Python module for applying sun glint corrections to optical reflectance image data. The following three commonly used sunglint correction algorithms are supported:
- Cox and Munk (1954) statistical/geometry approach;
- Hedley et al. (2005) correlation approach, and;
- NIR/SWIR subtraction approach (e.g. Dierssen et al., 2015)
Please go through the usage examples (below) as they contain handy hints
Notes
- This module has been tested on image data from Landsat-5, -7 and -8, and Sentinel-2A/B. Further testing is required on WV2 data.
- Cox and Munk approach requires an input wind speed (not yet automated)
- The Hedley approach requires a user-selected region of interest. This module creates an interactive matplotlib figure that a user can manipulate (add/delete vertices) to create the desiged polygon over the region of interest.
Comments on deglinting S2A/B using Hedley et al. (2005) and NIR subtraction:
- This module can use band 8 (10 m) to deglint bands 1, 2, 3, 4, 5 and 6, that have varying spatial resolutions. In this case band 8 will be downscaled to 60 m (for band 1) or 20 m (for bands 5 and 6). Here, rasterio is used for the resampling (bilinear interpolation). However, artefacts in the in the deglinted bands 5, 6 have been found when band 8 (10 m) is used rather band 7 or 8A (20 m) as the NIR band. Further work will be required to investigate the best practice of downscaling higher resolution bands.
References
-
Cox, C., Munk, W. 1954. Statistics of the Sea Surface Derived from Sun Glitter. J. Mar. Res., 13, 198-227.
-
Cox, C., Munk, W. 1954. Measurement of the Roughness of the Sea Surface from Photographs of the Suns Glitter. J. Opt. Soc. Am., 44, 838-850.
-
Dierssen, H.M., Chlus, A., Russell, B. 2015. Hyperspectral discrimination of floating mats of seagrass wrack and the macroalgae Sargassum in coastal waters of Greater Florida Bay using airborne remote sensing. Remote Sens. Environ., 167(15), 247-258, doi: https://doi.org/10.1016/j.rse.2015.01.027
-
Hedley, J. D., Harborne, A. R., Mumby, P. J. (2005). Simple and robust removal of sun glint for mapping shallow-water benthos. Int. J. Remote Sens., 26(10), 2107-2112.
Log into NCI and clone this repository into a dir/workspace in you NCI home dir:
cd ~/<your project dir>
git clone git@github.com:GeoscienceAustralia/sun-glint-correction.git
cd sun-glint-correction
git checkout -b develop
# set up a local Python 3.6 runtime environment
source configs/sunglint.env # should be error/warning free
export CUSTOM_PY_INSTALL=~/.digitalearthau/dea-env/20200713/local
mkdir -p $CUSTOM_PY_INSTALL/lib/python3.6/site-packages/
python setup.py install --prefix=$CUSTOM_PY_INSTALLpython setup.py install --prefix=<install-dir>
Here, --prefix is optional.
Linux
- Sentinel-2A/B
- Landsat-5 TM
- Landsat-7 ETM
- Landsat-8 OLI
- Worldview-2
Perform the various sunglint corrections on Sentinel-2B data. This code works on S2A and S2B. First source configs/sunglint.env
#!/usr/bin/env python3
import os
import sys
import numpy as np
import matplotlib.pyplot as plt
from os.path import join as pjoin
import datacube
from sungc import deglint
dc = datacube.Datacube()
# until wagl-water-atcor products become available,
# test with nbart products.
product = "nbart"
dc_name = "s2b_ard_granule"
# Rottnest - Perth, WA
avail_datasets = dc.find_datasets(
product=dc_name,
lon=(115.40, 115.80),
lat=(-31.60, -32.40),
time=("2019-01-31", "2019-02-02")
)
working_dir = "/path/to/somewhere/"
def main():
# for this test select the first dataset
ds = avail_datasets[0]
g = deglint.GlintCorr(ds, product)
# Create output directories
# sub_folderName = SYYYY_MM_DD
overpass_dt = g.overpass_datetime.strftime("%Y-%m-%d %H:%M:%S.%f")
sub_folderName = (
g.sensor[0].upper()
+ "_".join(overpass_dt.split()[0].split("-"))
)
mnir_dir = pjoin(working_dir, "MINUS_NIR", dc_name, sub_folderName)
hedley_dir = pjoin(working_dir, "HEDLEY", dc_name, sub_folderName)
cox_munk_dir = pjoin(working_dir, "COX_MUNK", dc_name, sub_folderName)
# make output directories
os.makedirs(mnir_dir, exist_ok=True)
os.makedirs(hedley_dir, exist_ok=True)
os.makedirs(cox_munk_dir, exist_ok=True)
