RFMIP-IRF
Testing of radiation code parameterisations
Reference line-by-line (LBL) radiative transfer models are known to be highly accurate compared to laboratory experiments. However, their high spectral resolution precludes their use in climate models which require fast radiation codes. Essentially, the entire shortwave and longwave spectra need to be parameterised by a handful of wavelength intervals or quadrature points to enable acceptable run times in global climate models (GCMs).
IRF from greenhouse gases (RFMIP-IRF-GHG)
The greenhouse gas (GHG) forcing component of RFMIP seeks to characterise the climate-relevant accuracy of radiative transfer parameterisations in present day atmosphere and cloud- and aerosol-free conditions.
Sources of error can include the base climatology used (as temperature and humidity profiles, which vary between models, affect the radiative forcing), the spectral absorption profiles of greenhouse gases in the GCMs, and the parameterisation error between the GCM and LBL radiation codes. A set of specified GHG experiments will be prescribed (table 1) and climate modelling centres will be invited to provide their model output with these forcing experiments.
For detailed comparison, we are preparing a library of around 100 atmospheric profiles which are representative of present-day conditions and will provide globally-averaged radiative fluxes under a suitable weighting. For these experiments we request that modelling groups perform the following offline experiments with their native radiative transfer model in aerosol- and cloud-free skies using atmospheric profiles specified by the RFMIP team.
Experiment title RFMIP-IRF, CMIP6 label rad-irf for all below experiments.
Table 1: RFMIP-IRF-GHG experiments
Atmospheric conditions | Gas concentrations | Relevant experiments |
---|---|---|
Present day (2014) | Present day | |
Present day | Pre-industrial | Historical |
Present day | pre-industrial CO2 × 4 | abrupt4xCO2 |
Present day | “Future” | RCP8.5 at 2100 |
Present day | Pre-industrial CO2 × 0.5 | |
Present day | Pre-industrial CO2 × 1 | |
Present day | Pre-industrial CO2 × 2 | |
Present day | Pre-industrial CO2 × 3 | |
Present day | Pre-industrial CO2 × 8 | |
Present day | Pre-industrial CH4 | |
Present day | Pre-industrial N2O | |
Present day | Pre-industrial HFC | |
Present day +4K | Present day | |
Present day +4K | Present day with water vapour increased to maintain relative humidity at +4K | |
Pre-industrial | Pre-industrial | |
“Future” | “Future” | |
Present day | Last Glacial Maximum, per PMIP (Kageyama et al. (2016)) |
IRF from aerosols (RFMIP-IRF-AER)
The second largest radiative forcing in the industrial era is from aerosols. While total aerosol radiative forcing is very likely negative, the uncertainty range is larger than for greenhouse gases and thus contributes a large proportion of the uncertainty in the total radiative forcing. Aerosols both affect the radiation budget directly due to absorption and scattering of shortwave and longwave radiation (the aerosol-radiation interaction; RFari), and indirectly due to the modification of cloud properties (the aerosol-cloud interaction; RFaci).
Our focus is quantifying and better understanding the uncertainty in RFari by running simulations in cloudless skies. There are several different reasons why different climate models may show different radiation responses to aerosols. The total aerosol burden and aerosol spatial distribution depends on the model emission/concentration and transport scheme. The radiative properties of aerosols (extinction coefficient, single scattering albedo, and asymmetry factor) may vary between aerosol schemes. As with the well-mixed GHGs, the different parameterisations used in each model’s radiative transfer code also contributes to the uncertainty.
To compare aerosol radiative forcing between climate models, we will determine estimates of shortwave, clear-sky IRF from aerosols. We request instantaneous 3-hr atmospheric state variables, aerosol optical properties and clear-sky/clean-clear-sky fluxes on your model’s native grid (For a full list of variables see E3hrPt in (link to http://clipc-services.ceda.ac.uk/dreq/data/tabs02/cmvmm_RFMIP_CMIP_3_3.xlsx)) from the historical run for only the following days: Jan 1, Apr 1, July 1, Oct 1 and only the following years: 1850, 2005, 1980 and 1992 making a total of 16 days of data. Make sure to use a “double call” structure within your radiative code to ensure the fluxes with and without aerosol have identical underlying meteorology. Please see this document with more details about the protocol.