add BGC-Argo+ dataset support to ArgoFloat.open_dataset()

argopy/stores/float/argo_float.py

@@ -52,9 +52,16 @@ class ArgoFloat(FloatStore):
         ds = af.open_dataset('tech')
         ds = af.open_dataset('Rtraj')
         ds = af.open_dataset('Sprof')
-
         ds = af.open_dataset('Sprof', netCDF4=True)  # Return a netCDF4 Dataset instead of an xarray
 
+    .. code-block:: python
+        :caption: Load the BGC-Argo+ dataset (https://www.bgc-argo-plus.info)
+
+        # Fetch QC-processed, outlier-removed BGC data from the SOEST FTP server:
+        ds = af.open_dataset('BGCArgoPlus')
+        # Pin to a specific version:
+        ds = af.open_dataset('BGCArgoPlus', bgcplus_version='v0.1_2025_12')
+
     .. code-block:: python
         :caption: Other attributes and methods
 

argopy/stores/float/bgcargo_plus.py

@@ -0,0 +1,163 @@
+"""
+BGC-Argo+ dataset store for :class:`argopy.ArgoFloat`.
+
+The BGC-Argo+ dataset (https://www.bgc-argo-plus.info) is a quality-controlled,
+outlier-removed version of BGC-Argo float data curated at SOEST / University of
+Hawaiʻi at Mānoa (Bushinsky et al, 2025, Bushinsky et al, submitted).  Individual float files are served on the SOEST FTP server::
+
+    ftp://ftp.soest.hawaii.edu/bgc_argo_plus/outliers_removed/
+
+Usage
+-----
+The typical access path is through :class:`argopy.ArgoFloat`::
+
+    from argopy import ArgoFloat
+    ds = ArgoFloat(6903091).open_dataset('BGCArgoPlus')
+
+You can also use the store directly::
+
+    from argopy.stores.float.bgcargo_plus import BGCArgoPlusStore
+    store = BGCArgoPlusStore(6903091)
+    ds = store.open_dataset()
+    url = store.url 
+"""
+
+from __future__ import annotations
+
+import logging
+
+import xarray as xr
+
+from ...stores.implementations.ftp import ftpstore
+from ...utils import check_wmo
+
+log = logging.getLogger("argopy.stores.BGCArgoPlusStore")
+
+#: Root of the BGC-Argo+ FTP tree (no trailing slash)
+BGCARGO_PLUS_FTP_HOST = "ftp.soest.hawaii.edu"
+
+#: Path template on the FTP server.
+#: ``{version}`` can be overridden via :attr:`BGCArgoPlusStore.version`.
+BGCARGO_PLUS_PATH_TEMPLATE = (
+    "/bgc_argo_plus/outliers_removed/{version}/{wmo}_Sprof_BGCArgoPlus.nc"
+)
+
+#: Default version tag that maps to the latest production release on the FTP.
+BGCARGO_PLUS_DEFAULT_VERSION = "v0.1_2026_04"
+
+
+def bgcargo_plus_url(wmo: int, version: str = BGCARGO_PLUS_DEFAULT_VERSION) -> str:
+    """Return the FTP URL for a BGC-Argo+ float file.
+
+    Parameters
+    ----------
+    wmo : int
+        Float WMO number.
+    version : str, optional
+        Dataset version string, e.g. ``"v0.1_2026_04"``.
+
+    Returns
+    -------
+    str
+        Full FTP URL
+
+    Examples
+    --------
+    >>> from argopy.stores.float.bgcargo_plus import bgcargo_plus_url
+    >>> bgcargo_plus_url(6903091)
+    'ftp://ftp.soest.hawaii.edu/bgc_argo_plus/outliers_removed/v0.1_2026_04/6903091_Sprof_BGCArgoPlus.nc'
+    """
+    path = BGCARGO_PLUS_PATH_TEMPLATE.format(wmo=wmo, version=version)
+    return f"ftp://{BGCARGO_PLUS_FTP_HOST}{path}"
+
+
+class BGCArgoPlusStore:
+    """Store that fetches BGC-Argo+ individual-float files from SOEST FTP.
+
+    Parameters
+    ----------
+    wmo : int or str
+        Float WMO number.
+    version : str, optional
+        BGC-Argo+ dataset version, default :data:`BGCARGO_PLUS_DEFAULT_VERSION`.
+    cache : bool, optional
