ECMWF COPERNICUS REPORT

Copernicus Atmosphere Monitoring Service

Documentation of CAMS Climate Forcing products, version 1.5

Issued by: University of Reading / Nicolas Bellouin Date: 06/02/2020 Updated: 01/02/2021

This document has been produced in the context of the Copernicus Atmosphere Monitoring Service (CAMS). The activities leading to these results have been contracted by the European Centre for Medium-Range Weather Forecasts, operator of CAMS on behalf of the European Union (Delegation Agreement signed on 11/11/2014). All information in this document is provided "as is" and no guarantee or warranty is given that the information is fit for any particular purpose. The user thereof uses the information at its sole risk and liability. For the avoidance of all doubts, the European Commission and the European Centre for Medium-Range Weather Forecasts has no liability in respect of this document, which is merely representing the authors view.

Copernicus Atmosphere Monitoring Service

Contributors

UNIVERSITY OF READING N. Bellouin W. Davies K. Shine

UNIVERSITY OF LEIPZIG J. Mülmenstädt J. Quaas

UNIVERSITY OF LEEDS P. Forster L. Regayre C. Smith

MPI METEOROLOGIE HAMBURG G. Brasseur N. Sudarchikova I. Bouarar

Documentation of CAMS Climate Forcing products, version 1.5 Page 3 of 64

Copernicus Atmosphere Monitoring Service

Table of Contents

1. Alphabetical list of products 9

2. File access and naming convention 15

3. Carbon dioxide and methane 18 3.1 Product specifications 18 3.1.1 Carbon dioxide 18 3.1.2 Methane 21 3.2 Methods 25 3.2.1 Radiative transfer calculations 25 3.2.2 Inputs to the radiative transfer code 26 3.2.3 Pre-industrial atmospheric concentrations 29 3.3 Uncertainties 29

4. Tropospheric and stratospheric ozone 29 4.1 Product specifications 29 4.1.1 Tropospheric ozone 29 4.2 Methods 33 4.2.1 Radiative transfer calculations 33 4.2.2 Inputs to the radiative transfer code 33 4.2.3 Pre-industrial atmospheric concentrations 33 4.3 Uncertainties 34

5. Speciated aerosol optical depths 34 5.1 Product specifications 34 5.2 Methods 35 5.2.1 Identification of aerosol origin 35 5.2.2 Inputs to the algorithm 37 5.3 Uncertainties 38

6. Aerosol-radiation interactions 38 6.1 Product specifications 38 6.2 Methods 39 6.2.1 Inputs to the algorithm 41 6.2.2 Pre-industrial aerosol optical depths 41 6.3 Uncertainties 41

7. Aerosol-cloud interactions 41 7.1 Product specifications 41 7.2 Methods 42

Documentation of CAMS Climate Forcing products, version 1.5 Page 4 of 64

Copernicus Atmosphere Monitoring Service

7.2.1 Inputs to the algorithm 42 7.2.2 Pre-industrial reference state 42 7.3 Uncertainties 42

8. Uncertainties 43 8.1 Data and methods 43 8.2 Uncertainty from methodological choices 44 8.2.1 Timestepping and averaging 44 8.2.2 Resolution of reanalysis data 50 8.2.3 Tropopause definition 50 8.2.4 Radiation code 52 8.3 Uncertainty from aerosol optical properties and climatology 53 8.4 Combined uncertainty 56

9. List of acronyms 58

10. References 59

11. User support and contacts 63

Documentation of CAMS Climate Forcing products, version 1.5 Page 5 of 64

Copernicus Atmosphere Monitoring Service

Introduction

Please refer to section 9 for the definition of all acronyms.

Radiative forcing (RF) measures the imbalance in the Earth’s energy budget caused by a perturbation of the climate system, for example changes in atmospheric composition driven by human activities (Myhre et al., 2013). RF is a useful predictor of globally averaged temperature change, especially when rapid adjustments of atmospheric temperature and moisture profiles are taken into account (Boucher et al., 2013). RF has therefore become a quantitative metric to compare the potential climate response to different perturbations. Increases in greenhouse gas concentrations over the industrial era exerted a positive RF, causing a gain of energy in the climate system. In contrast, concurrent changes in atmospheric aerosol concentrations are thought to exert a negative RF leading to a loss of energy. Chapter 8 of the IPCC AR5 estimates net total industrial- era RF at +2.3 W m−2 with a broad confidence interval of +1.1 to +3.3 W m−2 (Myhre et al., 2013). Estimates for key forcing agents from the IPCC AR5 are summarised in Table I.1 below. The methods used to estimate RF of different species in the IPCC AR5 are diverse: global modelling of atmospheric composition, line-by-line radiative transfer calculations, simplified calculations, or observational-based calculations. The CAMS Climate Forcing service aims at refining those estimates by providing in a consistent way the distributions, global averages, and uncertainties of the RF of key atmospheric constituent.

