Skip to content

Latest commit

 

History

History
101 lines (86 loc) · 14.2 KB

cmip6_notes.md

File metadata and controls

101 lines (86 loc) · 14.2 KB
Raw

CMIP6

Citation

Get the dataset DOI from the lookup tables. There is also a "citation" link when you search for data files on the ESGF node websites.

The recommended acknowledgement when using CMIP6 data:

"We acknowledge the World Climate Research Programme, which, through its Working Group on Coupled Modelling, coordinated and promoted CMIP6. We thank the climate modeling groups for producing and making available their model output, the Earth System Grid Federation (ESGF) for archiving the data and providing access, and the multiple funding agencies who support CMIP6 and ESGF."

Resources

Data search and access at NCI

The general NCI data catalogue only has high level information on CMIP6 data. Instead, it's best to use the CleF command line tool or the NCI ESGF node webpage to search data.

  • Project fs38 for published Australian data, oi10 for replica data for other models.
  • At the ESGF node you can get the Thredds URL to access data directly via URL (in general, just replace /g/data/fs38/publications with https://esgf.nci.org.au/thredds/dodsC/master). This is a great option if you want a small subset of the data (e.g. small spatial region) because downloading a whole file just to get a small region is inefficient, but not so great for a large region/globe if bandwidth is slow.
  • For CleF:
    • At the moment $ clef --request sends to Paola Petrelli instead of NCI.
    • Instead, use the data download online form or send the output files from $clef --request (which are produced if you say no instead of yes at the end of the process) to help@nci.org.au or attach the files to a new issue at the NCI Help Portal.

Downloading data directly

The wget scripts from any of the ESGF nodes need to be run with a -H option. That makes it ask for your ESGF login credentials.

$ bash wget.sh -H

Models

A number of modelling centres have collections published with the The Journal of Advances in Modeling Earth Systems, which document various aspects of their CMIP6 models:

