Merge pull request #25 from brightbandtech/taylor/ope-61-cleanup

general cleanup

src/extremeweatherbench/case.py

@@ -1,4 +1,4 @@
-"""Utiltiies for defining individual units of case studies for analysis.
+"""Classes for defining individual units of case studies for analysis.
 Some code similarly structured to WeatherBench (Rasp et al.)."""
 
 import dataclasses

src/extremeweatherbench/config.py

@@ -42,9 +42,6 @@ class Config:
     cache: bool = False
 
 
-# TODO(daniel): Convert to just a dictionary mapping, since that's all this is. No need
-# for a dataclass here. Can also use a defaultdict that reflexively maps to keys added
-# to the mapping unless otherwise specified.
 @dataclasses.dataclass
 class ForecastSchemaConfig:
     """A mapping between standard variable names used across EWB, and their counterpart

src/extremeweatherbench/evaluate.py

@@ -181,6 +181,7 @@ def _open_forecast_dataset(
     eval_config: config.Config,
     forecast_schema_config: config.ForecastSchemaConfig = DEFAULT_FORECAST_SCHEMA_CONFIG,
 ):
+    """Open the forecast dataset specified for evaluation."""
     logging.info("Opening forecast dataset")
     if eval_config.forecast_dir.startswith("s3://"):
         fs = fsspec.filesystem("s3")

src/extremeweatherbench/events.py

@@ -5,9 +5,6 @@
 from typing import List, Optional
 from extremeweatherbench import case
 
-# TODO(taylor): don't need link here
-CLIMATOLOGY_LINK = "/home/taylor/data/era5_2m_temperature_85th_by_hour_dayofyear.zarr"
-
 
 @dataclasses.dataclass
 class EventContainer:

src/extremeweatherbench/metrics.py

@@ -11,8 +11,6 @@
 logger = logging.getLogger(__name__)
 logger.setLevel(logging.INFO)
 
-T2M_85TH_PERCENTILE_CLIMATOLOGY_PATH = "gs://brightband-scratch/taylor/climatology/era5_2m_temperature_85th_rolling_by_hour_dayofyear.zarr"
-
 
 @dataclasses.dataclass
 class Metric:

src/extremeweatherbench/utils.py

@@ -46,10 +46,7 @@ def convert_longitude_to_180(
 
 
 def generate_json_from_nc(u, so, fs, fs_out, json_dir):
-    """Generate a kerchunk JSON file from a NetCDF file.
-
-    TODO(taylor): Define function signature and docstring.
-    """
+    """Generate a kerchunk JSON file from a NetCDF file."""
     with fs.open(u, **so) as infile:
         h5chunks = SingleHdf5ToZarr(infile, u, inline_threshold=300)
 
@@ -65,9 +62,6 @@ def generate_json_from_nc(u, so, fs, fs_out, json_dir):
 
 
 def clip_dataset_to_bounding_box(
-    # NOTE(daniel): given its use here, "case.Location" should be moved to this module
-    # or something else stand-alone; high likelihood of inadvertently introducing a
-    # circular import dependency here.
     dataset: xr.Dataset,
     location_center: Location,
     length_km: float,
@@ -140,8 +134,6 @@ def remove_ocean_gridpoints(dataset: xr.Dataset) -> xr.Dataset:
     Returns:
         The dataset masked to only land gridpoints.
     """
-    # TODO(taylor): Extend this so that the user may pass their own land-sea mask,
-    # best suited the dataset they're analyzing.
     land = regionmask.defined_regions.natural_earth_v5_0_0.land_110
     land_sea_mask = land.mask(dataset.longitude, dataset.latitude)
     land_mask = land_sea_mask == 0

tests/test_evaluate.py

@@ -37,7 +37,6 @@ def test_open_obs_datasets_no_forecast_paths():
         evaluate._open_forecast_dataset(invalid_config)
 
 
-# TODO: test for discordant datetimes not in line with forecast dataset
 def test_evaluate_base_case(mock_forecast_dataset, mock_gridded_obs_dataset):
     base_case = case.IndividualCase(
         id=1,