Pandas 2.2 and xarray 2023.11

CHANGES.rst

@@ -20,6 +20,7 @@ New features and enhancements
 
 Breaking changes
 ^^^^^^^^^^^^^^^^
+* With pandas 2.2 and xarray 2023.11.0, many frequency strings where changed : Y becomes YE, M -> ME, Q -> QE. A and AS are removed. T, L, U, N become min, ms, us and ns.
 * `bump2version` has been replaced with `bump-my-version` to bump the version number using configurations set in the `pyproject.toml` file. (:issue:`1557`, :pull:`1569`).
 * `xclim`'s units registry and units formatting are now extended from `cf-xarray`. The exponent sign "^" is now never added in the ``units`` attribute. For example, square meters are given as "m2" instead of "m^2" by xclim, both are still accepted as input. (:issue:`1010`, :pull:`1590`).
 * `yamale` is now listed as a core dependency (was previously listed in the `dev` installation recipe). (:issue:`1595`, :pull:`1596`).

environment.yml

@@ -16,14 +16,14 @@ dependencies:
     - lmoments3
     - numba
     - numpy >=1.16
-    - pandas >=0.23,<2.2
+    - pandas >=2.2
     - pint >=0.9
     - poppler >=0.67
     - pyyaml
     - scikit-learn >=0.21.3
     - scipy >=1.2
     - statsmodels
-    - xarray >=2022.06.0,<2023.11.0
+    - xarray >=2023.11.0
     - yamale
     # Extras
     - eofs

pyproject.toml

@@ -44,14 +44,13 @@ dependencies = [
   "lmoments3>=1.0.5",
   "numba",
   "numpy>=1.16",
-  "pandas>=0.23,<2.0; python_version == '3.8'",
-  "pandas>=0.23,<2.2; python_version >= '3.9'",
+  "pandas>=2.2; python_version >= '3.9'",
   "pint>=0.10",
   "pyyaml",
   "scikit-learn>=0.21.3",
   "scipy>=1.2",
   "statsmodels",
-  "xarray>=2022.06.0,<2023.11.0",
+  "xarray>=2023.11.0",
   "yamale"
 ]
 

tests/test_atmos.py

@@ -272,7 +272,7 @@ def test_wind_power_potential_from_3h_series():
     from xclim.testing.helpers import test_timeseries
 
     w = test_timeseries(
-        np.ones(96) * 15, variable="sfcWind", start="7/1/2000", units="m s-1", freq="3H"
+        np.ones(96) * 15, variable="sfcWind", start="7/1/2000", units="m s-1", freq="3h"
     )
     out = atmos.wind_power_potential(wind_speed=w)
 

tests/test_bootstrapping.py

@@ -26,9 +26,9 @@ class Test_bootstrap:
         "var,p,index,freq, cftime",
         (
             ["tas", 98, tg90p, "MS", False],
-            ["tasmin", 98, tn90p, "A-JUL", False],
-            ["tasmax", 98, tx90p, "Q-APR", False],
-            ["tasmax", 98, tx90p, "Q-APR", True],
+            ["tasmin", 98, tn90p, "YE-JUL", False],
+            ["tasmax", 98, tx90p, "QE-APR", False],
+            ["tasmax", 98, tx90p, "QE-APR", True],
             ["tasmin", 2, tn10p, "MS", False],
             ["tasmax", 2, tx10p, "YS", False],
             ["tasmax", 2, tx10p, "YS", True],

tests/test_calendar.py

@@ -58,7 +58,7 @@ def da(index):
     )
 
 
-@pytest.mark.parametrize("freq", ["6480H", "302431T", "23144781S"])
+@pytest.mark.parametrize("freq", ["6480h", "302431min", "23144781s"])
 def test_time_bnds(freq, datetime_index, cftime_index):
     da_datetime = da(datetime_index).resample(time=freq)
     da_cftime = da(cftime_index).resample(time=freq)
@@ -91,11 +91,11 @@ def test_time_bnds_irregular(typ):
         start = xr.cftime_range("1990-01-01", periods=24, freq="MS")
         # Well. xarray string parsers do not support sub-second resolution, but cftime does.
         end = xr.cftime_range(
-            "1990-01-01T23:59:59", periods=24, freq="M"
+            "1990-01-01T23:59:59", periods=24, freq="ME"
         ) + pd.Timedelta(0.999999, "s")
     elif typ == "pd":
         start = pd.date_range("1990-01-01", periods=24, freq="MS")
-        end = pd.date_range("1990-01-01 23:59:59.999999999", periods=24, freq="M")
+        end = pd.date_range("1990-01-01 23:59:59.999999999", periods=24, freq="ME")
 
