# -*- coding: utf-8 -*-
# Copyright 2022, SERTIT-ICube - France, https://sertit.unistra.fr/
# This file is part of sertit-utils project
# https://github.com/sertit/sertit-utils
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Raster tools
You can use this only if you have installed sertit[full] or sertit[rasters_rio]
"""
import logging
import os
import tempfile
from functools import wraps
from pathlib import Path
from typing import Any, Callable, Optional, Union
import numpy as np
from cloudpathlib import AnyPath, CloudPath
from sertit.logs import SU_NAME
try:
import geopandas as gpd
import rasterio
from rasterio import MemoryFile, features
from rasterio import mask as rmask
from rasterio import merge, warp
from rasterio.enums import Resampling
from shapely.geometry import Polygon
except ModuleNotFoundError as ex:
raise ModuleNotFoundError(
"Please install 'rasterio' and 'geopandas' to use the 'rasters_rio' package."
) from ex
from sertit import files, misc, strings, vectors
np.seterr(divide="ignore", invalid="ignore")
MAX_CORES = os.cpu_count() - 2
PATH_ARR_DS = Union[str, tuple, rasterio.DatasetReader]
LOGGER = logging.getLogger(SU_NAME)
DEG_2_RAD = np.pi / 180
[docs]def bigtiff_value(arr: Any) -> str:
"""
Returns YES if array is larger than 4 GB, IF_NEEDED otherwise.
Args:
arr (Any): Numpy array or xarray
Returns:
str: YES or IF_NEEDED
"""
try:
itemsize = arr.itemsize
except AttributeError:
itemsize = arr.data.itemsize
# Check size
if arr.size * itemsize / 1024 / 1024 / 1024 > 4:
bigtiff = "YES"
else:
bigtiff = "IF_NEEDED"
return bigtiff
[docs]def path_arr_dst(function: Callable) -> Callable:
"""
Path, :code:`xarray`, (array, metadata) or dataset decorator.
Allows a function to ingest:
- a path
- a :code:`xarray`
- a :code:`rasterio` dataset
- :code:`rasterio` open data: (array, meta)
.. code-block:: python
>>> # Create mock function
>>> @path_or_dst
>>> def fct(dst):
>>> read(dst)
>>>
>>> # Test the two ways
>>> read1 = fct("path/to/raster.tif")
>>> with rasterio.open("path/to/raster.tif") as dst:
>>> read2 = fct(dst)
>>>
>>> # Test
>>> read1 == read2
True
Args:
function (Callable): Function to decorate
Returns:
Callable: decorated function
"""
@wraps(function)
def path_or_arr_or_dst_wrapper(
path_or_arr_or_ds: Union[str, rasterio.DatasetReader], *args, **kwargs
) -> Any:
"""
Path or dataset wrapper
Args:
path_or_arr_or_ds (Union[str, rasterio.DatasetReader]): Raster path or its dataset
*args: args
**kwargs: kwargs
Returns:
Any: regular output
"""
if isinstance(path_or_arr_or_ds, (str, Path, CloudPath)):
with rasterio.open(str(path_or_arr_or_ds)) as dst:
out = function(dst, *args, **kwargs)
elif isinstance(path_or_arr_or_ds, tuple):
arr, meta = path_or_arr_or_ds
with MemoryFile() as memfile:
with memfile.open(**meta, BIGTIFF=bigtiff_value(arr)) as dst:
dst.write(arr)
out = function(dst, *args, **kwargs)
else:
out = None
# Try if xarray is importable
try:
import xarray as xr
if isinstance(path_or_arr_or_ds, (xr.DataArray, xr.Dataset)):
meta = {
"driver": "GTiff",
"dtype": path_or_arr_or_ds.dtype,
"nodata": path_or_arr_or_ds.rio.encoded_nodata,
"width": path_or_arr_or_ds.rio.width,
"height": path_or_arr_or_ds.rio.height,
"count": path_or_arr_or_ds.rio.count,
"crs": path_or_arr_or_ds.rio.crs,
"transform": path_or_arr_or_ds.rio.transform(),
}
with MemoryFile() as memfile:
with memfile.open(
**meta, BIGTIFF=bigtiff_value(path_or_arr_or_ds)
) as dst:
if path_or_arr_or_ds.rio.encoded_nodata is not None:
path_or_arr_or_ds = path_or_arr_or_ds.fillna(
path_or_arr_or_ds.rio.encoded_nodata
)
dst.write(path_or_arr_or_ds.data)
out = function(dst, *args, **kwargs)
except ModuleNotFoundError:
out = None
if out is None:
out = function(path_or_arr_or_ds, *args, **kwargs)
return out
return path_or_arr_or_dst_wrapper
[docs]@path_arr_dst
def get_new_shape(
dst: PATH_ARR_DS, resolution: Union[tuple, list, float], size: Union[tuple, list]
) -> (int, int):
"""
Get the new shape (height, width) of a resampled raster.
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
resolution (Union[tuple, list, float]): Resolution of the wanted band, in dataset resolution unit (X, Y)
size (Union[tuple, list]): Size of the array (width, height). Not used if resolution is provided.