Table I.1 - Estimates of Radiative Forcing and Effective Radiative Forcing for changes in atmospheric composition between 1750 and 2011, taken from Tables 8.2 and 8.6 of the IPCC AR5 (Myhre et al., 2013a).

Forcing agent Radiative Forcing Effective Radiative Forcing RF (W m−2) ERF (W m−2) CO2 +1.82 ± 0.19 CH4 +0.48 ± 0.05 N2O +0.17 ± 0.03 Halocarbons +0.360 ± 0.036 Total well-mixed greenhouse gases +2.83 (2.54 to 3.12) +2.83 (2.26 to 3.40) Tropospheric ozone +0.40 (0.20 to 0.60) Stratospheric ozone −0.05 (−0.15 to +0.05) Stratospheric water vapour from +0.07 (+0.02 to +0.12) methane Aerosol-radiation interactions −0.35 (−0.85 to +0.15) −0.45 (−0.95 to +0.05) Aerosol-cloud interactions −0.45 (−1.2 to 0.0)

CAMS Climate Forcing provides RF separately for:

• carbon dioxide • methane • tropospheric ozone

Documentation of CAMS Climate Forcing products, version 1.5 Page 6 of 64

Copernicus Atmosphere Monitoring Service

• stratospheric ozone • interactions between anthropogenic aerosols and radiation • interactions between anthropogenic aerosols and clouds and their uncertainties. For aerosols, CAMS Climate Forcing also produces distributions of aerosol optical depths, distinguishing natural from anthropogenic aerosols.

This documentation describes “version 1.5” RF estimates, which have been quality-controlled for scientific use. However, version 1.5 products include several simplifying assumptions that will be improved upon in the future: needs for further improvements are clearly listed in the boxes labelled “Future improvements” in this document.

CAMS Climate Forcing estimates follow the definitions for instantaneous and stratospherically- adjusted RF given in the IPCC AR5 (Myhre et al., 2013):

• Instantaneous RF is the “instantaneous change in net (down minus up) radiative flux (shortwave plus longwave; in W m–2) due to an imposed change.” • Stratospherically-adjusted RF, hereafter referred to simply as adjusted RF, is “the change in net irradiance at the tropopause after allowing for stratospheric temperatures to readjust to radiative equilibrium, while holding surface and tropospheric temperatures and state variables such as water vapour and cloud cover fixed at the unperturbed values”.

The reference state is taken by CAMS Climate Forcing products to be year 1750, except for aerosol RF, which is defined with respect to present-day natural aerosol and scaled to pre-industrial conditions. (See section 6.) Radiative effect (RE) is also quantified for aerosols: the difference with RF is that RE is defined with respect to an atmosphere that contains no aerosols.

CAMS instantaneous and adjusted RF are quantified in terms of flux changes at the top of the atmosphere (TOA), the surface, and the climatological tropopause for carbon dioxide, methane, and ozone. Adjusted RF is not estimated for aerosol perturbations because it differs only slightly from instantaneous RF at the TOA.

CAMS Climate Forcing products are quantified by default in “all-sky” conditions, meaning that the radiative effects of clouds are included in the radiative transfer calculations. Products denoted “clear-sky” are computed by excluding clouds in the radiative transfer calculations.

IPCC AR5 further defines effective radiative forcing (ERF) as “the change in net TOA downward radiative flux after allowing for atmospheric temperatures, water vapour and clouds to adjust, but with surface temperature or a portion of surface conditions unchanged.” ERF estimates by the IPCC AR5 are given in Table I.1 above. ERF estimates by CAMS Climate Forcing are planned as part of a future evolution of RF products.

Documentation of CAMS Climate Forcing products, version 1.5 Page 7 of 64

Copernicus Atmosphere Monitoring Service

Figure I.1 illustrates the RF production chain that was used for version 1.5 RF products.

