roms-tools 1.7.0__py3-none-any.whl → 2.1.0__py3-none-any.whl

roms_tools/setup/nesting.py

@@ -0,0 +1,575 @@
+import numpy as np
+import xarray as xr
+from scipy.interpolate import griddata
+from dataclasses import dataclass, field
+from typing import Dict, Union
+from pathlib import Path
+from roms_tools.setup.grid import Grid
+from roms_tools.setup.utils import (
+    interpolate_from_rho_to_u,
+    interpolate_from_rho_to_v,
+    get_boundary_coords,
+    wrap_longitudes,
+    save_datasets,
+    _to_yaml,
+    _from_yaml,
+)
+from roms_tools.setup.plot import _plot_nesting
+import logging
+from scipy.interpolate import interp1d
+
+
+@dataclass(frozen=True, kw_only=True)
+class Nesting:
+    """Represents the relationship between a parent and a child grid in nested ROMS
+    simulations. This class facilitates mapping the boundaries of the child grid onto
+    the parent grid indices and modifying the child grid topography such that it matches
+    the parent topography at the boundaries.
+
+    Parameters
+    ----------
+    parent_grid : Grid
+        The parent grid object, containing information about the larger domain.
+    child_grid : Grid
+        The child grid object, containing information about the nested domain.
+    boundaries : Dict[str, bool], optional
+        Dictionary specifying which boundaries of the child grid are used
+        in the nesting process. Keys are "south", "east", "north", and "west",
+        with boolean values indicating inclusion. Defaults to all boundaries (True).
+    child_prefix : str, optional
+        Prefix added to variable names in the generated dataset to distinguish
+        child grid information. Defaults to "child".
+    period : float, optional
+        The temporal resolution or output period for boundary variables in the child grid.
+        Defaults to hourly.
+
+    Attributes
+    ----------
+    ds : xr.Dataset
+        An xarray Dataset containing the index mappings between the child and parent grids
+        for each specified boundary. Includes metadata about grid points, angles,
+        and boundary variable outputs.
+    """
+
+    parent_grid: Grid
+    child_grid: Grid
+    boundaries: Dict[str, bool] = field(
+        default_factory=lambda: {
+            "south": True,
+            "east": True,
+            "north": True,
+            "west": True,
+        }
+    )
+    child_prefix: str = "child"
+    period: float = 3600.0
+
+    def __post_init__(self):
+
+        parent_grid_ds = self.parent_grid.ds
+        child_grid_ds = self.child_grid.ds
+
+        # Adjust longitude for dateline crossing to prevent interpolation artifacts
+        for grid_ds in [parent_grid_ds, child_grid_ds]:
+            grid_ds = wrap_longitudes(grid_ds, straddle=self.parent_grid.straddle)
+
+        # Map child boundaries onto parent grid indices
+        ds = map_child_boundaries_onto_parent_grid_indices(
+            parent_grid_ds,
+            child_grid_ds,
+            self.boundaries,
+            self.child_prefix,
+            self.period,
+        )
+        object.__setattr__(self, "ds", ds)
+
+        # Modify child topography and mask to match the parent grid
+        child_grid_ds = modify_child_topography_and_mask(
+            parent_grid_ds, child_grid_ds, self.boundaries
+        )
+
+        # Convert longitudes back to [0, 360] range
+        for grid_ds in [parent_grid_ds, child_grid_ds]:
+            grid_ds = wrap_longitudes(grid_ds, straddle=False)
+        object.__setattr__(self.parent_grid, "ds", parent_grid_ds)
+        object.__setattr__(self.child_grid, "ds", child_grid_ds)
+
+    def plot(self, with_dim_names=False) -> None:
+        """Plot the parent and child grids in a single figure.
+
+        Returns
+        -------
+        None
+            This method does not return any value. It generates and displays a plot.
