Ensemble percentiles

src/valenspy/__init__.py

@@ -11,7 +11,7 @@
 from valenspy.diagnostic import Diagnostic, Model2Self, Model2Ref, Ensemble2Ref, Ensemble2Self
 from valenspy.diagnostic.visualizations import *
 #Utility
-from valenspy._utilities import is_cf_compliant, cf_status
+from valenspy._utilities import is_cf_compliant, cf_status, datatree_to_dataset, datatree_to_dataframe, restructure_by_level, split_by_level
 
 # =============================================================================
 # Version

src/valenspy/_utilities/__init__.py

@@ -5,6 +5,7 @@
     load_yml, 
     generate_parameters_doc
 )
-from._formatting import create_named_regex, parse_string_to_time_period
+from ._formatting import create_named_regex, parse_string_to_time_period
 from .cf_checks import is_cf_compliant, cf_status
 from .unit_converter import CORDEX_VARIABLES, _convert_all_units_to_CF
+from ._datatree import datatree_to_dataset, datatree_to_dataframe, restructure_by_level, split_by_level

src/valenspy/_utilities/_datatree.py

@@ -0,0 +1,98 @@
+import pandas as pd
+import xarray as xr
+from copy import deepcopy
+
+def split_by_level(dt: xr.DataTree, level: int):
+    """
+    Split a DataTree into multiple DataTrees based on the unique values at a given level in the node paths.
+
+    Parameters
+    ----------
+    dt : xr.DataTree
+        The DataTree to split.
+    level : int
+        The level in the node paths to split the DataTree by. Level 0 is the root level.
+
+    Returns
+    -------
+    dict
+        A dictionary where the keys are the unique values at the specified level in the node paths, and the values are the corresponding DataTrees containing only the nodes with that value at the specified level.
+    """
+    #Make a deep copy of the datatree to avoid modifying the original one. This is needed as we are going to orphan the datasets in the new datatrees, which would also orphan them in the original datatree if we don't make a copy.
+    dt = deepcopy(dt)
+    if level >= 2:
+        dt = restructure_by_level(dt, level)
+    result = {value: dt[value] for value in set(dt.children)}
+    for value in result:
+        result[value].orphan()
+    return result
+
+def restructure_by_level(dt: xr.DataTree, level: int):
+    """
+    Restructure a DataTree such that level n in the node paths becomes the new root level.
+
+    Parameters
+    ----------
+    dt : xr.DataTree
+        The DataTree to restructure.
+    level : int
+        The level in the node paths to restructure the DataTree by. Level 0 is the root level.
+
+    Returns
+    -------
+    xr.DataTree
+        A restructured DataTree where level n in the node paths becomes the new root level.
+    """
+    if level < 2:
+        raise ValueError("Level must be greater than or equal to 2 as reording the first level leaves the tree unchanged")
+    level = level - 1 #As the root level is not included in the path split, we need to subtract 1 from the level to get the correct index.
+    reorganized_nodes = {
+        "/".join([path.split("/")[level]] + path.split("/")[:level] + path.split("/")[level+1:]): node.dataset
+        for path, node in dt.subtree_with_keys
+        if len(path.split("/")) > level #Strict
+    }
+    return xr.DataTree.from_dict(reorganized_nodes)
+
+def datatree_to_dataset(dt: xr.DataTree, **kwargs):
+    """
+    Convert a DataTree to a xarray Dataset.
+
+    Parameters
+    ----------
+    dt : xr.DataTree
+        The DataTree to convert to a xarray Dataset.
+    **kwargs : dict
+        Keyword arguments to pass to the xarray concat function.
+    """
+    datasets = []
+    for key, ds in dt.to_dict().items():
+        if ds:
+            ds_copy = ds.copy()
+            ds_copy = ds_copy.expand_dims({"id": [str(key)]})
+            datasets.append(ds_copy)
+
+    return xr.concat(datasets, dim="id", **kwargs)
+
+def datatree_to_dataframe(dt: xr.DataTree, add_attributes=False):
+    """
+    Convert a DataTree to a pandas DataFrame.
+    """
+    data_frames = []
+    for key, ds in dt.to_dict().items():
+        if ds:
+            if not ds.dims:  # Non-dimensional datasets
+                df = pd.DataFrame({var: float(ds[var].values) for var in ds.data_vars}, index=[key])
+            else:  # Dimensional datasets
+                df = ds.to_dataframe()
+            df["id"] = str(key)
+
+            if add_attributes:
+                for attr in add_attributes:
+                    #if attr is a substring of any attribute in ds.attrs:
+                    for ds_attr in ds.attrs:
+                        if attr in ds_attr:
+                            df[attr] = ds.attrs[ds_attr]
+                            break
+            data_frames.append(df)
+
+    return pd.concat(data_frames, axis=0).reset_index()

src/valenspy/diagnostic/_ensemble2ref.py

@@ -2,7 +2,26 @@
 from valenspy.diagnostic.functions import *
 from valenspy.diagnostic.visualizations import *
 
-__all__ = ["MetricsRankings"]
+__all__ = [
+    "ClimateChangeSignal",
+    "ClimateChangeSignalOfSpatialMean",
+    "MetricsRankings",
+    ]
+
+ClimateChangeSignalOfSpatialMean = Ensemble2Ref(
+    climate_change_signal_of_spatial_mean,
+    plot_map,
+    "Climate Change Signal of the spatial means",
+    "The spatial climate change signal as the difference between the temporal average of two periods",
+    plot_type="facetted"
+)
+
+ClimateChangeSignal = Ensemble2Ref(
+    mean_climate_change_signal,
+    lambda ds : ds,
+    "Climate Change Signal",
+    "The climate change signal as the difference between the spatial and temporal average of two periods."
+)
 
 MetricsRankings = Ensemble2Ref(
     calc_metrics_dt,

src/valenspy/diagnostic/_ensemble2self.py

@@ -0,0 +1,14 @@
+from valenspy.diagnostic.diagnostic import Ensemble2Self
+from valenspy.diagnostic.functions import *
+from valenspy.diagnostic.visualizations import *
+
+__all__ = [
+    "Ensemble_Quantile_Spatial_Mean"
+]
+
+Ensemble_Quantile_Spatial_Mean = Ensemble2Self(
+    ensemble_quantile_of_spatial_mean,
+    plot_map_per_dimension,
+    "Ensemble quantiles of spatial mean",
+    "The quantiles accross the ensembles spatial mean."
+)