Support lidar data

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "res-wind-up"
-version = "0.4.4"
+version = "0.4.5"
 authors = [
     { name = "Alex Clerc", email = "alex.clerc@res-group.com" }
 ]

wind_up/interface.py

@@ -13,7 +13,7 @@
 from wind_up.constants import REANALYSIS_WD_COL
 from wind_up.northing import add_wf_yawdir, apply_northing_corrections
 from wind_up.optimize_northing import auto_northing_corrections
-from wind_up.reanalysis_data import ReanalysisDataset, add_reanalysis_data
+from wind_up.reanalysis_data import MastOrLiDARDataset, ReanalysisDataset, add_reanalysis_data
 from wind_up.scada_funcs import filter_scada_df, get_raw_scada_and_cfg_from_file
 from wind_up.scada_power_curve import calc_pc_and_rated_ws
 from wind_up.smart_data import add_smart_lat_long_to_cfg
@@ -143,6 +143,7 @@ class AssessmentInputs:
     cfg: WindUpConfig
     plot_cfg: PlotConfig
     pre_post_splitter: PrePostSplitter
+    mast_or_lidar_datasets: list[MastOrLiDARDataset] | None = None
 
     @classmethod
     def from_cfg(
@@ -155,6 +156,7 @@ def from_cfg(
         toggle_df: pd.DataFrame | None = None,
         reanalysis_datasets: list[ReanalysisDataset],
         cache_dir: Path | None = None,
+        mast_or_lidar_datasets: list[MastOrLiDARDataset] | None = None,
     ) -> AssessmentInputs:
         """Construct instance of AssessmentInputs from configuration objects and data.
 
@@ -183,6 +185,7 @@ def from_cfg(
             cfg=cfg,
             plot_cfg=plot_cfg,
             pre_post_splitter=pre_post_splitter,
+            mast_or_lidar_datasets=mast_or_lidar_datasets,
         )
 
 

wind_up/main_analysis.py

@@ -12,7 +12,6 @@
 import wind_up
 from wind_up.circular_math import circ_diff
 from wind_up.constants import (
-    PROJECTROOT_DIR,
     RANDOM_SEED,
     REANALYSIS_WD_COL,
     REANALYSIS_WS_COL,
@@ -44,6 +43,7 @@
 
 if TYPE_CHECKING:
     from wind_up.models import PlotConfig, Turbine, WindUpConfig
+    from wind_up.reanalysis_data import MastOrLiDARDataset
 
 logger = logging.getLogger(__name__)
 
@@ -183,6 +183,7 @@ def _get_ref_df(
     test_wtg: Turbine,
     toggle_df: pd.DataFrame | None = None,
     keep_only_toggle_off: bool = True,
+    mast_or_lidar_datasets: list[MastOrLiDARDataset] | None = None,
 ) -> pd.DataFrame:
     if ref_name in [x.name for x in cfg.asset.wtgs]:
         ref_df = wf_df.loc[ref_name].copy()
@@ -201,9 +202,10 @@ def _get_ref_df(
             original_wd_col = REANALYSIS_WD_COL
         elif ref_name in [x.name for x in cfg.asset.masts_and_lidars]:
             ref_obj = next(x for x in cfg.asset.masts_and_lidars if x.name == ref_name)
-            ref_df = pd.read_parquet(
-                PROJECTROOT_DIR / "input_data" / "masts_and_lidars" / cfg.asset.name / f"{ref_obj.data_file_name}",
-            )
+            if mast_or_lidar_datasets is None:
+                msg = f"No mast or lidar datasets provided for {ref_name}"
+                raise ValueError(msg)
+            ref_df = next(x for x in mast_or_lidar_datasets if x.id == ref_name).data
             northing_df = wf_df.loc[cfg.test_wtgs[0].name, [REANALYSIS_WS_COL, REANALYSIS_WD_COL, WINDFARM_YAWDIR_COL]]
             ref_df = ref_df.merge(northing_df, how="left", left_index=True, right_index=True)
             original_ws_col = ref_obj.wind_speed_column
@@ -449,6 +451,7 @@ def _calc_test_ref_results(
     plot_cfg: PlotConfig,
     random_seed: int,
     toggle_df: pd.DataFrame | None = None,
+    mast_or_lidar_datasets: list[MastOrLiDARDataset] | None = None,
 ) -> dict:
     test_name = test_wtg.name
     (plot_cfg.plots_dir / test_name / ref_name).mkdir(exist_ok=True, parents=True)
@@ -478,6 +481,7 @@ def _calc_test_ref_results(
         test_wtg=test_wtg,
         toggle_df=toggle_df,
         keep_only_toggle_off=keep_only_toggle_off,
+        mast_or_lidar_datasets=mast_or_lidar_datasets,
     )
     if len(ref_df) == 0:
         result_manager.warning(f"ref_df is empty for {ref_name}")
@@ -925,6 +929,7 @@ def run_wind_up_analysis(
                 cfg=cfg,
                 plot_cfg=plot_cfg,
                 random_seed=random_seed,
+                mast_or_lidar_datasets=inputs.mast_or_lidar_datasets,
             )
             test_ref_results = test_ref_results | test_results
             logger.info(test_ref_results)

wind_up/models.py

@@ -97,10 +97,6 @@ class MastOrLidar(BaseModel):
     name: str = Field(description="Object name", min_length=2)
     latitude: float = Field(default=np.nan, ge=-90, le=90)
     longitude: float = Field(default=np.nan, ge=-180, le=180)
-    data_file_name: str = Field(
-        min_length=2,
-        description="Name of data timeseries file",
-    )
     wind_speed_column: str = Field(min_length=2, description="Name of wind speed column in data timeseries file")
     wind_direction_column: str = Field(
         min_length=2,

wind_up/reanalysis_data.py

@@ -222,3 +222,11 @@ def add_reanalysis_data(
     if plot_cfg is not None:
         plot_wf_and_reanalysis_sample_timeseries(wf_df=wf_and_reanalysis_df, plot_cfg=plot_cfg)
     return wf_and_reanalysis_df
+
+
+@dataclass
+class MastOrLiDARDataset:
+    """Class to store mast or LiDAR data."""
+
+    id: str
+    data: pd.DataFrame