Channel Attention Mixture Density Network (CA-MDN) Hyperspectral Retrievals of Chlorophyll-a Using EMIT

This repo contains files for the paper  "Global High-Resolution Hyperspectral Retrievals of Chlorophyll-a Using EMIT", by Li et al., submitted to Geophysical Research Letters.

This repository provides source code that is updated to include the newly added Channel Attention Mixture Density Network (CA-MDN) model for retrieving chlorophyll-a concentrations from hyperspectral sensors, such asNASA's Earth Surface Mineral Dust Source Investigation (EMIT).  Users can process EMIT .nc files, estimate chlorophyll-a using pre-trained models, and run workflows via Python scripts or command line. The package is modular, allowing replacement of the core MDN with advanced variants. Full usage examples, Jupyter notebooks, and installation instructions are provided for seamless integration into remote sensing workflows.
In addition, the GLObal Reflectance community dataset for Imaging and optical sensing of Aquatic environments (GLORIA) (https://www.nature.com/articles/s41597-023-01973-y) chlorophyll-a data used in the model training and testing is provided as CSV files based on water types: