Regrid NWP inputs

.github/workflows/conda-pytest.yaml

@@ -0,0 +1,32 @@
+name: Conda pytest
+on: push
+
+jobs:
+  push_to_registry:
+    name: Build docker file for testing
+    runs-on: ubuntu-latest
+    steps:
+      - name: Check out the repo
+        uses: actions/checkout@v2
+
+      - uses: conda-incubator/setup-miniconda@v2
+        with:
+          miniconda-version: "latest"
+          channels: bioconda, conda-forge, defaults
+          auto-update-conda: true
+          auto-activate-base: true
+
+      - name: Install the conda
+        shell: bash -l {0}
+        run: |
+          conda install -c conda-forge xesmf esmpy -y
+
+      - name: Install the repo
+        shell: bash -l {0}
+        run: |
+          pip install -e .
+
+      - name: run pytest
+        shell: bash -l {0}
+        run: |
+          python -m pytest

Dockerfile

@@ -1,40 +1,36 @@
-FROM python:3.10-slim
+FROM continuumio/miniconda3
 
 ARG TESTING=0
 
-# make sure it doesnt fail if the docker file doesnt know the git commit
-ARG GIT_PYTHON_REFRESH=quiet
+SHELL ["/bin/bash", "-l", "-c"]
 
 RUN apt-get update
 RUN apt-get install git -y
 RUN apt-get install g++ gcc libgeos++-dev libproj-dev proj-data proj-bin -y
 
-# copy files
+# Copy files
 COPY setup.py app/setup.py
 COPY README.md app/README.md
 COPY requirements.txt app/requirements.txt
-RUN pip install git+https://github.com/SheffieldSolar/PV_Live-API#pvlive_api
-
-
-# install requirements
-RUN pip install torch --index-url https://download.pytorch.org/whl/cpu
-RUN pip install -r app/requirements.txt
-
-# copy library files
 COPY pvnet_app/ app/pvnet_app/
 COPY tests/ app/tests/
 COPY scripts/ app/scripts/
+COPY data/ app/data/
 
-# change to app folder
+# Install requirements
+RUN conda install python=3.10
+RUN conda install -c conda-forge xesmf esmpy -y
+RUN echo "export ESMFMKFILE='/opt/conda/lib/esmf.mk'" >> ~/.bashrc
+RUN pip install torch --index-url https://download.pytorch.org/whl/cpu
+RUN pip install -r app/requirements.txt
+RUN pip install git+https://github.com/SheffieldSolar/PV_Live-API#pvlive_api
+
+# Change to app folder
 WORKDIR /app
 
-# install library
+# Install library
 RUN pip install -e .
 
-# download models so app can used cached
-RUN python scripts/cache_default_models.py
-
-
 RUN if [ "$TESTING" = 1 ]; then pip install pytest pytest-cov coverage; fi
 
-CMD ["python", "-u","pvnet_app/app.py"]
+CMD ["python", "-u","pvnet_app/app.py"]

pvnet_app/app.py

@@ -19,6 +19,7 @@
 import torch
 import typer
 import xarray as xr
+import xesmf as xe
 from nowcasting_datamodel.connection import DatabaseConnection
 from nowcasting_datamodel.models import (
     ForecastSQL,
@@ -102,6 +103,42 @@
 # ---------------------------------------------------------------------------
 # HELPER FUNCTIONS
 