     time = start + (end - start) / 2
 
@@ -147,7 +147,7 @@ def test_percentile_doy_invalid():
     tas = xr.DataArray(
         [0, 1],
         dims=("time",),
-        coords={"time": pd.date_range("2000-01-01", periods=2, freq="H")},
+        coords={"time": pd.date_range("2000-01-01", periods=2, freq="h")},
     )
     with pytest.raises(ValueError):
         percentile_doy(tas)
@@ -156,10 +156,10 @@ def test_percentile_doy_invalid():
 @pytest.mark.parametrize(
     "freqA,op,freqB,exp",
     [
-        ("D", ">", "H", True),
+        ("D", ">", "h", True),
         ("2YS", "<=", "QS-DEC", False),
         ("4W", "==", "3W", False),
-        ("24H", "==", "D", True),
+        ("24h", "==", "D", True),
     ],
 )
 def test_compare_offsets(freqA, op, freqB, exp):
@@ -276,8 +276,8 @@ def test_get_calendar_errors(obj):
         ("standard", "noleap", True, "D"),
         ("noleap", "default", True, "D"),
         ("noleap", "all_leap", False, "D"),
-        ("proleptic_gregorian", "noleap", False, "4H"),
-        ("default", "noleap", True, "4H"),
+        ("proleptic_gregorian", "noleap", False, "4h"),
+        ("default", "noleap", True, "4h"),
     ],
 )
 def test_convert_calendar(source, target, target_as_str, freq):
@@ -312,7 +312,7 @@ def test_convert_calendar(source, target, target_as_str, freq):
     [
         ("standard", "360_day", "D"),
         ("360_day", "default", "D"),
-        ("proleptic_gregorian", "360_day", "4H"),
+        ("proleptic_gregorian", "360_day", "4h"),
     ],
 )
 @pytest.mark.parametrize("align_on", ["date", "year"])
@@ -332,17 +332,17 @@ def test_convert_calendar_360_days(source, target, freq, align_on):
 
     if align_on == "date":
         np.testing.assert_array_equal(
-            conv.time.resample(time="M").last().dt.day,
+            conv.time.resample(time="ME").last().dt.day,
             [30, 29, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30],
         )
     elif target == "360_day":
         np.testing.assert_array_equal(
-            conv.time.resample(time="M").last().dt.day,
+            conv.time.resample(time="ME").last().dt.day,
             [30, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30, 29],
         )
     else:
         np.testing.assert_array_equal(
-            conv.time.resample(time="M").last().dt.day,
+            conv.time.resample(time="ME").last().dt.day,
             [30, 29, 30, 30, 31, 30, 30, 31, 30, 31, 29, 31],
         )
     if source == "360_day" and align_on == "year":
@@ -357,7 +357,7 @@ def test_convert_calendar_360_days_random():
         dims=("time",),
         coords={
             "time": date_range(
-                "2004-01-01", "2004-12-31T23:59:59", freq="12H", calendar="default"
+                "2004-01-01", "2004-12-31T23:59:59", freq="12h", calendar="default"
             )
         },
     )
@@ -366,7 +366,7 @@ def test_convert_calendar_360_days_random():
         dims=("time",),
         coords={
             "time": date_range(
-                "2004-01-01", "2004-12-30T23:59:59", freq="12H", calendar="360_day"
+                "2004-01-01", "2004-12-30T23:59:59", freq="12h", calendar="360_day"
             )
         },
     )
@@ -395,8 +395,8 @@ def test_convert_calendar_360_days_random():
     "source,target,freq",
     [
         ("standard", "noleap", "D"),
-        ("noleap", "default", "4H"),
-        ("noleap", "all_leap", "M"),
+        ("noleap", "default", "4h"),
+        ("noleap", "all_leap", "ME"),
         ("360_day", "noleap", "D"),
         ("noleap", "360_day", "D"),
     ],
@@ -556,7 +556,7 @@ def test_clim_mean_doy(tas_series):
 
 def test_doy_to_days_since():
     # simple test
-    time = date_range("2020-07-01", "2022-07-01", freq="AS-JUL")
+    time = date_range("2020-07-01", "2022-07-01", freq="YS-JUL")
     da = xr.DataArray(
         [190, 360, 3],
         dims=("time",),
@@ -587,7 +587,7 @@ def test_doy_to_days_since():
     xr.testing.assert_identical(da, da2)
 
