Data information and known issues

Laurent Franchisteguy — 2021-05-20T08:00:00Z

Unless a precision is given, informations/issues listed here apply to all of the CNRM-CERFACS data files, whatever the model configuration (AOCGM, ESM, ...), whatever the MIP and the experiment. Updated and detailed informations are also available on ES-DOC dataset errata, see specific link for CNRM-CERFACS models.

– Variant label r1i1p1f2 : given our simulations starting date, we used a version of forcing dataset not earlier than 6.2.0 . That's why we set forcing_index=2, according to the CMIP6 guidelines at this date. Other modelling groups use a different number for the forcing_index (commonly Â«Â f1Â Â»). Note that this does not necessarily means that forcing version is different from our. Since the CMIP6 forcing version is unfortunately not documented by the CMIP6 file attributes, the only way to know is to ask the modelling centers or refer to the ES-DOC external model documentation (when available)

– April 2021 : all CNRM-CM6-1 amip-lfmip experiments (v20200408 and v20200218) have been unpublished because we ran the simulations relaxing the soil moisture more loosely than intended (24h relaxation time). Also, there was an issue with the way we computed the soil moisture running mean daily climatology for the amip-lfmip-rmLC experiments (it is fine at the monthly time-scale but there is an issue on the day-to-day values).
In the new experiments published on ESGF (v20210512), the liquid soil moisture is prescribed (the relaxation time equals the time step) and the running mean daily climatology is corrected.

– February 2021 : the errors detected in CNRM-CM6-1/CNRM-ESM2-1 output for several COSP variables ( albisccp, pctisccp, clcalipso2, climodis, cltmodis and clwmodis (Tables Emon and E3hrPt) ) have been corrected and republished on ESGF. A few simulations have been rerun to provide corrected variables. This concerns the following experiments for CNRM-CM6-1 only: amip, amip-piForcing, amip-4xCO2, amip-p4K, amip-lwoff, amip-p4K-lwoff, amip-m4K, amip-future4K.

– September 2020 : note that variables sivoln and sivols are overestimated (around 1%)

– September 2020 : all CNRM-ESM2-1 chlorophyll-a data are 3 orders of magnitude (1e3) too high because related variables are in g/m3 instead of kg/m3 as expected in CMIP6 data request. Users are invited to apply a factor of 10^-3 (list of concerned datasets detailed on https://errata.es-doc.org/static/view.html?uid=09b33977-b6a7-6447-b5f4-05457eb4f9da ).

– June 2020 : the carbon cycle diagnostics rh and nep have been wrongly produced by CNRM-CM6-1 and CNRM-CM6-1-HR. We have unpublished them from the ESGF (https://errata.es-doc.org/static/view.html?uid=2add4cda-d3af-6bb5-9557-73fc58b892f7). Note that these variables have been correctly produced by CNRM-ESM2-1.

– June 2020 : we have identified a unit error for the variable burntFractionAll for the table Lmon. The variable is given in % day-1 whereas the unit in the monthly files should be given in % month-1. This unit error can be easily fixed by multiplying burntFractionAll by the number of days of a given month.

– June 2020 : we have identified an issue in volcanic aerosols used in CNRM-CM6-1-HR. These aerosols are included through a monthly and interannual stratospheric aerosol optical depth dataset, whose latitudes have been unfortunately reversed in the following simulations :
- DECK : CNRM-CM6-1-HR_historical_r1i1p1f2
- DECK : CNRM-CM6-1-HR_amip_r1i1p1f2
- HighResMIP : CNRM-CM6-1-HR_hist-1950_r1i1p1f2
- HighResMIP : CNRM-CM6-1-HR_highresSST-present_r1i1p1f2
However the total content (global average) of volcanic aerosols is correct, only the spatial repartition between the two hemispheres is wrong. Note also that the other simulations using CNRM-CM6-1-HR are not concerned by this issue.

