PMIP4 and the CMIP6 DRS

• CMIP6 document: CMIP6 Global Attributes, DRS, Filenames, Directory Structure, and CV’s (version 1.0)
  • Note: CMIP6 allowed CV values

• Legacy CMIP5 documents:
  • CMIP5 Data Reference Syntax (DRS) and Controlled Vocabularies (Version 1.3.1 - 13 June 2012)
  • CMIP5 Model Output Requirements: File Contents and Format, Data Structure and Metadata (7 January 2010)

• activity_id = activity labels
• mip_era = activity’s associated CMIP cycle
• Note: The project_id used in CMIP5 is being replaced in CMIP6 with two global attributes: 1) an activity_id, and 2) a mip_era (a label indicating the cycle of CMIP that this activity falls under which will be set to “CMIP6” for the 6th CMIP cycle). In a few cases it may be appropriate to include multiple activities in the activity_id (with multiple activities allowed, separated by single spaces). An example of this is “LUMIP AerChemMIP” for one of the land-use change experiments.

You can find all the referenced experiment names on es-doc search site: select Project=CMIP6-DRAFT and Type=Experiment

• experiment_id = root experiment identifier
• experiment = short expt. description

• sub_experiment_id=none ⇒ needed for CMIP6 hindcast and forecast experiments to indicate “start year”. For other experiments, this should be set to none
• sub_experiment=none ⇒ needed for CMIP6 hindcast and forecast experiments. For other experiments, this should be set to none

Reminder: DECK = Diagnostic, Evaluation and Characterization of Klima. More information on the CMIP6 experiments is available in Eyring et al 2017

You can find specific details about the experiments by visiting the PMIP4 experimental design section, or by directly clicking on one of the experiments below

• Allowed characters: CMIP6 experiment_id values are similar to CMIP5 (…so the permitted characters will be: a-z, A-Z, 0-9, and “-”), but a compound structure is allowed (segments separated by hyphens; e.g., “abrupt-4xCO2”); in a few cases multiple sub-experiments will be defined belonging to this root and constituting an ensemble.

• Being wise when creating/proposing a new name

  • Global attributes that label experiments are needed to construct file names and directories and can generally be used as search facets.  Together, they should have the following characteristics:
    
        * Uniquely label each experiment within CMIP6 and distinguish experiments with specified conditions that differ in any way 
        * Easily be interpreted and remembered
        * Facilitate representations of groups of experiments that are closely related (e.g., same forecast conditions but different start dates, or experiment with an “offline” model driven by output from various models)

• Planning for groups of related simulations

  • Often several simulations will be performed that satisfy the conditions specified for each experiment.  For example simulations of the historical period can branch from various points in a control run, and each of these will satisfy the conditions defining the experiment.  Together such simulations constitute a “conforming ensemble” with member all satisfying the same “root” experiment specifications.  There are also occasional cases where the experiment designers (MIP leaders) define a family of related simulations and choose to label these with a common “root” experiment name.  An example of this is the set of decadal prediction hindcasts that are all run similarly but started from different start dates (with each simulation identified by a different sub-experiment label).   Such “defined ensembles” of experiments will be labeled with a “root” experiment name, and a “sub-experiment_id” will be used to distinguish among members in the ensemble.

  • Ensemble of simulations usually share a common experiment_id and have different ripf variant labels.

You can find specific details about the experiments by visiting the PMIP4 experimental design section, or by directly clicking on one of the experiments below

experiment_id	experiment	Status
LDv1-LGMspin	Last Glacial Maximum spinup	Work
LDv1-transpin	Transient orbit and trace gases spinup (26-21 ka)	Work
LDv1	Transient deglaciation (21-0 ka)	Work
	Early Holocene (9.5 ka? 8.5 ka?)	Work
lig116k ?	Transition from the LIG to the glacial (2 experiments)	Work
	MIS11 (2 experiments)	Work
	Transient Holocene (6 ka to 0)	Work
	Transient LIG (130 ka to 125 ka)	Work
	DeepMIP	Work
	More experiments to come…