Returns:
(int, int): Height, width
"""
def _get_new_dim(dim: int, res_old: float, res_new: float) -> int:
"""
Get the new dimension in pixels
Args:
dim (int): Old dimension
res_old (float): Old resolution
res_new (float): New resolution
Returns:
int: New dimension
"""
return int(np.round(dim * res_old / res_new))
# By default keep original shape
new_height = dst.height
new_width = dst.width
# Compute new shape
if resolution is not None:
if isinstance(resolution, (int, float)):
new_height = _get_new_dim(dst.height, dst.res[1], resolution)
new_width = _get_new_dim(dst.width, dst.res[0], resolution)
elif resolution is None:
pass
else:
try:
if len(resolution) != 2:
raise ValueError(
"We should have a resolution for X and Y dimensions"
)
if resolution[0] is not None:
new_width = _get_new_dim(dst.width, dst.res[0], resolution[0])
if resolution[1] is not None:
new_height = _get_new_dim(dst.height, dst.res[1], resolution[1])
except (TypeError, KeyError):
raise ValueError(
f"Resolution should be None, 2 floats or a castable to a list: {resolution}"
)
elif size is not None:
try:
new_height = size[1]
new_width = size[0]
except (TypeError, KeyError):
raise ValueError(f"Size should be None or a castable to a list: {size}")
return new_height, new_width
[docs]def get_nodata_mask(
array: Union[np.ma.masked_array, np.ndarray],
has_nodata: bool,
default_nodata: int = 0,
) -> np.ndarray:
"""
Get nodata mask from a masked array.
The nodata may not be set before, then pass a nodata value that will be evaluated on the array.
.. code-block:: python
>>> diag_arr = np.diag([1,2,3])
array([[1, 0, 0],
[0, 2, 0],
[0, 0, 3]])
>>> get_nodata_mask(diag_arr, has_nodata=False)
array([[1, 0, 0],
[0, 1, 0],
[0, 0, 1]], dtype=uint8)
>>> get_nodata_mask(diag_arr, has_nodata=False, default_nodata=1)
array([[0, 1, 1],
[1, 1, 1],
[1, 1, 1]], dtype=uint8)
Args:
array (np.ma.masked_array): Array to evaluate
has_nodata (bool): If the array as its nodata specified. If not, using default_nodata.
default_nodata (int): Default nodata used if the array's nodata is not set
Returns:
np.ndarray: Pixelwise nodata array
"""
# Nodata mask
if not has_nodata or not isinstance(
array, np.ma.masked_array
): # Unspecified nodata is set to None by rasterio
if np.isnan(default_nodata):
nodata_mask = np.where(np.isnan(array), 0, 1).astype(np.uint8)
else:
nodata_mask = np.where(array != default_nodata, 1, 0).astype(np.uint8)
else:
nodata_mask = np.where(array.mask, 0, 1).astype(np.uint8)
return nodata_mask
[docs]@path_arr_dst
def rasterize(
dst: PATH_ARR_DS,
vector: Union[gpd.GeoDataFrame, Path, CloudPath, str],
value_field: str = None,
default_nodata: int = 0,
default_value: int = 1,
**kwargs,
) -> (np.ma.masked_array, dict):
"""
Rasterize a vector into raster format.
Note that passing `merge_alg = MergeAlg.add` will add the vector values to the given a raster
See: https://pygis.io/docs/e_raster_rasterize.html
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
vector (Union[gpd.GeoDataFrame, Path, CloudPath, str]): Vector to be rasterized
value_field (str): Field of the vector with the values to be burnt on the raster (should be scalars). If let to None, the raster will be binary (`default_nodata`, `default_value`).
default_nodata (int): Default nodata of the raster (outside the vector in the raster extent)
default_value (int): Used as value for all geometries, if `value_field` not provided
Returns:
np.ma.masked_array, dict: Rasterized vector and its metadata
"""
if not isinstance(vector, gpd.GeoDataFrame):
vector = vectors.read(vector)
# Manage nodata
if dst.nodata:
nodata = dst.nodata
else:
nodata = default_nodata
# Manage vector values
if value_field:
geom_value = (
(geom, value) for geom, value in zip(vector.geometry, vector[value_field])
)
dtype = kwargs.pop("dtype", vector[value_field].dtype)
else:
geom_value = vector.geometry
dtype = kwargs.pop("dtype", np.uint8)
# Rasterize vector
mask = features.rasterize(
geom_value,
out_shape=(dst.height, dst.width),
fill=nodata, # Outside vector
default_value=default_value, # Inside vector
transform=dst.transform,
dtype=dtype,
all_touched=kwargs.get("all_touched", True),
**kwargs,
)
meta = dst.meta.copy()
meta["dtype"] = dtype
meta["nodata"] = nodata
return mask, meta
@path_arr_dst
def _vectorize(
dst: PATH_ARR_DS,
values: Union[None, int, list] = None,
keep_values: bool = True,
dissolve: bool = False,
get_nodata: bool = False,
default_nodata: int = 0,
) -> gpd.GeoDataFrame:
"""
Vectorize a raster, both to get classes or nodata.
If dissolved is False, it returns a GeoDataFrame with a GeoSeries per cluster of pixel value,
with the value as an attribute. Else it returns a GeoDataFrame with a unique polygon.
.. WARNING::
If :code:`get_nodata` is set to False:
- Please only use this function on a classified raster.
- This could take a while as the computing time directly depends on the number of polygons to vectorize.
Please be careful.
Else:
- You will get a classified polygon with data (value=0)/nodata pixels.
