Feature Request: Add ERA5 Pressure-Level and Model-Level Data in Cloud-Optimized Zarr Format

[Mohammad Mohseni Aref (mohseniaref)]

Hi Planetary Computer team,

Thanks for making ERA5 available through Microsoft Planetary Computer.

I have been exploring the ERA5 collection and noticed that it currently provides mainly single-level atmospheric variables.

Would it be possible to extend this collection to include ERA5 pressure-level and model-level data?

Why this would be useful

I work on InSAR time-series analysis and atmospheric delay correction at the University of Potsdam. For these applications, we need vertical profiles of atmospheric temperature, humidity, pressure, and geopotential to calculate refractivity and integrate atmospheric delays along the radar line of sight.

At present, obtaining these data through the Copernicus Climate Data Store (CDS) can be quite slow, particularly when processing hundreds of satellite acquisition dates. Downloading and managing large GRIB/NetCDF files also adds considerable overhead.

I suspect other researchers working on GNSS tropospheric delays, atmospheric modelling, hydrology, and climate applications face similar difficulties.

Requested additions

ERA5 pressure levels (37 levels): Temperature (t), specific humidity (q), and geopotential (z).

ERA5 model levels (137 levels): Temperature, specific humidity, logarithm of surface pressure (lnsp), surface geopotential, and the hybrid-level coefficients needed to reconstruct pressure.

Hourly temporal resolution: To match satellite acquisition times.

Efficient spatial and temporal subsetting: Access to selected geographic regions, atmospheric levels, and acquisition dates without retrieving entire global datasets.

Cloud-optimized Zarr: Direct access through xarray, dask, and zarr, with appropriate chunking for regional analysis and time-series processing.

A possible first step

If hosting all 137 model levels is too demanding in terms of storage and maintenance, providing the 37 ERA5 pressure levels would already be a major improvement.

It would also be helpful to know whether these datasets are already planned or available through another cloud-hosted service.

I would be interested to hear the team's thoughts on feasibility, and whether other researchers would find this useful.

Thanks!

Mohammad Mohseni Aref