aka r<N>i<M>p<L> or rip

• b) ⇒ realization = an integer (≥1) distinguishing among members of an ensemble of simulations (e.g., 1, 2, 3, etc.). If only a single simulation was performed, then it is recommended that realization=1.
  For fields appearing in table “fx” in the CMIP5 Requested Output, set realization=0 (violating the general rule that it should be a positive definite integer).
  Note that if two different simulations were started from the same initial conditions, the same realization number should be used for both simulations. For example if a historical run with “natural forcing” only and another historical run that includes anthropogenic forcing were initiated from the same point in a control run, both should be assigned the same realization. Also, each so-called RCP (future scenario) simulation should normally be assigned the same realization integer as the historical run from which it was initiated. This will allow users to easily splice together the appropriate historical and future runs. A similar convention should be followed, when appropriate, with other simulations (e.g., the decadal simulations).
  Note that the realization can be used in constructing the “ensemble member” called for by the DRS document; it is the value of N in r<N>i<M>p<L>.
  [Note that for the “Transpose AMIP” project, the “realization” number is used to distinguish among the 16 members of each of 4 suites of runs (i.e., the 4 “seasons”) generated from different observed conditions, spaced 30 hours apart. So, for example, the 16-member ensemble of runs initialized at 00Z on 15 Oct 2008, 06Z 16 Oct 2008, 12Z 17 Oct 2008, and so-on, would be assigned “r1”, “r2”, “r3”, etc.]

• b) ⇒ initialization_method = an integer (≥1) referring to the initialization method used or different observational datasets used to initialize.
  If only a single method and dataset was used to initialize the model, then this argument should normally be given the value 1.
  For fields appearing in table “fx” in the CMIP5 Requested Output, set initialization_method=0 (violating the general rule that it should be a positive definite integer).
  See the DRS document for guidance on assigning initialization_method. Note that the initialization_method is used in constructing the “ensemble member” called for in the DRS document; it is the value of M in r<N>i<M>p<L>.

• b) ⇒ physics_version = an integer (≥1) referring to the physics version used by the model If there is only one physics version of the model, then this argument should be normally given the value 1.
  Note that model versions that are substantially different should be given a different “model_id”; assigning a different “physics_version” should be reserved for closely-related model versions (e.g., as in a “perturbed physics” ensemble) or for the same model, but with different forcing or feedbacks active. In CMIP5, one would distinguish, for example, among runs forced by different combinations of “forcing” agents (as called for under the “historicalMisc” experiment – experiment 7.3) by assigning different values to physics_version.
  For fields appearing in table “fx” in the CMIP5 Requested Output, set physics_version=0 (violating the general rule that it should be a positive definite integer). Note that the physics_version is used in constructing the “ensemble member” called for by the DRS document; it is the value of L in r<N>i<M>p<L>

• a) ⇒ Ensemble member (r<N>i<M>p<L>) =

  This triad of integers (N, M, L), formatted as
  shown above (e.g., “r3i1p21”) distinguishes among closely related simulations by a single
  model. All three are required even if only a single simulation is performed.
  
  The so-called “realization” number (a positive integer value of “N”) is used to distinguish among
  members of an ensemble typically generated by initializing a set of runs with different, but
  equally realistic, initial conditions. CMIP5 historical runs initialized from different times of a
  control run, for example, would be identified by “r1”, “r2”, “r3”, etc.). The data supplier must
  assign a realization number to each atomic dataset. It is generally recommended that the numbers
  be assigned sequentially starting with 1 (but other recommendations, specified below, may
  override this recommendation). In CMIP5, time-independent variables (i.e., those with
  frequency=”fx”) are not expected to differ across ensemble members, so for these N should be
  invariably assigned the value zero (“r0”). For TAMIP (“the Transpose AMIP activity), the
  “realization” number is used to distinguish among the 16 members of each of 4 ensembles (one
  for each of 4 “seasons”) generated from different observed conditions, spaced 30 hours apart.
  So, for example, the 16-member ensemble of runs initialized at 00Z on 15 Oct 2008, 06Z 16 Oct
  2008, 12Z 17 Oct 2008, and so-on, would be assigned “r1”, “r2”, “r3”, etc.
  
