CNRM contribution to CMIP5

Known issues with CNRM-CM5 CMIP5 data

senesi — 2018-02-23T07:30:00Z

The known and documented problems with datasets are, or were :

1 (~~not an issue~~) The oceanic fields do not have coordinated latitude and longitude, because the oceanic grid used by the model is a curvilinear one; this complies with CMIP5 requirements and actually implements CMIP5 recommendation regarding oceanic grids; grid vertices coordinates are included in all files
2 (~~20010718~~) The grid area of CNRMCM5 (in areacella_fx_CNRM-CM5_historical_r0i0p0.nc) was wrong before version 20110718
3 (20111207) parent_experiment_rip is wrong for all historical experiments (including Nat, GHG, Misc, Ext), for member numbers beyond 1. The parent_experiment realization number has been set equal to the realization number of the child, while it should be equal to 1 (there is only one -common- parent experiment member, piControl r1i1p1). Thanks to J.Painter for raising the issue.
4 (20111207) branch_time is wrong for all RCP experiments; it is set to 1850-01-01 while it is of course 2006-01-01. Thanks to J.Painter for raising the issue.
5 ~~(20111207) some datasets did appear in the gateways interface and datanode catalog but were not accessible~~
6 (20111212) base time for oceanic data files is not constant over the years for piControl and all historicalXXX experiments from year 1870, and RCP runs from year 2016. However, values of time coordinates in data files are correct w.r.t. these varying references. Regarding the branch time, please refer to the summary page for branch time. Thanks to D.Pierce for raising the issue.
7 (20111214) base time for SeaIce files is uniformly set to 1800-1-1, while it should be set to the date of simulation start. However, values of time coordinates in data files are correct w.r.t. this reference.
8 - (~~withdrawn : oceanic fixed fields are now published~~)
9 ~~(20120201)~~ For all CNRM-CM5 cf3hr and cfDay 3D variables on model level, surface pressure was in hPa instead of Pa. Affected versions 20111006 (cfDay : amip, aquaControl, aqua4K, aqua4xCO2) 20111010 (cfDay : amip4K) 20111012 (cfDay : amip4xCO2, amipFuture) 20111018 (cf3hr : amip)
10 - ESG gateways do not show all datafiles/variables which have been published to the PCMDI gateway. Please let us know at contact.cmip5@meteo.fr about any data you are missing
11 (20120213) scale factor error for streamfunction variables msftmyz et msftmyzba. They are overscaled by 10**6 and are not masked
12 (20120301) variable sfcWind from Amon table is published as 'output2' instead of 'output1', due to an error in the publisher software
13 (20120321) Clear-sky shortwave radiation at the surface : the breakdown of net radiation between upward and downward components is wrong over LAND for all experiments; a proxy can be computed by interested users, using the net flux (which is correct when derived from published data) and deriving:
- the downward component as the net divided by 1 minus the all-sky (rather than clear-sky) surface albedo (i.e. the ratio between upward and downward all-sky radiation at the surface);
- the upward component as the downward minus the net flux.
  This breakdown is only a proxy and will not be published.
14 (20120402) RCP simulations - warm bias on ice sheets/shelves from 2150 : the ice sheet modeling does not include ice dynamics, and the ice sheet locations where snow mass budget is negative do end up as bare ground with a quite low albedo; there, low level warming is excessive; this applies to a varying fraction of the Greenland and Antartic extent, and this fraction becomes significant for Greenland from 2150 onwards; Antarctic ice shelves (Ronne and Ross) are also affected after 2150.
15 (20120416) cfSites table data for Dome C - Antarctica (station #119) contain negative relative humidity values ; they are due to very low specific humidity values which are transformed to very low negative values, at a compression/decompression stage in the processing suite. This data will not be corrected. Please apply a lower threshold of 0 % relative humidity when using the data.
16 (20120510) Description of forcings in metadata is ambiguous with respect to ozone : ozone is never quoted as an externally varying forcing, while actually, whenever chlorine concentration evolves, this has a direct impact on ozone concentration, because chlorine is an input of the our ozone prognostic scheme; so, for each simulation with 'GHG' quoted in forcing metadata, one could read 'SO, TO' as well.
17 (20120616) The version of NEMO (ocean model) in the metadata is misleading : it is labeled as NEMO3.3v10.6.6P but is based on NEMO release V3.2
18 (201308) : areacella was too low by a factor of 100 for all experiments for versions before 20130826.
19 (201401) : Fixed fields publication version 20130826 show to issues :
- Only atmos field should have been republished, because only areacella was actually changed
- orog field for lgm is wrong; version 20140201 of atmos fields fixes that.
20 (201402) : Error in the time coordinate for the six hourly atmos data (6 hrLev) of the rcp45 simulations (affected version 20121001). Version 20140226 fixes this issue.
21 (20140528) branch_time is wrong for historical experiment of CNRM-CM5-2. It is reported as 1850-01-01, in version 20130401 of the relevant datasets, while it actually is 1960-01-01.
22 (20141106) Variable wfo (water flux into the ocean) is positive upward, which is not fully consistent with the variable description
23 (20141106) : averaged or cumulative variables, as e.g. precipitation, have wrong time values : they are time-stamped with the time for the end of the accumulation period rather than the middle of the period
24 (20141107) Variables hfyba and hfydiff are erroneous in all CNRM-CM5.1 and CNRM-CM5.2 simulations, they should not be used. Variable hfy is erroneous but can be corrected off-line. The error is mainly important near the Equator where the ratio of the grid cell isizes n X and Y directions is far from 1. Here is the netcdf file of the correction factor to apply on the hfy filed stored on ESG servers
25 (20150115) The period 2006-2012 is not entirely covered in all members of historicalExt daily atmospherical simulations (affected version 20121001). The version 20141217 fixes this issue for all members of historicalExt simulations
26 (20150204) All CNRM-CM5 (except AMIP) and CNRM-CM5-2 variable tsl (temperature of soil) from Lmon table are erroneous : please substract 273.15 before using the data
27 (20160216) Variable ta from Amon table for members r8 (1850-1899 period) and r3 (1900-1949 period) of historicalGHG simulations are corrupted for first month and first level of pressure for part of the model domain (30N-65N)
28 (20170224) Latitude and longitude for variables uo and vo from Omon table for all members of historical simulations (v20130101) are erroneous.Values are reported on the T grid. U values actually apply to a location which is half a mesh farther east (and resp. V values to a location which is half a mesh farther north)
29 (20170821) Time axis is corrupted in the file va_day_CNRM-CM5_rcp85_r1i1p1_20360101-20401231.nc (version 20130101) of CNRM5 simulation. The time values are random at indices 119-1382, even holding negative values at times (simple issue of renumbering the time axis).
30 (20180223) An error has been detected in some files of the member r1i1p1 of historical simulation of the CNRM-CM5 CMIP5 model. The atmosphere data at the 6-hourly frequency on the model level (6hLev) for the variables ps, ta, ua, va, hus do not come for the member r1i1p1. We therefore advise not to use them. For more details, please read the attached file

