v1.2.0

This release changes the default behaviour of to_dataframe. Bound type variables will now be ignored, as they are typically redundant once a dataset has been converted to a dataframe. For example a variable like lon_bnds will now be ignored. To retain the old behaviour use to_dataframe(drop_bnds=False).

There are additional improvements to internals that do not modify package usage.

v1.1.11

This is a minor release that makes improvements to pub_plot.

The pub_plot method will now display arrows bars on the colour bar when limits are supplied to make it clear values can exceed those shown in the colour bar.

v1.1.10

Version 1.1.10

Quieter warnings on Linux. Improvements have been made to reduce the number of repetitive warnings on Linux.

The pub_plot method now lets you set the dpi in saved plots.

v1.1.8

This is a minor release that fixes a bug that caused errors when norm was set using pub_plot. Everything else should work as before.

v1.1.7

Minor bug fix release

This is a minor release that improves how pub_plot handles some arguments to provide more consistent behaviour.

v1.1.6

Adding the ability to create vertical level bounds

This release adds a new dataset method generate_level_bounds, which will generate vertical bounds from the vertical levels. This is useful in case this info is not provided in the original data and they need to be estimated for vertical averaging etc.

v1.1.5

This is a minor release that improves the package import speed.

Faster import speeds

Previously, the package tried to import geoviews at startup. This was a bottleneck, as gevoviews takes a long time to import. Geoviews is now imported when first used.

v1.1.3

This is a minor release with some improvements to temporal methods, such as tmean.

Enhanced temporal methods

You can now supply a window argument to statistical temporal methods. For example, if you wanted to calculate a weekly mean value, you could do the following: ds.tmean(window=7).

v1.1.2

This is a minor release.

The subset method has been improved internally. Previously, if you ran, for example, ds.subset(month=1) and one or more of the files in a dataset had no data for January, an error was thrown. This has now been changed, so that a warning is thrown and the files are removed from the dataset.

v1.1.1

This is a minor release with some improvements to internals. Code written for the previous release will be fully compatible.

The annual_anomaly method is now able to run in parallel will multiple cores are defined using options.

The match_points method now has a quiet argument.