     # with start
-    time = date_range("2020-12-31", "2022-12-31", freq="Y")
+    time = date_range("2020-12-31", "2022-12-31", freq="YE")
     da = xr.DataArray(
         [190, 360, 3],
         dims=("time",),
@@ -624,10 +624,10 @@ def test_doy_to_days_since():
 @pytest.mark.parametrize(
     "freq,em,eb,es,ea",
     [
-        ("4AS-JUL", 4, "A", True, "JUL"),
-        ("M", 1, "M", False, None),
-        ("YS", 1, "A", True, "JAN"),
-        ("3A", 3, "A", False, "DEC"),
+        ("4YS-JUL", 4, "Y", True, "JUL"),
+        ("ME", 1, "M", False, None),
+        ("YS", 1, "Y", True, "JAN"),
+        ("3YE", 3, "Y", False, "DEC"),
         ("D", 1, "D", True, None),
         ("3W", 21, "D", True, None),
     ],
@@ -649,8 +649,8 @@ def test_parse_offset_invalid():
 @pytest.mark.parametrize(
     "m,b,s,a,exp",
     [
-        (1, "A", True, None, "AS-JAN"),
-        (2, "Q", False, "DEC", "2Q-DEC"),
+        (1, "Y", True, None, "YS-JAN"),
+        (2, "Q", False, "DEC", "2QE-DEC"),
         (1, "D", False, None, "D"),
     ],
 )
@@ -694,7 +694,7 @@ def test_convert_doy():
         dims=("time",),
         coords={
             "time": xr.date_range(
-                "2000-01-01", periods=5, freq="AS-JUL", calendar="standard"
+                "2000-01-01", periods=5, freq="YS-JUL", calendar="standard"
             )
         },
         attrs={"is_dayofyear": 1, "calendar": "standard"},

tests/test_checks.py

@@ -108,15 +108,15 @@ def test_assert_daily(self, date_range):
     def test_bad_frequency(self, date_range):
         with pytest.raises(ValidationError):
             n = 365
-            times = date_range("2000-01-01", freq="12H", periods=n)
+            times = date_range("2000-01-01", freq="12h", periods=n)
             da = xr.DataArray(np.arange(n), [("time", times)], attrs=self.tas_attrs)
             tg_mean(da)
 
     # Decreasing index
     def test_decreasing_index(self, date_range):
         with pytest.raises(ValidationError):
             n = 365
-            times = date_range("2000-01-01", freq="12H", periods=n)
+            times = date_range("2000-01-01", freq="12h", periods=n)
             da = xr.DataArray(
                 np.arange(n), [("time", times[::-1])], attrs=self.tas_attrs
             )
@@ -149,25 +149,25 @@ def test_check_hourly(self, date_range, random):
         }
 
         n = 100
-        time = date_range("2000-01-01", freq="H", periods=n)
+        time = date_range("2000-01-01", freq="h", periods=n)
         da = xr.DataArray(random.random(n), [("time", time)], attrs=tas_attrs)
-        datachecks.check_freq(da, "H")
+        datachecks.check_freq(da, "h")
 
-        time = date_range("2000-01-01", freq="3H", periods=n)
+        time = date_range("2000-01-01", freq="3h", periods=n)
         da = xr.DataArray(random.random(n), [("time", time)], attrs=tas_attrs)
         with pytest.raises(ValidationError):
-            datachecks.check_freq(da, "H")
+            datachecks.check_freq(da, "h")
 
         with pytest.raises(ValidationError):
-            datachecks.check_freq(da, ["H", "D"])
+            datachecks.check_freq(da, ["h", "D"])
 
-        datachecks.check_freq(da, "H", strict=False)
-        datachecks.check_freq(da, ["H", "D"], strict=False)
-        datachecks.check_freq(da, "3H")
-        datachecks.check_freq(da, ["H", "3H"])
+        datachecks.check_freq(da, "h", strict=False)
+        datachecks.check_freq(da, ["h", "D"], strict=False)
+        datachecks.check_freq(da, "3h")
+        datachecks.check_freq(da, ["h", "3h"])
 
         with pytest.raises(ValidationError, match="Unable to infer the frequency of"):
-            datachecks.check_freq(da.where(da.time.dt.dayofyear != 5, drop=True), "3H")
+            datachecks.check_freq(da.where(da.time.dt.dayofyear != 5, drop=True), "3h")
 
     def test_common_time(self, tas_series, date_range, random):
         tas_attrs = {
@@ -176,7 +176,7 @@ def test_common_time(self, tas_series, date_range, random):
         }
 
         n = 100
-        time = date_range("2000-01-01", freq="H", periods=n)
+        time = date_range("2000-01-01", freq="h", periods=n)
         da = xr.DataArray(random.random(n), [("time", time)], attrs=tas_attrs)
 
         # No freq
@@ -187,7 +187,7 @@ def test_common_time(self, tas_series, date_range, random):
             datachecks.check_common_time([db, da])
 
