Merge pull request #152 from openclimatefix/fix-ecmwf

fix to use ecmwf model only

pvnet_app/model_configs/all_models.yaml

@@ -63,10 +63,10 @@ models:
   - name: pvnet_ecmwf # this name is important as it used for blending
     pvnet:
         repo: openclimatefix/pvnet_uk_region
-        version: 35d55181a82440bdd087f380d650bfd0b64bd322
+        version: c14f7427d9854d63430aa936ce45f55d3818d033
     summation:
         repo: openclimatefix/pvnet_v2_summation
-        version: 9002baf1e9dc1ec141f3c4a1fa8447b6316a4558
+        version: 4fe6b1441b6dd549292c201ed85eee156ecc220c
     ecmwf_only: True
     uses_satellite_data: False
     uses_ocf_data_sampler: False

tests/test_app.py

@@ -17,13 +17,12 @@
 from pvnet_app.model_configs.pydantic_models import get_all_models
 
 
-
 def test_app(
     db_session, nwp_ukv_data, nwp_ecmwf_data, sat_5_data, gsp_yields_and_systems, me_latest
 ):
 
     """Test the app running the intraday models"""
-    
+
     with tempfile.TemporaryDirectory() as tmpdirname:
 
         os.chdir(tmpdirname)
@@ -93,7 +92,7 @@ def test_app_day_ahead_model(
     """Test the app running the day ahead model"""
 
     with tempfile.TemporaryDirectory() as tmpdirname:
-        
+
         os.chdir(tmpdirname)
 
         temp_nwp_path = "temp_nwp_ukv.zarr"
@@ -152,13 +151,14 @@ def test_app_day_ahead_model(
         == expected_forecast_results * expected_forecast_timesteps
     )
 
+
 def test_app_no_sat(
     db_session, nwp_ukv_data, nwp_ecmwf_data, sat_5_data, gsp_yields_and_systems, me_latest
 ):
     """Test the app for the case when no satellite data is available"""
 
     with tempfile.TemporaryDirectory() as tmpdirname:
-        
+
         os.chdir(tmpdirname)
 
         temp_nwp_path = "temp_nwp_ukv.zarr"
@@ -168,8 +168,8 @@ def test_app_no_sat(
         temp_nwp_path = "temp_nwp_ecmwf.zarr"
         os.environ["NWP_ECMWF_ZARR_PATH"] = temp_nwp_path
         nwp_ecmwf_data.to_zarr(temp_nwp_path)
-        
-        # There is no satellite data available at the environ path
+
+        # There is no satellite data available at the environ path
         os.environ["SATELLITE_ZARR_PATH"] = "nonexistent_sat.zarr.zip"
 
         os.environ["RUN_EXTRA_MODELS"] = "True"
@@ -182,7 +182,7 @@ def test_app_no_sat(
         from pvnet_app.app import app
 
         app(gsp_ids=list(range(1, 318)), num_workers=2)
-        
+
     # Only the models which don't use satellite will be run in this case
     # The models below are the only ones which should have been run
     all_models = get_all_models(run_extra_models=True)
@@ -218,11 +218,75 @@ def test_app_no_sat(
     assert len(db_session.query(ForecastValueSevenDaysSQL).all()) == expected_forecast_results * 16
 