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
values (Union[None, int, list]): Get only the polygons concerning this/these particular values
keep_values (bool): Keep the passed values. If False, discard them and keep the others.
dissolve (bool): Dissolve all the polygons into one unique. Only works if values are given.
get_nodata (bool): Get nodata vector (raster values are set to 0, nodata values are the other ones)
default_nodata (int): Default values for nodata in case of non existing in file
Returns:
gpd.GeoDataFrame: Vector
"""
# Get the shapes
array = dst.read(masked=True)
# Manage nodata value
has_nodata = dst.nodata is not None
nodata = dst.nodata if has_nodata else default_nodata
# Manage values
if values is not None:
if not isinstance(values, list):
values = [values]
# If we want a dissolved vector, just set 1instead of real values
arr_vals = 1 if dissolve else array
if keep_values:
true = arr_vals
false = nodata
else:
true = nodata
false = arr_vals
data = np.where(np.isin(array, values), true, false).astype(array.dtype)
else:
data = array.data
# Get nodata mask
nodata_mask = get_nodata_mask(data, has_nodata=False, default_nodata=nodata)
# Get shapes (on array or on mask to get nodata vector)
shapes = features.shapes(
nodata_mask if get_nodata else data,
mask=None if get_nodata else nodata_mask,
transform=dst.transform,
)
# Convert to geodataframe
gdf = vectors.shapes_to_gdf(shapes, dst.crs)
# Return valid geometries
gdf = vectors.make_valid(gdf)
# Dissolve if needed
if dissolve:
gdf = gpd.GeoDataFrame(geometry=gdf.geometry, crs=gdf.crs).dissolve()
return gdf
[docs]@path_arr_dst
def vectorize(
dst: PATH_ARR_DS,
values: Union[None, int, list] = None,
keep_values: bool = True,
dissolve: bool = False,
default_nodata: int = 0,
) -> gpd.GeoDataFrame:
"""
Vectorize a raster to get the class vectors.
If dissolved is False, it returns a GeoDataFrame with a GeoSeries per cluster of pixel value,
with the value as an attribute. Else it returns a GeoDataFrame with a unique polygon.
.. WARNING::
- Please only use this function on a classified raster.
- This could take a while as the computing time directly depends on the number of polygons to vectorize.
Please be careful.
.. code-block:: python
>>> raster_path = "path/to/raster.tif" # Classified raster, with no data set to 255
>>> vec1 = vectorize(raster_path)
>>> # or
>>> with rasterio.open(raster_path) as dst:
>>> vec2 = vectorize(dst)
>>> vec1 == vec2
True
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
values (Union[None, int, list]): Get only the polygons concerning this/these particular values
keep_values (bool): Keep the passed values. If False, discard them and keep the others.
dissolve (bool): Dissolve all the polygons into one unique. Only works if values are given.
default_nodata (int): Default values for nodata in case of non existing in file
Returns:
gpd.GeoDataFrame: Classes Vector
"""
return _vectorize(
dst,
values=values,
get_nodata=False,
keep_values=keep_values,
dissolve=dissolve,
default_nodata=default_nodata,
)
[docs]@path_arr_dst
def get_valid_vector(dst: PATH_ARR_DS, default_nodata: int = 0) -> gpd.GeoDataFrame:
"""
Get the valid data of a raster as a vector.
Pay attention that every nodata pixel will appear too.
If you want only the footprint of the raster, please use :code:`get_footprint`.
.. code-block:: python
>>> raster_path = "path/to/raster.tif" # Classified raster, with no data set to 255
>>> nodata1 = get_nodata_vec(raster_path)
>>> # or
>>> with rasterio.open(raster_path) as dst:
>>> nodata2 = get_nodata_vec(dst)
>>> nodata1 == nodata2
True
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
default_nodata (int): Default values for nodata in case of non existing in file
Returns:
gpd.GeoDataFrame: Nodata Vector
"""
nodata = _vectorize(
dst, values=None, get_nodata=True, default_nodata=default_nodata
)
return nodata[
nodata.raster_val != 0
] # 0 is the values of not nodata put there by rasterio
[docs]@path_arr_dst
def get_nodata_vector(dst: PATH_ARR_DS, default_nodata: int = 0) -> gpd.GeoDataFrame:
"""
Get the nodata vector of a raster as a vector.
Pay attention that every nodata pixel will appear too.
If you want only the footprint of the raster, please use :code:`get_footprint`.
.. code-block:: python
>>> raster_path = "path/to/raster.tif" # Classified raster, with no data set to 255
>>> nodata1 = get_nodata_vec(raster_path)
>>> # or
>>> with rasterio.open(raster_path) as dst:
>>> nodata2 = get_nodata_vec(dst)
>>> nodata1 == nodata2
True
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
default_nodata (int): Default values for nodata in case of non existing in file
Returns:
gpd.GeoDataFrame: Nodata Vector
"""
nodata = _vectorize(
dst, values=None, get_nodata=True, default_nodata=default_nodata
)
return nodata[
nodata.raster_val == 0
] # 0 is the values of not nodata put there by rasterio
@path_arr_dst
def _mask(
dst: PATH_ARR_DS,
shapes: Union[gpd.GeoDataFrame, Polygon, list],
nodata: Optional[int] = None,
do_crop: bool = False,
**kwargs,
) -> (np.ma.masked_array, dict):
"""
Overload of rasterio mask function in order to create a masked_array.
The :code:`mask` function docs can be seen `here <https://rasterio.readthedocs.io/en/latest/api/rasterio.mask.html>`_.
It basically masks a raster with a vector mask, with the possibility to crop the raster to the vector's extent.
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
shapes (Union[gpd.GeoDataFrame, Polygon, list]): Shapes with the same CRS as the dataset
(except if a :code:`GeoDataFrame` is passed, in which case it will automatically be converted.
nodata (int): Nodata value. If not set, uses the ds.nodata. If doesnt exist, set to 0.
do_crop (bool): Whether to crop the raster to the extent of the shapes. Default is False.
