product argument from read_data_list
• Fix #11 by relying on shapely exception which moved
• [JASCIS-375] Ensure the mask is retained when expanding coordinates from 1d to 2d
• The Cloudsat reader no longer expands coordinate arrays for 1-d datasets (such as ice-water path)

v1.5.4

CIS 1.5.4 fixes

• Minor fix for the info command on Windows

v1.5.3

Bug fixes

• Fixed a (potentially serious) bug in unit parsing which would convert any string to lowercase.
• [JASCIS-367] Make the name() method more consistent between gridded and ungridded data
• Minor fix when reading variables from PP files with spaces in the name

v1.5.2

New features

• Gridded and ungridded datasets can now be subset to an arbitrary lat/lon (shapely) shape.

Bug fixes

• Slicing and copying Coords now preserves the axis attribute
• Fixed an issue where subsetting gridded data over multiple coordinates sometimes resulted in an error
• CIS will now catch errors when writing out metadata values which might have special types and can't be safely
  cast (e.g. VALID_RANGE).
• Minor fix for log scale color bars
• Minor fix for parsing the command aliases
• Minor fix for creating data lists from iterators

v1.5.1

Bug fixes

• Minor fix in interpreting units when reading some NetCDF data in Python 2
• Fixed an issue where line and scatter plots weren't respecting the yaxis keyword

v1.5.0

CIS 1.5 features

• The biggest change is that CIS can now be used as a Python library, all of the command line tools are now easily
  available through Python. This allows commands to be run sequentially in memory, slicing of gridded or ungridded
  datasets and easy integration with other Python packages such as Iris and Pandas.
• Taylor diagrams - CIS is now able to plot Taylor diagrams which are an excellent way of quantitatively comparing two
  or more (collocated) datasets
• All map plots are now able to be plotted in any of the available Cartopy projections, see
  http://scitools.org.uk/cartopy/docs/latest/crs/projections.html for a full list.

Incompatible changes

• Since aggregation of gridded datasets has quite a different set of options as compared to the aggregation of
  ungridded datasets, the aggregate command has been deprecated for gridded datasets. It is still supported through
  command line for the time being, but will be removed in future releases. Please use the collapse command instead.

Bugs fixed

• [JASCIS-268] The plotting routines have been re-architected to allow easier testing and extension.
• [JASCIS-357] Added deprecation for the aggregation of gridded datasets
• [JASCIS-329] Metadata objects now attempt to use cf_units for all units, but will fall back to strings if needed. In
  future releases we may insist on plugins providing standard units.

v1.4.0

CIS 1.4 features

• CIS now includes full support for Python => 3.4, as well as Python 2.7
• New verbose and quiet flags allow for control over how much CIS commands output to the screen. The default verbosity has also changed so that by default only warnings and errors will be output to the screen. The full debug output remains for the cis.log file.
• Significant optimizations have been made in gridded -> ungridded collocation which should now be considerably faster. Also, when collocating multiple gridded source datasets the interpolation indices are now cached internally leading to further time savings.
• Any valid_range attributes in supported NetCDF or HDF files (including MODIS, CALIOP and CloudSat) files are now automatically respected by CIS. All data values outside of the valid range are masked. Data from NetCDF files with valid_min or valid_max attributes is also masked appropriately.
• CloudSat missing and missop attributes are now read and combined to mask out values which don't conform to the inequality defined.
• [JASCIS-342] The extrapolation modes are now consistent across both gridded->gridded and gridded->ungridded collocation modes. The default is no extrapolation (gridded->gridded would previously extrapolate). This can still be overridden by the user.
• [JASCIS-128] If the output file already exists the user is now prompted to overwrite it. This prompt can be disabled by using the --force-overwrite argument, or setting the CIS_FORCE_OVERWRITE environment variable to 'TRUE'.

Incompatible changes

• To accommodate the new verbose flags (-v) the info command now takes a single datagroup argument, and optional variable names, as reflected in the updated documentation.

• CIS no longer prepends ungridded output files with 'cis-'. Instead CIS creates a global attribute in the output file called source which contains 'CIS'. This is checked in the updated CIS plugin when reading any NetCDF file.

  Note - While this is much neater going forward and will hopefully save a lot of head scratching it will mean CIS is unable to read old files produced by CIS automatically. All commands can be forced to use the CIS product by including the product=cis keyword argument. Alternatively you can update the data file manually using the following command:
  ncatted -O -a source,global,a,c,"CIS" in.nc

Bugs fixed

• [JASCIS-34] MODIS L3 data is now correctly treated as gridded data.
• [JASCIS-345] Product regular expression matching now matches the whole string rather than just the start.
• [JASCIS-360] Collocation now correctly applies the 'missing_data_for_missing_sample' logic for all collocations.
• [JASCIS-361] Fixed the CloudSat scale and offset transformation so that they are now applied correctly.
• [JASCIS-281] Fixed a caching error when aggregating multiple ungridded datasets which could lead to incorrect values
• CIS no longer crashes when the CIS_PLUGIN_HOME path cannot be found

v1.3.4

Fixed a minor test issue.