ContainersNVIDIADeepSeek-R1NVIDIALlama-3.1-Nemotron-70B-InstructNVIDIAPyTorch
CollectionsNVIDIAOmniverse Kit (FB)NVIDIADeepStream SDKNVIDIAOmniverse Kit App Streaming
ModelsNVIDIAStyleGAN3 pretrained modelsNVIDIAPeopleNetNVIDIATrafficCamNet
ResourcesNVIDIARiva Skills Quick Start
Helm ChartsNVIDIAGPU OperatorNVIDIANVIDIA NIM Operator
NVIDIA
NVIDIA
Earth-2 FourCastNet 3
Model
NVIDIA
NVIDIA
Earth-2 FourCastNet 3

FourCastNet 3 is a probabilistic global weather modeling that uses geometric machine learning.

This model is backed by NVIDIA's Plus Plus (++) Promise
to learn more about the quality of the datasets used to train this model.
FieldResponse
Intended Task/Domain:Weather Forecasting and Simulation
Model Type:Spherical Neural Operator
Intended Users:Weather Forecasters, Researchers, Related Industrial users.
Output:Tensor representing 72 atmospheric variables
Describe how the model works:Input variables are fed through a spherical convolutional architecture to produce the weather state in the next 6 hours.
Name the adversely impacted groups this has been tested to deliver comparable outcomes regardless of:Not Applicable (N/A)
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 1980-2016 training dataset. There is no mechanism to enforce physical consistency for predictions.
Verified to have met prescribed NVIDIA quality standards:Yes
Performance Metrics:Accuracy (ACC), Error (RMSE), and Probabilistic Calibration (CRPS)
Potential Known Risks:This model may incorrectly predict future weather states and phenomenon
Licensing:Apache 2.0 license.