Add support for month-avgs of grazing & growth var groups

Extended the make_averaged_dataset worker and related test cases to include
grazing and growth dataset variable groups. Updated configuration files and
added parameter-specific validations for these new variable groups.

Author: douglatornell

config/nowcast.yaml

@@ -442,6 +442,12 @@ averaged datasets:
     chemistry:
       reshapr config: month-average_202111_chemistry.yaml
       file pattern: "SalishSeaCast_1m_chem_T_{yyyymmdd}_{yyyymmdd}.nc"
+    grazing:
+      reshapr config: month-average_202111_grazing_mortality.yaml
+      file pattern: "SalishSeaCast_1m_graz_T_{yyyymmdd}_{yyyymmdd}.nc"
+    growth:
+      reshapr config: month-average_202111_bio_growth_rates.yaml
+      file pattern: "SalishSeaCast_1m_prod_T_{yyyymmdd}_{yyyymmdd}.nc"
     physics:
       reshapr config: month-average_202111_physics.yaml
       file pattern: "SalishSeaCast_1m_grid_T_{yyyymmdd}_{yyyymmdd}.nc"
@@ -1587,6 +1593,10 @@ message registry:
       failure month biology: biology dataset month-averaging failed
       success month chemistry: chemistry dataset month-averaged
       failure month chemistry: chemistry dataset month-averaging failed
+      success month grazing: grazing dataset month-averaged
+      failure month grazing: grazing dataset month-averaging failed
+      success month growth: biology growth rates dataset month-averaged
+      failure month growth: biology growth rates dataset month-averaging failed
       success month physics: physics dataset month-averaged
       failure month physics: physics dataset month-averaging failed
       crash: make_averaged_dataset worker crashed

nowcast/workers/make_averaged_dataset.py

@@ -54,7 +54,7 @@ def main():
     )
     worker.cli.add_argument(
         "reshapr_var_group",
-        choices={"biology", "chemistry", "physics"},
+        choices={"biology", "chemistry", "grazing", "growth", "physics"},
         help="Dataset variable group to run extraction for",
     )
     worker.cli.add_date_option(
@@ -154,6 +154,12 @@ def make_averaged_dataset(parsed_args, config, *args):
             f"run_date = {run_date.format('YYYY-MM-DD')}"
         )
         raise WorkerError
+    if avg_time_interval == "day" and reshapr_var_group in {"grazing", "growth"}:
+        logger.error(
+            f"Day-average {reshapr_var_group} datasets are calculated by NEMO; "
+            f"use this worker for month-averaging"
+        )
+        raise WorkerError
     reshapr_config_dir = Path(config["averaged datasets"]["reshapr config dir"])
     reshapr_config_yaml = config["averaged datasets"][avg_time_interval][
         reshapr_var_group

tests/workers/test_make_averaged_dataset.py

@@ -59,6 +59,12 @@ def config(base_config):
                     chemistry:
                       reshapr config: month-average_202111_chemistry.yaml
                       file pattern: "SalishSeaCast_1m_chem_T_{yyyymmdd}_{yyyymmdd}.nc"
+                    grazing:
+                      reshapr config: month-average_202111_grazing_mortality.yaml
+                      file pattern: "SalishSeaCast_1m_graz_T_{yyyymmdd}_{yyyymmdd}.nc"
+                    growth:
+                      reshapr config: month-average_202111_bio_growth_rates.yaml
+                      file pattern: "SalishSeaCast_1m_prod_T_{yyyymmdd}_{yyyymmdd}.nc"
                     physics:
                       reshapr config: month-average_202111_physics.yaml
                       file pattern: "SalishSeaCast_1m_grid_T_{yyyymmdd}_{yyyymmdd}.nc"
@@ -100,6 +106,8 @@ def test_add_reshapr_var_group_arg(self, mock_worker):
         assert worker.cli.parser._actions[4].choices == {
             "biology",
             "chemistry",
+            "grazing",
+            "growth",
             "physics",
         }
         assert worker.cli.parser._actions[4].help
@@ -142,12 +150,16 @@ def test_message_registry_keys(self, prod_config):
             "failure month biology",
             "success month chemistry",
             "failure month chemistry",
+            "success month grazing",
+            "failure month grazing",
+            "success month growth",
+            "failure month growth",
             "success month physics",
             "failure month physics",
             "crash",
         ]
 
-    def test_averaged_datasets(self, prod_config):
+    def test_reshapr_configs(self, prod_config):
         averaged_datasets = prod_config["averaged datasets"]
         expected = "/SalishSeaCast/SalishSeaNowcast/config/reshapr/"
 
@@ -196,6 +208,16 @@ def test_day_averaged_datasets(
                 "month-average_202111_chemistry.yaml",
                 "SalishSeaCast_1m_chem_T_{yyyymmdd}_{yyyymmdd}.nc",
             ),
+            (
+                "grazing",
+                "month-average_202111_grazing_mortality.yaml",
+                "SalishSeaCast_1m_graz_T_{yyyymmdd}_{yyyymmdd}.nc",
+            ),
+            (
+                "growth",
+                "month-average_202111_bio_growth_rates.yaml",
+                "SalishSeaCast_1m_prod_T_{yyyymmdd}_{yyyymmdd}.nc",
+            ),
             (
                 "physics",
                 "month-average_202111_physics.yaml",
@@ -247,6 +269,8 @@ def test_day_average_success(self, avg_time_interval, reshapr_var_group, caplog)
         (
             ("month", "biology"),
             ("month", "chemistry"),
+            ("month", "grazing"),
+            ("month", "growth"),
             ("month", "physics"),
         ),
     )
@@ -301,6 +325,8 @@ def test_day_average_failure(self, avg_time_interval, reshapr_var_group, caplog)
         (
             ("month", "biology"),
             ("month", "chemistry"),
+            ("month", "grazing"),
+            ("month", "growth"),
             ("month", "physics"),
         ),
     )
@@ -385,6 +411,8 @@ def mock_extract_netcdf(reshapr_config, reshapr_config_yaml):
         (
             ("month", "biology"),
             ("month", "chemistry"),
+            ("month", "grazing"),
+            ("month", "growth"),
             ("month", "physics"),
         ),
     )
@@ -446,3 +474,22 @@ def test_bad_month_avg_run_date(self, caplog, config):
         assert caplog.records[0].levelname == "ERROR"
         expected = f"Month-averaging must start on the first day of a month but run_date = 2022-11-10"
         assert caplog.messages[0] == expected
+
+    @pytest.mark.parametrize("reshapr_var_group", ("grazing", "growth"))
+    def test_month_avg_only_var_groups(self, reshapr_var_group, caplog, config):
+        parsed_args = SimpleNamespace(
+            avg_time_interval="day",
+            run_date=arrow.get("2024-09-24"),
+            reshapr_var_group=reshapr_var_group,
+        )
+        caplog.set_level(logging.DEBUG)
+
+        with pytest.raises(WorkerError):
+            make_averaged_dataset.make_averaged_dataset(parsed_args, config)
+
+        assert caplog.records[0].levelname == "ERROR"
+        expected = (
+            f"Day-average {reshapr_var_group} datasets are calculated by NEMO; "
+            f"use this worker for month-averaging"
+        )
+        assert caplog.messages[0] == expected