from copy import deepcopy
from pathlib import Path
from textwrap import dedent
from typing import Any, Optional, TextIO
import numpy as np
import pandas as pd
import xarray as xr
import imod
from imod.mf6.utilities.regrid import RegridderWeightsCache
from imod.msw.fixed_format import VariableMetaData
from imod.msw.pkgbase import MetaSwapPackage
from imod.msw.utilities.common import find_in_file_list
from imod.prepare import common
from imod.typing import GridDataArray, Imod5DataDict, IntArray
from imod.util.regrid_method_type import RegridMethodType
def _is_parsable_and_existing_path(potential_path: str, mete_grid_path: Path) -> bool:
"""
mete_grid.inp can contain values like "0.", which are converted to float by
MetaSWAP. String is converted to path and checked if existing path.
"""
try:
float(potential_path)
return False
except ValueError:
# Resolve paths relative to mete_grid.inp path.
path = mete_grid_path / ".." / Path(potential_path)
return path.is_file()
def open_first_meteo_grid(mete_grid_path: str | Path, column_nr: int) -> xr.DataArray:
"""
Find and open first meteo grid path in mete_grid.inp. This grid is enough to
generate meteomappings. There can be floats before in the column which
should be skipped.
"""
if column_nr not in [2, 3]:
raise ValueError("Column nr should be 2 or 3")
mete_grid_path = Path(mete_grid_path)
with open(mete_grid_path, "r") as f:
lines = f.readlines()
potential_paths = [line.split(",")[column_nr].replace('"', "") for line in lines]
for potential_path in potential_paths:
if _is_parsable_and_existing_path(potential_path, mete_grid_path):
resolved_path = mete_grid_path / ".." / Path(potential_path)
return imod.rasterio.open(resolved_path)
error_message = dedent(f"""
Did not find parsable path to existing .ASC file in column {column_nr}. Got
values (printing first 10): {potential_paths[:10]}.""")
raise ValueError(error_message)
def open_first_meteo_grid_from_imod5_data(imod5_data: Imod5DataDict, column_nr: int):
paths = imod5_data["extra"]["paths"]
metegrid_path = find_in_file_list("mete_grid.inp", paths)
return open_first_meteo_grid(metegrid_path, column_nr=column_nr)
class MeteoMapping(MetaSwapPackage):
"""
This class provides common methods for creating mappings between
meteorological data and MetaSWAP grids. It should not be instantiated
by the user but rather be inherited from within imod-python to create
new packages.
"""
meteo: GridDataArray
def __init__(self):
super().__init__()
def _render(
self,
file: TextIO,
index: IntArray,
svat: xr.DataArray,
*args: Any,
):
data_dict = {"svat": svat.values.ravel()[index]}
row, column = self.grid_mapping(svat, self.meteo)
data_dict["row"] = row[index]
data_dict["column"] = column[index]
dataframe = pd.DataFrame(
data=data_dict, columns=list(self._metadata_dict.keys())
)
self._check_range(dataframe)
return self.write_dataframe_fixed_width(file, dataframe)
@staticmethod
def grid_mapping(svat: xr.DataArray, meteo_grid: xr.DataArray) -> pd.DataFrame:
flip_meteo_x = meteo_grid.indexes["x"].is_monotonic_decreasing
flip_meteo_y = meteo_grid.indexes["y"].is_monotonic_decreasing
nrow = meteo_grid["y"].size
ncol = meteo_grid["x"].size
# Convert to cell boundaries for the meteo grid
# Method always returns monotonic increasing edges
meteo_x = common._coord(meteo_grid, "x")
meteo_y = common._coord(meteo_grid, "y")
# Create the SVAT grid
svat_grid_y, svat_grid_x = np.meshgrid(svat.y, svat.x, indexing="ij")
svat_grid_y = svat_grid_y.ravel()
svat_grid_x = svat_grid_x.ravel()
# Determine where the svats fit in within the cell boundaries of the meteo grid
row = np.searchsorted(meteo_y, svat_grid_y)
column = np.searchsorted(meteo_x, svat_grid_x)
# Find out of bounds members
if (column == 0).any() or (column > ncol).any():
raise ValueError("Some values are out of bounds for column")
if (row == 0).any() or (row > nrow).any():
raise ValueError("Some values are out of bounds for row")
# Flip axis when meteofile bound are flipped, relative to the coords
if flip_meteo_y:
row = (nrow + 1) - row
if flip_meteo_x:
column = (ncol + 1) - column
n_subunit = svat["subunit"].size
return np.tile(row, n_subunit), np.tile(column, n_subunit)
def regrid_like(
self,
target_grid: GridDataArray,
regrid_context: RegridderWeightsCache,
regridder_types: Optional[RegridMethodType] = None,
):
return deepcopy(self)
[docs]
class PrecipitationMapping(MeteoMapping):
"""
This contains the data to connect precipitation grid cells to MetaSWAP
svats. The precipitation grid does not have to be equal to the metaswap
grid: connections between the precipitation cells to svats will be
established using a nearest neighbour lookup.