– May 2020 : In LS3MIP, the amip-lfmip-rmLC and amip-lfmip-pdLC transient simulations (1980-2100 period) were each run twice, following the ssp126 and ssp585 scenarios. To distinguish the simulation names, it was decided to use the f1 variant label for the experiments following the ssp126 scenario, and the f2 variant label for the experiments following the ssp585 scenario. But since we did not want to use the f1 variant label in any of our CMIP6 experiments, we used the f11 label instead.
Hence the following names for the CNRM-CERFACS lfmip experiments :
-CNRM-CM6-1-amip-lfmip-rmLC_r1i1p1f11 follows the ssp126 scenario ; it corresponds to the amip-rmLC_r1i1p1f1 experiments from other modelling centers.
- CNRM-CM6-1-amip-lfmip-pdLC_r1i1p1f11 follows the ssp126 scenario ; it corresponds to the amip-rmLC_r1i1p1f1 experiments from other modelling centers.
- CNRM-CM6-1-amip-lfmip-rmLC_r1i1p1f2 follows the ssp585 scenario, as specified by LS3MIP.
- CNRM-CM6-1-amip-lfmip-pdLC_r1i1p1f2 follows the ssp585 scenario, as specified by LS3MIP.
- The present-day (1980-2014) simulation CNRM-CM6-1-amip-pObs_r1i1p1f2 corresponds to the amip-pObs_r1i1p1f1 experiments from other modelling centers.

– February 2020 : We have identified issues of discontinuity for some diagnostics related to the ocean heat content in two simulations of CNRM-CM6-1 (historical member 29 v20191004 and hist-GHG member 3 v20190308). This issue is due to a corrupted restart file used in the course of the simulation. The discontinuity is not seable in many diagnostics but impact the reproducibility of the results.
Besides, we would like to raise your attention that some of the impacted variables might have been used as lateral boundary conditions for regional climate models. It is important for these users to employ the most up-to-date version of these simulations that have been republished on ESGF (historical member 29 version v20200529 and hist-GHG version v20201208 ). Note that in these up-to-date versions, the data will be the same for the first years of the simulation but diverge from the former version from the date of corrupted restart file and are thus different for a long period of the simulation.

– January 2020 : CNRM-ESM2-1 so-called *4co2* diagnostics, instantaneous radiation fields under 4*CO2 conditions, have been incorrectly produced as identical diagnostics as the corresponding fields under CO2 conditions. These diagnostics consist in 12 variables named rld4co2*, rldcs4co2*, rlu4co2*, rlucs4co2*, rlut4co2*, rlutcs4co2*, rsd4co2*, rsdcs4co2*, rsu4co2*, rsucs4co2*, rsut4co2*, rsutcs4co2*. We have unpublished them from the ESGF (see https://errata.es-doc.org/static/view.html?uid=ba6c8e9c-94a3-2122-913e-e339786e9b28 ) . Note also that the CNRM-CM6-1 *4co2* diagnostics have been correctly produced.

– November 2019 : The amip-a4SST-4xCO2 simulation (v20190912) has been unpublished due to a wrong fixed CO2 concentration that was quadrupled compared to the piControl CO2 level. The revised simulation wich uses time-evolving CO2 concentrations that are quadrupled compared to the amip reference simulation has been republished on ESGF (v20191218).

– September 2019 : We have identified issues in the reproductibility of two ScenarioMIP CNRM-ESM2-1 simulation (ssp585 and ssp370). This issue leads to inconsistencies between some variables from the same simulation. It only affects the results of first ensemble members (r1i1p1f2) of ssp585 (v20190328) from 2047 onwards and of ssp370 (v20190328) from 2059 onwards. Before that time, the results of both simulations are not impacted. After that time, the scientific results of both simulations make sense and are reasonable but we prefer to commit an update version of these simulations on ESGF as soon as possible.
Besides, we would like to raise your attention that some of the impacted variables might have been used as lateral boundary conditions for regional climate models. It is important for these users to employ the most up-to-date version of these two scenarios. Updated version (v20191021) of these scenarios have been published on ESGF.

