dlesym-v0-isccp-era5

PhysicsNeMo Checkpoints: DLESyM-v0-ISCCP-ERA5

Description:

DLESyM-V0-ISCCP-ERA5 is a data-driven global earth system model, shown to have stable long-term rollouts producing realistic climate variability. This model includes an atmosphere and ocean component, using atmospheric variables as well as the sea-surface temperature on a HEALPix nside=64 (approximately 1 degree) resolution grid. The model architecture is a U-Net with padding operations modified to support using the HEALPix grid.

This set of checkpoints corresponds to the original model weights presented in Cresswell et al. 2024. Notably, the model takes as input the outgoing longwave radiation (OLR) variable from the International Satellite Cloud Climatology Project (ISCCP) in addition to several ERA5 variables.

The model package also includes a precipitation model that can operate in diagnostic mode on top the core atmospheric model backbone.

For inference see NVIDIA Earth2Studio.

This model is ready for commercial or non-commercial use.

License/Terms of Use:

Governing Terms: Use of this model is governed by the Linux Foundation OpenMDW License Agreement, version 1.1.

Deployment Geography:

Global

Use Case:

Global atmosphere/ocean earth system modeling.

Release Date:

Hugging Face 06/22/2026 via https://huggingface.co/nvidia/dlesym-v0-isccp-era5

Model Architecture

Architecture Type: DLESyM uses two UNet architectures adapted to the HEALPix grid, one for each of the atmosphere and ocean components. <br> Network Architecture: HEALPix UNet <br>

This model provides the following checkpoints:

• Atmosphere model -- 3.5M Parameters
• Precipitation diagnostic model -- 1.5M Parameters
• Ocean model -- 0.77M Parameters

Input:

Input Type(s):

• Tensor (10 surface and pressure-level variables)

Input Format: PyTorch Tensor <br> Input Parameters:

• Six Dimensional (6D) (batch, lead time, variable, face, height, width) <br>

Other Properties Related to Input:

• Input latitude/longitude grid: 0.25 degree 721 x 1440, regridded to HEALPix nside=64 grid in "XY" format with a north origin and clockwise order
  • See HEALPIX_PAD_XY in the earth2grid package for specific details
• Input state weather variables: z500, tau300-700, z1000, t2m, tcwv, t850, z250, ws10m, olr, sst
  • tau300-700 (geopotential thickness) is defined as the difference between z300 and z700 geopotential levels.
  • ws10m (wind speed at 10m above surface) is defined as the square root of the sum of the squared zonal and meridional wind components, i.e. sqrt(u10m **2 + v10m **2).
  • olr is the OLR variable from the ISCCP dataset

For variable name information, review the HRRR Lexicon at Earth2Studio. Review the config.yaml provided in the model package for information on the input lead times required by the model.

Output:

Output Type: Tensor (11 surface and pressure-level variables) <br> Output Format: Pytorch Tensor <br> Output Parameters: Six Dimensional (6D) (batch, lead time, variable, face, height, width) <br> Other Properties Related to Output:

• Output latitude/longitude grid: HEALPix nside=64 grid in "XY" format with a north origin and clockwise order.
  • See HEALPIX_PAD_XY in the earth2grid package for specific details
• Output state weather variables: z500, tau300-700, z1000, t2m, tcwv, t850, z250, ws10m, olr, tp, sst <br>
  • tau300-700 (geopotential thickness) is defined as the difference between z300 and z700 geopotential levels.
  • ws10m (wind speed at 10m above surface) is defined as the square root of the sum of the squared zonal and meridional wind components, i.e. sqrt(u10m **2 + v10m **2).

Our AI models are designed and/or optimized to run on NVIDIA GPU-accelerated systems. By leveraging NVIDIA’s hardware (e.g. GPU cores) and software frameworks (e.g., CUDA libraries), the model achieves faster training and inference times compared to CPU-only solutions.

