Merge branch 'ecdr-at-25km' into ecdr-at-25km-ss

CHANGELOG.md

@@ -1,3 +1,10 @@
+# v0.2.0
+
+* Produce 25km CDR instead of 12.5km.
+* Refactor how platforms are handled to support overriding platform start dates
+  via yaml configuration files.
+
+
 # v0.1.0
 
 * Initial version of the ECDR.

pyproject.toml

@@ -1,9 +1,9 @@
 [project]
 name = "seaice_ecdr"
-version = "0.1.0"
+version = "0.2.0"
 
 [tool.bumpversion]
-current_version = "0.1.0"
+current_version = "0.2.0"
 commit = false
 tag = false
 

seaice_ecdr/__init__.py

@@ -6,7 +6,7 @@
 
 from seaice_ecdr.constants import LOGS_DIR
 
-__version__ = "v0.1.0"
+__version__ = "v0.2.0"
 
 DEFAULT_LOG_LEVEL = "INFO"
 

seaice_ecdr/_types.py

@@ -2,14 +2,3 @@
 
 # In kilometers.
 ECDR_SUPPORTED_RESOLUTIONS = Literal["12.5", "25"]
-
-# Supported sats
-SUPPORTED_SAT = Literal[
-    "am2",  # AMSR2
-    "ame",  # AMSRE
-    "F17",  # SSMIS F17
-    "F13",  # SSMI F13
-    "F11",  # SSMI F11
-    "F08",  # SSMI F08
-    "n07",  # Nimbus-7 SMMR
-]

seaice_ecdr/ancillary.py

@@ -19,7 +19,8 @@
 from seaice_ecdr._types import ECDR_SUPPORTED_RESOLUTIONS
 from seaice_ecdr.constants import CDR_ANCILLARY_DIR
 from seaice_ecdr.grid_id import get_grid_id
-from seaice_ecdr.platforms import SUPPORTED_SAT, get_platform_by_date
+from seaice_ecdr.platforms import PLATFORM_CONFIG, Platform
+from seaice_ecdr.platforms.config import N07_PLATFORM
 
 ANCILLARY_SOURCES = Literal["CDRv4", "CDRv5"]
 
@@ -96,7 +97,6 @@ def get_surfacetype_da(
     hemisphere: Hemisphere,
     resolution: ECDR_SUPPORTED_RESOLUTIONS,
     ancillary_source: ANCILLARY_SOURCES,
-    platform: SUPPORTED_SAT,
 ) -> xr.DataArray:
     """Return a dataarray with surface type information for this date."""
     ancillary_ds = get_ancillary_ds(
@@ -111,7 +111,8 @@ def get_surfacetype_da(
     polehole_surface_type = 100
     if "polehole_bitmask" in ancillary_ds.data_vars.keys():
         polehole_bitmask = ancillary_ds.polehole_bitmask
-        polehole_bitlabel = f"{platform}_polemask"
+        platform = PLATFORM_CONFIG.get_platform_by_date(date)
+        polehole_bitlabel = f"{platform.id}_polemask"
         polehole_bitvalue = bitmask_value_for_meaning(
             var=polehole_bitmask,
             meaning=polehole_bitlabel,
@@ -174,8 +175,8 @@ def nh_polehole_mask(
     *,
     date: dt.date,
     resolution: ECDR_SUPPORTED_RESOLUTIONS,
-    sat=None,
     ancillary_source: ANCILLARY_SOURCES,
+    platform: Platform | None = None,
 ) -> xr.DataArray:
     """Return the northern hemisphere pole hole mask for the given date and resolution."""
     ancillary_ds = get_ancillary_ds(
@@ -186,12 +187,12 @@ def nh_polehole_mask(
 
     polehole_bitmask = ancillary_ds.polehole_bitmask
 
-    if sat is None:
-        sat = get_platform_by_date(
+    if platform is None:
+        platform = PLATFORM_CONFIG.get_platform_by_date(
             date=date,
         )
 
-    polehole_bitlabel = f"{sat}_polemask"
+    polehole_bitlabel = f"{platform.id}_polemask"
     polehole_bitvalue = bitmask_value_for_meaning(
         var=polehole_bitmask,
         meaning=polehole_bitlabel,
@@ -280,22 +281,21 @@ def get_invalid_ice_mask(
     hemisphere: Hemisphere,
     date: dt.date,
     resolution: ECDR_SUPPORTED_RESOLUTIONS,
-    platform: SUPPORTED_SAT,
     ancillary_source: ANCILLARY_SOURCES,
+    platform: Platform,
 ) -> xr.DataArray:
     """Return an invalid ice mask for the given date.
 
