AI Weather Analytics with Earth-2
This blueprint is a starter application for weather analysis and forecasting that’s meant to feel like a real product: a UI that can visualize multiple layers of geospatial data and a backend that can plug into AI forecasting models. On the NVIDIA Build page it’s tagged around Earth-2 and weather simulation, and it highlights FourCastNet as the associated NIM.
The interesting angle here isn’t “a model that predicts weather” (that’s a long-running research area) — it’s the workflow glue: getting forecast outputs, uncertainty layers, and derived analytics into a form that domain users can explore. If you’re building dashboards for climate risk, energy planning, agriculture, or logistics, the hard part is usually turning model output tensors into consistent map layers and time-series views.
What to try first: skim the prerequisites on the blueprint page, then run the blueprint end-to-end once using the default setup. After that, swap in your own data source for one layer at a time (for example: historical observations, radar tiles, or a custom region). The page also links a GitHub repo, which is a good starting point if you want to treat the blueprint as code rather than a click-through demo.
Source listing: https://build.nvidia.com/blueprints?filters=publisher%3Anvidia