Software Integration

Runtime Engine(s): Not Applicable <br> Supported Hardware Microarchitecture Compatibility: <br>

• NVIDIA Ampere <br>
• NVIDIA Blackwell <br>
• NVIDIA Hopper <br>

Supported Operating System(s):

• Linux <br>

The integration of foundation and fine-tuned models into AI systems requires additional testing using use-case-specific data to ensure safe and effective deployment. Following the V-model methodology, iterative testing and validation at both unit and system levels are essential to mitigate risks, meet technical and functional requirements, and ensure compliance with safety and ethical standards before deployment.

Model Version(s):

Model Version: v0 <br>

Training, Testing, and Evaluation Datasets:

Training Dataset:

Link: ERA5 <br>

Data Collection Method by dataset: <br>

• Automatic/Sensors <br>

Labeling Method by dataset: <br>

• Automatic/Sensors <br>

Data Modality:

• Gridded geophysical time series

Data Size:

• 170 GB subset used for model training

Properties: ERA5 data for the period January 1983 - December 2016. ERA5 provides hourly estimates of various atmospheric, land, and oceanic climate variables. The data covers the Earth on a 30km grid and resolves the atmosphere at 137 levels. <br>

Link: ISCCP <br>

Data Collection Method by dataset <br>

• Automatic/Sensors <br>

Labeling Method by dataset <br>

• Automatic/Sensors <br>

Data Modality:

• Gridded geophysical time series

Data Size:

• 18 GB subset used for model training

Properties: ISCCP OLR data for the period January 1983 - December 2016. The International Satellite Cloud Climatology Project (ISCCP) calibrates and compiles observations from a large suite of satellites into a single dataset.

Testing Dataset:

Link: ERA5 <br>

Data Collection Method by dataset: <br>

• Automatic/Sensors <br>

Labeling Method by dataset: <br>

• Automatic/Sensors <br>

Properties: ERA5 data for the period January 2016 - December 2017. ERA5 provides hourly estimates of various atmospheric, land, and oceanic climate variables. The data covers the Earth on a 30km grid and resolves the atmosphere at 137 levels. <br>

Link: ISCCP <br>

Data Collection Method by dataset <br>

• Automatic/Sensors <br>

Labeling Method by dataset <br>

• Automatic/Sensors <br>

Data Modality:

• Gridded geophysical time series

Properties: ISCCP OLR data for the period January 2016 - December 2017. The International Satellite Cloud Climatology Project (ISCCP) calibrates and compiles observations from a large suite of satellites into a single dataset.

Evaluation Dataset:

Link: ERA5 <br>

Data Collection Method by dataset: <br>

• Automatic/Sensors <br>

Labeling Method by dataset: <br>

• Automatic/Sensors <br>

Properties: ERA5 data for the period January 2017 onwards. ERA5 provides hourly estimates of various atmospheric, land, and oceanic climate variables. The data covers the Earth on a 30km grid and resolves the atmosphere at 137 levels. <br>

Link: ISCCP <br>

Data Collection Method by dataset <br>

• Automatic/Sensors <br>

Labeling Method by dataset <br>

• Automatic/Sensors <br>

Data Modality:

• Gridded geophysical time series

Properties: ISCCP OLR data for the period January 2017 onwards. The International Satellite Cloud Climatology Project (ISCCP) calibrates and compiles observations from a large suite of satellites into a single dataset.

Inference:

Engine: PyTorch <br> Test Hardware:

• NVIDIA Ampere (A100) <br>
• NVIDIA Hopper (H100) <br>
• NVIDIA Ada Lovelace (L40S) <br>

Ethical Considerations:

NVIDIA believes Trustworthy AI is a shared responsibility and we have established policies and practices to enable development for a wide array of AI applications. Developers should work with their internal model team to ensure this model meets requirements for the relevant industry and use case and addresses unforeseen product misuse. For more detailed information on ethical considerations for this model, please see the Model Card++ Bias, Explainability, Safety & Security, and Privacy Subcards.

Please report model quality, risk, security vulnerabilities or NVIDIA AI Concerns here.