We have published a copy of the NA-CORDEX dataset on Amazon Web Services (AWS). We thank the AWS Open Data Sponsorship Program and the Amazon Sustainability Data Initiative for providing free storage and egress for these data.

The AWS S3 bucket name is "ncar-na-cordex", in the us-west-2 region.


The NA-CORDEX data on AWS is stored in Zarr format. This format supports the same data model as netCDF and is well suited to object storage and distributed computing in the cloud using the Pangeo libraries in Python.

Available Data

An Intake-ESM Catalog listing all available data can be found at:

There are independent Zarr stores for each variable, scenario, output frequency, grid resolution, and bias correction method. The naming convention is:

  • frequency = {day}
  • variable = one of the variable names listed in the table below
  • scenario = {eval, hist, rcp45, rcp85, hist-rcp45, hist-rcp85} (see below)
  • grid = {NAM-22i, NAM-44i} (0.25-degree / 0.50-degree resolution)
  • bias-correction = {raw, mbcn-Daymet, mbcn-gridMET} (see Dataset Description for details on bias-correction)

VariableLong Name
hurs Near-Surface Relative Humidity *
huss Near-Surface Specific Humidity
pr Precipitation (kg m-2 s-1)
prec Precipitation (mm/day)
ps Surface Air Pressure
rsds Surface Downwelling Shortwave Radiation
sfcWind Near-Surface Wind Speed **
tas Near-Surface Air Temperature (K) †
tasmax Daily Maximum Near-Surface Air Temperature (K) †
tasmin Daily Minimum Near-Surface Air Temperature (K) †
temp Near-Surface Air Temperature (C) †
tmax Daily Maximum Near-Surface Air Temperature (C) †
tmin Daily Minimum Near-Surface Air Temperature (C) †
uas Eastward Near-Surface Wind Velocity
vas Northward Near-Surface Wind Velocity

"Near-Surface" = 2m for temperature and humidity, 10m for winds
* calculated from huss, ps, and tas
** calculated from uas and vas
† temp, tmax, and tmin are the same as tas, tasmax, and tasmin except for the units (degrees Celsius instead of degrees Kelvin); this is for user convenience and compatibility with the observed datasets used in bias-correction. The same is true of prec and pr (mm/day instead of kg/m^2/s).

eval 1979-2015driven by ERA-Interim reanalysis
hist 1950-2005driven by Historical GCM
rcp45 2006-2100driven by RCP 4.5 GCM
rcp85 2006-2100driven by RCP 8.5 GCM
hist-rcp451950-2100concatenation of hist and rcp45
hist-rcp851950-2100concatenation of hist and rcp85

Zarr Stores

Zarr stores are available for all combinations of variable, frequency, etc. listed above. For example, the Zarr store day/ contains daily precipitation data from simulations driven by GCMs under historical and RCP 8.5 emissions for the years 1950-2100 at quarter-degree resolution, not bias-corrected.

Common Time Coordinates

Combining data from different simulations into a single zarr store requires that it have a common time coordinate. To accommodate varying calendars, we inserted days filled with missing_value to pad out the shorter calendars to match the real-world Gregorian calendar. This approach retains all data and allows the use of standard date-time libraries.

Simulations with 365-day / noleap calendars had missing days inserted on Feb. 29th of leap years. Simulations with a 360-day calendar additionally had missing days inserted on January, March, May, August, and October 31st of every year.

Known Issues

Known issues with the data:
Descriptions of how the data have been revised over time:

Dataset Citation

Use of NA-CORDEX data is subject to the Terms of Use. When publishing research based on NA-CORDEX data, be sure to include a citation for the dataset. Information on citing the NA-CORDEX dataset can be found here:

The citation for the AWS version of the dataset is:
Bonnlander, B., S. McGinnis, A. Banihirwe, E. Nienhouse, and J. de La Beaujardière, 2021. "NA-CORDEX dataset on AWS", version 1.0, UCAR/NCAR Computational and Informations Systems Lab, accessed [date], DOI:

We recommend citing both the parent dataset and the AWS version.