– September 2019 : an error was detected in the aqua-p4K CNRM-C6-1 experiment (v20190820). It was rerun and republished on the ESGF (v20191004).

– September 2019 : the amip-piForcing experiment based on the CNRM-CM6-1 standard configuration of the CNRM model (v20190820) has been mistakenly initialized on January 1st 1979 (like the AMIP experiment from the DECK). It was rerun starting as requested by CFMIP on January 1st 1870 and republished on the ESGF (v20191114).

– September 2019 : the solar constant has been wrongly set to 2014 in the piClim-ghg and piClim-anthro simulations, and is therefore different from the solar constant in the piClim-control simulation where is has been set to 1850. The rsdt fields therefore differ in the control and these perturbed simulations

– August 2019 : an error in the cfadDbze94 variable (Table Emon and E3hrPt) was detected in CNRM-CM6-1/CNRM-ESM2-1 output (bug within the online coupling between the model and the COSP/CloudSat Radar simulator). The erroneous output have been removed from the ESGF archive. The amip simulations, which have produced this diagnostic, will be rerun as soon as possible with a corrected diagnostic (see https://errata.es-doc.org/static/view.html?uid=fedee017-40b6-fcb3-8fca-48f9730de0da).

– August 2019 : please consider the following information that sheds some light on some diagnostics from CNRM-ESM2-1 simulations such as for instance the piClim-4xCO2 simulation. In CNRM-ESM2-1, chemical evolutions are computed by the chemistry scheme from the top of the atmosphere down to 560 hPa. Below that level, concentrations of a number of species (i.e., N2O, CH4, CO, CO2, CFC11, CFC12, CFC113, CCl4, CH3CCl3, CH3Cl, HCFC22, CH3Br, H1211, H1301) are relaxed towards the CMIP6 yearly evolving global mean abundances. In the specific cases of the piClim-4xCO2 or of the abrupt4xCO2 simulations, as the start of the simulation is a 1850 state the CO2 concentration will be 4 times that of 1850 throughout the entire atmosphere after about 15 years of simulations. So please consider this time of adjustment, that depends of course on the chemical species, and on its initial and final states.

– As the CNRM-ESM2-1 chemistry scheme does not parametrize the low troposphere ozone chemistry, and therefore does not consider emissions of ozone precursors, all CNRM-ESM2-1 *NTCF* simulations consider only modifications of aerosol (or aerosol precursor) emissions. As a result, for example, the piClim-aer (RFMIP) and piClim-NTCF (AerChemMIP) simulations performed with the atmospheric component of CNRM-ESM2-1 are identical.

– June 2019 : for the sake of clarification, note that our CNRM-CM6-1 piClim-aer simulation has all aerosols at 2014, including dust and sea-salt. We performed an additional simulation piClim-aerant with dust and sea-salt at 1850 values, whose diagnostics are available upon request to contact.cmip(at)meteo.fr .
Note also that our CNRM-CM6-1 piClim-ghg and CNRM-ESM2-1 piClim-ghg simulations has all GHGs, including CFCs, at 2014. Thus, our piClim-ghg simulations consider 2014 stratospheric ozone. As for the low tropospheric ozone, our CNRM climate models don't parameterize it.

– May 2019 : the definition of field vt100 as requested in CMIP6 Data Request version 01.00.21 (used for all runs) is "Northward Heat Flux due to Eddies" ; it was overlooked, which resulted in the publication of a field which is a raw temperature advection (i.e. a raw product of temperature and northward wind component); furthermore, the units indicated is W/m*2, which is erroneous. The erroneous output were removed from the ESGF archive (see https://errata.es-doc.org/static/view.html?uid=ff701570-b976-b9eb-8b88-348836e14b09)

– April 2019 : time sampling of pressure-level data is badly described in all CNRM-CM6 and CNRM-ESM2-1 data on pressure levels. CMIP6 does not describe in detail how time-averaged fields of pressure-level data should be computed. CNRM-CERFACS chosed to perform interpolation to pressure levels first (using instantaneous fields), followed by time averaging; the frequency at which interpolation to pressure level was done si uniformly 3h; hence, the meta-data of pressure-level fields, which states, e.g. :
va:interval_operation = "900 s" ;
va:interval_write = "1 month" ;
is not meaningful for 'interval_operation'

