A computationally efficient, open source feature tracking algorithm, called ORB, is adopted and tuned for retrieval of the first guess sea ice drift from Sentinel-1 SAR images. Pattern matching algorithm based on MCC calculation is used further to retrieve sea ice drift on a regular grid.
- Korosov A.A. and Rampal P., A Combination of Feature Tracking and Pattern Matching with Optimal Parametrization for Sea Ice Drift Retrieval from SAR Data, Remote Sens. 2017, 9(3), 258; doi:10.3390/rs9030258
- Muckenhuber S., Korosov A.A., and Sandven S., Open-source feature-tracking algorithm for sea ice drift retrieval from Sentinel-1 SAR imagery, The Cryosphere, 10, 913-925, doi:10.5194/tc-10-913-2016, 2016
- Nansat - scientist friendly open-source Python toolbox for processing 2D satellite earth observation data)
- OpenCV - open-source computer vision
# Install Python, OpenCV and other requirements using miniconda (http://conda.pydata.org/miniconda.html):
conda create -q --yes -n py3drift -c conda-forge python=3.6 numpy scipy matplotlib netcdf4 gdal opencv nose
# Activate the conda environment with installed requirements:
source activate py3drift
# Use pip to install from the repo:
pip install https://github.com/nansencenter/sea_ice_drift/archive/v0.7.tar.gz
# download example datasets
wget ftp://ftp.nersc.no/pub/nansat/test_data/generic/S1A_EW_GRDM_1SDH_20161005T142446_20161005T142546_013356_0154D8_C3EC.SAFE.tif
wget ftp://ftp.nersc.no/pub/nansat/test_data/generic/S1B_EW_GRDM_1SDH_20161005T101835_20161005T101935_002370_004016_FBF1.SAFE.tif
# start Python and import relevant libraries
import matplotlib.pyplot as plt
from nansat import Nansat
from sea_ice_drift import SeaIceDrift
# open pair of satellite images using Nansat and SeaIceDrift
filename1='S1B_EW_GRDM_1SDH_20161005T101835_20161005T101935_002370_004016_FBF1'
filename2='S1A_EW_GRDM_1SDH_20161005T142446_20161005T142546_013356_0154D8_C3EC'
sid = SeaIceDrift(filename1, filename2)
# run ice drift retrieval using Feature Tracking
uft, vft, lon1ft, lat1ft, lon2ft, lat2ft = sid.get_drift_FT()
# plot
plt.quiver(lon1ft, lat1ft, uft, vft);plt.show()
# define a grid (e.g. regular)
lon1pm, lat1pm = np.meshgrid(np.linspace(-3, 2, 50),
np.linspace(86.4, 86.8, 50))
# run ice drift retrieval for regular points using Pattern Matching
# use results from the Feature Tracking as the first guess
upm, vpm, rpm, apm, lon2pm, lat2pm = sid.get_drift_PM(
lon1pm, lat1pm,
lon1ft, lat1ft,
lon2ft, lat2ft)
# select high quality data only
gpi = rpm > 0.4
# plot high quality data on a regular grid
plt.quiver(lon1pm[gpi], lat1pm[gpi], upm[gpi], vpm[gpi], rpm[gpi])
Full example here