  Models used for forecasts that depend on the initial conditions might be initialized from
  observations using different methods or different observational datasets. These should be
  distinguished by assigning different positive integer values of “M” in the “initialization method
  indicator” (i<M>). For CMIP5 this indicator might in some cases be needed to distinguish
  among runs in the decadal-prediction suite of experiments (1.1-1.6). The data supplier must
  assign an initialization method number to each atomic dataset. It is recommended that the
  numbers be assigned sequentially starting with 1. In CMIP5, time-independent variables (i.e.,
  6
  those with frequency=”fx”) are not expected to differ across ensemble members, so for these M
  should invariably be assigned the value zero (“i0”). A key (i.e., a table) should be made
  available that associates each value of M with a particular initialization method and/or
  observational dataset.
  
  If there are many closely related model versions, which, as a group, are generally referred to as a
  perturbed physics ensemble (e.g., QUMP or climateprediction.net ensembles), then these should
  be distinguishable by a “perturbed physics” number, p<L>, where the positive integer value of L
  is uniquely associated with a particular set of model parameters (e.g., r3i1p78 is a third
  realization of the seventy-eighth version of the perturbed physics model). If there are different
  “forcing” combinations prescribed in experiment 7.3 in CMIP5 (the “historicalMisc” runs), then
  each of these different runs are also assigned different values of L (in “p<L>”). Note that the
  data supplier must assign a physics version number to each atomic dataset. It is recommended
  that the numbers be assigned sequentially starting with 1. In CMIP5, time-independent variables
  (i.e., those with frequency=”fx”) are not expected to differ across ensemble members, so for
  these L should always be assigned the value zero (“p0”). A key (i.e., a table) should be made
  available that associates each value of L with a particular set of model parameter values and/or,
  in the case of the “historicalMisc” experiment, a particular suite of “forcing” agents.
  
  Note that for a single model and experiment N, M, and L should be interpretable independently;
  for all members of the ensemble, the correspondence between the values of N, M, and L and the
  simulation characteristics they represent should be consistent. For example the two different
  ensemble members, r3i1p7 and r3i1p8, should both be initialized from exactly the same initial
  conditions using the same method (because the “r” and “i” values are identical) although the
  subsequent evolution of the simulations will presumably differ since they were produced by two
  different “perturbed physics” versions of the same model. Note that there may be cases where
  “gaps” could occur in the list of ensemble members. If, for example, two different initialization
  procedures were used, but the second procedure was tested with only a subset of the initial
  condition cases of the first procedure (say, every other case). Then the list of ensemble members
  would look like: r1i1p1, r2i1p1 r3i1p1, r4i1p1, r5i1p1, r6i1p1, …, r1i2p1, r3i2p1, r5i2p1, …
  
  A recommendation for CMIP5 is that each so-called RCP (future scenario) simulation should
  when possible be assigned the same realization integer as the historical run from which it was
  initiated. This will allow users to easily splice together the appropriate historical and future runs.
  Thus, for example, suppose a 3-member ensemble of historical runs of a model exists, and a
  single rcp45 simulation was produced, initialized from the third member of the historical
  ensemble. The recp45 simulation would be designated “r3” (rather than “r1”), even though it is
  the only existing ensemble member, in order to indicate that it was spawned from member 3 of
  the historical ensemble. A similar convention should be followed, when appropriate, with other
  simulations (e.g., the decadal simulations).

aka r<k>i<l>p<m>f<n> or ripf

• realization_index = realization number (integer >0)
• initialization_index = index for variant of initialization method (integer >0)
• physics_index = index for model physics variant (integer >0)
• forcing_index = index for variant of forcing (integer >0)
  • Note: the information stored in the forcing attribute in CMIP5 may in CMIP6 appear in the variant_info attribute
• variant_label = a label constructed from 4 indices stored as global attributes

  • r<k>i<l>p<m>f<n>
                      where
                         k = realization_index
                         l = initialization_index
                         m = physics_index
                         n = forcing_index

• variant_info = brief description of what is unique about this ripf variant
  • Example: “forcing: black carbon aerosol only”, “realization 1”, “realization 1; initialized using anomaly approach (method 2)”

Reminder: each option in r<k>i<l>p<m>f<n> has to be a strictly positive integer

The long PMIP4 simulations are going to require both a lot of processing power and a lot of storage. It is quite likely that there will be only one realization, for a given set of i<l>p<m>f<n> and that the variant label will always start with r1

Depending on available resources, the PMIP4 groups may choose to perform several simulations for the same experiment, using different combinations of forcings. The forcings used will have to be carefully described in the documentation (and in the metadata inside each NetCDF file) and be encoded in the integer value of the forcing_index.