+        Cache downloaded files locally (passed to :class:`ftpstore`).
+    cachedir : str, optional
+        Local cache directory (passed to :class:`ftpstore`).
+    timeout : int, optional
+        FTP connection timeout in seconds (passed to :class:`ftpstore`).
+
+    Examples
+    --------
+    >>> from argopy.stores.float.bgcargo_plus import BGCArgoPlusStore
+    >>> store = BGCArgoPlusStore(6903091)
+    >>> ds = store.open_dataset()
+    >>> store.url
+    'ftp://ftp.soest.hawaii.edu/bgc_argo_plus/...'
+    """
+
+    def __init__(
+        self,
+        wmo: int | str,
+        version: str = BGCARGO_PLUS_DEFAULT_VERSION,
+        cache: bool = False,
+        cachedir: str = "",
+        timeout: int = 0,
+    ):
+        self.WMO = check_wmo(wmo)[0]
+        self.version = version
+        self.cache = cache
+        self.cachedir = cachedir
+        self.timeout = timeout
+
+        # Build the FTP store pointing at the SOEST server
+        ftp_root = f"{BGCARGO_PLUS_FTP_HOST}"
+        self._fs = ftpstore(
+            host=ftp_root,
+            cache=self.cache,
+            cachedir=self.cachedir if self.cachedir else None,
+            timeout=self.timeout if self.timeout else None,
+        )
+
+    @property
+    def url(self) -> str:
+        """Full FTP URL of this float's BGC-Argo+ file."""
+        return bgcargo_plus_url(self.WMO, version=self.version)
+
+    def open_dataset(self, **kwargs) -> xr.Dataset:
+        """Download and open the BGC-Argo+ netCDF file for this float.
+
+        Parameters
+        ----------
+        **kwargs
+            Additional keyword arguments forwarded to
+            :meth:`argopy.stores.ftpstore.open_dataset` (e.g. ``lazy=True``,
+            ``xr_opts={"decode_times": False}``).
+
+        Returns
+        -------
+        :class:`xarray.Dataset`
+
+        Examples
+        --------
+        >>> from argopy.stores.float.bgcargo_plus import BGCArgoPlusStore
+        >>> ds = BGCArgoPlusStore(6903091).open_dataset()
+        """
+        log.debug("BGCArgoPlusStore: fetching %s", self.url)
+        try:
+            ds = self._fs.open_dataset(self.url, **kwargs)
+        except FileNotFoundError as exc:
+            raise FileNotFoundError(
+                f"Could not retrieve BGC-Argo+ file for WMO {self.WMO} "
+                f"(version='{self.version}') from {self.url}.\n"
+                f"Original error: {exc}"
+            ) from exc
+        return ds
+
+    def __repr__(self) -> str:
+        return (
+            f"<BGCArgoPlusStore>\n"
+            f"  WMO     : {self.WMO}\n"
+            f"  version : {self.version}\n"
+            f"  URL     : {self.url}\n"
+        )

argopy/stores/float/spec.py

@@ -14,6 +14,7 @@
 from ...utils import check_wmo, argo_split_path, shortcut2gdac
 from ...options import OPTIONS
 from .. import ArgoIndex
+from .bgcargo_plus import BGCArgoPlusStore, BGCARGO_PLUS_DEFAULT_VERSION
 
 
 log = logging.getLogger("argopy.stores.ArgoFloat")
@@ -301,30 +302,71 @@ def ls_dataset(self) -> dict:
                     avail.update({name: file})
         return dict(sorted(avail.items()))
 
-    def open_dataset(
-        self, name: str = "prof", cast: bool = True, **kwargs
-    ) -> xr.Dataset:
+    def open_dataset(self, name: str = "prof", cast: bool = True, **kwargs) -> xr.Dataset:
         """Open and decode a dataset
 