+def regrid_nwp_data(nwp_path):
+    """This function loads the NWP data, then regrids and saves it back out if the data is not on
+    the same grid as expected. The data is resaved in-place.
+    """
+    ds_raw = xr.open_zarr(nwp_path)
+
+    # These are the coords we are aiming for
+    ds_target_coords = xr.load_dataset(f"{this_dir}/../data/nwp_target_coords.nc")
+
+    # Check if regridding step needs to be done
+    needs_regridding = not (
+        ds_raw.latitude.equals(ds_target_coords.latitude) and
+         ds_raw.longitude.equals(ds_target_coords.longitude)
+
+    )
+
+    if not needs_regridding:
+        logger.info("No NWP regridding required - skipping this step")
+        return
+
+    logger.info("Regridding NWP to expected grid")
+    # Its more efficient to regrid eagerly
+    ds_raw = ds_raw.compute()
+
+    # Regrid
+    regridder = xe.Regridder(ds_raw, ds_target_coords, method="bilinear")
+    ds_regridded = regridder(ds_raw)
+
+    # Re-save - including rechunking
+    os.system(f"rm -fr {nwp_path}")
+    ds_regridded["variable"] = ds_regridded["variable"].astype(str)
+    ds_regridded.chunk(dict(step=12, x=100, y=100)).to_zarr(nwp_path)
+
+    return
+
+
 def populate_data_config_sources(input_path, output_path):
     """Resave the data config and replace the source filepaths
 
@@ -319,6 +356,9 @@ def app(
     fs = fsspec.open(os.environ["NWP_ZARR_PATH"]).fs
     fs.get(os.environ["NWP_ZARR_PATH"], "nwp.zarr", recursive=True)
 
+    # Regrid the nwp data if needed
+    regrid_nwp_data("nwp.zarr")
+
     # ---------------------------------------------------------------------------
     # 2. Set up data loader
     logger.info("Creating DataLoader")

requirements.txt

@@ -7,10 +7,11 @@ fsspec[s3]
 xarray
 zarr
 numpy
-pandas==2.0
+pandas
 sqlalchemy
 torchdata
 pytest
 pytest-cov
 typer
 testcontainers
+xesmf

tests/conftest.py

@@ -88,7 +88,7 @@ def nwp_data():
         f"{os.path.dirname(os.path.abspath(__file__))}/test_data/nwp_shell.zarr"
     )
 
-    # Last init time was at least 2 hours ago and hour to 3-hour interval
+    # Last init time was at least 2 hours ago and floor to 3-hour interval
     t0_datetime_utc = time_before_present(timedelta(hours=2)).floor(timedelta(hours=3))
     ds.init_time.values[:] = pd.date_range(
         t0_datetime_utc - timedelta(hours=3 * (len(ds.init_time) - 1)),
@@ -104,17 +104,16 @@ def nwp_data():
     for v in list(ds.variables.keys()):
         if ds[v].dtype == object:
             ds[v].encoding.clear()
-
+    
     # Add data to dataset
     ds["UKV"] = xr.DataArray(
-        np.zeros([len(ds[c]) for c in ds.coords]),
-        coords=ds.coords,
+        np.zeros([len(ds[c]) for c in ds.xindexes]),
+        coords=[ds[c] for c in ds.xindexes],
     )
 
     # Add stored attributes to DataArray
     ds.UKV.attrs = ds.attrs["_data_attrs"]
     del ds.attrs["_data_attrs"]
-
     return ds
 
 

tests/test_data/nwp_shell.zarr/.zattrs

@@ -4,7 +4,9 @@
         "GRIB_centre": "egrr",
         "GRIB_centreDescription": "U.K. Met Office - Exeter",
         "GRIB_edition": 2,
-        "GRIB_subCentre": 0,
+        "GRIB_subCentre": 5,
+        "history": "2023-10-26T12:56 GRIB to CDM+CF via cfgrib-0.9.10.3/ecCodes-2.31.0 with {\"source\": \"../tmp/nwpc/ground_snow-depth-water-equivalent_2023102609.grib\", \"filter_by_keys\": {}, \"encode_cf\": [\"parameter\", \"time\", \"geography\", \"vertical\"]}",
         "institution": "U.K. Met Office - Exeter"
-    }
-}
+    },
+    "coordinates": "latitude longitude"
+}

tests/test_data/nwp_shell.zarr/.zgroup

@@ -1,3 +1,3 @@
 {
     "zarr_format": 2
-}
+}