There are several ways to proceed. The easiest way is to let each group choose its own way of numbering the forcings combinations (and document it!), but all groups should try to use a common scheme for and associate the same combination of forcings with the same integer

The contact people for each experiment determine which forcings combinations are most likely to be used and associate them with a predefined number. If necessary, a group can later ask for a new forcing combination to be registered

The following scheme will create bigger integers, but the values will be more meaningful

If there are 10 or less options for each type of forcing, we can assign a power of 10 to each type, multiply it with the forcing option and add everything

Tentative example for the lgm experiment:

Example: GLAC-1D + Hopcroft et al + interactive vegetation = 2 * 100 + 1 * 10 + 1 ⇒ f211

historical simulations that are the continuation of a past1000 simulation should use 1000 for the initialization_method ⇒ i1000

There are at least 2 sensitivity experiments

There are at least 3 sensitivity experiments

There may be other experiments listed in Proposed PMIP4 experiment_id values and PMIP4 experiments

The DRS defines (among other things) how the different attributes will be combined to generate unambiguous directories and file names, in the ESGF distributed database

Directory structure = <mip_era>/
                        <activity_id>/
                          <institution_id>/
                            <source_id>/
                              <experiment_id>/
                                <member_id>/         <== variant_label
                                  <table_id>/
                                    <variable_id>/
                                      <grid_label>/
                                        <version>/

file name = <variable_id>_<table_id>_<experiment_id >_<source_id>_<member_id>_<grid_label>[_<time_range>].nc

For PMIP4, we have sub_experiment_id == none (because we don't use forecast and hindcast), and therefore member_id == variant_label

Examples:

• Directory = CMIP6/CMIP/NCAR/CCSM2-1/1pctCO2/r1i1p1f1/Amon/tas/gn/v20150320/
  File = tas_Amon_CCSM2-1_1pctCO2_r1i1p1f1_gn_202001-202912.nc

• Directory = CMIP6/DCPP/NCAR/CCSM2-1/dcppA-hindcast/s1960-r1i2p1f1/Amon/tas/gr/v20150320/
  File = tas_Amon_CCSM2-1_hindcast_s1960-r1i2p1f1_gn_198001-198412.nc

The data end users have to follow the Terms of Use and licensing information detailed on the CMIP6: Proper citation and acknowledgement page.

The Terms of Use are also detailed in the license global attribute available in each data file created by the providers

Note: you can get the latest version of the license on github.

The “license” attribute should record the following statement (with segments in square brackets optional, and with required, appropriate text entered in place of <*> ): 

“CMIP6 model data produced by <Your Centre Name> is licensed under a Creative Commons Attribution-[NonCommercial-]ShareAlike 4.0 International License (https://creativecommons.org/licenses). Use of the data must be acknowledged following guidelines found at https://pcmdi.llnl.gov/home/CMIP6/CitationRequirements6-0.html. Further information about this data, including some limitations, can be found via the further_info_url (recorded as a global attribute in this file)[ and at <some URL maintained by modeling group>]. The data producers and data providers make no warranty, either express or implied, including, but not limited to, warranties of merchantability and fitness for a particular purpose. All liabilities arising from the supply of the information (including any liability arising in negligence) are excluded to the fullest extent permitted by law.”

The [*] indicates that institutions may choose to use the Non-commercial version of this license by inserting the words “NonCommercial” at this point, but this will significantly limit the use of the data in downstream climate mitigation and adaptation applications.  Please do not simply copy the statement above when writing data; Some text must be entered, some text is optional and the symbols “[*]” should not appear in the licensing text.

You can read the The Licenses section of About The Licenses if you want to understand the Creative Commons copyright licenses. You can also read the Six licenses for sharing your work summary pdf.

More specifically, CMIP6 data will be distributed under the following 2 licenses (each institute has to choose one license)

Logo	Abbreviation	Full name	Details
	CC BY-SA 4.0	Attribution-ShareAlike 4.0 International	https://creativecommons.org/licenses/by-sa/4.0/
	CC BY-NC-SA 4.0	Attribution-NonCommercial-ShareAlike 4.0 International	https://creativecommons.org/licenses/by-nc-sa/4.0/

Note: CC buttons and logos are available from the CC site Downloads page.