     SMMR (n07) uses a day-of-year based climatology. All other platforms use a
     month-based mask.
     """
     # SMMR / n07 case:
-    if platform == "n07":
+    if platform == N07_PLATFORM:
         # TODO: Daily (SMMR) mask is used at end for cleanup,
         #       but not for initial TB field generation
         # Skip the smmr invalid ice mask for now...
         print("WARNING: Using non-SMMR invalid ice masks")
-        # return get_smmr_invalid_ice_mask(hemisphere=hemisphere, date=date, resolution=resolution, ancillary_source=ancillary_source)
-
+        # return get_smmr_invalid_ice_mask(hemisphere=hemisphere, date=date)
     # All other platforms:
     ancillary_ds = get_ancillary_ds(
         hemisphere=hemisphere,

seaice_ecdr/complete_daily_ecdr.py

@@ -31,9 +31,7 @@
     date_in_nh_melt_season,
     melting,
 )
-from seaice_ecdr.platforms import (
-    get_platform_by_date,
-)
+from seaice_ecdr.platforms import PLATFORM_CONFIG
 from seaice_ecdr.set_daily_ncattrs import finalize_cdecdr_ds
 from seaice_ecdr.spillover import LAND_SPILL_ALGS
 from seaice_ecdr.temporal_composite_daily import get_tie_filepath, make_tiecdr_netcdf
@@ -76,19 +74,19 @@ def get_ecdr_filepath(
     is_nrt: bool,
 ) -> Path:
     """Return the complete daily eCDR file path."""
-    platform = get_platform_by_date(date)
+    platform = PLATFORM_CONFIG.get_platform_by_date(date)
     if is_nrt:
         ecdr_filename = nrt_daily_filename(
             hemisphere=hemisphere,
             date=date,
-            sat=platform,
+            platform_id=platform.id,
             resolution=resolution,
         )
     else:
         ecdr_filename = standard_daily_filename(
             hemisphere=hemisphere,
             date=date,
-            sat=platform,
+            platform_id=platform.id,
             resolution=resolution,
         )
 
@@ -443,12 +441,10 @@ def _add_surfacetype_da(
     #       The methodology here should be reviewed to see if there is
     #       a "better" way to add a geo-referenced dataarray to an existing
     #       xr Dataset.
-    platform = get_platform_by_date(date)
     surfacetype_da = get_surfacetype_da(
         date=date,
         hemisphere=hemisphere,
         resolution=resolution,
-        platform=platform,
         ancillary_source=ancillary_source,
     )
     # Force use of the cde_ds coords instead of the x, y, time vars

seaice_ecdr/daily_aggregate.py

@@ -19,10 +19,10 @@
 from seaice_ecdr.constants import DEFAULT_BASE_OUTPUT_DIR
 from seaice_ecdr.nc_attrs import get_global_attrs
 from seaice_ecdr.nc_util import concatenate_nc_files
-from seaice_ecdr.platforms import get_first_platform_start_date
+from seaice_ecdr.platforms import PLATFORM_CONFIG
 from seaice_ecdr.util import (
     get_complete_output_dir,
-    sat_from_filename,
+    platform_id_from_filename,
     standard_daily_aggregate_filename,
 )
 
