Morro
Forecasting droughts and cyclones, then comparing simulated cloud-seeding interventions.

The project
Morro pairs ten-day weather forecasts with a dashboard for comparing possible interventions. It uses GraphCast and NVIDIA Earth2Studio, with classifiers that flag drought severity and cyclone formation.
My contribution
I worked on the physics-based perturbation pipeline and trained the Random Forest classifiers used to detect drought severity and cyclone formation. The team built Morro at TreeHacks 2026.
Context
Winner of a Stanford Ecopreneurship Sustainability award at TreeHacks 2026. Built with Tej Kosaraju and Kieran P.
This is a hackathon simulation. Changes in a forecast are not evidence that an intervention will work in the real atmosphere.