+        """
+
+        _plot_nesting(
+            self.parent_grid.ds,
+            self.child_grid.ds,
+            self.parent_grid.straddle,
+            with_dim_names,
+        )
+
+    def save(
+        self,
+        filepath: Union[str, Path],
+        filepath_child_grid: Union[str, Path],
+        np_eta: int = None,
+        np_xi: int = None,
+    ) -> None:
+        """Save the nesting and child grid file to netCDF4 files. The child grid file is
+        required because the topography and mask of the child grid has been modified.
+
+        This method allows saving the nesting and child grid data either each as a single file or each partitioned into multiple files, based on the provided options. The dataset can be saved in two modes:
+
+        1. **Single File Mode (default)**:
+            - If both `np_eta` and `np_xi` are `None`, the entire dataset is saved as a single netCDF4 file.
+            - The file is named based on the provided `filepath`, with `.nc` automatically appended to the filename.
+
+        2. **Partitioned Mode**:
+            - If either `np_eta` or `np_xi` is specified, the dataset is partitioned spatially along the `eta` and `xi` axes into tiles.
+            - Each tile is saved as a separate netCDF4 file. Filenames will be modified with an index to represent each partition, e.g., `"filepath_YYYYMM.0.nc"`, `"filepath_YYYYMM.1.nc"`, etc.
+
+        Parameters
+        ----------
+        filepath : Union[str, Path]
+            The base path and filename for the output files. The filenames will include the specified path and the `.nc` extension.
+            If partitioning is used, additional indices will be appended to the filenames, e.g., `"filepath.0.nc"`, `"filepath.1.nc"`, etc.
+
+        filepath_child_grid : Union[str, Path]
+            The base path and filename for saving the childe grid file.
+
+        np_eta : int, optional
+            The number of partitions along the `eta` direction. If `None`, no spatial partitioning is performed along the `eta` axis.
+
+        np_xi : int, optional
+            The number of partitions along the `xi` direction. If `None`, no spatial partitioning is performed along the `xi` axis.
+
+        Returns
+        -------
+        List[Path]
+            A list of `Path` objects for the saved files. Each element in the list corresponds to a file that was saved.
+        """
+
+        # Ensure filepath is a Path object
+        filepath = Path(filepath)
+        filepath_child_grid = Path(filepath_child_grid)
+
+        # Remove ".nc" suffix if present
+        if filepath.suffix == ".nc":
+            filepath = filepath.with_suffix("")
+        if filepath_child_grid.suffix == ".nc":
+            filepath_child_grid = filepath_child_grid.with_suffix("")
+
+        dataset_list = [self.ds, self.child_grid.ds]
+        output_filenames = [str(filepath), str(filepath_child_grid)]
+
+        saved_filenames = save_datasets(
+            dataset_list, output_filenames, np_eta=np_eta, np_xi=np_xi
+        )
+
+        return saved_filenames
+
+    def to_yaml(self, filepath: Union[str, Path]) -> None:
+        """Export the parameters of the class to a YAML file, including the version of
+        roms-tools.
+
+        Parameters
+        ----------
+        filepath : Union[str, Path]
+            The path to the YAML file where the parameters will be saved.
+        """
+
+        _to_yaml(self, filepath)
+
+    @classmethod
+    def from_yaml(cls, filepath: Union[str, Path]) -> "Nesting":
+        """Create an instance of the Nesting class from a YAML file.
+
+        Parameters
+        ----------
+        filepath : Union[str, Path]
+            The path to the YAML file from which the parameters will be read.
+
+        Returns
+        -------
+        Nesting
+            An instance of the Nesting class.
