Skip to content

Releases: ihesp/IPART

IPART v3.6.1

13 Jul 06:59
Compare
Choose a tag to compare

Same as v3.6.0. Just v3.6.0 didn't manage to update into conda-forge so I create a new build.

IPART v3.6.0

12 Jul 06:47
Compare
Choose a tag to compare

Minor bug fix:

  • Force tuple(slicer) for ndarray slicing: 8b4d186, 47a805a
  • Version compatibility fix for networkx: g.edge fails for later versions of networks, use g.adj if the former failed: bccbaff
  • Fix a zonal cyclic check bug: check only if zonal_cyclic parameter is set to True: bccbaff
  • Change selem to footprint in skimage.morphology funcs: 544aec0

IPART v3.5.0

20 Jun 08:47
Compare
Choose a tag to compare

Minor fix:

  • When data resolution is higher than 1.0 degrees, put axis-finding using down-sampled AR mask in a try block. If it failed, revert back to axis-finding using original resolution.

IPART v3.4.0

04 Mar 03:15
Compare
Choose a tag to compare

Minor fixes:

  • fix a bug in latitudinal range filtering when data cover both of the Northern and Southern Hemispheres.
  • more robust handling of zonally cyclic data.
  • (related to a change in v3.3.0) a better way to prevent potential matplotlib memory leaking.

IPART v3.3.0

03 Mar 07:00
Compare
Choose a tag to compare
  • Minor fixes

Use agg backend of matplotlib in utils/funcs.py to prevent memory leaking.

Allow specifying the calendar type (e.g. noleap) when reading netCDF data using readNC():
readNC(data_path, varid, calendar='noleap').

IPART v3.2.0

02 Mar 04:15
Compare
Choose a tag to compare

Speed optimization for the AR detection task.

For computations in scripts/detect_ARs.py and scripts/detect_ARs_generator_version.py, expect to see a 200 - 300 % speed up (only when the data resolution is higher than 1.0 degree latitude/longitude).

If the opencv module is also installed, up to 300 - 500 % speed gain (tested with 0.25 degree resolution data).

IPART v3.1.0

31 Aug 02:16
Compare
Choose a tag to compare

Fix a few bugs in the nc data metadata fetching utilities:

  1. Convert dimension names to lower cases (e.g. 'TIME' -> 'time') when reading metadata. This was found to be an issue with NCEP reanalysis.
  2. Deal with issues when the time axis has a units of 'days since 0001-01-01' (e.g. in case of 20CR reanalysis).
  3. Fix a bug in setting the equator-ward latitudinal bound for the Southern Hemisphere.

IPART v3.0.9

11 Jan 01:10
Compare
Choose a tag to compare

Fix an issue in netcdf _FillValue attribute when creating variable attributes.

IPART v3.0.8

31 Oct 06:57
3d8abc8
Compare
Choose a tag to compare

Tagged release for JOSS acceptance.

IPART v3.0.7

20 Oct 12:45
Compare
Choose a tag to compare
  • Fix a bug in the AR track filtering function. If you find some tracking missing, this might be the reason.
  • Fix related tests.