Figure I.1 - Diagram of the radiative forcing production chain (light orange), which takes inputs from the the CAMS Global Reanalysis (blue) and produces radiative forcing estimates and their uncertainties (dark orange). Green boxes indicate observational constraints. BB stands for biomass burning, AOD for aerosol optical depth, and CCN for cloud condensation nuclei. ecRad is the radiative transfer code used by the ECMWF IFS.

ecRad

The source of atmospheric composition data is the CAMS Global Reanalysis performed with the ECMWF IFS (Inness et al., 2019). That reanalysis includes assimilation of satellite retrievals of atmospheric composition, thus improving RF estimates compared to free-running models. Improvements derive directly from observational constraints on reactive gas columns and aerosol optical depths and, for ozone, vertical profiles. Data assimilation also constrains gaseous and biomass-burning aerosol emissions, leading to indirect improvements in the simulation of atmospheric concentrations. The RF production chain therefore relies in priority on variables improved by the data assimilation process (gas mixing ratios, total aerosol optical depth). However, it is not possible to solely rely on assimilated variables because other characteristics of the model affect RF directly (vertical profiles of aerosols and gases, speciation of total aerosol mass) or indirectly (cloud cover and cloud type, surface albedo, temperature and moisture profiles). In addition, parameters required by the RF estimate but not simulated by the Global Reanalysis (e.g. aerosol size distributions) are provided by ancillary datasets.

The version 1.5 production chain takes most distributions of atmospheric properties from the CAMS Global Reanalysis dataset, which currently covers 2003—2017. Concentrations of carbon dioxide and methane are taken from third-party datasets because scaling to preindustrial condition is still being trialed.

Documentation of CAMS Climate Forcing products, version 1.5 Page 8 of 64

Copernicus Atmosphere Monitoring Service

Version 1.5 computes RF using two different radiative transfer codes:

1. Streamer (Key and Schweiger 1988), a wrapper to DISORT, for aerosol-radiation and aerosol- cloud RF; 2. the IFS radiative transfer code ecRad (see section 3.2.1) (Hogan and Bozzo 2018).

1. Alphabetical list of products

As of February 2020, the CAMS Climate Forcing service publishes 117 different products: - 24 for carbon dioxide; - 24 for methane; - 24 for tropospheric ozone; - 24 for stratospheric ozone; - 20 for aerosol-radiation interactions, including speciated aerosol optical depths and radiative effects; - 1 for aerosol-cloud interactions.

Table 1.1 lists the main characteristics of the CAMS Climate Forcing products and the section in this document where more detailed information can be found about each product.

Table 1.1 - Alphabetical list by variable name of CAMS Climate Forcing products, as of February 2020. All products cover the period 2003—2017. # Variable name Forcing Description RE/RF type and Section agent spectrum 1 abs550anth Aerosols Anthropogenic N/A 5 absorption AOD 2 abs550dust Aerosols Mineral dust N/A 5 absorption AOD 3 abs550landnat Aerosols Land-based N/A 5 natural absorption AOD 4 abs550marine Aerosols Marine N/A 5 absorption AOD 5 arf_ch4_srf_lw_cs Methane RF at surface Adjusted, LW, clear-sky 3 6 arf_ch4_srf_lw Methane RF at surface Adjusted, LW 3 7 arf_ch4_srf_sw_cs Methane RF at surface Adjusted, SW, clear-sky 3 8 arf_ch4_srf_sw Methane RF at surface Adjusted, SW 3 9 arf_ch4_srf_lw_cs Methane RF at TOA Adjusted, LW, clear-sky 3 10 arf_ch4_srf_lw Methane RF at TOA Adjusted, LW 3