@@ -37,7 +37,9 @@ def _get_daily_complete_filepaths_for_year(
     resolution: ECDR_SUPPORTED_RESOLUTIONS,
 ) -> list[Path]:
     data_list = []
-    start_date = max(dt.date(year, 1, 1), get_first_platform_start_date())
+    start_date = max(
+        dt.date(year, 1, 1), PLATFORM_CONFIG.get_first_platform_start_date()
+    )
     for period in pd.period_range(start=start_date, end=dt.date(year, 12, 31)):
         expected_fp = get_ecdr_filepath(
             date=period.to_timestamp().date(),
@@ -115,7 +117,7 @@ def _update_ncrcat_daily_ds(
         temporality="daily",
         aggregate=True,
         source=", ".join([fp.name for fp in daily_filepaths]),
-        sats=[sat_from_filename(fp.name) for fp in daily_filepaths],
+        platform_ids=[platform_id_from_filename(fp.name) for fp in daily_filepaths],
     )
     ds.attrs = daily_aggregate_ds_global_attrs
 

seaice_ecdr/initial_daily_ecdr.py

@@ -31,7 +31,7 @@
 from pm_tb_data._types import NORTH, Hemisphere
 from pm_tb_data.fetch.nsidc_0001 import NSIDC_0001_SATS
 
-from seaice_ecdr._types import ECDR_SUPPORTED_RESOLUTIONS, SUPPORTED_SAT
+from seaice_ecdr._types import ECDR_SUPPORTED_RESOLUTIONS
 from seaice_ecdr.ancillary import (
     ANCILLARY_SOURCES,
     get_empty_ds_with_time,
@@ -42,7 +42,7 @@
 from seaice_ecdr.cli.util import datetime_to_date
 from seaice_ecdr.constants import CDR_ANCILLARY_DIR, DEFAULT_BASE_OUTPUT_DIR
 from seaice_ecdr.grid_id import get_grid_id
-from seaice_ecdr.platforms import get_platform_by_date
+from seaice_ecdr.platforms import PLATFORM_CONFIG, SUPPORTED_PLATFORM_ID
 from seaice_ecdr.regrid_25to12 import reproject_ideds_25to12
 from seaice_ecdr.spillover import LAND_SPILL_ALGS, land_spillover
 from seaice_ecdr.tb_data import (
@@ -54,8 +54,8 @@
 from seaice_ecdr.util import get_intermediate_output_dir, standard_daily_filename
 
 
-def platform_is_smmr(platform):
-    return platform in ("n07", "s36")
+def platform_is_smmr(platform_id: SUPPORTED_PLATFORM_ID):
+    return platform_id in ("n07", "s36")
 
 
 def cdr_bootstrap_raw(
@@ -64,7 +64,7 @@ def cdr_bootstrap_raw(
     tb_h37: npt.NDArray,
     tb_v19: npt.NDArray,
     bt_coefs,
-    platform: SUPPORTED_SAT,
+    platform: SUPPORTED_PLATFORM_ID,
 ):
     """Generate the raw bootstrap concentration field.
     Note: tb fields should already be transformed before
@@ -205,7 +205,7 @@ def _setup_ecdr_ds(
     ecdr_ide_ds.attrs["data_source"] = tb_data.data_source
 
     # Set the platform
-    ecdr_ide_ds.attrs["platform"] = tb_data.platform
+    ecdr_ide_ds.attrs["platform"] = tb_data.platform_id
 
     file_date = dt.date(1970, 1, 1) + dt.timedelta(
         days=int(ecdr_ide_ds.variables["time"].data)
@@ -407,7 +407,7 @@ def compute_initial_daily_ecdr_dataset(
             # The CDRv4 calculation causes TB to be zero/missing where
             # no sea ice can occur because of invalid region or land
             logger.debug(f"Applying invalid ice mask to TB field: {tb_si_varname}")
-            platform = get_platform_by_date(date)
+            platform = PLATFORM_CONFIG.get_platform_by_date(date)
             invalid_ice_mask = get_invalid_ice_mask(
                 hemisphere=hemisphere,
                 date=date,
@@ -561,26 +561,26 @@ def compute_initial_daily_ecdr_dataset(
     )
     logger.debug("Initialized spatial_interpolation_flag with TB fill locations")
 