 
+# test legacy models
+# Its nice to have this here, so we can run the latest version in production, but still use the old models
+# Once we have re trained PVnet summation models we can remove this
+def test_app_ecwmf_only(db_session, nwp_ecmwf_data, gsp_yields_and_systems, me_latest):
+    """Test the app for the case running model just on ecmwf"""
+
+    with tempfile.TemporaryDirectory() as tmpdirname:
+
+        os.chdir(tmpdirname)
+
+        temp_nwp_path = "temp_nwp_ecmwf.zarr"
+        os.environ["NWP_ECMWF_ZARR_PATH"] = temp_nwp_path
+        nwp_ecmwf_data.to_zarr(temp_nwp_path)
+
+        # There is no satellite or ukv data available at the environ path
+        os.environ["SATELLITE_ZARR_PATH"] = "nonexistent_sat.zarr.zip"
+        os.environ["NWP_UKV_ZARR_PATH"] = "nonexistent_nwp.zarr.zip"
+
+        os.environ["RUN_EXTRA_MODELS"] = "False"
+        os.environ["SAVE_GSP_SUM"] = "True"
+        os.environ["DAY_AHEAD_MODEL"] = "False"
+        os.environ["USE_OCF_DATA_SAMPLER"] = "False"
+        os.environ["USE_ECMWF_ONLY"] = "True"
+
+        # Run prediction
+        # Thes import needs to come after the environ vars have been set
+        from pvnet_app.app import app
+
+        app(gsp_ids=list(range(1, 318)), num_workers=2)
+
+    # Only the models which don't use satellite will be run in this case
+    # The models below are the only ones which should have been run
+    all_models = get_all_models(get_ecmwf_only=True)
+
+    # Check correct number of forecasts have been made
+    # (317 GSPs + 1 National + maybe GSP-sum) = 318 or 319 forecasts
+    # Forecast made with multiple models
+    expected_forecast_results = 0
+    for model_config in all_models:
+        expected_forecast_results += 318 + model_config.save_gsp_sum
+
+    forecasts = db_session.query(ForecastSQL).all()
+    # Doubled for historic and forecast
+    assert len(forecasts) == expected_forecast_results * 2
+
+    # Check probabilistic added
+    assert "90" in forecasts[0].forecast_values[0].properties
+    assert "10" in forecasts[0].forecast_values[0].properties
+
+    # 318 GSPs * 16 time steps in forecast
+    assert len(db_session.query(ForecastValueSQL).all()) == expected_forecast_results * 16
+    assert len(db_session.query(ForecastValueLatestSQL).all()) == expected_forecast_results * 16
+
+    expected_forecast_results = 0
+    for model_config in all_models:
+        # National
+        expected_forecast_results += 1
+        # GSP
+        expected_forecast_results += 317 * model_config.save_gsp_to_recent
+        expected_forecast_results += model_config.save_gsp_sum  # optional Sum of GSPs
+
+    assert len(db_session.query(ForecastValueSevenDaysSQL).all()) == expected_forecast_results * 16
+
+
 # test legacy models
 # Its nice to have this here, so we can run the latest version in production, but still use the old models
 # Once we have re trained PVnet summation models we can remove this
 def test_app_ocf_datapipes(
-        db_session, nwp_ukv_data, nwp_ecmwf_data, sat_5_data, gsp_yields_and_systems, me_latest
+    db_session, nwp_ukv_data, nwp_ecmwf_data, sat_5_data, gsp_yields_and_systems, me_latest
 ):
     """Test the app running the day ahead model"""
 
@@ -245,6 +309,7 @@ def test_app_ocf_datapipes(
         os.environ["DAY_AHEAD_MODEL"] = "False"
         os.environ["RUN_EXTRA_MODELS"] = "False"
         os.environ["USE_OCF_DATA_SAMPLER"] = "False"
+        os.environ["USE_ECMWF_ONLY"] = "False"
 
         # Run prediction
         # Thes import needs to come after the environ vars have been set
@@ -273,14 +338,14 @@ def test_app_ocf_datapipes(
     expected_forecast_timesteps = 16
 
     assert (
-            len(db_session.query(ForecastValueSQL).all())
-            == expected_forecast_results * expected_forecast_timesteps
+        len(db_session.query(ForecastValueSQL).all())
+        == expected_forecast_results * expected_forecast_timesteps
     )
     assert (
-            len(db_session.query(ForecastValueLatestSQL).all())
-            == expected_forecast_results * expected_forecast_timesteps
+        len(db_session.query(ForecastValueLatestSQL).all())
+        == expected_forecast_results * expected_forecast_timesteps
     )
     assert (
-            len(db_session.query(ForecastValueSevenDaysSQL).all())
-            == expected_forecast_results * expected_forecast_timesteps
-    )
+        len(db_session.query(ForecastValueSevenDaysSQL).all())
+        == expected_forecast_results * expected_forecast_timesteps
+    )