+        """
+        filepath = Path(filepath)
+
+        parent_grid = Grid.from_yaml(filepath, "ParentGrid")
+        child_grid = Grid.from_yaml(filepath, "ChildGrid")
+        params = _from_yaml(cls, filepath)
+
+        return cls(parent_grid=parent_grid, child_grid=child_grid, **params)
+
+
+def map_child_boundaries_onto_parent_grid_indices(
+    parent_grid_ds,
+    child_grid_ds,
+    boundaries={"south": True, "east": True, "north": True, "west": True},
+    child_prefix="child",
+    period=3600.0,
+    update_land_indices=True,
+):
+
+    bdry_coords_dict = get_boundary_coords()
+
+    # add angles at u- and v-points
+    child_grid_ds["angle_u"] = interpolate_from_rho_to_u(child_grid_ds["angle"])
+    child_grid_ds["angle_v"] = interpolate_from_rho_to_v(child_grid_ds["angle"])
+
+    ds = xr.Dataset()
+
+    for direction in ["south", "east", "north", "west"]:
+        if boundaries[direction]:
+            for grid_location in ["rho", "u", "v"]:
+                names = {
+                    "latitude": f"lat_{grid_location}",
+                    "longitude": f"lon_{grid_location}",
+                    "mask": f"mask_{grid_location}",
+                    "angle": f"angle_{grid_location}",
+                }
+                bdry_coords = bdry_coords_dict[grid_location]
+                if grid_location == "rho":
+                    suffix = "r"
+                else:
+                    suffix = grid_location
+
+                lon_child = child_grid_ds[names["longitude"]].isel(
+                    **bdry_coords[direction]
+                )
+                lat_child = child_grid_ds[names["latitude"]].isel(
+                    **bdry_coords[direction]
+                )
+
+                mask_child = child_grid_ds[names["mask"]].isel(**bdry_coords[direction])
+
+                i_eta, i_xi = interpolate_indices(
+                    parent_grid_ds, lon_child, lat_child, mask_child
+                )
+
+                if update_land_indices:
+                    i_eta, i_xi = update_indices_if_on_parent_land(
+                        i_eta, i_xi, grid_location, parent_grid_ds
+                    )
+
+                var_name = f"{child_prefix}_{direction}_{suffix}"
+                if grid_location == "rho":
+                    ds[var_name] = xr.concat([i_xi, i_eta], dim="two")
+                    ds[var_name].attrs[
+                        "long_name"
+                    ] = f"{grid_location}-points of {direction}ern child boundary mapped onto parent (absolute) grid indices"
+                    ds[var_name].attrs["units"] = "non-dimensional"
+                    ds[var_name].attrs["output_vars"] = "zeta, temp, salt"
+                else:
+                    angle_child = child_grid_ds[names["angle"]].isel(
+                        **bdry_coords[direction]
+                    )
+                    ds[var_name] = xr.concat([i_xi, i_eta, angle_child], dim="three")
+                    ds[var_name].attrs[
+                        "long_name"
+                    ] = f"{grid_location}-points  of {direction}ern child boundary mapped onto parent grid (absolute) indices and angle"
+                    ds[var_name].attrs["units"] = "non-dimensional and radian"
+
+                    if grid_location == "u":
+                        ds[var_name].attrs["output_vars"] = "ubar, u, up"
+                    elif grid_location == "v":
+                        ds[var_name].attrs["output_vars"] = "vbar, v, vp"
+
+                ds[var_name].attrs["output_period"] = period
+
+    vars_to_drop = ["lat_rho", "lon_rho", "lat_u", "lon_u", "lat_v", "lon_v"]
+    vars_to_drop_existing = [var for var in vars_to_drop if var in ds]
+    ds = ds.drop_vars(vars_to_drop_existing)
+
+    # Rename dimensions
+    dims_to_rename = {
+        dim: f"{child_prefix}_{dim}" for dim in ds.dims if dim not in ["two", "three"]
+    }
+    ds = ds.rename(dims_to_rename)
+
+    ds = ds.assign_coords(
+        {
+            "indices_rho": ("two", ["xi", "eta"]),
+            "indices_vel": ("three", ["xi", "eta", "angle"]),
+        }
+    )
+
+    return ds
+
+
+def interpolate_indices(parent_grid_ds, lon, lat, mask):
+    """Interpolate the parent indices to the child grid.
