3D Cloud Reconstruction in Multi-Angle Polarimetry with a Foundation Model

Using AI/ML for remote sensing analysis is challenged by the expense of training a neural network (NN) and adapting it to be useful for various measurement types. Foundation models address this by creating a universal NN, only requiring training of a downstream (adaptation) model. Building upon a previously funded IRAD effort which created a MODIS all-sky radiance based foundation model (led by J. Gong, 613), we propose to test if it can benefit multi-angle polarimeter observations as well. A successful result will position GSFC to lead in AI/ML applications in the Earth Sciences, thus aligning with Strategic Focus Area (SFA) priorities. We will leverage our expertise with the recently launched PACE/HARP2 multi-angle polarimeter, and as a side benefit will make a 3D cloud structure product.

Foundation models are highly adaptable and can be applied to multiple datasets. AI/ML techniques are rapidly evolving, so this early-stage innovation IRAD will strengthen GSFC’s dominance in the use of AI/ML for remote sensing of the atmosphere and provide for ROI in proposals to NASA ESTO’s Advanced Information Systems Technology (AIST) Program, specifically the Analytic Collaborative Framework (ACF), which seeks to fully utilize diverse datasets (among other programs). A new 3D cloud mask product will be generated for PACE, and in the longer term, AOS, both GSFC managed missions. Because of adaptability, the proposed work will also position GSFC scientists to better compete in a variety of mission or topic specific ROSES solicitations.

This is early/mid-stage (TRL 3) — the most realistic path in is NASA SBIR/STTR, which funds small businesses and research institutions to develop technology aligned with NASA's needs (equity-free, phased funding). Check whether a current SBIR/STTR solicitation topic overlaps with this project's technology area, or contact the project directly (above) to ask.

None of these are guaranteed paths for this specific project — TechPort itself doesn't have an "apply" button. Reaching out to the contact(s) above with a specific question is usually the fastest way to find out what's actually open.