         Parameters
         ----------
         name: str, optional, default = "prof"
-            Name of the dataset to open. It can be any key from the dictionary returned by :class:`ArgoFloat.ls_dataset`.
+            Name of the dataset to open. It can be any key from the dictionary
+            returned by :class:`ArgoFloat.ls_dataset`, or the special value
+            ``"BGCArgoPlus"`` to load the QC-processed file from the BGC-Argo+
+            dataset (https://www.bgc-argo-plus.info) served on the SOEST FTP.
         cast: bool, optional, default = True
-            Determine if the dataset variables should be cast or not. This is similar to opening the dataset directly with :class:`xr.open_dataset` using the ``engine=`argo``` option.
-            This will be ignored if the ``netCDF4` kwarg is set to True.
+            Determine if the dataset variables should be cast or not. This is similar
+            to opening the dataset directly with :class:`xr.open_dataset` using the
+            ``engine="argo"`` option.
+            This will be ignored if the ``netCDF4`` kwarg is set to True.
+            Note: ``cast`` is **not** applied when ``name="BGCArgoPlus"``.
+        bgcplus_version: str, optional
+            BGC-Argo+ dataset version tag (only used when ``name='BGCArgoPlus'``).
+            Defaults to :data:`argopy.stores.float.bgcargo_plus.BGCARGO_PLUS_DEFAULT_VERSION`.
         **kwargs
-            All the other parameters are passed to the GDAC store `open_dataset` method.
+            All the other parameters are passed to the GDAC store ``open_dataset``
+            method (or to :class:`argopy.stores.float.bgcargo_plus.BGCArgoPlusStore`
+            when ``name="BGCArgoPlus"``).
 
         Returns
         -------
         :class:`xarray.Dataset`
 
         Notes
         -----
-        Use the ``netCDF4=True`` option to return a :class:`netCDF4.Dataset` object instead of a :class:`xarray.Dataset`.
+        Use the ``netCDF4=True`` option to return a :class:`netCDF4.Dataset` object
+        instead of a :class:`xarray.Dataset`.
+
+        When ``name="BGCArgoPlus"``, the file is fetched from the SOEST FTP server::
+
+            ftp://ftp.soest.hawaii.edu/bgc_argo_plus/Individual_Floats/outliers_removed/
+
+        and a :class:`FileNotFoundError` is raised if the float is not (yet) part
+        of the BGC-Argo+ dataset.
+
+        Examples
+        --------
+        .. code-block:: python
+
+            from argopy import ArgoFloat
+            ds = ArgoFloat(6903091).open_dataset('BGCArgoPlus')
+            # Pin to a specific version:
+            ds = ArgoFloat(6903091).open_dataset('BGCArgoPlus', bgcplus_version='v0.1_2025_12')
 
         """
+        # ---- BGC-Argo+ special case -----------------------------------------
+        if name == "BGCArgoPlus":
+            version = kwargs.pop("bgcplus_version", BGCARGO_PLUS_DEFAULT_VERSION)
+            store = BGCArgoPlusStore(
+                self.WMO,
+                version=version,
+                cache=self.cache,
+                cachedir=self.cachedir,
+                timeout=self.timeout,
+            )
+            ds = store.open_dataset(**kwargs)
+            self._dataset["BGCArgoPlus"] = ds
+            return ds
+        # ---------------------------------------------------------------------
+
         if name not in self.ls_dataset():
             raise ValueError(
                 "Dataset '%s' not found. Available dataset for this float are: %s"
@@ -404,6 +446,8 @@ def __repr__(self):
         summary.append("Number of cycles: %s" % self.N_CYCLES)
         if self._online:
             summary.append("Dashboard: %s" % dashboard(wmo=self.WMO, url_only=True))
-        summary.append("Netcdf dataset available: %s" % list(self.ls_dataset().keys()))
+        summary.append(
+            "Netcdf dataset available: %s" % list(self.ls_dataset().keys())
+        )
 
         return "\n".join(summary)