+
+    Parameters
+    ----------
+    parent_grid_ds : xarray.Dataset
+        Grid information of parent grid.
+    lon : xarray.DataArray
+        Longitudes of the child grid where interpolation is desired.
+    lat : xarray.DataArray
+        Latitudes of the child grid where interpolation is desired.
+    mask: xarray.DataArray
+        Mask for the child longitudes and latitudes under consideration.
+    Returns
+    -------
+    i : xarray.DataArray
+        Interpolated i-indices for the child grid.
+    j : xarray.DataArray
+        Interpolated j-indices for the child grid.
+    """
+    i_eta = np.arange(-0.5, len(parent_grid_ds.eta_rho) + -0.5, 1)
+    i_xi = np.arange(-0.5, len(parent_grid_ds.xi_rho) + -0.5, 1)
+
+    parent_grid_ds = parent_grid_ds.assign_coords(
+        i_eta=("eta_rho", i_eta)
+    ).assign_coords(i_xi=("xi_rho", i_xi))
+
+    lon_parent = parent_grid_ds.lon_rho
+    lat_parent = parent_grid_ds.lat_rho
+    i_parent = parent_grid_ds.i_eta
+    j_parent = parent_grid_ds.i_xi
+
+    # Create meshgrid
+    j_parent, i_parent = np.meshgrid(j_parent.values, i_parent.values)
+
+    # Flatten the input coordinates and indices for griddata
+    points = np.column_stack((lon_parent.values.ravel(), lat_parent.values.ravel()))
+    i_parent_flat = i_parent.ravel()
+    j_parent_flat = j_parent.ravel()
+
+    # Interpolate the i and j indices
+    i = griddata(points, i_parent_flat, (lon.values, lat.values), method="linear")
+    j = griddata(points, j_parent_flat, (lon.values, lat.values), method="linear")
+
+    i = xr.DataArray(i, dims=lon.dims)
+    j = xr.DataArray(j, dims=lon.dims)
+
+    # Check for NaN values
+    if np.sum(np.isnan(i)) > 0 or np.sum(np.isnan(j)) > 0:
+        raise ValueError(
+            "Some points are outside the grid. Please choose either a bigger parent grid or a smaller child grid."
+        )
+
+    # Check whether indices are close to border of parent grid
+    nxp, nyp = lon_parent.shape
+    if np.min(i) < 0 or np.max(i) > nxp - 2:
+        logging.warning(
+            "Some boundary points of the child grid are very close to the boundary of the parent grid."
+        )
+    if np.min(j) < 0 or np.max(j) > nyp - 2:
+        logging.warning(
+            "Some boundary points of the child grid are very close to the boundary of the parent grid."
+        )
+
+    return i, j
+
+
+def update_indices_if_on_parent_land(i_eta, i_xi, grid_location, parent_grid_ds):
+    """Finds points that are in the parent land mask but not land masked in the child
+    and replaces parent indices with nearest neighbor wet points.
+
+    Parameters
+    ----------
+    i_eta : xarray.DataArray
+        Interpolated i_eta-indices for the child grid.
+    i_xi : xarray.DataArray
+        Interpolated i_xi-indices for the child grid.
+    mask: xarray.DataArray
+        Mask for the child longitudes and latitudes under consideration.
+    grid_location : str
+        Location type ('rho', 'u', 'v').
+    parent_grid_ds : xarray.Dataset
+        Grid information of parent grid.
+
+    Returns
+    -------
+    i_eta : xarray.DataArray
+        Updated i_eta-indices for the child grid.
+    i_xi : xarray.DataArray
+        Updated i_xi-indices for the child grid.
