# Nonlocal Gravity Wave Parameterization Training Data from ERA5

## Facts

- Provider: Aman Gupta et al., NASA-IBM Prithvi WxC team
- License: CC-BY-4.0 (https://forum.ecmwf.int/t/cc-by-licence-to-replace-licence-to-use-copernicus-products-on-02-july-2025/13464)
- Size: 4 files
- Format: PNG, NetCDF (in sampled files)
- Stored at: Hugging Face, https://huggingface.co/datasets/ibm-nasa-geospatial/gravity-wave-parameterization
- Access: public, readable without an account
- Category: Climate & Weather
- Source: https://arxiv.org/abs/2409.13598
- Page: https://scigantic.com/archive/08dfbd11-e192-5562-aa9d-84229964dde7
- Keywords: Climate & Weather, weather forecasting, PNG, NetCDF, Hugging Face

## About

Input and output training pairs for learning a nonlocal parameterization of atmospheric gravity waves, computed from ECMWF ERA5 reanalysis and used to fine-tune the Prithvi WxC weather and climate foundation model. The publishers computed the pairs for 2010, 2012, 2014 and 2015; this repository provides one month from the validation set for testing. Data are on a 64 latitude by 128 longitude grid at hourly steps. The input has 491 channels: latitude, longitude, surface elevation, and zonal wind u, meridional wind v, temperature T and pressure P on 122 vertical levels each. The output has 366 channels: potential temperature theta and the zonal and meridional fluxes of vertical momentum (u'omega' and v'omega') on 122 levels. Listing the repository (measured 2026-11-13) shows a single 20.9 GB netCDF file (wxc_input_u_v_t_p_output_theta_uw_vw_era5_training_data_hourly_2015_constant_mu_sigma_scaling05.nc) whose attributes record the scaling recipe per variable. Open it with xarray or netCDF4 (ncdump -h shows the attributes). ERA5 is CC-BY-4.0 and that licence governs this derived data; credit the Copernicus Climate Change Service (C3S) and ECMWF.

## Load the data

```python
from huggingface_hub import hf_hub_download
path = hf_hub_download(repo_id="ibm-nasa-geospatial/gravity-wave-parameterization", filename="prithvi_schematic.png", repo_type="dataset")
print(path)
```

## Data structure

File types in sampled files: PNG, NetCDF

Example file: `prithvi_schematic.png`

## Open it

- In a hosted notebook with the data mounted: https://scigantic.com/archive/08dfbd11-e192-5562-aa9d-84229964dde7
- Full record as JSON (schema card included): https://api.scigantic.com/api/archives/08dfbd11-e192-5562-aa9d-84229964dde7
- Over MCP: `get_schema_card` with id `08dfbd11-e192-5562-aa9d-84229964dde7` at https://api.scigantic.com/mcp

## Related topics

- [Climate & Weather datasets](https://scigantic.com/topics/climate-and-weather)
- [PNG datasets](https://scigantic.com/topics/png)
- [NetCDF datasets](https://scigantic.com/topics/netcdf)
- [Weather forecasting datasets](https://scigantic.com/topics/weather-forecasting)
