CESM LENS on AWS

Overview

The National Center for Atmospheric Research (NCAR) Community Earth System Model Large Ensemble (CESM LENS) dataset includes a 40-member ensemble of climate simulations for the period 1920-2100. All model runs were subject to the same radiative forcing scenario: historical up to 2005, and RCP8.5 thereafter. (RCP8.5 - Representative Concentration Pathway 8.5 - refers to the worst-case scenario considered in the Fifth Assessment Report of the Intergovernmental Panel on Climate Change - IPCC). Each of the 40 runs begins from a slightly different initial atmospheric state (created by randomly perturbing temperatures at the level of round-off error). The data comprise both surface (2D) and volumetric (3D) variables in the atmosphere, ocean, land, and ice domains.

The total LENS data volume is ~500 TB, and is traditionally accessible through the NCAR Climate Data Gateway (CDG) for download or via web services. A subset (currently ~70 TB compressed) including the most useful variables is now freely available on S3 thanks to the AWS Public Dataset Program.

Accessing CESM LENS on AWS

S3 bucket: ncar-cesm-lens in us-west-2 region
Amazon Resource Name arn:aws:s3:::ncar-cesm-lens
Bucket contents list: https://ncar-cesm-lens.s3.amazonaws.com/

The intent is for users to compute directly on the data in AWS rather than downloading from S3 (traditional download is available from the NCAR CDG link above).

Code Repository

A Jupyter Notebook illustrating how to read the LENS data on AWS, and reproducing Figures 2 and 4 from Kay et al. (2015), has been developed. This Notebook and other resources on GitHub will be gradually improved and augmented.

Available Data

Zarr format: The LENS data on AWS are structured according to the Zarr storage format. There are independent Zarr stores for each component, frequency, experiment, and variable. The naming convention is:
{component}/{frequency}/cesmLE-{experiment}-{variable}.zarr
where:

Zarr Stores

The table below shows the available Zarr stores, including the experiments, variables, time ranges, and 2D or 3D nature (3D means multiple atmosphere levels or ocean depths are present). See also text file listing all Zarr stores (in alphabetical order)

Prefix Experiments Variables Start End Vertical
atm/monthly/CESM-LE-* 20C, RCP85, HIST, CTRL, CTRL_SLAB, CTRL_AMIP 2D: FLNS, FLNSC, FLUT, FSNS, FSNSC, FSNTOA, ICEFRAC, LHFLX, PRECC, PRECL, PRECSC, PRECSL, PSL, SHFLX, TMQ, TREFHT, TREFHTMN, TREFHTMX, TS
3D: T, U, V, Q, Z3
1920-01 (20C)
2006-01 (RCP8.5)
1850-01 (HIST)
0400-01 (CTRL)
0001-01 (CTRL_SLAB)
0001-01 (CTRL_AMIP)
2005-12 (20C)
2100-12 (RCP8.5)
1919-12 (HIST)
2200-12 (CTRL)
1000-12 (CTRL_SLAB)
2600-12 (CTRL_AMIP)
2D, 3D
lnd/monthly/CESM-LE-* 20C, RCP85 FSNO, H2OSNO, QRUNOFF, RAIN, SNOW, SOILLIQ, SOILWATER_10CM 2D
ocn/monthly/CESM-LE-* 20C, RCP85, CTRL 2D: SSH, SST
3D: SALT
2D, 3D
ice_nh/monthly/CESM-LE-*, ice_sh/monthly/CESM-LE-* 20C, RCP85, CTRL aice, hi 2D
atm/daily/CESM-LE-* 20C, RCP85 FLNS, FLNSC, FLUT, FSNS, FSNSC, FSNTOA, ICEFRAC, LHFLX, PRECL, PRECSC, PRECSL, PRECT, PRECTMX, PSL, Q850, SHFLX, TMQ, TREFHT, TREFHTMN, TREFHTMX, TS, UBOT, WSPDSRFAV, Z500 1920-01-01 (20C)
2006-01-01 (RCP8.5)
2005-12-31 (20C)
2100-12-31 (RCP8.5)
2D
lnd/daily/CESM-LE-* 20C, RCP85 FSNO, H2OSNO, QRUNOFF, RAIN, SNOW, SOILWATER_10CM 2D
ice_nh/daily/CESM-LE-*, ice_sh/daily/CESM-LE-* 20C, RCP85 aice_d, hi_d 2D
atm/hourly6-1990-2005/CESM-LE-* 20C 2D: PRECT, PS, PSL, QREFHT, TREFHT, TS
3D: T, U, V, Q, Z3
1990-01-01T00 (20C) 2005-12-31T18 (20C) 2D, 3D
atm/hourly6-2026-2035/CESM-LE-* RCP85 2D: PRECT, PS, PSL, QREFHT, TREFHT, TS
3D: T, U, V, Q, Z3
2026-01-01T00 (RCP8.5) 2035-12-31T18 (RCP8.5) 2D, 3D
atm/hourly6-2071-2080/CESM-LE-* RCP85 2D: PRECT, PS, PSL, QREFHT, TREFHT, TS
3D: T, U, V, Q, Z3
2071-01-01T00 (RCP8.5) 2080-12-31T18 (RCP8.5) 2D, 3D