+    """
+
+    if grid_location == "rho":
+        i_eta_rho = i_eta + 0.5
+        i_xi_rho = i_xi + 0.5
+        mask_rho = parent_grid_ds.mask_rho
+        summed_mask = np.zeros_like(i_eta_rho)
+
+        for i in range(len(i_eta_rho)):
+            i_eta_lower = int(np.floor(i_eta_rho[i]))
+            i_xi_lower = int(np.floor(i_xi_rho[i]))
+            mask = mask_rho.isel(
+                eta_rho=slice(i_eta_lower, i_eta_lower + 2),
+                xi_rho=slice(i_xi_lower, i_xi_lower + 2),
+            )
+            summed_mask[i] = np.sum(mask)
+
+    elif grid_location in ["u", "v"]:
+        i_eta_u = i_eta + 0.5
+        i_xi_u = i_xi
+
+        mask_u = parent_grid_ds.mask_u
+        summed_mask_u = np.zeros_like(i_eta_u)
+
+        for i in range(len(i_eta_u)):
+            i_eta_lower = int(np.floor(i_eta_u[i]))
+            i_xi_lower = int(np.floor(i_xi_u[i]))
+            mask = mask_u.isel(
+                eta_rho=slice(i_eta_lower, i_eta_lower + 2),
+                xi_u=slice(i_xi_lower, i_xi_lower + 2),
+            )
+            summed_mask_u[i] = np.sum(mask)
+
+        i_eta_v = i_eta
+        i_xi_v = i_xi + 0.5
+
+        mask_v = parent_grid_ds.mask_v
+        summed_mask_v = np.zeros_like(i_xi_v)
+
+        for i in range(len(i_eta_v)):
+            i_eta_lower = int(np.floor(i_eta_v[i]))
+            i_xi_lower = int(np.floor(i_xi_v[i]))
+            mask = mask_v.isel(
+                eta_v=slice(i_eta_lower, i_eta_lower + 2),
+                xi_rho=slice(i_xi_lower, i_xi_lower + 2),
+            )
+            summed_mask_v[i] = np.sum(mask)
+
+        summed_mask = summed_mask_u * summed_mask_v
+
+    # Filter out points where summed_mask is 0
+    valid_points = summed_mask != 0
+    x_mod = np.arange(len(summed_mask))[valid_points]
+    i_eta_mod = i_eta[valid_points]
+    i_xi_mod = i_xi[valid_points]
+
+    # Handle indices where summed_mask is 0
+    indx = np.where(summed_mask == 0)[0]
+    if len(indx) > 0:
+        i_eta_interp = interp1d(
+            x_mod, i_eta_mod, kind="nearest", fill_value="extrapolate"
+        )
+        i_xi_interp = interp1d(
+            x_mod, i_xi_mod, kind="nearest", fill_value="extrapolate"
+        )
+
+        i_eta[indx] = i_eta_interp(indx)
+        i_xi[indx] = i_xi_interp(indx)
+
+    return i_eta, i_xi
+
+
+def modify_child_topography_and_mask(
+    parent_grid_ds,
+    child_grid_ds,
+    boundaries={"south": True, "east": True, "north": True, "west": True},
+):
+
+    # regrid parent topography and mask onto child grid
+    points = np.column_stack(
+        (parent_grid_ds.lon_rho.values.ravel(), parent_grid_ds.lat_rho.values.ravel())
+    )
+    xi = (child_grid_ds.lon_rho.values, child_grid_ds.lat_rho.values)
+
+    values = parent_grid_ds["h"].values.ravel()
+    h_parent_interpolated = griddata(points, values, xi, method="linear")
+    h_parent_interpolated = xr.DataArray(
+        h_parent_interpolated, dims=("eta_rho", "xi_rho")
+    )
+
+    values = parent_grid_ds["mask_rho"].values.ravel()
+    mask_parent_interpolated = griddata(points, values, xi, method="linear")
+    mask_parent_interpolated = xr.DataArray(
+        mask_parent_interpolated, dims=("eta_rho", "xi_rho")
+    )
+
+    # compute weight based on distance
+    alpha = compute_boundary_distance(child_grid_ds["mask_rho"], boundaries)
+    # update child topography and mask to be weighted sum between original child and interpolated parent
+    child_grid_ds["h"] = (
+        alpha * child_grid_ds["h"] + (1 - alpha) * h_parent_interpolated
+    )
+
+    child_mask = (
+        alpha * child_grid_ds["mask_rho"] + (1 - alpha) * mask_parent_interpolated
+    )
+    child_grid_ds["mask_rho"] = xr.where(child_mask >= 0.5, 1, 0)
+
+    return child_grid_ds
+
+
+def compute_boundary_distance(
+    child_mask, boundaries={"south": True, "east": True, "north": True, "west": True}
+):
+    """Computes a normalized distance field from the boundaries of a grid, given a mask
+    and boundary conditions. The normalized distance values range from 0 (boundary) to 1
+    (inner grid).
