CMIP6 recommendations

Recommendations for CMIP6 solar forcing data

Version 3.2 available (final version)

The final version of the dataset has been published in 

Matthes, K., Funke, B., Anderson, M. E., Barnard, L., Beer, J., Charbonneau, P., Clilverd, M. A., Dudok de Wit, T., Haberreiter, M., Hendry, A., Jackman, C. H., Kretschmar, M., Kruschke, T., Kunze, M., Langematz, U., Marsh, D. R., Maycock, A., Misios, S., Rodger, C. J., Scaife, A. A., Seppälä, A., Shangguan, M., Sinnhuber, M., Tourpali, K., Usoskin, I., van de Kamp, M., Verronen, P. T., and S. Versick, 2017: Solar Forcing for CMIP6 (v3.2). Geosci. Model Dev., 10, doi:10.5194/gmd-10-2247-2017.

30 Nov 2017

See bottom of this page for release notes.

21.06.2016: A mailing list to inform you about news and updates regarding the solar forcing recommendations is now available. If you like to receive news and updated information on the solar forcing recommendations for CMIP6, please visit our subscription page to add your email to our mailing list. It is a manually managed low-volume list to avoid SPAM send via this list.

An overview of the CMIP6 solar forcing dataset (irradiance as well as particle-related parameters) and guidelines for their usage are given here.

Irradiance forcing

TSI:     Total solar irradiance
F10.7: F10.7 cm solar radio flux
SSI:    Spectral solar irradiance for 10-100,000 nm range

Particle forcing:

Ap:   Daily planetary Ap-index
Kp:   Daily planetary Kp-index
iprp:  Ion-pair production rate by solar protons
iprg:  Ion-pair production rate by galactic cosmic rays
iprm: Ion-pair production rate by medium-energy electrons

Please note the following differences to the solar forcing recommended for CMIP5:

  • New, lower TSI value: 1361.0 ± 0.5 W m−2  (Mamajek et al., 2015).
  • Time-varying solar forcing is provided in one file from 1850-2300 in daily as well as monthly resolution separately
  • Particle forcing (due to protons, medium-energy electrons, and galactic cosmic rays) included in daily resolution files

A detailed description of how the data were constructed as well as a “technical” documentation of the files’ structures, dimensions, variable names, etc. can be found here.

All data are provided in zipped netcdf (HDF5) files. For those models that do require only a subset of the forcings, these are to be selected from the provided files by the user.

Auxililiary software packages for data pre-prossessing are available at the bottom of this page, including:

  • a routine (currently only available in MATLAB) that pre-processes solar irradiance data by selecting only TSI and SSI (plus wavelength information) to integrate the latter over spectral bands to be defined by the user according to the needs of his/her model.
  • A routine (MATLAB and IDL) for generation of a particle-induced NOy upper boundary condition for chemistry-climate models with upper lid at 0.01-1 hPa (based on the Ap index included in the “reference”, “extreme”, and piControl forcing datasets).
  •  A routine (MATLAB) for projection of particle-induced ionization rates (i.e., iprg, iprm, iprp, all provided on geomagnetic latitudes) onto geographic coordinates as function of pressure level and time.

What to prescribe in the CMIP6 pre-industrial control simulation (part of CMIP6 DECK)?

The pre-industrial control forcing (pictrontrol) is constructed of time-averaged historical data (see below) corresponding to 1850-1873 (solar cycle 9+10) mean conditions. See metadata file for more details.

In addition, we provide for sensitivity studies a variable pe-industrial  control forcing which includes an 11-year solar cycle but without longterm trend (see below).  Note that it is not officially part of a CMIP6 MIP proposal.

What to prescribe in transient simulations, such as the DECK AMIP Historical and CMIP6 Historical experiments?

The standard solar forcing dataset recommended for usage is the solar reference scenario dataset which consists of historical reconstructions (1850-2014) and the most likely scenario for future solar forcing (2015-2299; see metadata file for more details) which – unlike CMIP5 future solar forcing – is still variable over time:

Please note that in order to account for solar cycle effects, not only radiative effects but also photochemical effects have to be taken into account!

CMIP6 models that do not have interactive chemistry should in addition to adapting the irradiance changes also prescribe the CMIP6 recommended ozone forcing data set. The solar signal in this ozone dataset is consistent with the CMIP6 solar forcing.

What to prescribe in future simulations (DAMIP, ScenarioMIP, DCPP)?

The future solar forcing recommendation is part of the solar reference scenario (ref) as described above and the outcome of an ISSI team "Scenarios of Future Solar Activity for Climate Modelling" in 2015. The solar reference scenario provides the most likely scenario for future solar forcing (2015-2099) and is described in more detail in Matthes et al., GMD, 2017.