CNRM-CM technical doc

voldoire — 2011-09-02T16:51:12Z

Link on private site CNRM-CM technical documentation

CNRM-CM5 reference paper

voldoire — 2011-09-02T14:28:56Z

The CNRM-CM5.1 global climate model: description and basic evaluation

A. Voldoire, E. Sanchez-Gomez, D. Salas y MÃ©lia, B. Decharme, C. Cassou, S.SÃ©nÃ©si, S. Valcke, I. Beau, A. Alias, M. Chevallier, M. DÃ©quÃ©, J. Deshayes, H. Douville, E. Fernandez, G. Madec, E. Maisonnave
, M.-P. Moine, S. Planton, D.Saint-Martin, S. Szopa, S. Tyteca, R. Alkama, S. Belamari, A. Braun, L. Coquart, F. Chauvin.

Climate Dynamics, vol. 40(9): 2091-2121, DOI:10.1007/s00382-011-1259-y.

Atlas de simulations

senesi — 2011-07-11T13:55:08Z

On trouvera sous ce lien les altlas de simulations partagÃ©s avec le Cerfacs

- CNRM-CERFACS

Model description

senesi — 2011-05-18T10:30:26Z

Please see either the reference paper for CNRM-CM5 or the reference location for model description

Boundary conditions

senesi — 2011-05-17T17:08:57Z

...(to be completed)...Please refer to upper level page for more explanations

Accessing CMIP5 data

senesi — 2011-05-16T12:12:58Z

( CNRM-CM5 runs for CMIP5 are described elsewhere )

You can access CNRM-CM5 CMIP5 data using one of the ESG data portals which list shows on CMIP5 data accessibility page. A beginner's guide is available, and support is provided at this adress

You can also access CNRM-CM5 data directly on the CNRM datanode. At the time of writing, this latter method ensures that all CNRM-CM5 published data can be accessed. It is however less user-friendly. Issues with this system should be directed to contact.cmip5@meteo.fr

We devised a set of bash functions to query and download ESG CMIP5 data. Just source this file in your bash shell and follow the directions which will be displayed

Additionaly, for those reading french, this presentation provides a very short overwiew of CMIP5, describes CMIP5 experiments run by CNRM,and provides some hints on accessing data on the ESG, at the date of the talk. This other presentation addresses only practical aspects of CMIP5 data access, from a february 2012 talk

