AI4Arctic - snow

Estimation of snow-parameters from sentinel-3 data using deep learning

The goal of the AI4Arctic project were to develop AI/deep learning - based estimation of snow and ice parameters from Sentinel data. The project was funded by ESA. This repository contains code for the snow-part of the project. The code for the ice-part can be found here.

The deep learning model is based on the UNet architecture. The input data is Sentinel-3 data from the SLSTR sensor. The model outputs the following snow-parameters as geotiff files:

• Fractional snow cover (FSC) with values; FSC:0-100, cloud:-1, no data:-2
• Snow grain size (SGS) with (uncalibrated) values; SGS:60-100, cloud:-1, no data:-2
• Snow surface wetness (SSW) classes with values; Dry, cold snow: 0, Dry, moderate cold snow: 1, Dry, warming snow: 2, Moist snow: 3, Moist, warming snow: 4, Wet snow: 4, cloud:-1, no data:-2

Setup

Make sure you are running the code on a computer with python 3 and GDAL installed. Type the following commands in the terminal to setup the repository and python environment:

git clone git@github.com:NorskRegnesentral/ai4artic_snow.git
cd ai4artic_snow
python -m virtualenv env
source env/bin/activate
pip install -r REQUIREMENTS.txt

Save creodias-credentials to a text file to enable data-download

Create an account at creodias.eu (if you dont already have one). Create a new file at the root of this repository named

creodias_credentials.txt

Make two lines in the text file, the first with your username and the second with your password.

Usage

python main.py YYYYMMDD

Where YYYYMMDD is a date for which you desire snow products for. If omitted, the date for yesterday will be used.

• main.py: A script to run the entire snow-pipeline

  1. user specify which date to process
  2. data-downloading
  3. preprocessing (conversion to reflectance)
  4. deep learning prediction
  5. mosaicing and export

Contact

For questions, contact Anders U. Waldeland at anders@nr.no