Documentation of CAMS Climate Forcing products, version 1.5 Page 9 of 64

Copernicus Atmosphere Monitoring Service

11 arf_ch4_srf_sw_cs Methane RF at TOA Adjusted, SW, clear-sky 3 12 arf_ch4_srf_sw Methane RF at TOA Adjusted, SW 3 13 arf_ch4_trop_lw_cs Methane RF at Adjusted, LW, clear-sky 3 tropopause 14 arf_ch4_trop_lw Methane RF at Adjusted, LW 3 tropopause 15 arf_ch4_trop_sw_cs Methane RF at Adjusted, SW, clear-sky 3 tropopause 16 arf_ch4_trop_sw Methane RF at Adjusted, SW 3 tropopause 17 arf_co2_srf_lw_cs Carbon RF at surface Adjusted, LW, clear-sky 3 dioxide 18 arf_co2_srf_lw Carbon RF at surface Adjusted, LW 3 dioxide 19 arf_co2_srf_sw_cs Carbon RF at surface Adjusted, SW, clear-sky 3 dioxide 20 arf_co2_srf_sw Carbon RF at surface Adjusted, SW 3 dioxide 21 arf_co2_srf_lw_cs Carbon RF at TOA Adjusted, LW, clear-sky 3 dioxide 22 arf_co2_srf_lw Carbon RF at TOA Adjusted, LW 3 dioxide 23 arf_co2_srf_sw_cs Carbon RF at TOA Adjusted, SW, clear-sky 3 dioxide 24 arf_co2_srf_sw Carbon RF at TOA Adjusted, SW 3 dioxide 25 arf_co2_trop_lw_cs Carbon RF at Adjusted, LW, clear-sky 3 dioxide tropopause 26 arf_co2_trop_lw Carbon RF at Adjusted, LW 3 dioxide tropopause 27 arf_co2_trop_sw_cs Carbon RF at Adjusted, SW, clear-sky 3 dioxide tropopause 28 arf_co2_trop_sw Carbon RF at Adjusted, SW 3 dioxide tropopause 29 arf_strato3_srf_lw_cs Stratospheri RF at surface Adjusted, LW, clear-sky 4 c ozone 30 arf_strato3_srf_lw Stratospheri RF at surface Adjusted, LW 4 c ozone 31 arf_strato3_srf_sw_cs Stratospheri RF at surface Adjusted, SW, clear-sky 4 c ozone 32 arf_strato3_srf_sw Stratospheri RF at surface Adjusted, SW 4 c ozone 33 arf_strato3_srf_lw_cs Stratospheri RF at TOA Adjusted, LW, clear-sky 4

Documentation of CAMS Climate Forcing products, version 1.5 Page 10 of 64

Copernicus Atmosphere Monitoring Service

c ozone 34 arf_strato3_srf_lw Stratospheri RF at TOA Adjusted, LW 4 c ozone 35 arf_strato3_srf_sw_cs Stratospheri RF at TOA Adjusted, SW, clear-sky 4 c ozone 36 arf_strato3_srf_sw Stratospheri RF at TOA Adjusted, SW 4 c ozone 37 arf_strato3_trop_lw_cs Stratospheri RF at Adjusted, LW, clear-sky 4 c ozone tropopause 38 arf_strato3_trop_lw Stratospheri RF at Adjusted, LW 4 c ozone tropopause 39 arf_strato3_trop_sw_cs Stratospheri RF at Adjusted, SW, clear-sky 4 c ozone tropopause 40 arf_strato3_trop_sw Stratospheri RF at Adjusted, SW 4 c ozone tropopause 41 arf_tropo3_srf_lw_cs Tropospheri RF at surface Adjusted, LW, clear-sky 4 c ozone 42 arf_tropo3_srf_lw Tropospheri RF at surface Adjusted, LW 4 c ozone 43 arf_tropo3_srf_sw_cs Tropospheri RF at surface Adjusted, SW, clear-sky 4 c ozone 44 arf_tropo3_srf_sw Tropospheri RF at surface Adjusted, SW 4 c ozone 45 arf_tropo3_srf_lw_cs Tropospheri RF at TOA Adjusted, LW, clear-sky 4 c ozone 46 arf_tropo3_srf_lw Tropospheri RF at TOA Adjusted, LW 4 c ozone 47 arf_tropo3_srf_sw_cs Tropospheri RF at TOA Adjusted, SW, clear-sky 4 c ozone 48 arf_tropo3_srf_sw Tropospheri RF at TOA Adjusted, SW 4 c ozone 49 arf_tropo3_trop_lw_cs Tropospheri RF at Adjusted, LW, clear-sky 4 c ozone tropopause 50 arf_tropo3_trop_lw Tropospheri RF at Adjusted, LW 4 c ozone tropopause 51 arf_tropo3_trop_sw_cs Tropospheri RF at Adjusted, SW, clear-sky 4 c ozone tropopause 52 arf_tropo3_trop_sw Tropospheri RF at Adjusted, SW 4 c ozone tropopause 53 od550aer Aerosols Total AOD N/A 5 54 od550anth Aerosols Anthropogenic N/A 5 AOD 55 od550dust Aerosols Mineral dust N/A 5

