Add a days_in_month accessor to CFTimeIndex

doc/examples/monthly-means.ipynb

@@ -29,89 +29,9 @@
     "import numpy as np\n",
     "import pandas as pd\n",
     "import xarray as xr\n",
-    "from netCDF4 import num2date\n",
     "import matplotlib.pyplot as plt "
    ]
   },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "#### Some calendar information so we can support any netCDF calendar.  "
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "ExecuteTime": {
-     "end_time": "2018-11-28T20:51:35.991620Z",
-     "start_time": "2018-11-28T20:51:35.960336Z"
-    }
-   },
-   "outputs": [],
-   "source": [
-    "dpm = {'noleap': [0, 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31],\n",
-    "       '365_day': [0, 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31],\n",
-    "       'standard': [0, 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31],\n",
-    "       'gregorian': [0, 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31],\n",
-    "       'proleptic_gregorian': [0, 31, 28, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31],\n",
-    "       'all_leap': [0, 31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31],\n",
-    "       '366_day': [0, 31, 29, 31, 30, 31, 30, 31, 31, 30, 31, 30, 31],\n",
-    "       '360_day': [0, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30, 30]} "
-   ]
-  },
-  {
-   "cell_type": "markdown",
-   "metadata": {},
-   "source": [
-    "#### A few calendar functions to determine the number of days in each month\n",
-    "If you were just using the standard calendar, it would be easy to use the `calendar.month_range` function."
-   ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {
-    "ExecuteTime": {
-     "end_time": "2018-11-28T20:51:36.015151Z",
-     "start_time": "2018-11-28T20:51:35.994079Z"
-    }
-   },
-   "outputs": [],
-   "source": [
-    "def leap_year(year, calendar='standard'):\n",
-    "    \"\"\"Determine if year is a leap year\"\"\"\n",
-    "    leap = False\n",
-    "    if ((calendar in ['standard', 'gregorian',\n",
-    "        'proleptic_gregorian', 'julian']) and\n",
-    "        (year % 4 == 0)):\n",
-    "        leap = True\n",
-    "        if ((calendar == 'proleptic_gregorian') and\n",
-    "            (year % 100 == 0) and\n",
-    "            (year % 400 != 0)):\n",
-    "            leap = False\n",
-    "        elif ((calendar in ['standard', 'gregorian']) and\n",
-    "                 (year % 100 == 0) and (year % 400 != 0) and\n",
-    "                 (year < 1583)):\n",
-    "            leap = False\n",
-    "    return leap\n",
-    "\n",
-    "def get_dpm(time, calendar='standard'):\n",
-    "    \"\"\"\n",
-    "    return a array of days per month corresponding to the months provided in `months`\n",
-    "    \"\"\"\n",
-    "    month_length = np.zeros(len(time), dtype=np.int)\n",
-    "    \n",
-    "    cal_days = dpm[calendar]\n",
-    "    \n",
-    "    for i, (month, year) in enumerate(zip(time.month, time.year)):\n",
-    "        month_length[i] = cal_days[month]\n",
-    "        if leap_year(year, calendar=calendar) and month == 2:\n",
-    "            month_length[i] += 1\n",
-    "    return month_length"
-   ]
-  },
   {
    "cell_type": "markdown",
    "metadata": {},
@@ -131,7 +51,7 @@
    "outputs": [],
    "source": [
     "ds = xr.tutorial.open_dataset('rasm').load()\n",
-    "print(ds)"
+    "ds"
    ]
   },
   {
@@ -143,7 +63,17 @@
     "- calculate the month lengths for each monthly data record\n",
     "- calculate weights using `groupby('time.season')`\n",
     "\n",
-    "Finally, we just need to multiply our weights by the `Dataset` and sum allong the time dimension.  "
+    "Finally, we just need to multiply our weights by the `Dataset` and sum allong the time dimension.  Creating a `DataArray` for the month length is as easy as using the `days_in_month` accessor on the time coordinate.  The calendar type, in this case `'noleap'`, is automatically considered in this operation."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "month_length = ds.time.dt.days_in_month\n",
+    "month_length"
    ]
   },
   {
@@ -157,13 +87,8 @@
    },
    "outputs": [],
    "source": [
-    "# Make a DataArray with the number of days in each month, size = len(time)\n",
-    "month_length = xr.DataArray(get_dpm(ds.time.to_index(), calendar='noleap'),\n",
-    "                            coords=[ds.time], name='month_length')\n",
-    "\n",
     "# Calculate the weights by grouping by 'time.season'.\n",
-    "# Conversion to float type ('astype(float)') only necessary for Python 2.x\n",
-    "weights = month_length.groupby('time.season') / month_length.astype(float).groupby('time.season').sum()\n",
+    "weights = month_length.groupby('time.season') / month_length.groupby('time.season').sum()\n",
     "\n",
     "# Test that the sum of the weights for each season is 1.0\n",
     "np.testing.assert_allclose(weights.groupby('time.season').sum().values, np.ones(4))\n",
@@ -183,7 +108,7 @@
    },
    "outputs": [],
    "source": [
-    "print(ds_weighted)"
+    "ds_weighted"
    ]
   },
   {
@@ -262,13 +187,9 @@
    "source": [
     "# Wrap it into a simple function\n",
     "def season_mean(ds, calendar='standard'):\n",
-    "    # Make a DataArray of season/year groups\n",
-    "    year_season = xr.DataArray(ds.time.to_index().to_period(freq='Q-NOV').to_timestamp(how='E'),\n",
-    "                               coords=[ds.time], name='year_season')\n",
-    "\n",
     "    # Make a DataArray with the number of days in each month, size = len(time)\n",
-    "    month_length = xr.DataArray(get_dpm(ds.time.to_index(), calendar=calendar),\n",
-    "                                coords=[ds.time], name='month_length')\n",
+    "    month_length = ds.time.dt.days_in_month\n",
+    "\n",
     "    # Calculate the weights by grouping by 'time.season'\n",
     "    weights = month_length.groupby('time.season') / month_length.groupby('time.season').sum()\n",
     "\n",
@@ -278,13 +199,6 @@
     "    # Calculate the weighted average\n",
     "    return (ds * weights).groupby('time.season').sum(dim='time')"
    ]
-  },
-  {
-   "cell_type": "code",
-   "execution_count": null,
-   "metadata": {},
-   "outputs": [],
-   "source": []
   }
  ],
  "metadata": {
@@ -304,7 +218,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.6.8"
+   "version": "3.7.3"
   },
   "toc": {
    "base_numbering": 1,
@@ -321,5 +235,5 @@
   }
  },
  "nbformat": 4,
- "nbformat_minor": 2
+ "nbformat_minor": 4
 }