**kwargs: Other rasterio.mask options
Returns:
(np.ma.masked_array, dict): Cropped array as a masked array and its metadata
"""
if isinstance(shapes, (gpd.GeoDataFrame, gpd.GeoSeries)):
shapes = shapes.to_crs(dst.crs).geometry
elif not isinstance(shapes, list):
shapes = [shapes]
# Set nodata
if not nodata:
if dst.nodata:
nodata = dst.nodata
else:
nodata = 0
# Crop dataset
msk, trf = rmask.mask(dst, shapes, nodata=nodata, crop=do_crop, **kwargs)
# Create masked array
nodata_mask = np.where(msk == nodata, 1, 0).astype(np.uint8)
mask_array = np.ma.masked_array(msk, nodata_mask, fill_value=nodata)
# Update meta
out_meta = update_meta(mask_array, dst.meta)
out_meta["transform"] = trf
return mask_array, out_meta
[docs]@path_arr_dst
def mask(
dst: PATH_ARR_DS,
shapes: Union[gpd.GeoDataFrame, Polygon, list],
nodata: Optional[int] = None,
**kwargs,
) -> (np.ma.masked_array, dict):
"""
Masking a dataset:
setting nodata outside of the given shapes, but without cropping the raster to the shapes extent.
Overload of rasterio mask function in order to create a masked_array.
The :code:`mask` function docs can be seen `here <https://rasterio.readthedocs.io/en/latest/api/rasterio.mask.html>`_.
It basically masks a raster with a vector mask, with the possibility to crop the raster to the vector's extent.
.. code-block:: python
>>> raster_path = "path/to/raster.tif"
>>> shape_path = "path/to/shapes.geojson" # Any vector that geopandas can read
>>> shapes = gpd.read_file(shape_path)
>>> masked_raster1, meta1 = mask(raster_path, shapes)
>>> # or
>>> with rasterio.open(raster_path) as dst:
>>> masked_raster2, meta2 = mask(dst, shapes)
>>> masked_raster1 == masked_raster2
True
>>> meta1 == meta2
True
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
shapes (Union[gpd.GeoDataFrame, Polygon, list]): Shapes with the same CRS as the dataset
(except if a :code:`GeoDataFrame` is passed, in which case it will automatically be converted.
nodata (int): Nodata value. If not set, uses the ds.nodata. If doesnt exist, set to 0.
**kwargs: Other rasterio.mask options
Returns:
(np.ma.masked_array, dict): Masked array as a masked array and its metadata
"""
return _mask(dst, shapes=shapes, nodata=nodata, do_crop=False, **kwargs)
[docs]@path_arr_dst
def crop(
dst: PATH_ARR_DS,
shapes: Union[gpd.GeoDataFrame, Polygon, list],
nodata: Optional[int] = None,
**kwargs,
) -> (np.ma.masked_array, dict):
"""
Cropping a dataset:
setting nodata outside of the given shapes AND cropping the raster to the shapes extent.
**HOW:**
Overload of rasterio mask function in order to create a masked_array.
The :code:`mask` function docs can be seen `here <https://rasterio.readthedocs.io/en/latest/api/rasterio.mask.html>`_.
It basically masks a raster with a vector mask, with the possibility to crop the raster to the vector's extent.
.. code-block:: python
>>> raster_path = "path/to/raster.tif"
>>> shape_path = "path/to/shapes.geojson" # Any vector that geopandas can read
>>> shapes = gpd.read_file(shape_path)
>>> cropped_raster1, meta1 = crop(raster_path, shapes)
>>> # or
>>> with rasterio.open(raster_path) as dst:
>>> cropped_raster2, meta2 = crop(dst, shapes)
>>> cropped_raster1 == cropped_raster2
True
>>> meta1 == meta2
True
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
shapes (Union[gpd.GeoDataFrame, Polygon, list]): Shapes with the same CRS as the dataset
(except if a :code:`GeoDataFrame` is passed, in which case it will automatically be converted.
nodata (int): Nodata value. If not set, uses the ds.nodata. If doesnt exist, set to 0.
**kwargs: Other rasterio.mask options
Returns:
(np.ma.masked_array, dict): Cropped array as a masked array and its metadata
"""
return _mask(dst, shapes=shapes, nodata=nodata, do_crop=True, **kwargs)
[docs]@path_arr_dst
def read(
dst: PATH_ARR_DS,
resolution: Union[tuple, list, float] = None,
size: Union[tuple, list] = None,
resampling: Resampling = Resampling.nearest,
masked: bool = True,
**kwargs,
) -> (np.ma.masked_array, dict):
"""
Read a raster dataset from a :code:`rasterio.Dataset` or a path.
The resolution can be provided (in dataset unit) as:
- a tuple or a list of (X, Y) resolutions
- a float, in which case X resolution = Y resolution
- None, in which case the dataset resolution will be used
Tip:
Use index with a list of one element to keep a 3D array
.. code-block:: python
>>> raster_path = "path/to/raster.tif"
>>> raster1, meta1 = read(raster_path)
>>> # or
>>> with rasterio.open(raster_path) as dst:
>>> raster2, meta2 = read(dst)
>>> raster1 == raster2
True
>>> meta1 == meta2
True
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
resolution (Union[tuple, list, float]): Resolution of the wanted band, in dataset resolution unit (X, Y)
size (Union[tuple, list]): Size of the array (width, height). Not used if resolution is provided.
resampling (Resampling): Resampling method (nearest by default)
masked (bool): Get a masked array, :code:`True` by default (whereas it is False by default in rasterio)
**kwargs: Other dst.read() arguments such as indexes.
Returns:
np.ma.masked_array, dict: Masked array corresponding to the raster data and its meta data
"""
# Get new height and width
new_height, new_width = get_new_shape(dst, resolution, size)
# Manage out_shape
if "indexes" in kwargs:
if isinstance(kwargs["indexes"], int):
out_shape = (new_height, new_width)
else:
out_shape = (len(kwargs["indexes"]), new_height, new_width)
else:
out_shape = (dst.count, new_height, new_width)
# Read data
array = dst.read(
out_shape=out_shape,
resampling=resampling,
masked=masked,
**kwargs,
)
# Update meta
dst_transform = dst.transform * dst.transform.scale(
(dst.width / new_width), (dst.height / new_height)
)
dst_meta = dst.meta.copy()
dst_meta.update(
{
"height": new_height,
"width": new_width,
"transform": dst_transform,
"dtype": array.dtype,
"nodata": dst.nodata,
}
)
return array, dst_meta
[docs]def write(
raster: Union[np.ma.masked_array, np.ndarray],
meta: dict,
path: Union[str, CloudPath, Path],
**kwargs,
) -> None:
"""
Write raster to disk (encapsulation of rasterio's function)
Metadata will be copied and updated with raster's information (ie. width, height, count, type...)
The driver is GTiff by default, and no nodata value is provided.
The file will be compressed if the raster is a mask (saved as uint8)
.. code-block:: python
>>> raster_path = "path/to/raster.tif"
>>> raster_out = "path/to/out.tif"
>>> # Read raster
>>> raster, meta = read(raster_path)
>>> # Rewrite it on disk
>>> write(raster, meta, raster_out)
Args:
raster (Union[np.ma.masked_array, np.ndarray]): Raster to save on disk
meta (dict): Basic metadata that will be copied and updated with raster's information
path (Union[str, CloudPath, Path]): Path where to save it (directories should be existing)
**kwargs: Overloading metadata, ie :code:`nodata=255`
"""
raster_out = raster.copy()
# Manage raster type (impossible to write boolean arrays)
if raster_out.dtype == bool:
raster_out = raster_out.astype(np.uint8)
# Update metadata
out_meta = meta.copy()
# Update raster to be sure to write down correct nodata pixels
nodata = kwargs.get("nodata")
if nodata is None:
nodata = meta.get("nodata")
if nodata is None:
if isinstance(raster_out, np.ma.masked_array):
nodata = raster_out.fill_value
# TODO: change this with rasterio 1.3.0 (masked option in write)
if isinstance(raster_out, np.ma.masked_array):
raster_out[raster_out.mask] = nodata
out_meta["nodata"] = nodata
# Force compression and driver (but can be overwritten by kwargs)
out_meta["driver"] = "GTiff"
# Compress to LZW by default
out_meta["compress"] = kwargs.get("compress", "lzw")
# Bigtiff if needed (more than 4Go)
out_meta["BIGTIFF"] = bigtiff_value(raster_out)
# Set more threads
out_meta["NUM_THREADS"] = MAX_CORES
# Update metadata with array data
out_meta = update_meta(raster_out, out_meta)
# Update metadata with additional params
for key, val in kwargs.items():
out_meta[key] = val
# Manage raster shape
if len(raster_out.shape) == 2:
raster_out = np.expand_dims(raster_out, axis=0)
# Write product
with rasterio.open(str(path), "w", **out_meta) as dst:
dst.write(raster_out)
[docs]def collocate(
master_meta: dict,
slave_arr: Union[np.ma.masked_array, np.ndarray],
slave_meta: dict,
resampling: Resampling = Resampling.nearest,
) -> (Union[np.ma.masked_array, np.ndarray], dict):
"""
Collocate two georeferenced arrays:
forces the *slave* raster to be exactly georeferenced onto the *master* raster by reprojection.
.. code-block:: python
>>> master_path = "path/to/master.tif"
>>> slave_path = "path/to/slave.tif"
>>> col_path = "path/to/collocated.tif"
>>> # Just open the master data
>>> with rasterio.open(master_path) as master_dst:
>>> # Read slave
>>> slave, slave_meta = read(slave_path)
>>> # Collocate the slave to the master
>>> col_arr, col_meta = collocate(master_dst.meta,
>>> slave,
>>> slave_meta,
>>> Resampling.bilinear)
>>> # Write it
>>> write(col_arr, col_path, col_meta)
Args:
master_meta (dict): Master metadata
slave_arr (np.ma.masked_array): Slave array to be collocated
slave_meta (dict): Slave metadata
resampling (Resampling): Resampling method
Returns:
np.ma.masked_array, dict: Collocated array and its metadata
"""
collocated_arr = np.zeros(
(master_meta["count"], master_meta["height"], master_meta["width"]),
dtype=master_meta["dtype"],
)
warp.reproject(
source=slave_arr,
destination=collocated_arr,
src_transform=slave_meta["transform"],
src_crs=slave_meta["crs"],
dst_transform=master_meta["transform"],
dst_crs=master_meta["crs"],
src_nodata=slave_meta["nodata"],
dst_nodata=slave_meta["nodata"],
resampling=resampling,
num_threads=MAX_CORES,
)
meta = master_meta.copy()
meta.update(
{
"dtype": slave_meta["dtype"],
"driver": slave_meta["driver"],
"nodata": slave_meta["nodata"],
}
)
if isinstance(slave_arr, np.ma.masked_array):
collocated_arr = np.ma.masked_array(
collocated_arr, slave_arr.mask, fill_value=slave_meta["nodata"]
)
return collocated_arr, meta
[docs]def sieve(
array: Union[np.ma.masked_array, np.ndarray],
out_meta: dict,
sieve_thresh: int,
connectivity: int = 4,
) -> (Union[np.ma.masked_array, np.ndarray], dict):
"""
Sieving, overloads rasterio function with raster shaped like (1, h, w).