Short and long names of each variable:

Atmosphere Variables
FLNSNet longwave flux at surface
FLNSCClearsky net longwave flux at surface
FLUTUpwelling longwave flux at top of model
FSNSNet solar flux at surface
FSNSCClearsky net solar flux at surface
FSNTOANet solar flux at top of atmosphere
ICEFRACFraction of sfc area covered by sea-ice
LHFLXSurface latent heat flux
PRECCConvective precipitation rate (liq + ice)
PRECLLarge-scale (stable) precipitation rate (liq + ice)
PRECSCConvective snow rate (water equivalent)
PRECSLLarge-scale (stable) snow rate (water equivalent)
PRECTTotal (convective and large-scale) precipitation rate (liq + ice)
PRECTMXMaximum (convective and large-scale) precipitation rate (liq+ice)
PSSurface pressure
PSLSea level pressure
Q3D Specific humidity
Q850Specific Humidity at 850 mbar pressure surface
QREFHTReference height humidity
SHFLXSurface sensible heat flux
T3D Temperature
TMQTotal (vertically integrated) precipitable water
TREFHTReference height temperature
TREFHTMNMinimum reference height temperature over output period
TREFHTMXMaximum reference height temperature over output period
TSSurface temperature (radiative)
U3D Zonal wind
UBOTLowest model level zonal wind
V3D Meridional wind
WSPDSRFAVHorizontal total wind speed average at the surface
Z33D Geopotential Height (above sea level)
Z500Geopotential Z at 500 mbar pressure surface
Land Variables
FSNOfraction of ground covered by snow
H2OSNOsnow depth (liquid water)
QRUNOFFtotal liquid runoff (does not include QSNWCPICE)
RAINatmospheric rain
SNOWatmospheric snow
SOILLIQsoil liquid water (vegetated land units only)
SOILWATER_10CMsoil liquid water + ice in top 10cm of soil (veg land units only)
Ocean Variables
SALT3D Salinity
SSHSea Surface Height
SSTPotential Temperature
Ice Variables (Note: northern and southern hemisphere grids are stored separately as ice_nh and ice_sh)
aiceice area (aggregate)
aice_dice area (aggregate)
higrid cellmean ice thickness
hi_dgrid cellmean ice thickness

Known Issues

Users should consult the list of known issues in the CESM LENS dataset.

Data Citation and Updates

Data are freely available and reusable under the terms of the CC-BY-4.0 license. See Terms of Use. If you use these data, we request that you provide attribution in any derived products. The original, complete LENS dataset and the AWS-hosted subset have different DOIs (Digital Object Identifiers) to reflect their differing scope and format, so please cite whichever version of the dataset used, as well as the Kay et al. (2015) paper:

We also urge users to watch the CESM LENS GitHub repo for additions or changes to data and tools, and to notify cesm-lens-aws at ucar.edu if you are using these data so that we can justify continued maintenance on S3. All LENS model runs have been completed, so we do not expect to add data on an ongoing basis, but may be able to include additional fields upon request.