# ---------------------- #
# Cox and Munk (1954) #
# ---------------------- #
# deglint Bands 1, 2, 3, 4, 5 and 6 with Cox and Munk (1954).
# you need to provide a wind-speed (m/s), here its been
# set to 5 m/s
cm_bandList = g.cox_munk(
vis_band_ids=["1", "2", "3", "4", "5", "6"],
odir=cox_munk_dir,
wind_speed=5,
)
# the deglinted bands are saved as geotifs in cox_munk_dir
# see cm_bandList
# ---------------------- #
# Hedley et al. (2019) #
# ---------------------- #
# deglint Bands 2, 3 and 4 using Hedley et al. (2005)
# 1) create shapefile with an interactive matplotlib figure
shp_file = pjoin(hedley_dir, g.obase_shp)
if not os.path.exists(shp_file):
g.create_roi_shp(shp_file=shp_file, dwnscaling_factor=10)
# 2) deglint the 10 m bands using band 8, and plot correlations
deglint_BList_10m = g.hedley_2005(
vis_band_ids=["2", "3", "4"],
nir_band_id="8",
odir=hedley_dir,
roi_shpfile=shp_file,
plot=True,
)
# the deglinted bands are saved as geotifs in hedley_dir,
# see deglint_BList_10m. The correlation plots are also
# saved in hedley_dir
# 3) deglint the 20- and 60- m bands using band 8A (20 m)
deglint_BList_other = g.hedley_2005(
vis_band_ids=["1", "5", "6"],
nir_band_id="8A",
odir=hedley_dir,
roi_shpfile=shp_file,
plot=True,
)
# In Hedley et al. (2019), band 9 was used to deglint band 1
# However, band 9 isn't currently distruibuted because it's
# affected by water vapour. As such band 8A (20 m), in this
# example, is downscaled to 60 m and used to deglint Band 1.
# ---------------------- #
# NIR SUBTRACTION #
# ---------------------- #
# deglint Bands 2, 3, and 4 by subtracting Band 8
mnir_bandList_10m = g.nir_subtraction(
vis_band_ids=["2", "3", "4"], nir_band_id="8", odir=mnir_dir,
)
# the deglinted bands are saved as geotifs in mnir_dir,
# see mnir_bandList_10m
# deglint bands 1, 5 and 6 using band 8A
mnir_bandList_other = g.nir_subtraction(
vis_band_ids=["1", "5", "6"], nir_band_id="8A", odir=mnir_dir,
)
if __name__ == "__main__":
main()Perform the various sunglint corrections on Landsat-8 OLI data. First source configs/sunglint.env
#!/usr/bin/env python3
import os
import sys
import numpy as np
import matplotlib.pyplot as plt
from os.path import join as pjoin
import datacube
from sungc import deglint
dc = datacube.Datacube()
# until wagl-water-atcor products become available,
# test with nbart products.
product = "nbart"
dc_name = "ga_ls8c_ard_3"
# Rottnest - Perth, WA
avail_datasets = dc.find_datasets(
product=dc_name,
lon=(115.40, 115.80),
lat=(-31.60, -32.40),
time=("2020-02-02", "2020-02-04")
)
working_dir = "/path/to/somewhere/"
def main():
# for this test select the first dataset
ds = avail_datasets[0]
g = deglint.GlintCorr(ds, product)
# Create output directories
# sub_folderName = LYYYY_MM_DD
overpass_dt = g.overpass_datetime.strftime("%Y-%m-%d %H:%M:%S.%f")
sub_folderName = (
g.sensor[0].upper()
+ "_".join(overpass_dt.split()[0].split("-"))
)
mnir_dir = pjoin(working_dir, "MINUS_NIR", dc_name, sub_folderName)
hedley_dir = pjoin(working_dir, "HEDLEY", dc_name, sub_folderName)
cox_munk_dir = pjoin(working_dir, "COX_MUNK", dc_name, sub_folderName)
# make output directories
os.makedirs(mnir_dir, exist_ok=True)
os.makedirs(hedley_dir, exist_ok=True)
os.makedirs(cox_munk_dir, exist_ok=True)
# ---------------------- #
# Cox and Munk (1954) #
# ---------------------- #
# deglint Bands 1, 2, 3 and 4 with Cox and Munk (1954).
# you need to provide a wind-speed (m/s), here its been
# set to 5 m/s
cm_bandList = g.cox_munk(
vis_band_ids=["1", "2", "3", "4"],
odir=cox_munk_dir,
wind_speed=5,
)
# the deglinted bands are saved as geotifs in cox_munk_dir
# see cm_bandList
# ---------------------- #
# Hedley et al. (2019) #
# ---------------------- #
# deglint Bands 1, 2, 3 and 4 using Hedley et al. (2005)
# 1) create shapefile with an interactive matplotlib figure
shp_file = pjoin(hedley_dir, g.obase_shp)
if not os.path.exists(shp_file):
g.create_roi_shp(shp_file=shp_file, dwnscaling_factor=10)
# 2) deglint the 30 m bands using band 6, and plot correlations
deglint_BList = g.hedley_2005(
vis_band_ids=["1", "2", "3", "4"],
nir_band_id="6",
odir=hedley_dir,
roi_shpfile=shp_file,
plot=True,
)
# the deglinted bands are saved as geotifs in hedley_dir,
# see deglint_BList. The correlation plots are also
# saved in hedley_dir
# ---------------------- #
# NIR SUBTRACTION #
# ---------------------- #
# deglint Bands 1, 2, 3, and 4 by subtracting Band 6
mnir_bandList = g.nir_subtraction(
vis_band_ids=["1", "2", "3", "4"], nir_band_id="6", odir=mnir_dir,
)
# the deglinted bands are saved as geotifs in mnir_dir,
# see mnir_bandList
if __name__ == "__main__":
main()A pre-commit config is provided to automatically format and check your code changes. This allows you to immediately catch and fix issues before you raise a failing pull request (which run the same checks under Travis).
If you don't use Conda, install pre-commit from pip:
pip install pre-commit
If you do use Conda, install from conda-forge (required because the pip version uses virtualenvs which are incompatible with Conda's environments)
conda install pre_commit
Now install the pre-commit hook to the current repository:
pre-commit install
Your code will now be formatted and validated before each commit. You can also
invoke it manually by running pre-commit run