doc/weather-climate.rst

@@ -95,7 +95,7 @@ For data indexed by a :py:class:`~xarray.CFTimeIndex` xarray currently supports:
 
 - Access of basic datetime components via the ``dt`` accessor (in this case
   just "year", "month", "day", "hour", "minute", "second", "microsecond",
-  "season", "dayofyear", and "dayofweek"):
+  "season", "dayofyear", "dayofweek", and "days_in_month"):
 
 .. ipython:: python
 
@@ -104,6 +104,7 @@ For data indexed by a :py:class:`~xarray.CFTimeIndex` xarray currently supports:
    da.time.dt.season
    da.time.dt.dayofyear
    da.time.dt.dayofweek
+   da.time.dt.days_in_month
 
 - Rounding of datetimes to fixed frequencies via the ``dt`` accessor:
 

doc/whats-new.rst

@@ -48,6 +48,13 @@ New Features
   By `Todd Jennings <https://github.com/toddrjen>`_
 - Allow plotting of boolean arrays. (:pull:`3766`)
   By `Marek Jacob <https://github.com/MeraX>`_
+- A ``days_in_month`` accessor for :py:class:`xarray.CFTimeIndex`, analogous to
+  the ``days_in_month`` accessor for a :py:class:`pandas.DatetimeIndex`, which
+  returns the days in the month each datetime in the index.  Now days in month
+  weights for both standard and non-standard calendars can be obtained using
+  the :py:class:`~core.accessor_dt.DatetimeAccessor` (:pull:`3935`).  This
+  feature requires cftime version 1.1.0 or greater.  By
+  `Spencer Clark <https://github.com/spencerkclark>`_.
 
 Bug fixes
 ~~~~~~~~~
@@ -69,7 +76,10 @@ Documentation
   :py:meth:`DataArray.diff` so it does document the ``dim``
   parameter as required. (:issue:`1040`, :pull:`3909`)
   By `Justus Magin <https://github.com/keewis>`_.
-
+- Updated :doc:`Calculating Seasonal Averages from Timeseries of Monthly Means
+  <examples/monthly-means>` example notebook to take advantage of the new
+  ``days_in_month`` accessor for :py:class:`xarray.CFTimeIndex`
+  (:pull:`3935`). By `Spencer Clark <https://github.com/spencerkclark>`_.
 
 Internal Changes
 ~~~~~~~~~~~~~~~~

xarray/coding/cftimeindex.py

@@ -243,6 +243,9 @@ class CFTimeIndex(pd.Index):
         "dayofyr", "The ordinal day of year of the datetime", "1.0.2.1"
     )
     dayofweek = _field_accessor("dayofwk", "The day of week of the datetime", "1.0.2.1")
+    days_in_month = _field_accessor(
+        "daysinmonth", "The number of days in the month of the datetime", "1.1.0.0"
+    )
     date_type = property(get_date_type)
 
     def __new__(cls, data, name=None):

xarray/tests/__init__.py

@@ -62,6 +62,7 @@ def LooseVersion(vstring):
 has_pynio, requires_pynio = _importorskip("Nio")
 has_pseudonetcdf, requires_pseudonetcdf = _importorskip("PseudoNetCDF")
 has_cftime, requires_cftime = _importorskip("cftime")
+has_cftime_1_1_0, requires_cftime_1_1_0 = _importorskip("cftime", minversion="1.1.0.0")
 has_dask, requires_dask = _importorskip("dask")
 has_bottleneck, requires_bottleneck = _importorskip("bottleneck")
 has_nc_time_axis, requires_nc_time_axis = _importorskip("nc_time_axis")

xarray/tests/test_cftimeindex.py

@@ -15,7 +15,7 @@
 )
 from xarray.tests import assert_array_equal, assert_identical
 
-from . import raises_regex, requires_cftime
+from . import raises_regex, requires_cftime, requires_cftime_1_1_0
 from .test_coding_times import (
     _ALL_CALENDARS,
     _NON_STANDARD_CALENDARS,
@@ -229,6 +229,13 @@ def test_cftimeindex_dayofweek_accessor(index):
     assert_array_equal(result, expected)
 
 
+@requires_cftime_1_1_0
+def test_cftimeindex_days_in_month_accessor(index):
+    result = index.days_in_month
+    expected = [date.daysinmonth for date in index]
+    assert_array_equal(result, expected)
+
+
 @requires_cftime
 @pytest.mark.parametrize(
     ("string", "date_args", "reso"),