snowlaps is a Python package built on a deep learning emulator of
the two-stream radiative transfer model
biosnicar v2.1. It was originally
developed to study the impact of different Light Absorbing Particles
(LAPs) on snow spectral albedo as part of a research
project in Southern
Norway. Now, snowlaps has grown into a library with two main use
cases.
- forward mode: predict snow spectral albedo from prescribed surface properties as a fast alternative to biosnicar.
- inverse mode: infer surface properties from snow spectral albedo observations.
snowlaps can be installed via the command line with
conda and
pip:
# clone repository in the folder of your choice
git clone [email protected]:openosmia/snowlaps.git
# move into snowlaps directory
cd snowlaps
# create conda environment
conda env create -f environment.yml
# activate conda environment
conda activate snowlaps
# install snowlaps
pip install -e .Installation can be sped up using the fast cross-platform package
manager mamba
(reimplementation of the conda package manager in C++), simply use
mamba instead of conda in the instructions above.
snowlaps can be used directly as a Python package or interactively
via a Streamlit app.
Example scripts are provided in
snowlaps/examples.
- example 1: forward run of the snowlaps emulator
- example 2: inversion of hyperspectral albedo measurements
- example 3: comparison of snowlaps and biosnicar predictions
The Streamlit app can be run locally via the terminal:
# move into snowlaps directory
cd snowlaps
# start Streamlit app on http://localhost:8501
./start_app.sh
Any contribution to snowlaps is welcome! Feel free to add new issues, open pull requests or ask questions in the discussion forum.
If you use this code, please cite the associated publication:
Chevrollier, L.-A., Wehrlé, A., Cook, J. M., Pirk, N., Benning, L. G., Anesio, A. M., and Tranter, M.: Separating the albedo reducing effect of different light absorbing particles on snow using deep learning, EGUsphere [preprint], https://doi.org/10.5194/egusphere-2024-2583, 2024.