Issue/solar night time (#130)

* make sure night time filter is working

* tidy

* lint

india_forecast_app/models/pvnet/model.py

@@ -95,8 +95,9 @@ def predict(self, site_id: str, timestamp: dt.datetime):
                 # Run batch through model
                 device_batch = copy_batch_to_device(batch_to_tensor(batch), DEVICE)
                 preds = self.model(device_batch).detach().cpu().numpy()
+
                 # filter out night time
-                if self.asset_type == SiteAssetType.pv:
+                if self.asset_type == SiteAssetType.pv.name:
                     preds = set_night_time_zeros(batch, preds)
 
                 # Store predictions
@@ -173,17 +174,12 @@ def predict(self, site_id: str, timestamp: dt.datetime):
                     0.0,
                 ]
                 log.debug(f"Previous values are {values_df['forecast_power_kw']}")
-                zero_values = values_df["forecast_power_kw"] == 0
                 for idx in range(8):
                     values_df["forecast_power_kw"][idx] -= (
                         values_df["forecast_power_kw"][idx] - final_gen_points
                     ) * smooth_values[final_gen_index + idx]
                 log.debug(f"New values are {values_df['forecast_power_kw']}")
 
-            if self.asset_type == "solar":
-                # make sure previous zero values are still zero
-                values_df["forecast_power_kw"][zero_values] = 0
-
         if self.asset_type == "wind":
             # Smooth with a 1 hour rolling window
             # Only smooth the wind else we introduce too much of a lag in the solar
@@ -244,9 +240,7 @@ def _prepare_data_sources(self):
             # if generation_da is still empty make nans
             if len(generation_da) == 0:
                 cols = [str(col) for col in self.generation_data["data"].columns]
-                generation_df = pd.DataFrame(
-                    index=forecast_timesteps, columns=cols, data=0.0001
-                )
+                generation_df = pd.DataFrame(index=forecast_timesteps, columns=cols, data=0.0001)
                 generation_da = generation_df.to_xarray()
             generation_da.to_netcdf(wind_netcdf_path, engine="h5netcdf")
 

india_forecast_app/models/pvnet/utils.py

@@ -51,7 +51,7 @@ def populate_data_config_sources(input_path, output_path):
         "wind": {"filename": wind_netcdf_path, "metadata_filename": wind_metadata_path},
         "pv": {"filename": pv_netcdf_path, "metadata_filename": pv_metadata_path},
         "nwp": {"ecmwf": nwp_ecmwf_path, "gfs": nwp_gfs_path},
-        "satellite": {"filepath": satellite_path}
+        "satellite": {"filepath": satellite_path},
     }
 
     if "nwp" in config["input_data"]:
@@ -89,7 +89,7 @@ def populate_data_config_sources(input_path, output_path):
 def process_and_cache_nwp(source_nwp_path: str, dest_nwp_path: str):
     """Reads zarr file, renames t variable to t2m and saves zarr to new destination"""
 
-    log.info(f'Processing and caching NWP data for {source_nwp_path}')
+    log.info(f"Processing and caching NWP data for {source_nwp_path}")
 
     # Load dataset from source
     ds = xr.open_zarr(source_nwp_path)
@@ -105,7 +105,7 @@ def process_and_cache_nwp(source_nwp_path: str, dest_nwp_path: str):
 
     is_gfs = "gfs" in source_nwp_path.lower()
 
-    if not is_gfs: # this is for ECMWF NWP
+    if not is_gfs:  # this is for ECMWF NWP
         # Rename t variable to t2m
         variables = list(ds.variable.values)
         new_variables = []
@@ -119,12 +119,12 @@ def process_and_cache_nwp(source_nwp_path: str, dest_nwp_path: str):
             else:
                 new_variables.append(var)
         ds.__setitem__("variable", new_variables)
-    
+
     # Hack to resolve some NWP data format differences between providers
     elif is_gfs:
         data_var = ds[list(ds.data_vars.keys())[0]]
         # # Use .to_dataset() to split the data variable based on 'variable' dim
-        ds = data_var.to_dataset(dim='variable') 
+        ds = data_var.to_dataset(dim="variable")
         ds = ds.rename({"t2m": "t"})
     # Save destination path
     ds.to_zarr(dest_nwp_path, mode="a")
@@ -136,18 +136,23 @@ def download_satellite_data(satellite_source_file_path: str) -> None:
     # download satellite data
     fs = fsspec.open(satellite_source_file_path).fs
     if fs.exists(satellite_source_file_path):
-        log.info(f"Downloading satellite data from {satellite_source_file_path} "
-                 f"to sat_15_min.zarr.zip")
+        log.info(
+            f"Downloading satellite data from {satellite_source_file_path} "
+            f"to sat_15_min.zarr.zip"
+        )
         fs.get(satellite_source_file_path, "sat_15_min.zarr.zip")
         log.info(f"Unzipping sat_15_min.zarr.zip to {satellite_path}")
         os.system(f"unzip -qq sat_15_min.zarr.zip -d {satellite_path}")
     else:
         log.error(f"Could not find satellite data at {satellite_source_file_path}")
 
+
 def set_night_time_zeros(batch, preds, sun_elevation_limit=0.0):
     """
     Set all predictions to zero for night time values
     """
+
+    log.debug("Setting night time values to zero")
     # get sun elevation values and if less 0, set to 0
     if BatchKey.wind_solar_elevation in batch.keys():
         key = BatchKey.wind_solar_elevation
@@ -167,7 +172,8 @@ def set_night_time_zeros(batch, preds, sun_elevation_limit=0.0):
         sun_elevation = sun_elevation.detach().cpu().numpy()
 
     # expand dimension from (1,197) to (1,197,7), 7 is due to the number plevels
-    sun_elevation = np.repeat(sun_elevation[:, :, np.newaxis], 7, axis=2)
+    n_plevels = preds.shape[2]
+    sun_elevation = np.repeat(sun_elevation[:, :, np.newaxis], n_plevels, axis=2)
     # only take future time steps
     sun_elevation = sun_elevation[:, batch[t0_key] + 1 :, :]
     preds[sun_elevation < sun_elevation_limit] = 0