Model Information Ocean model Ocean model characteristics Issues
ACCESS-CM2 Website, Reference, ES-DOC
ACCESS-ESM1-5 Website, Reference, ES-DOC
AWI-CM-1-1-MR Website, Reference, ES-DOC FESOM (unstructured grid)
BCC-CSM2-MR Website, Reference, ES-DOC MOM 4 BO, FS, FWF 1. has two lat and lon coordinates (remove auxillary coords with fix_bcc_models.sh)
BCC-ESM1 Website, Reference, ES-DOC MOM 4 BO, FS, FWF 1. has two lat and lon coordinates (remove auxillary coords with fix_bcc_models.sh)
2. from sometime in 1930s onwards time values are zero in the so historical data files
CAMS-CSM1-0 Website, Reference, ES-DOC 1. I've downloaded the only areacello file (from 1pctCO2 experiment)
2. The branch_time_in_parent (in historical at least) is listed as 3025, which refers to the year as opposed to days since (the corresponding time is 45625 days since 2090-01-01).
3. The parent_time_units is wrong (in historical at least). It should be days since 2090-01-01, not days since 1850-01-01
CanESM5 Website, Reference, ES-DOC
CAS-ESM2-0 Website, Reference, ES-DOC 1. The surface ocean variables are split over two versions, e.g. /g/data/oi10/replicas/CMIP6/CMIP/CAS/CAS-ESM2-0/piControl/r1i1p1f1/Omon/wfo/gn/v2020030[6,7]/
2. wfo wrong sign (I think) and fluxes are way too small in magnitude
3. Branch time in historical is wrong. My visual estimate is year 80 (or 29200 days since 0001-01-01)
CESM2 Website, Reference, ES-DOC 1. Bogus global volcello (after adding missing land mask) and areacello totals - use gr and calculate own Ofx variables.
2. Missing history attribute in thetao (and probably other) files.
CESM2-FV2 Website, Reference, ES-DOC As per CESM2.
CESM2-FV2-WACCM Website, Reference, ES-DOC As per CESM2.
CESM2-WACCM Website, Reference, ES-DOC As per CESM2.
CNRM-CM6-1 Website, Reference, ES-DOC NEMO 3.6 BO, FS, FWF 1. wfo wrong sign
2. masso is not volo * rhozero
3. missing dimension coordinates for x and y
4. so and thetao files use different time chunks
CNRM-ESM2-1 Website, Reference, ES-DOC NEMO 3.6 BO, FS, FWF 1. wfo wrong sign
2. masso is not volo * rhozero
3. so and thetao files use different time chunks
E3SM-1-0 Website, Reference, ES-DOC 1. the missing value for land in so and thetao files (and possibly others) is 1.0
2. the historical experiment data doesn't span 1850-2014
3. They don't provide Ofx files but when you calculate it: Global ocean volume is 1.488020639104819e+18. Typical value is 1.3e+18 m3
EC-Earth3 Website, Reference, ES-DOC NEMO v? BO, FS, FWF 1. missing years in piControl thetaoga data
2. masso is not volo * rhozero
3. wfo wrong sign
4. unknown file format for some control files: /g/data/oi10/replicas/CMIP6/CMIP/EC-Earth-Consortium/EC-Earth3/piControl/r1i1p1f1/Omon/thetao/gn/v20200312/thetao_Omon_EC-Earth3_piControl_r1i1p1f1_gn_250101-250112.nc, /g/data/oi10/replicas/CMIP6/CMIP/EC-Earth-Consortium/EC-Earth3/piControl/r1i1p1f1/Omon/hfds/gn/v20200312/hfds_Omon_EC-Earth3_piControl_r1i1p1f1_gn_267801-267812.nc - I've downloaded new copies
EC-Earth3-Veg Website, Reference, ES-DOC NEMO v? BO, FS, FWF 1. missing years in control thetaoga data
2. wfo does not equal wfonocorr
3. masso is not volo * rhozero
4. wfo wrong sign
5. The branch_time is listed as the year 2030 when it should be 1930
6. Bogus netCDF format for v20200226/tos_Omon_EC-Earth3-Veg_piControl_r1i1p1f1_gn_188601-188612.nc
FGOALS-f3-L Website, Reference, ES-DOC 1. Bogus volcello file (global volume is 1.9e+18 m3, typical value is 1.3e+18 m3)
2. Masked points are 1e35 in areacello file (but when fixed can be used to create correct volcello)
wfo wrong sign
4. Might be something wrong with piControl hfds data?
FGOALS-g3 Website, Reference, ES-DOC 1. Branch times in DAMIP files are wrong. Should be the same as in historical: 134685
GFDL-CM4 Website, Reference, ES-DOC MOM 6 BO, FS, FWF 1. historical branch time is listed as 36500 (year 100) but should be 91250 (year 250)
GFDL-ESM4 Website, Reference, ES-DOC 1. sos data missing for ssp585
GISS-E2-1-G Website, Reference, ES-DOC 1. wfo wrong sign
2. surface ocean variables (e.g. tos, hfds but not wfo) on different grid to full depth ocean (thetao, so). I've created gn-surface basin and areacello files to handle this.
GISS-E2-1-G-CC Website, Reference, ES-DOC 1. wfo wrong sign
2. surface ocean variables (e.g. tos, hfds but not wfo) on different grid to full depth ocean (thetao, so). I've created gn-surface basin and areacello files to handle this.
GISS-E2-1-H Website, Reference, ES-DOC 1. The v20190410/so_Omon_GISS-E2-1-H_piControl_r1i1p1f1_gr_322001-323912.nc file is missing (so I downloaded it myself).
2. For both thetao and so the 358001-358012 control file needs to be removed as it overlaps with the correct 358001-359912 file.
HadGEM3-GC31-LL Website, Reference, ES-DOC NEMO 3.6 BO, FS, FWF
IPSL-CM6A-LR Website, Reference, ES-DOC NEMO 3.2 BO, FS, FWF 1. wfo wrong sign
2. masso is not volo * rhozero
3. missing dimension coordinates for x and y
4. Some control evspsbl files are water_evaporation_flux and others are water_evapotranspiration_flux
MCM-UA-1-0 Website, Reference, ES-DOC 1. global ocean area and volume (from areacello and volcello) is way too small
2. The long_name for thetao and so has a bunch of white space after it
MIROC6 Website, Reference, ES-DOC
MIROC-ES2L Website, Reference, ES-DOC
MPI-ESM1-2-HR Website, Reference, ES-DOC
MPI-ESM1-2-LR Website, Reference, ES-DOC
MRI-ESM2-0 Website, Reference, ES-DOC 1. No gn volcello and when calculated get global value which is slightly too high to be realistic
2. gn ocean files use Nanf instead of masked value
NESM3 Website, Reference, ES-DOC 1. historical data doesn't go beyond 1999
2. curvilinear ocean grid and no Ofx data, so must be regridded using regrid.py first
NorCPM1 Website, Reference, ES-DOC 1. can use the gn areacello data (which has depth coord in metres) with gr data
2. historical branch time is wrong in metadata, should be 0
NorESM2-LM Website, Reference, ES-DOC 1. bogus gn volcello data (so use gr data)
NorESM2-MM Website, Reference, ES-DOC 1. no gn volcello data (so use gr data)
2. Omon, piControl data has both 145001-145012 and 145002-145912 files. Need latter so use *912.nc
SAM0-UNICON Website, Reference, ES-DOC POP2 BO, FS, FWF 1. wfo values are zero everywhere
2. hfds wrong sign
UKESM1-0-LL Website, Reference, ES-DOC NEMO 3.6 BO, FS, FWF 1. constant wfo

Assumed characteristics.