VHR example#

Let’s use EOReader with Very High Resolution data.

Imports#

import os
import logging

# EOReader
from eoreader.reader import Reader
from eoreader.bands import GREEN, NDVI, TIR_1, CLOUDS, HILLSHADE, to_str
from eoreader.env_vars import DEM_PATH

Create the logger#

# Create logger

logger = logging.getLogger("eoreader")
logger.setLevel(logging.INFO)

# create console handler and set level to debug
ch = logging.StreamHandler()
ch.setLevel(logging.INFO)

# create formatter
formatter = logging.Formatter('%(message)s')

# add formatter to ch
ch.setFormatter(formatter)

# add ch to logger
logger.addHandler(ch)

Open the VHR product#

Please be aware that EOReader will always work in UTM projection.
So if you give WGS84 data, EOReader will reproject the stacks and this can be time-consuming

# Set a DEM
os.environ[DEM_PATH] = os.path.join(
    "/home", "data", "DS2", "BASES_DE_DONNEES", "GLOBAL",
    "MERIT_Hydrologically_Adjusted_Elevations", "MERIT_DEM.vrt"
)

# Open your product
path = os.path.join("/home", "data", "DATA", "PRODS", "PLEIADES", "5547047101", "IMG_PHR1A_PMS_001")
reader = Reader()
prod = reader.open(path, remove_tmp=True)
prod
eoreader.PldProduct 'PHR1A_PMS_202005110231585_ORT_5547047101'
Attributes:
	condensed_name: 20200511T023158_PLD_ORT_PMS_5547047101
	path: /home/data/DATA/PRODS/PLEIADES/5547047101/IMG_PHR1A_PMS_001
	constellation: Pleiades
	sensor type: Optical
	product type: Ortho Single Image
	default resolution: 0.5
	acquisition datetime: 2020-05-11T02:31:58
	band mapping:
		BLUE: 3
		GREEN: 2
		RED: 1
		NIR: 4
		NARROW_NIR: 4
	needs extraction: False
	cloud cover: 0.0
# Plot the quicklook
prod.plot()
/opt/conda/lib/python3.10/site-packages/rasterio/__init__.py:277: NotGeoreferencedWarning: Dataset has no geotransform, gcps, or rpcs. The identity matrix will be returned.
  dataset = DatasetReader(path, driver=driver, sharing=sharing, **kwargs)
/opt/conda/lib/python3.10/site-packages/rioxarray/_io.py:924: NotGeoreferencedWarning: Dataset has no geotransform, gcps, or rpcs. The identity matrix will be returned.
  warnings.warn(str(rio_warning.message), type(rio_warning.message))  # type: ignore
../_images/72e56e69f5ca1f61a73fc324d356a6619d606beacc4e3610c0ba93c8e4aee9b8.png
# Get the bands information
prod.bands
eoreader.SpectralBand 'RED'
Attributes:
	id: 1
	eoreader_name: RED
	common_name: red
	gsd (m): 0.5
	asset_role: reflectance
	Center wavelength (nm): 650.0
	Bandwidth (nm): 120.0
eoreader.SpectralBand 'GREEN'
Attributes:
	id: 2
	eoreader_name: GREEN
	common_name: green
	gsd (m): 0.5
	asset_role: reflectance
	Center wavelength (nm): 560.0
	Bandwidth (nm): 120.0
eoreader.SpectralBand 'BLUE'
Attributes:
	id: 3
	eoreader_name: BLUE
	common_name: blue
	gsd (m): 0.5
	asset_role: reflectance
	Center wavelength (nm): 495.0
	Bandwidth (nm): 70.0
eoreader.SpectralBand 'NIR'
Attributes:
	id: 4
	eoreader_name: NIR
	common_name: nir
	gsd (m): 0.5
	asset_role: reflectance
	Center wavelength (nm): 840.0
	Bandwidth (nm): 200.0
eoreader.SpectralBand 'NIR'
Attributes:
	id: 4
	eoreader_name: NIR
	common_name: nir
	gsd (m): 0.5
	asset_role: reflectance
	Center wavelength (nm): 840.0
	Bandwidth (nm): 200.0
print(f"Acquisition datetime: {prod.datetime}")
print(f"Condensed name: {prod.condensed_name}")
Acquisition datetime: 2020-05-11 02:31:58
Condensed name: 20200511T023158_PLD_ORT_PMS_5547047101
# Open here some more interesting geographical data: extent and footprint
extent = prod.extent()
footprint = prod.footprint()

base = extent.plot(color='cyan', edgecolor='black')
footprint.plot(ax=base, color='blue', edgecolor='black', alpha=0.5)
<AxesSubplot: >
../_images/feb7ed81bcec80ba5b0b0218ece2b040dac4c6823974c619c3a4c770edf3454f.png

Here, if you want to orthorectify or pansharpen your data manually, you can set your stack here.

prod.ortho_stack = "/path/to/ortho_stack.tif"

If you do not provide this stack, but you give a non-orthorectified product to EOReader (i.e. SEN or PRJ products for Pleiades), you must provide a DEM to orthorectify correctly the data.

