| Field | Response |
|---|---|
| Intended Task/Domain: | Global Global Weather Data Assimilation |
| Model Type: | Vision Transformer |
| Intended User: | Weather and Climate ML-based researchers / developers implementing on global data assimilation pipelines. |
| Output: | Global variables: u10m, v10m, u100m, v100m, t2m, msl, tcwv, sst, sic, u50, u100, u150, u200, u250, u300, u400, u500, u600, u700, u850, u925, u1000, v50, v100, v150, v200, v250, v300, v400, v500, v600, v700, v850, v925, v1000, z50, z100, z150, z200, z250, z300, z400, z500, z600, z700, z850, z925, z1000, t50, t100, t150, t200, t250, t300, t400, t500, t600, t700, t850, t925, t1000, q50, q100, q150, q200, q250, q300, q400, q500, q600, q700, q850, q925, q1000. |
| Describe how the model works: | Sensor-specific point cloud embedders perform scatter-reduce aggregation onto the HPX64 grid, and a ViT backbone then refines the aggregated features into a global output field. |
| Name the adversely impacted groups this has been tested to deliver comparable outcomes regardless of: | Not Applicable |
| Technical Limitations & Mitigation: | The model may perform poorly for systems that are not similar to those in the training data, namely for rare weather phenomena or weather behavior outside of the 2000–2021 training dataset. There is no mechanism to enforce physical consistency for predictions. |
| Verified to have met prescribed NVIDIA quality standards: | Yes |
| Performance Metrics: | Accuracy, Throughput and Latency |
| Potential Known Risks: | This model may incorrectly predict weather states and phenomenon |
| Licensing: | NVIDIA Open Model License. |