In CMIP5, climate projections were based on a stationary sun scenario, obtained by simply repeating solar cycle 23, which ran from 04/1996 to 06/2008 (Lean and Rind, 2009). In CMIP6, we include a more realistic forcing, and provide two different scenarios:

  1. A reference scenario (ref) with the most likely level of solar forcing, and

  2. an extreme scenario (ext) with an exceptionally low level of activity, corresponding to the lower 5th percentile of all forecasts equivalent to that of a Maunder Minimum (see below). This extreme scenario is meant to be used for sensitivity studies, and is not officially part of a CMIP6 MIP proposal.

For the future period, the solar forcing is derived from the heliospheric potential. We forecast the latter by using 9400 years of reconstruction of solar activity from cosmogenic isotopes (Steinhilber et al., 2012). The reference scenario is the weighted average of 3 forecasts obtained by: (i) analogue forecast (superposed epoch), (ii) deterministic forecast relying on periodicities in the observed heliospheric potential, and (iii) autoregressive model. The forecast skill does not exceed 70 years. Because we provide the most likely scenario, the forcing data does not decay towards a climatological mean, but keeps on varying. We do not provide any extreme scenario with high levels of activity because the Sun just left such an episode (called grand solar maximum), and so is very unlikely to have another. In addition, none of our ensemble of forecasts provides such a scenario above a 1% level (Abreu et al., 2009).

The standard solar forcing dataset recommended for usage is the solar reference scenario (ref) dataset which consists of historical reconstructions (1850-2014) and the most likely scenario for future solar forcing (2015-2299):

Please note that in order to account for solar cycle effects, not only radiative effects but also photochemical effects have to be taken into account! CMIP6 models that do not have interactive chemistry should in addition to adapting the irradiance changes also prescribe the CMIP6 recommended ozone forcing data set. The solar signal in this ozone dataset is consistent with the CMIP6 solar forcing.

Solar forcing datasets for sensitivity studies (not officially part of a CMIP6 MIP proposal)

Some modelling groups might additionally consider a second pi-control simulations with variable 11-year solar cycle variability included but without long-term trend. This time series still has slightly different solar cycle amplitudes and also preserves the variable phase of the solar cycle, however, the solar cycle mean activity level is held constant as compared to the reference scenario (see metadata file for more details):

This dataset is only available with daily resolution and covers the time period  1 Jan 1850 - 9 Sep 2053 (end of solar cycle 27). The dataset can be extended to cover 1000 years by multiple repetition of the solar cycle sequence 12–27.

Some modelling groups might additionally consider in future simulations the solar extreme scenario (ext), which is identical to the reference scenario for the historical period before 31 Dec 2014 but evolves towards a more extreme Maunder minimum-like activity in the future (see metadata file for more details):

Auxililiary software packages for data pre-prossessing

Routine for pre-processing solar irradiance

The SSI data recommended on this website for usage in CMIP6 is provided in high spectral resolution (1nm for 10-750nm range, 5nm for 750-5,000nm, 10nm for 5,000-10,000nm, 50nm for 10,000-100,000nm) as mean irradiance over each bin.

Many (chemistry) climate models require integrated irradiances for each spectral band, while the resolution of their radiation (and/or photolysis) schemes is typically much lower.

This package provides an (hopefully) easy to use routine to convert from one to another.

Given one of the available original input files (netcdf) downloaded from the links above, this routine creates a new netcdf-file containing the integrated SSI for all spectral bands (and all timesteps found in the input file) defined by the user plus the fraction of these bands\' irradiance (SSI_frac) compared to the TSI. As a matter of completeness, TSI is also written to the newly created file, though of course only copied from the original file.

Currently, this routine is available for MATLAB only.

Routine for generation of a EPP-NOy upper boundary condition

For consideration of Energetic Particle Precipitation (EPP) in chemistry climate models with upper lid in the mesosphere (i.e., below the EPP source region), a NOy Upper Boundary Condition (UBC) is required in order to account for the EPP indirect effect (polar winter descent of EPP-generated odd nitrogen).

This package contains a routine for generation of a flux or density NOy-UBC from geomagnetic Ap data which is included in the solar forcing datasets. Output data is written to a netcdf file as daily resolved NOy zonal mean densities (in units of cm-3) or molecular fluxes (in units of cm-2 s-1) on model-specific pressure levels and bin center latitudes. The routine is available in IDL or MATLAB.

Routine for projection of EPP ionization rates onto geographic coordinates

For consideration of solar proton and mid-energy electron precipitation, as well as galactic cosmic rays in chemistry climate models, particle-induced ion pair production rates (variable iprp, iprm, and iprg) are required. These rates, however, are provided on geomagnetic latitudes and need to be projected onto geographic coordinates.

This package contains a routine for time- and pressure level dependent  projection onto geographic coordinates, based on recommended geomagnetic reference field parameters (see metadata file for more details). It is available for MATLAB only.

