BIOMASS AI(NPO-53972-1)
environmental science earth air space exoplanet
BIOMASS AI
(NPO-53972-1)
Overview
The BIOMASS-AI software provides an extensible machine learning framework for analyzing harmonized multi-mission data, such as from ESA's BIOMASS and NASA's NISAR missions. The software addresses the challenge of applying machine learning to multi-mission SAR observations by implementing a common processing infrastructure that geocodes data from both missions onto a unified spatial reference system, creating standardized, cloud-optimized products suitable for ML analysis. The software implements an end-to-end research pipeline that harmonizes SAR data products, applies trained deep learning models to automatically extract surface features, and produces cloud-optimized geospatial outputs for distribution and analysis. The modular architecture allows researchers to incorporate additional sensors (such as Sentinel-1), or train and deploy models for different science applications using the same SAR harmonization infrastructure. The initial implementation demonstrates the framework with crevasse detection models for ice-covered regions using BIOMASS and NISAR data, but the software is designed to support a variety of remote sensing science applications. The software is designed for open-source release as a contribution to open science, providing the research community with transparent, reproducible tools for multi-mission SAR analysis using freely available data. All outputs are open data products for scientific research.
Software Details
Category
Environmental Science (Earth, Air, Space, Exoplanet)
Reference Number
NPO-53972-1
Release Type
Open Source
Operating System