Documentation of CAMS Climate Forcing products, version 1.5 Page 11 of 64

Copernicus Atmosphere Monitoring Service

AOD 56 od550landnat Aerosols Land-based N/A 5 natural AOD 57 od550marine Aerosols Marine AOD N/A 5 58 re_ari_anth_srf_sw_cs Aerosols Clear-sky Instant, SW 6 anthropogenic aerosol- radiation RE at surface 59 re_ari_anth_toa_sw_cs Aerosols Clear-sky Instant, SW 6 anthropogenic aerosol- radiation RE at TOA 60 re_ari_dust_srf_sw_cs Aerosols Clear-sky Instant, SW 6 mineral dust aerosol- radiation RE at surface 61 re_ari_dust_toa_sw_cs Aerosols Clear-sky Instant, SW 6 mineral dust aerosol- radiation RE at TOA 62 re_ari_landnat_srf_sw_ Aerosols Clear-sky land- Instant, SW 6 cs based natural aerosol- radiation RE at surface 63 re_ari_landnat_toa_sw Aerosols Clear-sky land- Instant, SW 6 _cs based natural aerosol- radiation RE at TOA 64 re_ari_marine_srf_sw_ Aerosols Clear-sky Instant, SW 6 cs marine aerosol- radiation RE at surface 65 re_ari_marine_toa_sw_ Aerosols Clear-sky Instant, SW 6 cs marine aerosol- radiation RE at TOA

Documentation of CAMS Climate Forcing products, version 1.5 Page 12 of 64

Copernicus Atmosphere Monitoring Service

66 rf_aci_toa_sw Aerosols Aerosol-cloud Instanta, SW 7 RF at TOA 67 rf_ari_srf_sw_cs Aerosols Clear-sky Instanta, SW 6 aerosol- radiation RF at surface 68 rf_ari_toa_sw_cs Aerosols Clear-sky Instanta, SW 6 aerosol- radiation RF at TOA 69 rf_ari_toa_sw Aerosols Aerosol- Instanta, SW 6 radiation RF at TOA 70 rf_ch4_srf_lw_cs Methane RF at surface Instant, LW, clear-sky 3 71 rf_ch4_srf_lw Methane RF at surface Instant, LW 3 72 rf_ch4_srf_sw_cs Methane RF at surface Instant, SW, clear-sky 3 73 rf_ch4_srf_sw Methane RF at surface Instant, SW 3 74 rf_ch4_srf_lw_cs Methane RF at TOA Instant, LW, clear-sky 3 75 rf_ch4_srf_lw Methane RF at TOA Instant, LW 3 76 rf_ch4_srf_sw_cs Methane RF at TOA Instant, SW, clear-sky 3 77 rf_ch4_srf_sw Methane RF at TOA Instant, SW 3 78 rf_ch4_trop_lw_cs Methane RF at Instant, LW, clear-sky 3 tropopause 79 rf_ch4_trop_lw Methane RF at Instant, LW 3 tropopause 80 rf_ch4_trop_sw_cs Methane RF at Instant, SW, clear-sky 3 tropopause 81 rf_ch4_trop_sw Methane RF at Instant, SW 3 tropopause 82 rf_co2_srf_lw_cs Carbon RF at surface Instant, LW, clear-sky 3 dioxide 83 rf_co2_srf_lw Carbon RF at surface Instant, LW 3 dioxide 84 rf_co2_srf_sw_cs Carbon RF at surface Instant, SW, clear-sky 3 dioxide 85 rf_co2_srf_sw Carbon RF at surface Instant, SW 3 dioxide 86 rf_co2_srf_lw_cs Carbon RF at TOA Instant, LW, clear-sky 3 dioxide 87 rf_co2_srf_lw Carbon RF at TOA Instant, LW 3 dioxide 88 rf_co2_srf_sw_cs Carbon RF at TOA Instant, SW, clear-sky 3 dioxide

Documentation of CAMS Climate Forcing products, version 1.5 Page 13 of 64

Copernicus Atmosphere Monitoring Service

89 rf_co2_srf_sw Carbon RF at TOA Instant, SW 3 dioxide 90 rf_co2_trop_lw_cs Carbon RF at Instant, LW, clear-sky 3 dioxide tropopause 91 rf_co2_trop_lw Carbon RF at Instant, LW 3 dioxide tropopause 92 rf_co2_trop_sw_cs Carbon RF at Instant, SW, clear-sky 3 dioxide tropopause 93 rf_co2_trop_sw Carbon RF at Instant, SW 3 dioxide tropopause 94 rf_strato3_srf_lw_cs Stratospheri RF at surface Instant, LW, clear-sky 4 c ozone 95 rf_strato3_srf_lw Stratospheri RF at surface Instant, LW 4 c ozone 96 rf_strato3_srf_sw_cs Stratospheri RF at surface Instant, SW, clear-sky 4 c ozone 97 rf_strato3_srf_sw Stratospheri RF at surface Instant, SW 4 c ozone 98 rf_strato3_srf_lw_cs Stratospheri RF at TOA Instant, LW, clear-sky 4 c ozone 99 rf_strato3_srf_lw Stratospheri RF at TOA Instant, LW 4 c ozone 100 rf_strato3_srf_sw_cs Stratospheri RF at TOA Instant, SW, clear-sky 4 c ozone 101 rf_strato3_srf_sw Stratospheri RF at TOA Instant, SW 4 c ozone 102 rf_strato3_trop_lw_cs Stratospheri RF at Instant, LW, clear-sky 4 c ozone tropopause 103 rf_strato3_trop_lw Stratospheri RF at Instant, LW 4 c ozone tropopause 104 rf_strato3_trop_sw_cs Stratospheri RF at Instant, SW, clear-sky 4 c ozone tropopause 105 rf_strato3_trop_sw Stratospheri RF at Instant, SW 4 c ozone tropopause 106 rf_tropo3_srf_lw_cs Tropospheri RF at surface Instant, LW, clear-sky 4 c ozone 107 rf_tropo3_srf_lw Tropospheri RF at surface Instant, LW 4 c ozone 108 rf_tropo3_srf_sw_cs Tropospheri RF at surface Instant, SW, clear-sky 4 c ozone 109 rf_tropo3_srf_sw Tropospheri RF at surface Instant, SW 4 c ozone 110 rf_tropo3_srf_lw_cs Tropospheri RF at TOA Instant, LW, clear-sky 4

Documentation of CAMS Climate Forcing products, version 1.5 Page 14 of 64

Copernicus Atmosphere Monitoring Service

c ozone 111 rf_tropo3_srf_lw Tropospheri RF at TOA Instant, LW 4 c ozone 112 rf_tropo3_srf_sw_cs Tropospheri RF at TOA Instant, SW, clear-sky 4 c ozone 113 rf_tropo3_srf_sw Tropospheri RF at TOA Instant, SW 4 c ozone 114 rf_tropo3_trop_lw_cs Tropospheri RF at Instant, LW, clear-sky 4 c ozone tropopause 115 rf_tropo3_trop_lw Tropospheri RF at Instant, LW 4 c ozone tropopause 116 rf_tropo3_trop_sw_cs Tropospheri RF at Instant, SW, clear-sky 4 c ozone tropopause 117 rf_tropo3_trop_sw Tropospheri RF at Instant, SW 4 c ozone tropopause

Notes: a. Stratospheric adjustment is negligible for tropospheric aerosol perturbations, so for aerosols instantaneous and adjusted RFs are equal.

2. File access and naming convention

Files are currently stored on the CAMS Atmospheric Data Store (ADS) at https://ads.atmosphere.copernicus.eu.

The dataset should be cited as: Bellouin, N., Davies, W., Shine, K.P., Quaas, J., Mülmenstädt, J., Forster, P.M., Smith, C., Lee, L., Regayre, L., Brasseur, G., Sudarchikova, N., Bouarar, I., Boucher, O., and Myhre, G.: Supplemental Data of Radiative forcing of climate change from the Copernicus reanalysis of atmospheric composition: ECMWF Data catalogue, doi:10.24380/ads.1hj3y896, 2020.

As of February 2020, there are 117 different variables (“products”) available. (See Section 1 for full list.) Products are made from the CAMS Global Reanalysis of atmospheric composition (Inness et al., 2019), and therefore cover the period 2003—2017 that the reanalysis currently provides. Spatial and temporal resolutions depend on the variable. There is one file per variable and per month. For variables given at monthly temporal resolution, there is only one distribution per file. For variables given at daily temporal resolution, there are as many distributions in a monthly file as there are days in that month.

Files are in netCDF format and follow the Climate and Forecast metadata convention version 1.6 (CF-1.6). Version 1.5 files are in netCDF4 format, except for aerosol files, which are in netCDF3 format.

Documentation of CAMS Climate Forcing products, version 1.5 Page 15 of 64

Copernicus Atmosphere Monitoring Service

Future improvement: All version 2 products will be in netCDF4 format.

The netCDF header for methane LW instantaneous RF at the surface for the month of December 2012 is given below as an example for the file metadata given in CAMS Climate Forcing products. Aspects that may be different for other products are highlighted in red: refer to the corresponding note for details. netcdf z_cams_l_uor_201212_v1.5_rf_ch4_srf_lwa { dimensions: latitude = 61b ; longitude = 120b ; time = UNLIMITED ; // (1c currently) variables: float latitude(latitude) ; latitude:units = "degrees_north" ; latitude:long_name = "latitude" ; latitude:standard_name = "latitude" ; latitude:point_spacing = "even" ; float longitude(longitude) ; longitude:units = "degrees_east" ; longitude:long_name = "longitude" ; longitude:standard_name = "longitude" ; longitude:modulo = "" ; longitude:point_spacing = "even" ; float rf_ch4_srf_lwa(time, latitude, longitude) ; rf_ch4_srf_lwa:units = "W m-2d" ; rf_ch4_srf_lwa:long_name = "Methane_lw_instantaneous_radiative_forcing_at_surfacee" ; rf_ch4_srf_lwa:standard_name = "surface_instantaneous_longwave_forcingf" ; rf_ch4_srf_lw:species = "ch4g" ; float time(time) ; time:units = "hours since 1900h-01-01 00:00:0.0" ; time:long_name = "time" ; time:standard_name = "time" ; time:calendar = "gregorian" ; // global attributes: :Conventions = "CF-1.6" ; :title = "Copernicus Atmosphere Monitoring Service 74 Climate Forcings" ; :description = "CAMS 74 production chain : Monthly mean Methane radiative forcing for year 2012. Will Davies UoR 2019, [email protected]" ; i :references = "Update to Bellouin et al., doi:10.5194/acp-13-2045-2013, 2013" ; j :source = "model-generated, CAMS74, version 2 . $Rev: 95 $ . 60" k ;

Documentation of CAMS Climate Forcing products, version 1.5 Page 16 of 64

Copernicus Atmosphere Monitoring Service

:institution = "Department of Meteorology, University of Reading" l ; :history = "Thu Jan 30 13:16:12 2020: ncks -d time,11 /storage/silver/cams74/gn907779/gn907779- metcloud/outputs/Rev95ins/cams74_rf_ch4_srf_lw_2012.nc /storage/silver/cams74/pv904464/pv904464- glusterfs/version1.5/z_cams_l_uor_201212_v1.5_rf_ch4_srf_lw.nc\nwri teMonthlyMean.py -o chi -y 2012 run on 2020-01-07 03:02:46" m ; :NCO = "netCDF Operators version 4.7.5 (Homepage = http://nco.sf.net, Code = http://github.com/nco/nco)" n ;

}

Notes: a. Each product is associated with a different netCDF variable name; b. The number of grid boxes depends on the product: 61x120 for carbon dioxide, methane, and ozone products (3°x3° resolution), 360x181 for aerosol products (1°x1° resolution); c. Aerosol-radiation products are at daily resolution, giving from 28 to 31 distributions per file. Other products are at monthly resolution, giving 1 distribution per file; d. Radiative effects and forcings are in W m−2. Absorption and extinction AODs are dimensionless. e. The long name is a free-form name that tries to give as much information as possible on the variable’s meaning. f. The standard name is taken, as much as possible, from the CMIP standard output table available at http://cmip-pcmdi.llnl.gov/cmip5/data_description.html . When a new name needed to be created, it follows the CMIP naming convention. g. Variable attribute species is used to identify the forcing agent; h. The reference year varies depending on product: 1900 for carbon dioxide, methane, and ozone RF products, current year for aerosol products; i. The short description and contact email address depend on the product. See also section 11 for contact information for all products; j. A peer-reviewed publication describing the products is available at https://doi.org/10.5194/essd-12-1649-2020; k. Aerosol products are based on the MACC production chain. See sections 3 to 7 for details about each product; l. The CAMS Climate Forcing service is delivered by three main partners: The University of Reading (UK), University of Leeds, and Universität Leipzig (Germany). But the service is led by the University of Reading, which is the main contact for users (see section 11); m. The history attribute lists the commands issued to generate the products and the corresponding time stamps; n. Products have undergone metadata post-processing using the netCDF operators (http://nco.sourceforge.net/nco.html).

Future improvement: - Latitude, longitude, and time boundaries will be added to clarify gridbox dimensions

Documentation of CAMS Climate Forcing products, version 1.5 Page 17 of 64

Copernicus Atmosphere Monitoring Service

and time averaging periods.