Forces the output to :code:`np.uint8` (as only classified rasters should be sieved)
.. code-block:: python
>>> raster_path = "path/to/raster.tif" # classified raster
>>> # Read raster
>>> raster, meta = read(raster_path)
>>> # Rewrite it
>>> sieved, sieved_meta = sieve(raster, meta, sieve_thresh=20)
>>> # Write it
>>> raster_out = "path/to/raster_sieved.tif"
>>> write(sieved, raster_out, sieved_meta)
Args:
array (Union[np.ma.masked_array, np.ndarray]): Array to sieve
out_meta (dict): Metadata to update
sieve_thresh (int): Sieving threshold in pixels
connectivity (int): Connectivity, either 4 or 8
Returns:
(Union[np.ma.masked_array, np.ndarray], dict): Sieved array and updated meta
"""
assert connectivity in [4, 8]
# Read extraction array
expand = False
if len(array.shape) == 3 and array.shape[0] == 1:
array = np.squeeze(array) # Use this trick to make the sieve work
expand = True
# Get nodata mask
if isinstance(array, np.ma.masked_array):
mask = ~array.mask
else:
mask = np.ones_like(array)
if expand:
mask = np.squeeze(mask)
# Convert to np.uint8 if needed
dtype = np.uint8
meta = out_meta.copy()
if meta["dtype"] != dtype:
array = array.astype(dtype)
meta["dtype"] = dtype
# Sieve
result_array = np.empty(array.shape, dtype=array.dtype)
features.sieve(
array, size=sieve_thresh, out=result_array, connectivity=connectivity, mask=mask
)
# Use this trick to get the array back to 'normal'
if expand:
result_array = np.expand_dims(result_array, axis=0)
return result_array, meta
[docs]def get_dim_img_path(
dim_path: Union[str, CloudPath, Path], img_name: str = "*"
) -> Union[CloudPath, Path]:
"""
Get the image path from a :code:`BEAM-DIMAP` data.
A :code:`BEAM-DIMAP` file cannot be opened by rasterio, although its :code:`.img` file can.
.. code-block:: python
>>> dim_path = "path/to/dimap.dim" # BEAM-DIMAP image
>>> img_path = get_dim_img_path(dim_path)
>>> # Read raster
>>> raster, meta = read(img_path)
Args:
dim_path (Union[str, CloudPath, Path]): DIM path (.dim or .data)
img_name (str): .img file name (or regex), in case there are multiple .img files (ie. for S3 data)
Returns:
Union[CloudPath, Path]: .img file
"""
dim_path = AnyPath(dim_path)
if dim_path.suffix == ".dim":
dim_path = dim_path.with_suffix(".data")
assert dim_path.suffix == ".data" and dim_path.is_dir()
return files.get_file_in_dir(dim_path, img_name, extension="img", exact_name=True)
[docs]@path_arr_dst
def get_extent(dst: PATH_ARR_DS) -> gpd.GeoDataFrame:
"""
Get the extent of a raster as a :code:`geopandas.Geodataframe`.
.. code-block:: python
>>> raster_path = "path/to/raster.tif"
>>> extent1 = get_extent(raster_path)
>>> # or
>>> with rasterio.open(raster_path) as dst:
>>> extent2 = get_extent(dst)
>>> extent1 == extent2
True
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
Returns:
gpd.GeoDataFrame: Extent as a :code:`geopandas.Geodataframe`
"""
return vectors.get_geodf(geometry=[*dst.bounds], crs=dst.crs)
[docs]def merge_vrt(
crs_paths: list,
crs_merged_path: Union[str, CloudPath, Path],
abs_path: bool = False,
**kwargs,
) -> None:
"""
Merge rasters as a VRT. Uses :code:`gdalbuildvrt`.
See here: https://gdal.org/programs/gdalbuildvrt.html
Creates VRT with relative paths !
.. WARNING::
They should have the same CRS otherwise the mosaic will be false !
.. code-block:: python
>>> paths_utm32630 = ["path/to/raster1.tif", "path/to/raster2.tif", "path/to/raster3.tif"]
>>> paths_utm32631 = ["path/to/raster4.tif", "path/to/raster5.tif"]
>>> mosaic_32630 = "path/to/mosaic_32630.vrt"
>>> mosaic_32631 = "path/to/mosaic_32631.vrt"
>>> # Create mosaic, one by CRS !
>>> merge_vrt(paths_utm32630, mosaic_32630)
>>> merge_vrt(paths_utm32631, mosaic_32631, {"-srcnodata":255, "-vrtnodata":0})
Args:
crs_paths (list): Path of the rasters to be merged with the same CRS)
crs_merged_path (Union[str, CloudPath, Path]): Path to the merged raster
abs_path (bool): VRT with absolute paths. If not, VRT with relative paths (default)
kwargs: Other gdlabuildvrt arguments
"""
# Copy crs_paths in order not to modify it in place (replacing str by Paths for example)
crs_paths_cp = crs_paths.copy()
for idp, path in enumerate(crs_paths_cp):
crs_paths_cp[idp] = path
# Manage cloud paths (gdalbuildvrt needs url or true filepaths)
crs_merged_path = AnyPath(crs_merged_path)
if isinstance(crs_merged_path, CloudPath):
crs_merged_path = AnyPath(crs_merged_path.fspath)
for i, crs_path in enumerate(crs_paths_cp):
path = AnyPath(crs_path)
if isinstance(path, CloudPath):
path = AnyPath(path.fspath)
crs_paths_cp[i] = path
# Create relative paths
vrt_root = os.path.dirname(crs_merged_path)
try:
if abs_path:
paths = [
strings.to_cmd_string(files.to_abspath(path)) for path in crs_paths_cp
]
vrt_path = strings.to_cmd_string(crs_merged_path.resolve())
else:
paths = [
strings.to_cmd_string(files.real_rel_path(path, vrt_root))
for path in crs_paths_cp
]
vrt_path = strings.to_cmd_string(
files.real_rel_path(crs_merged_path, vrt_root)
)
except ValueError:
# ValueError when crs_merged_path and crs_paths are not on the same disk
paths = [strings.to_cmd_string(str(path)) for path in crs_paths_cp]
vrt_path = strings.to_cmd_string(str(crs_merged_path))
# Run cmd
arg_list = [val for item in kwargs.items() for val in item]
try:
vrt_cmd = ["gdalbuildvrt", vrt_path, *paths, *arg_list]
misc.run_cli(vrt_cmd, cwd=vrt_root)
except RuntimeError:
# Manage too long command line
with tempfile.TemporaryDirectory() as tmp_dir:
tmp_file = os.path.join(tmp_dir, "list.txt")
with open(tmp_file, "w+") as f:
for path in paths:
path = path.replace('"', "")
f.write(f"{path}\n")
vrt_cmd = [
"gdalbuildvrt",
"-input_file_list",
tmp_file,
vrt_path,
*arg_list,
]
misc.run_cli(vrt_cmd, cwd=vrt_root)
[docs]def merge_gtiff(
crs_paths: list, crs_merged_path: Union[str, CloudPath, Path], **kwargs
) -> None:
"""
Merge rasters as a GeoTiff.
