What’s new

Only some of the major changes are tracked here.

v0.4.1 (07 October 2024)

s2stools.utils.wrap_time() and s2stools.utils.unwrap_time() are added to handle different dimensions of time (time versus winter/ timestepofseason)
ensure compatibility with xarray 2024.09.0, numpy 2 and python 3.12
added s2stools.process.sel_fc_around_dates(): select forecast data around a given list of dates, useful e.g., for selecting forecasts initialized around sudden stratospheric warming events

breaking changes to s2stools.clim.climatology()
- parameter groupby is now "leadtime" by default instead of "validtime"
- support for groupby="validtime" is dropped for now, due to incompatibility with recent xarray version
new function s2stools.process.reft_hc_year_to_fc_init_date() to convert s2s data reftime-hindcast_year to forecast initialisation date

s2stools.plot.cmap_spread() accepts to new arguments less_dark_blue and less_dark_red to control saturation limits
new function s2stools.plot.legend_colored_labels_no_lines() behaves like ax.legend, but instead of drawing handles and labels next to each other, only labels are created, but in the color of the respective handle
new functions s2stools.plot.set_xticks_with_colored_labels() and s2stools.plot.set_yticks_with_colored_labels() makes colored xticklabels
new function s2stools.plot.add_map_inset() adds a small map to an existing axis, which can be used to indicate the geographic domain of a particular analysis
experimental: new function s2stools.compute.register_sample_variance_aggregation_for_flox() registers a function "sample_variance" to flox, which behaves like variance, but with ddof=1
s2stools.clim.climatology() now accepts hindcast-only datasets
s2stools.clim.climatology() now raises a warning if used with flox (we should support flox in the future, do you want to implement it?)

internal changes:

s2stools.utils.lat_weighted_spatial_mean() makes it quicker to average a dataset over a spatial domain, because it directly accounts for latitudinal dependece of grid box areas
s2stools.utils.groupby_quantiles() implements an adaption of xarray’s groupby_bins: while groupby_bins by default splits the data into bins of the same size, groupby_quantiles splits data into groups which have the same subset size
s2stools.process.s2sparser() supports 6hrly data, as required for precipitation for example
s2stools.process.s2sparser() supports data without dimension “number”
two new colorbars, and the default kwargs for plotting on a North Atlantic map
new features: s2stools.plot.add_map() and s2stools.plot.add_box()
download nceps nao index to xarray: s2stools.indices.nao()
added s2stools.compute.css() to compute correlation skill score
fix an issue with s2stools.indices.download_mjo()