This code contains everything necessary to build neural networks for retrieving atmospheric parameters such as temperature and humidity profiles or planetary boundary layer height.
Submitted to IEEE TRANSACTIONS ON GEOSCIENCE AND REMOTE SENSING: Improved Planetary Boundary Layer Sounding Using Hyperspectral Microwave and Backscatter Lidar Data Fusion
Features include:
- Comprehensive and expandable data specifications system for defining model variations.
- Deep integration with PyTorch and PyTorch Lightning for training and serialization with checkpoints.
- Productive metrics system for deep analysis of model outputs and data.
- much more!
The primary module that is built with this project is located in the src directory.
This project uses uv to manage project dependencies. Installation instructions can be found here.
Use uv sync to install all dependencies into a new virtual environment.
Note: uv is not python dependent and thus can create virtual environments on it's own.