.. WARNING::
They should have the same CRS otherwise the mosaic will be false !
.. code-block:: python
>>> paths_utm32630 = ["path/to/raster1.tif", "path/to/raster2.tif", "path/to/raster3.tif"]
>>> paths_utm32631 = ["path/to/raster4.tif", "path/to/raster5.tif"]
>>> mosaic_32630 = "path/to/mosaic_32630.tif"
>>> mosaic_32631 = "path/to/mosaic_32631.tif"
# Create mosaic, one by CRS !
>>> merge_gtiff(paths_utm32630, mosaic_32630)
>>> merge_gtiff(paths_utm32631, mosaic_32631)
Args:
crs_paths (list): Path of the rasters to be merged with the same CRS)
crs_merged_path (Union[str, CloudPath, Path]): Path to the merged raster
kwargs: Other rasterio.merge arguments
More info `here <https://rasterio.readthedocs.io/en/latest/api/rasterio.merge.html#rasterio.merge.merge>`_
"""
# Open datasets for merging
crs_datasets = []
try:
for path in crs_paths:
crs_datasets.append(rasterio.open(str(path)))
# Merge all datasets
merged_array, merged_transform = merge.merge(crs_datasets, **kwargs)
merged_meta = crs_datasets[0].meta.copy()
merged_meta.update(
{
"driver": "GTiff",
"height": merged_array.shape[1],
"width": merged_array.shape[2],
"transform": merged_transform,
}
)
finally:
# Close all datasets
for dataset in crs_datasets:
dataset.close()
# Save merge datasets
write(merged_array, merged_meta, crs_merged_path)
[docs]def unpackbits(array: np.ndarray, nof_bits: int) -> np.ndarray:
"""
Function found
`here <https://stackoverflow.com/questions/18296035/how-to-extract-the-bits-of-larger-numeric-numpy-data-types>`_
.. code-block:: python
>>> bit_array = np.random.randint(5, size=[3,3])
array([[1, 1, 3],
[4, 2, 0],
[4, 3, 2]], dtype=uint8)
# Unpack 8 bits (8*1, as itemsize of uint8 is 1)
>>> unpackbits(bit_array, 8)
array([[[1, 0, 0, 0, 0, 0, 0, 0],
[1, 0, 0, 0, 0, 0, 0, 0],
[1, 1, 0, 0, 0, 0, 0, 0]],
[[0, 0, 1, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0, 0],
[0, 0, 0, 0, 0, 0, 0, 0]],
[[0, 0, 1, 0, 0, 0, 0, 0],
[1, 1, 0, 0, 0, 0, 0, 0],
[0, 1, 0, 0, 0, 0, 0, 0]]], dtype=uint8)
Args:
array (np.ndarray): Array to unpack
nof_bits (int): Number of bits to unpack
Returns:
np.ndarray: Unpacked array
"""
dtype = array.dtype
if dtype == np.uint8:
unpacked = np.unpackbits(
np.expand_dims(array, axis=-1), axis=-1, count=nof_bits, bitorder="little"
)
else:
xshape = list(array.shape)
array = array.reshape([-1, 1])
msk = 2 ** np.arange(nof_bits, dtype=array.dtype).reshape([1, nof_bits])
unpacked = (
(array & msk).astype(bool).astype(np.uint8).reshape(xshape + [nof_bits])
)
return unpacked
[docs]def read_bit_array(
bit_mask: np.ndarray, bit_id: Union[list, int]
) -> Union[np.ndarray, list]:
"""
Read bit arrays as a succession of binary masks (sort of read a slice of the bit mask, slice number bit_id)
.. code-block:: python
>>> bit_array = np.random.randint(5, size=[3,3])
array([[1, 1, 3],
[4, 2, 0],
[4, 3, 2]], dtype=uint8)
# Get the 2nd bit array
>>> read_bit_array(bit_array, 2)
array([[0, 0, 0],
[1, 0, 0],
[1, 0, 0]], dtype=uint8)
Args:
bit_mask (np.ndarray): Bit array to read
bit_id (int): Bit ID of the slice to be read
Example: read the bit 0 of the mask as a cloud mask (Theia)
Returns:
Union[np.ndarray, list]: Binary mask or list of binary masks if a list of bit_id is given
"""
# Get the number of bits
nof_bits = 8 * bit_mask.dtype.itemsize
# Read cloud mask as bits
msk = unpackbits(bit_mask, nof_bits)
# Only keep the bit number bit_id and reshape the vector
if isinstance(bit_id, list):
bit_arr = [msk[..., bid] for bid in bit_id]
else:
bit_arr = msk[..., bit_id]
return bit_arr
[docs]@path_arr_dst
def hillshade(
dst: PATH_ARR_DS, azimuth: float = 315, zenith: float = 45
) -> (np.ma.masked_array, dict):
"""
Compute the hillshade of a DEM from an azimuth and elevation angle (in degrees).
