Some other utility functions (`s2stools.utils`)

Functions

s2stools.utils.add_years(dt64, years)

Add year to date.

Parameters:

dt64 ([datetime64], datetime64) – date to change
years ([int], int) – how many years to add. If scalar, add same year to each date, if list then years must have same shape as dt64.

Returns:

new date(s) with same shape as dt64

s2stools.utils.groupby_quantiles(dataarray, groupby, q, q_dim, labels=None)

xr.core.dataset.Dataset.groupby_bins(…, bins=N) groups the data into N groups which span same interval sizes. This method groups the data into len(q) bins, which have equal number of subdatasets in the respective groups.

Parameters:

xr.DataArray) (dataarray (xr.Dataset |)
str) (groupby (xr.Dataset | xr.DataArray |)
(list) (q)
(str) (q_dim)
([str]) (labels)

Returns:

groupby_object

Return type:

xarray.core.groupby.DatasetGroupBy | xarray.core.groupby.DataArrayGroupBy

s2stools.utils.lat_weighted_spatial_mean(dataarray, lon_name='longitude', lat_name='latitude')

Spatial average over longitude, latitude. Gridpoints are weighted by cos(lat) to account for different areas on Gaussian grid on a sphere.

Parameters:

(xr.DataArray) (dataarray)
(str) (lat_name)
(str)

Return type:

weighted_dataarray (xr.DataArray)

s2stools.utils.month_int_to_abbr(months)

convert e.g. [12, 1, 2] to [“Dec”, “Jan”, “Feb”] :param months: one month or list of months as integers

Returns:: list of str

s2stools.utils.to_timedelta64(a, assume='D')

Convert to numpy timedelta64

Parameters:

a (int or np.timedelta64) – timedelta
() (assume) – timedelta64 format that is assumed if a is of type int.

Returns:

np.timedelta64

s2stools.utils.unwrap_time(xr_data)

Transform data back from (“season”, “timestepofseason”) to a linear time axis.

Parameters:: xr.DataArray) (xr_data (xr.Dataset |)
Returns:: xr_data
Return type:: xr.Dataset | xr.DataArray

See also

wrap_time()

s2stools.utils.wrap_time(xr_data, season_start_month, change_dims=True)

Transform data from a linear time axis to a reshaped pair of ("season", "timestepofseason"). Can for example be used to go from time to DJF and day since Dec 1.

Parameters:

xr.DataArray) (xr_data (xr.Dataset |)
(int) (season_start_month)
(bool) (change_dims) – If False, the time dimension, and season and timestepofseason are just added as new coordinates.

Returns:

xr_data

Return type:

xr.Dataset | xr.DataArray

Examples

>>> times = pd.date_range("2000-11-01", "2001-12-01", freq="D")
>>> nt = len(times)
>>> ds = xr.DataArray(
>>>     np.random.randint(low=0, high=100, size=(nt, 10)),
>>>     coords=dict(time=times, x=np.arange(10)),
>>>     dims=["time", "x"],
>>> )
>>> print(ds)
<xarray.DataArray (time: 396, x: 10)>
array([[27, 34, 84, ..., 68, 16, 61],
       [12, 98, 94, ..., 25, 38, 87],
       [ 4, 27, 91, ..., 78, 48, 55],
       ...,
       [ 1, 75, 64, ..., 21, 21,  7],
       [18, 96,  9, ..., 60, 85,  2],
       [31, 28, 36, ..., 64, 48, 97]])
Coordinates:
  * time     (time) datetime64[ns] 2000-11-01 2000-11-02 ... 2001-12-01
  * x        (x) int64 0 1 2 3 4 5 6 7 8 9
>>> ds_transformed = wrap_time(ds, season_start_month=11, change_dims=True)
>>> print(ds_transformed)
<xarray.DataArray (season: 2, x: 10, timestepofseason: 365)>
array([[[46.,  3., 62., ..., 63., 15., 69.],
        [22., 23., 89., ..., 40., 47., 63.],
        ....
        [91.,  7., 78., ..., nan, nan, nan],
        [87., 30., 88., ..., nan, nan, nan]]])
Coordinates:
  * x                   (x) int64 0 1 2 3 4 5 6 7 8 9
  * timestepofseason    (timestepofseason) int64 1 2 3 4 5 ... 362 363 364 365
    time                (season, timestepofseason) datetime64[ns] 2000-11-01 ...
  * season              (season) int64 2000 2001
    season_start_month  int64 11

Some other utility functions (s2stools.utils)

Functions

Some other utility functions (`s2stools.utils`)