+
+    Parameters
+    ----------
+    child_mask : xr.DataArray
+        A 2D xarray DataArray representing the land/sea mask of the grid (1 for sea, 0 for land),
+        with dimensions ("eta_rho", "xi_rho").
+    boundaries : dict, optional
+        A dictionary specifying which boundaries are open. Keys are "south", "east", "north", "west",
+        with boolean values indicating whether the boundary is open.
+
+    Returns
+    -------
+    xr.DataArray
+        A 2D DataArray with normalized distance values.
+    """
+    dist = np.full_like(child_mask, 1e6, dtype=float)
+    nx, ny = child_mask.shape
+    n = max(nx, ny)
+
+    x = np.arange(nx) / n
+    y = np.arange(ny) / n
+    x, y = np.meshgrid(x, y, indexing="ij")
+
+    trans = 0.05
+    width = int(np.ceil(n * trans))
+
+    if boundaries["south"]:
+        bx = x[:, 0][child_mask[:, 0] > 0]
+        by = y[:, 0][child_mask[:, 0] > 0]
+        for i in range(len(bx)):
+            dtmp = (x[:, :width] - bx[i]) ** 2 + (y[:, :width] - by[i]) ** 2
+            dist[:, :width] = np.minimum(dist[:, :width], dtmp)
+
+    if boundaries["east"]:
+        bx = x[-1, :][child_mask[-1, :] > 0]
+        by = y[-1, :][child_mask[-1, :] > 0]
+        for i in range(len(bx)):
+            dtmp = (x[nx - width : nx, :] - bx[i]) ** 2 + (
+                y[nx - width : nx, :] - by[i]
+            ) ** 2
+            dist[nx - width : nx, :] = np.minimum(dist[nx - width : nx, :], dtmp)
+
+    if boundaries["north"]:
+        bx = x[:, -1][child_mask[:, -1] > 0]
+        by = y[:, -1][child_mask[:, -1] > 0]
+        for i in range(len(bx)):
+            dtmp = (x[:, ny - width : ny] - bx[i]) ** 2 + (
+                y[:, ny - width : ny] - by[i]
+            ) ** 2
+            dist[:, ny - width : ny] = np.minimum(dist[:, ny - width : ny], dtmp)
+
+    if boundaries["west"]:
+        bx = x[0, :][child_mask[0, :] > 0]
+        by = y[0, :][child_mask[0, :] > 0]
+        for i in range(len(bx)):
+            dtmp = (x[:width, :] - bx[i]) ** 2 + (y[:width, :] - by[i]) ** 2
+            dist[:width, :] = np.minimum(dist[:width, :], dtmp)
+
+    dist = np.sqrt(dist)
+    dist[dist > trans] = trans
+    dist = dist / trans
+    alpha = 0.5 - 0.5 * np.cos(np.pi * dist)
+
+    alpha = xr.DataArray(alpha, dims=("eta_rho", "xi_rho"))
+
+    return alpha