PAZ-unsupervised

Cryosphere PAZ satellite MAchine Learning

Abstract: Satellite imagery plays a pivotal role in unraveling the complexities of Earth's landscapes, and our project is dedicated to curating a comprehensive reference dataset sourced from the PAZ satellite's X-band images over the Massif du Mont Blanc region. The advantages lie in the detailed spatial resolution of the X-band, enabling nuanced analysis, while considerations involve addressing atmospheric conditions for accurate interpretation. Leveraging the unique perspective offered by PAZ's X-band, our project aims to create an extensively annotated dataset, fostering advancements in environmental monitoring and climate studies. We focused on cryospheric surfaces (encompassing ablation zones, accumulation zones, ice aprons, and hanging glaciers) the dataset also captures conventional land cover types, including urban areas, forests, plains, and rocky terrain. Through baseline classification experiments, employing K-nearest neighbors (KNN) for supervised learning, we seek to provide insights into the intricacies of land cover classification in this high-resolution satellite imagery. In a nutshell, our project simplifies access to diverse satellite data, shedding light on the unique features of the Massif du Mont Blanc region, from icy terrains to familiar landscapes, providing a useful tool for a further range of studies and applications.

Keywords: SAR X/C bands Snow/Ice backscatter SAR Glaciers Cryospheric zones classification Glaciers Cryospheric zones classification

Dataset: https://zenodo.org/records/10401360

Citation:

@inproceedings{linkwongchon:hal-04628785,
  TITLE = {{Supervised Classification For Analysis Of Cryospheric Zones Using SAR Statistical Timeseries}},
  AUTHOR = {Lin-Kwong-Chon, Christophe and Gallet, Matthieu and Kaushik, Suvrat and Trouv{\'e}, Emmanuel},
  URL = {https://hal.science/hal-04628785},
  BOOKTITLE = {{International Geoscience and Remote Sensing Symposium (IGARSS 2023)}},
  ADDRESS = {Athens, Greece},
  YEAR = {2024},
  MONTH = Jul,
}

Libraries

(back to top)

Start

Here are some sample instructions for setting up your project locally. To set up a local copy and get it running, follow these simple steps:

Requirements

• Create and load a virtual environment

sudo apt install python3-virtualenv
python3 -m venv venv
source ./venv/bin/activate
#'deactivate' to close the virtual environment

• Activate the env

source ./[path_to_venv]/bin/activate # on linux based platform
.\[path_to_venv]\Script\activate.ps1 # on windows based platform

• Install requirements.txt

pip install -r requirements.txt

Installation

Here's an example of how to install and configure the package.

1. Clone the repo

   git clone https://github.com/Matthieu-Gallet/PAZ-unsupervised.git

3. Navigate to the project directory and start the package.

   cd PAZ-unsupervised
   python code/create_dataset.py # Unit tests in progress ...

(back to top)

Contact

• Suvrat Kaushik - suvrat.kaushik@univ-smb.fr
• Christophe Lin-Kwong-Chon - christophe.lin-kwong-chon@univ-smb.fr
• Matthieu Gallet - matthieu.gallet@univ-smb.fr
• Emmanuel Trouvé - emmanuel.trouve@univ-smb.fr

Acknowledgments

The authors would like to thank the Spanish Instituto Nacional de Tecnica Aerospacial (INTA) for the PAZ images (Project AO-001-051). The work was conducted as part of the SHARE CNES/PNTS project.

(back to top)