Goal: replace `gdaldem CLI <https://gdal.org/programs/gdaldem.html>`_
NB: altitude = 90 - zenith
.. WARNING::
- It uses a 2nd order gradient instead of Horn's or Zevenbergen & Thorne's formula
- z_factor is fixed to 1.0
- scale managed by dst resolution
`Reference <https://git.earthdata.nasa.gov/projects/GEE/repos/gdal-enhancements-for-esdis/browse/gdal-1.10.0/apps/gdaldem.cpp>`_
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
azimuth (float): Azimuth angle in degrees
zenith (float): Zenith angle in degrees
Returns:
(np.ma.masked_array, dict): Hillshade and its metadata
"""
array = dst.read(masked=True)
# Squeeze if needed
expand = False
if len(array.shape) == 3 and array.shape[0] == 1:
array = np.squeeze(array) # Use this trick to make the sieve work
expand = True
# Compute angles
az_rad = azimuth * DEG_2_RAD
alt_rad = (90 - zenith) * DEG_2_RAD
# Compute slope and aspect
dx, dy = np.gradient(np.where(array.mask, 0.0, array.data), *dst.res)
x2_y2 = dx ** 2 + dy ** 2
aspect = np.arctan2(dx, dy)
# Compute hillshade (GDAL algo)
hillshade = (
np.sin(alt_rad) + np.cos(alt_rad) * np.sqrt(x2_y2) * np.sin(aspect - az_rad)
) / np.sqrt(1 + x2_y2)
hillshade = np.where(hillshade <= 0, 1.0, 254.0 * hillshade + 1)
# Use this trick to get the array back to 'normal'
if expand:
hillshade = np.expand_dims(hillshade, axis=0)
# Convert to masked array
hillshade_msk = np.ma.masked_array(hillshade, array.mask, fill_value=dst.nodata)
# Meta
meta = update_meta(hillshade_msk, dst.meta)
return hillshade_msk, meta
[docs]@path_arr_dst
def slope(
dst: PATH_ARR_DS,
in_pct: bool = False,
in_rad: bool = False,
) -> (np.ma.masked_array, dict):
"""
Compute the slope of a DEM (in degrees).
Goal: replace `gdaldem CLI <https://gdal.org/programs/gdaldem.html>`_
.. WARNING::
- It uses a 2nd order gradient instead of Horn's or Zevenbergen & Thorne's formula
- z_factor is fixed to 1.0
- scale managed by dst resolution
`Reference <https://git.earthdata.nasa.gov/projects/GEE/repos/gdal-enhancements-for-esdis/browse/gdal-1.10.0/apps/gdaldem.cpp>`_
Args:
dst (PATH_ARR_DS): Path to the raster, its dataset, its :code:`xarray` or a tuple containing its array and metadata
in_pct (bool): Outputs slope in percents
in_rad (bool): Outputs slope in radians. Not taken into account if :code:`in_pct == True`
Returns:
(np.ma.masked_array, dict): Slope and its metadata
"""
array = dst.read(masked=True)
# Squeeze if needed
expand = False
if len(array.shape) == 3 and array.shape[0] == 1:
array = np.squeeze(array) # Use this trick to make the sieve work
expand = True
# Compute slope (on unmasked data)
dx, dy = np.gradient(np.where(array.mask, 0.0, array.data), *dst.res)
x2_y2 = dx ** 2 + dy ** 2
if in_pct:
slope = 100 * (np.sqrt(x2_y2))
else:
slope = np.arctan(np.sqrt(x2_y2))
# Convert into degrees
if not in_rad:
slope = slope / DEG_2_RAD
# Use this trick to get the array back to 'normal'
if expand:
slope = np.expand_dims(slope, axis=0)
# Convert to masked array
slope_msk = np.ma.masked_array(slope, array.mask, fill_value=dst.nodata)
# Meta
meta = update_meta(slope_msk, dst.meta)
return slope_msk, meta
[docs]def reproject_match(
dst_meta: dict,
src_arr: Union[np.ma.masked_array, np.ndarray],
src_meta: dict,
resampling: Resampling = Resampling.nearest,
**kwargs,
) -> (Union[np.ma.masked_array, np.ndarray], dict):
"""
Reproject a raster to match the resolution, projection, and region of another raster.
Matching rioxarray reproject_match.
Args:
dst_meta (dict): Destination metadata
src_arr (Union[np.ma.masked_array, np.ndarray]): Source raster's array
src_meta (dict): Source metadata
resampling (Resampling): Resampling method
**kwargs: Passing other kwargs to `calculate_default_transform` and `reproject`
Returns:
Union[np.ma.masked_array, np.ndarray], dict: Reprojected array and its metadata
"""
# Source metadata
src_crs = src_meta["crs"]
src_count = src_meta["count"]
src_dtype = src_meta["dtype"]
src_nodata = src_meta["nodata"]
# Destination metadata
dst_crs = dst_meta["crs"]
meta = dst_meta.copy()
# Reproject data
dst_arr = np.empty(
(src_count, dst_meta["height"], dst_meta["width"]), dtype=src_dtype
)
dst_arr, dst_tr = warp.reproject(
source=src_arr,
destination=dst_arr,
src_crs=src_crs,
dst_crs=dst_crs,
src_transform=src_meta["transform"],
dst_transform=dst_meta["transform"],
src_nodata=src_nodata,
dst_nodata=dst_meta["nodata"], # input data should be in integer
num_threads=MAX_CORES,
resampling=resampling,
**kwargs,
)
# Update metadata
meta["count"] = src_count
meta["nodata"] = src_nodata
meta["dtype"] = src_dtype
meta["transform"] = dst_tr
meta["crs"] = dst_crs
meta["driver"] = "GTiff" # Force GTiff
return dst_arr, meta