Experiments, runs, publication

senesi — 2011-04-14T23:02:45Z

The status of data publication as of february, 8th, 2012 is : all data is published (including data of interest for CORDEX), except for the Last Glacial Maximum experiment. The ESG gateways are not able to show all data in their search interface, but you can access all of it directly on our datanode

CNRM-CM5 runs for CMIP5 are shared between CNRM and CERFACS; CNRM deals with all experiments but the decadal ones; please refer to contact.cmip5@cerfacs.fr for decadal runs

Also : detailed information on the status of data publication for all models can be found on the relevant CMIP5 page

Note : experiment names below are CMIP5 ones; please refer to the CMIP5 design document for details

piControl

The 850-year run is published

Note on ensembles

Ensembles are formed by choosing initial states at various years in piControl simulation, separated by at least 50 years (see details further below ). CMIP5 rules are strictly respected regarding members numbering consistency between different experiments (e.g RCP85 #2 starts from historical #2, and historical #4 starts with same initial state as historicalGHG #4)).

Recent past family

historical and historicalExt ensembles have 10 members (# 1 to 10)
historicalNat ensemble has 6 members (# 1,2,3,4,5,8).
historicalGHG ensemble has 6 members (# 1,2,3,4,5,8).
historicalMisc ensemble has 10 members (# 1 to 10) ; they were run using all anthropogenic forcings.

RCP

RCP8.5 is an ensemble of 5 members (#1, 2, 4, 6 and 10). Member 1 has an extension to 2300, is recommended for Cordex BC, and includes 6hLev data for 21st century.
RCP4.5 includes a single member, and was run up to 2300. It includes 6hLev data for the 21st century.
RCP2.6 includes a single member.
extension of RCPs for RCP8.5 and RCP4.5 from 2100 to 2300 were run.

Paleo climate

midHolocene : this run starts at the fake year 1950
lgm : the run is in the spin-up phase; it is the only one not published

CFMIP :

amip, amip4xC02, amip4K, amipFuture
aquaControl, aqua4K and aqua4xCO2

Ensembles

All runs start on january, 1st.
Within each ensemble, members branch time are separated by at least 50 years.
Branch year varies monotonically with member number, except for members 1 and 8, as indicated below : member 1 branches from year 400, and member 8 from year 0 (i.e 1850-01-01).
For RCP runs, the branch year should be understood as "branch year in piControl of the historical run, from which this RCP run does branch (in year 2006)"

member	1	2	3	4	5	6	7	8	9	10
branch year in piControl	400	100	150	200	250	300	350	__0	450	500
year of start in piControl	2250	1950	2000	2050	2100	2150	2200	1850	2300	2350
historical	x	x	x	x	x	x	x	x	x	x
historicalNat	x	x	x	x	x			x
historicalGHG	x	x	x	x	x			x
historicalMisc	x	x	x	x	x	x	x	x	x	x

RCP26	x
RCP45	x
RCP85	x	x		x		x				x

CNRM-CM5 soil layers scheme

senesi — 2011-04-13T14:24:52Z

The Surfex surface scheme of CNRM-CM5 is a force-restore scheme which includes three soil layers. The first layer is 1 cm deep. The two other layers have a depth which is varying in space, depending on the mixture of vegetation types.

Soil depth for layer 2 (i.e. root depth); enlarge by clicking

Base depth show in meters

Soil depth for layer 3; enlarge by clicking

Base depth show in meters

For CMIP5, moisture content of soil layers (mrlsl) has been interpolated to a set of 8 uniform depth soil layers, with bases at 0.01,0.2,0.5,1,1.5,2.,3.,8. meters. This leads to missing values at some locations for the deep layers.

Regarding temperatures in layers (tsl) values provided are not strictly representative of uniform-depth soil layers. Better interpret valus for layer 2 and 3 as relaxation values for the diurnal and annual waves, value for layer 4 being a general relaxation value.

CNRM contribution to CMIP5

senesi — 2011-04-13T11:49:06Z

CNRM and CERFACS contribute to CMIP5 with the model CNRM-CM5, which is an AOGCM. CNRM contributed experiments are summarized below and described in section Experiments, runs, publication.

With respect to the CMIP5 design, CNRM and CERFACS do cover the core AOGCM simulations, except for two which may be run later. Cerfacs handles decadal simulations, while CNRM handles the long-term and AMIP experiments shown below (figure below can be enlarged)

On the technical and practical sides : Data access is described on another page. CNRM runs an ESG data node for publishing its results to the ESGF system through the PCMDI gateway. We expect to also operate an OpenDAP server on our data node quite soon.