         # Not same freq
-        time = date_range("2000-01-01", freq="6H", periods=n)
+        time = date_range("2000-01-01", freq="6h", periods=n)
         db = xr.DataArray(random.random(n), [("time", time)], attrs=tas_attrs)
         with pytest.raises(ValidationError, match="Inputs have different frequencies"):
             datachecks.check_common_time([db, da])
@@ -197,6 +197,6 @@ def test_common_time(self, tas_series, date_range, random):
         db["time"] = db.time + pd.Timedelta(30, "min")
         with pytest.raises(
             ValidationError,
-            match=r"All inputs have the same frequency \(H\), but they are not anchored on the same minutes",
+            match=r"All inputs have the same frequency \(h\), but they are not anchored on the same minutes",
         ):
             datachecks.check_common_time([db, da])

tests/test_ensembles.py

@@ -129,7 +129,7 @@ def test_create_unequal_times(self, ensemble_dataset_objects, open_dataset):
         [(xr.cftime_range, {"calendar": "360_day"}), (pd.date_range, {})],
     )
     def test_create_unaligned_times(self, timegen, calkw):
-        t1 = timegen("2000-01-01", periods=24, freq="M", **calkw)
+        t1 = timegen("2000-01-01", periods=24, freq="ME", **calkw)
         t2 = timegen("2000-01-01", periods=24, freq="MS", **calkw)
 
         d1 = xr.DataArray(

tests/test_ffdi.py

@@ -149,7 +149,7 @@ def test_ffdi_indicators(self, open_dataset, init_kbdi, limiting_func):
         # outputs look sensible
         test_data = open_dataset(data_url)
 
-        pr_annual = test_data["pr"].resample(time="A").mean().mean("time")
+        pr_annual = test_data["pr"].resample(time="YS").mean().mean("time")
         pr_annual.attrs["units"] = test_data["pr"].attrs["units"]
 
         if init_kbdi:

tests/test_generic.py

@@ -26,7 +26,7 @@ def test_season_default(self, q_series):
 
     def test_season(self, q_series):
         q = q_series(np.arange(1000))
-        o = generic.select_resample_op(q, "count", freq="AS-DEC", season="DJF")
+        o = generic.select_resample_op(q, "count", freq="YS-DEC", season="DJF")
         assert o[0] == 31 + 29
 
 
@@ -97,7 +97,7 @@ def test_calendars(self):
         )
 
         out = generic.aggregate_between_dates(
-            data_std, start_std, end_std, op="sum", freq="AS-JUL"
+            data_std, start_std, end_std, op="sum", freq="YS-JUL"
         )
 
         # expected output
@@ -110,7 +110,7 @@ def test_calendars(self):
 
         # check calendar conversion
         out_noleap = generic.aggregate_between_dates(
-            data_std, start_std, end_noleap, op="sum", freq="AS-JUL"
+            data_std, start_std, end_noleap, op="sum", freq="YS-JUL"
         )
 
         np.testing.assert_allclose(out, out_noleap)

tests/test_generic_indicators.py

@@ -104,7 +104,7 @@ def test_missing(self, ndq_series):
         np.testing.assert_array_equal(out.sel(time="1902").isnull(), True)
 
     def test_3hourly(self, pr_hr_series, random):
-        pr = pr_hr_series(random.random(366 * 24)).resample(time="3H").mean()
+        pr = pr_hr_series(random.random(366 * 24)).resample(time="3h").mean()
         out = generic.stats(pr, freq="MS", op="var")
         assert out.units == "kg2 m-4 s-2"
         assert out.long_name == "Variance of variable"

tests/test_helpers.py

@@ -88,7 +88,7 @@ def test_day_lengths(method):
 
 
 def test_cosine_of_solar_zenith_angle():
-    time = xr.date_range("1900-01-01T00:30", "1900-01-03", freq="H")
+    time = xr.date_range("1900-01-01T00:30", "1900-01-03", freq="h")
     time = xr.DataArray(time, dims=("time",), coords={"time": time}, name="time")
     lat = xr.DataArray(
         [0, 45, 70], dims=("site",), name="lat", attrs={"units": "degree_north"}

tests/test_indicators.py

@@ -816,7 +816,7 @@ def test_resampling_indicator_with_indexing(tas_series):
     np.testing.assert_allclose(out, [28, 29])
 
     out = xclim.atmos.tx_days_above(
-        tas, thresh="0 degC", freq="AS-JUL", doy_bounds=(1, 50)
+        tas, thresh="0 degC", freq="YS-JUL", doy_bounds=(1, 50)
     )
     np.testing.assert_allclose(out, [50, 50, np.NaN])