Merge branch 'main' into dynamic-time-selection

CHANGELOG.rst

@@ -4,7 +4,11 @@ Changelog
 
 v0.55.0 (unreleased)
 --------------------
-Contributors to this version: Juliette Lavoie (:user:`juliettelavoie`), Trevor James Smith (:user:`Zeitsperre`), Sascha Hofmann (:user:`saschahofmann`), Baptiste Hamon (:user:`baptistehamon`), Pascal Bourgault (:user:`aulemahal`).
+Contributors to this version: Juliette Lavoie (:user:`juliettelavoie`), Trevor James Smith (:user:`Zeitsperre`), Sascha Hofmann (:user:`saschahofmann`), Pascal Bourgault (:user:`aulemahal`), Baptiste Hamon (:user:`baptistehamon`).
+
+Breaking changes
+^^^^^^^^^^^^^^^^
+* Missing value method "WMO" was modified to remove the criterion that the timeseries needs to be continuous (without holes). (:pull:`2058`).
 
 Announcements
 ^^^^^^^^^^^^^
@@ -23,13 +27,18 @@ New features and enhancements
 * `xclim` now tracks energy usage and carbon emissions ("last run", "average", and "total") during CI workflows using the `eco-ci-energy-estimation` GitHub Action. (:pull:`2046`).
 * ``xclim.testing.helpers.test_timeseries`` now accepts a `calendar` argument that is forwarded to ``xr.cftime_range``. (:pull:`2019`).
 * New ``xclim.indices.fao_allen98``, exporting the FAO-56 Penman-Monteith equation for potential evapotranspiration (:issue:`2004`, :pull:`2067`).
-* Time selection in ``xclim.core.calendar.select_time`` and the ``**indexer`` argument of indicators now support day-of-year bounds given as DataArrays with spatial and/or temporal dimensions. (:issue:`1987`, :pull:`2055`).
+* Missing values method "pct" and "at_least_n" now accept a new "subfreq" option that allows to compute the missing mask in two steps. When given, the algorithm is applied at this "subfreq" resampling frequency first and then the result is resampled at the target indicator frequency. In the output, a period is invalid if any of its subgroup where flagged as invalid by the chosen method. (:pull:`2058`, :issue:`1820`).
+* Time selection in ``xclim.core.calendar.select_time`` and the ``**indexer`` argument of indicators now supports day-of-year bounds given as DataArrays with spatial and/or temporal dimensions. (:issue:`1987`, :pull:`2055`).
 
 Internal changes
 ^^^^^^^^^^^^^^^^
 * `sphinx-codeautolink` and `pygments` have been temporarily pinned due to breaking API changes. (:pull:`2030`).
 * Adjusted the ``TestOfficialYaml`` test to use a dynamic method for finding the installed location of `xclim`. (:pull:`2028`).
 * Adjusted two tests for better handling when running in Windows environments. (:pull:`2057`).
+* Refactor of the ``xclim.core.missing`` module, usage of the ``Missing`` objects has been broken. (:pull:`2058`, :issue:`1820`, :issue:`2000`).
+    - Objects are initialized with their options and then called with the data, input frequency, target frequency and indexer.
+    - Subclasses receive non-resampled DataArray in their ``is_missing`` methods.
+    - ``MissingWMO`` now uses ``xclim.indices.helpers.resample_map`` which should greatly improve performance in a dask context.
 
 Bug fixes
 ^^^^^^^^^

pyproject.toml

@@ -137,7 +137,7 @@ target-version = [
 ]
 
 [tool.bumpversion]
-current_version = "0.54.1-dev.10"
+current_version = "0.54.1-dev.11"
 commit = true
 commit_args = "--no-verify --signoff"
 tag = false

src/xclim/__init__.py

@@ -13,7 +13,7 @@
 
 __author__ = """Travis Logan"""
 __email__ = "[email protected]"
-__version__ = "0.54.1-dev.10"
+__version__ = "0.54.1-dev.11"
 
 
 with _resources.as_file(_resources.files("xclim.data")) as _module_data:

src/xclim/core/indicator.py

@@ -1529,12 +1529,6 @@ def __init__(self, **kwds):
                 "Cannot set `missing_options` with `missing` method being from context."
             )
 
-        # Validate hard-coded missing options
-        kls = MISSING_METHODS[self.missing]
-        self._missing = kls.execute
-        if self.missing_options:
-            kls.validate(**self.missing_options)
-
         super().__init__(**kwds)
 
     def _history_string(self, das, params):
@@ -1546,6 +1540,8 @@ def _history_string(self, das, params):
 
         if missing != "skip":
             mopts = self.missing_options or OPTIONS[MISSING_OPTIONS].get(missing)
+            if mopts.get("subfreq", "absent") is None:
+                mopts.pop("subfreq")  # impertinent default
             if mopts:
                 opt_str += f", missing_options={mopts}"
 
@@ -1560,17 +1556,19 @@ def _postprocess(self, outs, das, params):
         outs = super()._postprocess(outs, das, params)
 
         freq = self._get_missing_freq(params)
-        if self.missing != "skip" or freq is False:
+        method = (
+            self.missing if self.missing != "from_context" else OPTIONS[CHECK_MISSING]
+        )
+        if method != "skip" and freq is not False:
             # Mask results that do not meet criteria defined by the `missing` method.
             # This means all outputs must have the same dimensions as the broadcasted inputs (excluding time)
-            options = self.missing_options or OPTIONS[MISSING_OPTIONS].get(
-                self.missing, {}
-            )
+            options = self.missing_options or OPTIONS[MISSING_OPTIONS].get(method, {})
+            misser = MISSING_METHODS[method](**options)
 
             # We flag periods according to the missing method. skip variables without a time coordinate.
             src_freq = self.src_freq if isinstance(self.src_freq, str) else None
             miss = (
-                self._missing(da, freq, src_freq, options, params.get("indexer", {}))
+                misser(da, freq, src_freq, **params.get("indexer", {}))
                 for da in das.values()
                 if "time" in da.coords
             )