-    platform = get_platform_by_date(date)
-    if platform == "am2":
+    platform = PLATFORM_CONFIG.get_platform_by_date(date)
+    if platform.id == "am2":
         bt_coefs_init = pmi_bt_params_amsr2.get_ausi_amsr2_bootstrap_params(
             date=date,
             satellite="amsr2",
             gridid=ecdr_ide_ds.grid_id,
         )
-    elif platform == "ame":
+    elif platform.id == "ame":
         bt_coefs_init = pmi_bt_params_amsre.get_ausi_amsre_bootstrap_params(
             date=date,
             satellite="amsre",
             gridid=ecdr_ide_ds.grid_id,
         )
-    elif platform in get_args(NSIDC_0001_SATS):
+    elif platform.id in get_args(NSIDC_0001_SATS):
         bt_coefs_init = pmi_bt_params_0001.get_nsidc0001_bootstrap_params(
             date=date,
-            satellite=platform,
+            satellite=platform.id,
             gridid=ecdr_ide_ds.grid_id,
         )
-    elif platform_is_smmr(platform):
+    elif platform_is_smmr(platform.id):
         bt_coefs_init = get_smmr_params(hemisphere=hemisphere, date=date)
     else:
         raise RuntimeError(f"platform bootstrap params not implemented: {platform}")
@@ -618,14 +618,14 @@ def compute_initial_daily_ecdr_dataset(
         pole_mask = nh_polehole_mask(
             date=date,
             resolution=tb_data.resolution,
-            sat=platform,
             ancillary_source=ancillary_source,
+            platform=platform,
         )
         ecdr_ide_ds["pole_mask"] = pole_mask
 
     nt_params = get_cdr_nt_params(
         hemisphere=hemisphere,
-        platform=platform,
+        platform=platform.id,
     )
 
     nt_coefs = NtCoefs(
@@ -683,7 +683,7 @@ def compute_initial_daily_ecdr_dataset(
                 v19=bt_v19,
                 v22=bt_v22,
             ),
-            platform=platform.lower(),
+            platform=platform.id.lower(),
         )
         bt_v37 = transformed["v37"]
         bt_h37 = transformed["h37"]
@@ -713,7 +713,7 @@ def compute_initial_daily_ecdr_dataset(
         wintrc=bt_coefs_init["wintrc"],
         wslope=bt_coefs_init["wslope"],
         wxlimt=bt_coefs_init["wxlimt"],
-        is_smmr=platform_is_smmr(platform),
+        is_smmr=platform_is_smmr(platform.id),
     )
 
     # Note:
@@ -853,7 +853,7 @@ def compute_initial_daily_ecdr_dataset(
         tb_h37=bt_h37,
         tb_v19=bt_v19,
         bt_coefs=bt_coefs,
-        platform=platform,
+        platform=platform.id,
     )
 
     # Set any bootstrap concentrations below 10% to 0.
@@ -950,7 +950,6 @@ def compute_initial_daily_ecdr_dataset(
         tb_data=tb_data,
         algorithm=land_spillover_alg,
         land_mask=non_ocean_mask.data,
-        platform=platform,
         ancillary_source=ancillary_source,
         bt_conc=bt_asCDRv4_conc,
         nt_conc=nt_asCDRv4_conc,
@@ -1252,7 +1251,7 @@ def get_idecdr_dir(*, intermediate_output_dir: Path) -> Path:
 def get_idecdr_filepath(
     *,
     date: dt.date,
-    platform,
+    platform_id: SUPPORTED_PLATFORM_ID,
     hemisphere: Hemisphere,
     resolution: ECDR_SUPPORTED_RESOLUTIONS,
     intermediate_output_dir: Path,
@@ -1262,7 +1261,7 @@ def get_idecdr_filepath(
     standard_fn = standard_daily_filename(
         hemisphere=hemisphere,
         date=date,
-        sat=platform,
+        platform_id=platform_id,
         resolution=resolution,
     )
     idecdr_fn = "idecdr_" + standard_fn
@@ -1285,10 +1284,10 @@ def make_idecdr_netcdf(
     ancillary_source: ANCILLARY_SOURCES,
     overwrite_ide: bool = False,
 ) -> None:
-    platform = get_platform_by_date(date)
+    platform = PLATFORM_CONFIG.get_platform_by_date(date)
     output_path = get_idecdr_filepath(
         date=date,
-        platform=platform,
+        platform_id=platform.id,
         hemisphere=hemisphere,
         intermediate_output_dir=intermediate_output_dir,
         resolution=resolution,