– March 2019Â : airmass* : the airmass* field is a (time, lat, lon) field that corresponds to the mass of the atmosphere in the corresponding columns. The CMIP6 Data Request requested a (time, lev, lat, lon) airmass field. A script is provided to recompute the airmass in the CNRM model layers

– March 2019Â : the ap_bnds and b_bnds arrays in all CNRM CMIP6 files with atmospheric data on model levels are incorrectly ordered. Correctly ordered arrays are provided in the joint CNRM_models_vertical_coordinates.nc file

– February 2019Â : dimension â€˜landUse typeâ€™ missing for variables tasLut, mrsosLut and hussLut. The single provided field stand for Lut index=1 (i.e. tile primary_and_secondary_land)

– February 2019Â : an error in the monthly and daily averages of variables albisccp/pctisccp was detected in CNRM-CM6-1/CNRM-ESM2-1 output. These time-means were not weighted by the ISCCP Total Cloud Fraction (cltisccp) as requested by the CMIP6 Data Request. The erroneous output were removed from the ESGF archive (see https://errata.es-doc.org/static/view.html?uid=00f7fdab-c123-a55d-9788-bd73dfb8aa1a). Only a few simulations concerned by this issue will be rerun to provide a diagnostic consistent with the CFMIP data request. This will be advertised on this web page as soon as available. We might consider to rerun a few other experiments, if relevant for the community, and depending on our ressources

– February 2019Â : variables agesno and tsn time averages are purposely plain time averages of high frequency values (i.e. without any weighting). So, they do not strictly comply with CMIP6 Data Request content which states :"When computing the time-mean here, the time samples, weighted by the mass of snow on the land portion of the grid cell, are accumulated and then divided by the sum of the weights"

– December 2018Â : ozone (o3 variable) concentrations were first published in "kg kg-1" with a corresponding incorrect unit attribute in the netcdf files (o3:units = "mol mol-1"). See https://errata.es-doc.org/static/view.html?uid=a04ee6dc-e338-1e06-b187-096c6407ca85 . Corrected (converted in the standard unit "mol mol-1") and republished on ESGF (v20190408)

– November 2018Â : data published with a mispelling in the realm name (ocnBgChem instead of ocnBgchem). See https://errata.es-doc.org/static/view.html?uid=17e7ea87-f9b0-18aa-8e14-7b4ef721f232 . Corrected and republished on ESGF

– August 2018Â : missing periods for a list of variablesÂ of CNRM-CM6-1 LR piControl simulation :
* 1850-1949 for SImon (siage, sicompstren, sifb, siflcondbot, siflcondtop, sifllatstop, sifllwutop, siflsensupbot, siflswdbot, siflswdtop, siflswutop, sisali, sisnthick, sispeed, sitempbot, sitempsnic, sitemptop,sithick), SIday (sithick) and Limon (tsn)
* 1900-1942 and 2250-2260 periods for Omon (bigthetao)
* 1850-2049 and 2241-2260 periods for Omon (bigthetaoga) and Amon(evspsbl)

Data access

Laurent Franchisteguy — 2019-04-25T17:42:19Z

CNRM CMIP simulations are available on the ESGF (Earth System Grid Federation, https://esgf.llnl.gov/ ).

Access to data is possible through any of the federated ESGF-COG Nodes (https://esgf.llnl.gov/nodes.html) or more directly through the CNRM ESGF datanode https://esg1.umr-cnrm.fr/.

Search and access to CMIP6 simulation outputs is possible using the following direct links : https://esgf-node.ipsl.upmc.fr/search/cmip6-ipsl/ or https://esgf-node.llnl.gov/search/cmip6/ .

CNRM-CERFACS contribution to CMIP6

Data information and known issues

Data access