⚠⚠⚠ DIMAP SEN products are orthorectified using RPCs and not the rigorous sensor model. A shift of several meters can occur. Please refer to this issue.

Load some bands#

# Select the bands you want to load
bands = [GREEN, NDVI, TIR_1, CLOUDS, HILLSHADE]

# Be sure they exist for Pleiades sensor
ok_bands = [band for band in bands if prod.has_band(band)]
print(to_str(ok_bands))  # Pleiades doesn't provide TIR and SHADOWS bands
['GREEN', 'NDVI', 'CLOUDS', 'HILLSHADE']
# Load those bands as a dict of xarray.DataArray
band_dict = prod.load(ok_bands)
band_dict[GREEN]
Reprojecting band NIR to UTM with a 0.5 m resolution.
Reprojecting band RED to UTM with a 0.5 m resolution.
Reprojecting band GREEN to UTM with a 0.5 m resolution.
<xarray.DataArray 'GREEN' (band: 1, y: 18124, x: 16754)>
array([[[nan, nan, nan, ..., nan, nan, nan],
        [nan, nan, nan, ..., nan, nan, nan],
        [nan, nan, nan, ..., nan, nan, nan],
        ...,
        [nan, nan, nan, ..., nan, nan, nan],
        [nan, nan, nan, ..., nan, nan, nan],
        [nan, nan, nan, ..., nan, nan, nan]]], dtype=float32)
Coordinates:
  * band         (band) int64 1
  * x            (x) float64 7.024e+05 7.024e+05 ... 7.108e+05 7.108e+05
  * y            (y) float64 9.689e+06 9.689e+06 9.689e+06 ... 9.68e+06 9.68e+06
    spatial_ref  int64 0
Attributes:
    cleaning_method:   nodata
    long_name:         GREEN
    constellation:     Pleiades
    constellation_id:  PLD
    product_path:      /home/data/DATA/PRODS/PLEIADES/5547047101/IMG_PHR1A_PM...
    product_name:      PHR1A_PMS_202005110231585_ORT_5547047101
    product_filename:  IMG_PHR1A_PMS_001
    instrument:        HiRI
    product_type:      Ortho Single Image
    acquisition_date:  20200511T023158
    condensed_name:    20200511T023158_PLD_ORT_PMS_5547047101
    orbit_direction:   DESCENDING
    radiometry:        reflectance
    cloud_cover:       0.0
# The nan corresponds to the nodata you see on the footprint
# Plot a subsampled version
band_dict[GREEN][:, ::10, ::10].plot()
<matplotlib.collections.QuadMesh at 0x7f2813e41ed0>
../_images/4d16b827b1d0fa9c511f718a783e060bfb86653925178208b973c420da68b22c.png
# Plot a subsampled version
band_dict[NDVI][:, ::10, ::10].plot()
<matplotlib.collections.QuadMesh at 0x7f2810177d00>
../_images/a89507ea7d372c70e13f467f3620a8106d79d469af1d9850ec3e339f293a9bb8.png
# Plot a subsampled version
band_dict[CLOUDS][:, ::10, ::10].plot()
<matplotlib.collections.QuadMesh at 0x7f281007d450>
../_images/347127ad59548639c68788413be57dec253cf38998440f180717e0a3afbd48b7.png
# Plot a subsampled version
band_dict[HILLSHADE][:, ::10, ::10].plot()
<matplotlib.collections.QuadMesh at 0x7f28100f1bd0>
../_images/a904822a474a46bc36a043489f23ccc75367301a5c75a5005750a7f8d6be955a.png

Stack some bands#

# You can also stack those bands
stack = prod.stack(ok_bands)
# Plot a subsampled version
import matplotlib.pyplot as plt

nrows = len(stack)
fig, axes = plt.subplots(nrows=nrows, figsize=(2 * nrows, 6 * nrows), subplot_kw={"box_aspect": 1})

for i in range(nrows):
    stack[i, ::10, ::10].plot(x="x", y="y", ax=axes[i])
../_images/9e30e803cad62c15cd72f2732b4ec2a5281aa4c408ce9af317c3c359b224f0bc.png