 

Release notes

30 Nov 2017
Variable pi control solar forcing v3.2 now available and small bug fix in NOy UBC model (new version=1.4)

  • Daily resolution variable pi control solar forcing (v3.2) finally made available, all respective infos given in Matthes et al. (2017)
  • NOy UBC model v1.4 changes wrt v1.1 (idl) and v1.3 (matlab):ncdf output variables set to double => correct ouput time units

03 Jan 2017
New files of version 3.2 uploaded (only new filenames and attributes, no data changes)

24 Oct 2016
New version 3.2 of all forcing files and bug fix in NOy UBC model (new version=1.3)

  • Elimination of artefact in SSI forcing resulting from inprecise data processing:
    SSI for 70,000-100,000nm now ~9.3% lower which is equivalent to an integral value of ~0.00015 W/m². Because of the scaling to TSI (which is not changed, compared to v3.1 and older) of the whole spectrum, SSI for shorter wavelengths is now slightly higher, though only partly resolved given the single precision of the data as provided here.
  • Fixed scaling of iprg in variable picontrol forcing
  • All other forcing variables unchanged

8 Aug 2016
Bug fix in MATLAB code of NOy UBC model (new version=1.2), no change in forcing data

23 Jun 2016
Bug fix in NOy UBC model (new version=1.1), no change in forcing data

  • bug fix in NOy UBC model (new version=1.1):  avoids crash for frequency = fx (picontrol) when used with forcing data versions >= 3.0 (non-beta), no changes in UBC model output when used with transient forcing, no changes in forcing files!

14 Jun 2016
Updated variable pi control forcing file (version number 3.1 unchanged)

  • another bugfix in variable pi control solar forcing: previous file versions of solarforcing_picontrol_day_3.1.nc contained empty variable "scnum"; this is now filled

13 Jun 2016
Updated variable pi control forcing file (version number 3.1 unchanged)

  • bugfix in variable pi control solar forcing (solarforcing_picontrol_day_3.1.nc): core data not changed but corrected content of variable "calday" (calendar day); previous file\'s "calday" contained values of calendar year

06 Jun 2016
New files available (version number 3.1 unchanged):

  • daily resolution variable pi control solar forcing published
  • minor corrections in global attributes of some files

02 May 2016
Dataset version 3.1 available. This is an update of the final dataset that can be used for the CMIP production runs. No changes to the data itself (compared to v3.0). Changes (wrt v3.0):

  • all coordinate variables now in double precision
  • globale attribute "creation_date" now ISO 8601 compliant
  • updated MATLAB solar irradiance pre-processing routine
  • minor corrections in metadata description

26 Apr 2016
Dataset version 3.0 available. This is the final dataset that can be used for the CMIP production runs. Changes (wrt beta_v2):

  • an artefact obvious around the year 1878 has been removed. Note that this affects SSI at ALL wavelengths, TSI, and F10.7.
  • an artefact in GCR ionization rates in the ext scenario (after year 2050) has been removed.
  • bug in time_bnds in daily datasets has been removed.
  • Variable Pi control dataset added.
  • new file name convention introduced.
  • changes of variable attributes (see metadata description).
  • Changes of global attributes
  • metadata description file updated. 

18 Feb 2016

  • Metadata description file once again updated, now including a recommendation on how to project IPR data (as function of geomagnetic latitude) onto geographic coordinates

17 Feb 2016
Dataset version beta_v2 available. Changes (wrt beta_v1):

  • an artefact obvious at SSI wavelength bins > 10,000nm has been removed. Note that this affects SSI at ALL wavelengths. TSI (of transient forcing files) is not affected.
  • piControl forcing has been corrected; besides data artefact (see above), beta_v1 piControl forcing suffered from erroneous calculation routine; Note that this affects SSI at all wavelengths AND TSI (TSI of beta_v2 ~0.03 W/m^2 lower than beta_v1)
  • time_bounds added to daily resolution netcdf files. Note that all forcings should be kept constant within a time bin (i.e. day or month depending on used forcing resolution)
  • time_bounds in monthly resolution and piControl forcing files corrected (lower edge and upper edge had to be expanded for 0.5 days)
  • Meta data file updated (recommendations regarding the conversion of IPR data to chemical productions have been added)
  • MATLAB solar irradiance pre-processing routine updated (time_bounds are conveyed to custom output netcdf file)

Acknowledgements

The making of this solar forcing dataset directly benefitted from contributions by:

  • the project SOLIC, part of the research program ROMIC, funded by the Federal Ministry of Education and Research in Germany (BMBF; FKZ: 01LG1219)
  • the team on "Scenarios of Future Solar Activity for Climate Modelling" at the International Space Science Institute (ISSI, Bern),
  • the European Community’s Seventh Framework Programme (FP7 2012): Project SOLID (grant agreement no 313188),
  • the EU COST Action ES1005 (TOSCA)