This class is responsible for the file `svat2precgrid.inp`.
Parameters
----------
precipitation: array of floats (xr.DataArray)
Describes the precipitation data. The extend of the grid must be larger
than the MetaSvap grid. The data must also be coarser than the MetaSvap
grid.
"""
_file_name = "svat2precgrid.inp"
_metadata_dict = {
"svat": VariableMetaData(10, None, None, int),
"row": VariableMetaData(10, None, None, int),
"column": VariableMetaData(10, None, None, int),
}
[docs]
def __init__(
self,
precipitation: xr.DataArray,
):
super().__init__()
self.meteo = precipitation
[docs]
@classmethod
def from_imod5_data(cls, imod5_data: Imod5DataDict) -> "PrecipitationMapping":
"""
Construct a MetaSWAP PrecipitationMapping package from iMOD5 data in the
CAP package, loaded with the
:func:`imod.formats.prj.open_projectfile_data` function.
Opens first ascii grid in mete_grid.inp, which is used to construct
mappings to svats. The grids should not change in dimension over time.
No checks are done whether cells switch from inactive to active or vice
versa.
Parameters
----------
imod5_data: Imod5DataDict
iMOD5 data as returned by
:func:`imod.formats.prj.open_projectfile_data`
Returns
-------
imod.msw.PrecipitationMapping
"""
column_nr = 2
meteo_grid = open_first_meteo_grid_from_imod5_data(imod5_data, column_nr)
return cls(meteo_grid)
[docs]
class EvapotranspirationMapping(MeteoMapping):
"""
This contains the data to connect evapotranspiration grid cells to MetaSWAP
svats. The evapotranspiration grid does not have to be equal to the metaswap
grid: connections between the evapotranspiration cells to svats will be
established using a nearest neighbour lookup.
This class is responsible for the file `svat2etrefgrid.inp`.
Parameters
----------
evapotransporation: array of floats (xr.DataArray)
Describes the evapotransporation data. The extend of the grid must be
larger than the MetaSvap grid. The data must also be coarser than the
MetaSvap grid.
"""
_file_name = "svat2etrefgrid.inp"
_metadata_dict = {
"svat": VariableMetaData(10, None, None, int),
"row": VariableMetaData(10, None, None, int),
"column": VariableMetaData(10, None, None, int),
}
[docs]
def __init__(
self,
evapotranspiration: xr.DataArray,
):
super().__init__()
self.meteo = evapotranspiration
[docs]
@classmethod
def from_imod5_data(cls, imod5_data: Imod5DataDict) -> "EvapotranspirationMapping":
"""
Construct a MetaSWAP EvapotranspirationMapping package from iMOD5 data
in the CAP package, loaded with the
:func:`imod.formats.prj.open_projectfile_data` function.
Opens first ascii grid in mete_grid.inp, which is used to construct
mappings to svats. The grids should not change in dimension over time.
No checks are done whether cells switch from inactive to active or vice
versa.
Parameters
----------
imod5_data: Imod5DataDict
iMOD5 data as returned by
:func:`imod.formats.prj.open_projectfile_data`
Returns
-------
imod.msw.EvapotranspirationMapping
"""
column_nr = 3
meteo_grid = open_first_meteo_grid_from_imod5_data(imod5_data, column_nr)
return cls(meteo_grid)