3. Carbon dioxide and methane 3.1 Product specifications 3.1.1 Carbon dioxide

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_srf_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_longwave_forcing_assuming_clear_sky Long name Carbon Dioxide_lw_adjusted_radiative_forcing_at_surface_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_srf_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_longwave_forcing Long name Carbon Dioxide_lw_adjusted_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_srf_sw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_shortwave_forcing_assuming_clear_sky Long name Carbon Dioxide_sw_adjusted_radiative_forcing_at_surface_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_srf_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_shortwave_forcing Long name Carbon Dioxide_sw_adjusted_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_toa_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_adjusted_longwave_forcing_assuming_clear_sky Long name Carbon Dioxide_lw_adjusted_radiative_forcing_at_toa_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_toa_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_adjusted_longwave_forcing Long name Carbon Dioxide_lw_adjusted_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units

Documentation of CAMS Climate Forcing products, version 1.5 Page 18 of 64

Copernicus Atmosphere Monitoring Service

arf_co2_toa_sw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_adjusted_shortwave_forcing_assuming_clear_sky Long name Carbon Dioxide_sw_adjusted_radiative_forcing_at_toa_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_toa_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_adjusted_shortwave_forcing Long name Carbon Dioxide_sw_adjusted_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_trop_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard tropopause_adjusted_longwave_forcing_assuming_clear_sky name Long name Carbon Dioxide_lw_adjusted_radiative_forcing_at_tropopause_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_trop_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_adjusted_longwave_forcing Long name Carbon Dioxide_lw_adjusted_radiative_forcing_at_tropopause

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_trop_sw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard tropopause_adjusted_shortwave_forcing_assuming_clear_sky name Long name Carbon Dioxide_sw_adjusted_radiative_forcing_at_tropopause_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_co2_trop_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_adjusted_shortwave_forcing Long name Carbon Dioxide_sw_adjusted_radiative_forcing_at_tropopause

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_srf_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard surface_instantaneous_longwave_forcing_assuming_clear_sky name Long name Carbon Dioxide_lw_instantaneous_radiative_forcing_at_surface_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_srf_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_longwave_forcing

Documentation of CAMS Climate Forcing products, version 1.5 Page 19 of 64

Copernicus Atmosphere Monitoring Service

Long name Carbon Dioxide_lw_instantaneous_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_srf_sw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard surface_instantaneous_shortwave_forcing_assuming_clear_sky name Long name Carbon Dioxide_sw_instantaneous_radiative_forcing_at_surface_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_srf_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_shortwave_forcing Long name Carbon Dioxide_sw_instantaneous_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_toa_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_longwave_forcing_assuming_clear_sky Long name Carbon Dioxide_lw_instantaneous_radiative_forcing_at_toa_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_toa_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_longwave_forcing Long name Carbon Dioxide_lw_instantaneous_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_toa_sw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_forcing_assuming_clear_sky Long name Carbon Dioxide_sw_instantaneous_radiative_forcing_at_toa_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_toa_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_forcing Long name Carbon Dioxide_sw_instantaneous_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_trop_lw_c 3° x 3° Monthly (2003 – 2017) W m−2 s CMIP Standard tropopause_instantaneous_longwave_forcing_assuming_clear_sky name Long name Carbon Dioxide_lw_instantaneous_radiative_forcing_at_tropopause_assuming_clear_s

Documentation of CAMS Climate Forcing products, version 1.5 Page 20 of 64

Copernicus Atmosphere Monitoring Service

ky

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_trop_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_instantaneous_longwave_forcing Long name Carbon Dioxide_lw_instantaneous_radiative_forcing_at_tropopause

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_trop_sw_c 3° x 3° Monthly (2003 – 2017) W m−2 s CMIP Standard tropopause_instantaneous_shortwave_forcing_assuming_clear_sky name Long name Carbon Dioxide_sw_instantaneous_radiative_forcing_at_tropopause_assuming_clear_ sky

Variable Name Spatial Resolution Temporal Resolution Units rf_co2_trop_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_instantaneous_shortwave_forcing Long name Carbon Dioxide_sw_instantaneous_radiative_forcing_at_tropopause

3.1.2 Methane

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_srf_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_longwave_forcing_assuming_clear_sky Long name Methane_lw_adjusted_radiative_forcing_at_surface_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_srf_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_longwave_forcing Long name Methane_lw_adjusted_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_srf_sw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_shortwave_forcing_assuming_clear_sky Long name Methane_sw_adjusted_radiative_forcing_at_surface_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_srf_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_shortwave_forcing Long name Methane_sw_adjusted_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units

Documentation of CAMS Climate Forcing products, version 1.5 Page 21 of 64

Copernicus Atmosphere Monitoring Service

arf_ch4_toa_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_adjusted_longwave_forcing_assuming_clear_sky Long name Methane_lw_adjusted_radiative_forcing_at_toa_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_toa_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_adjusted_longwave_forcing Long name Methane_lw_adjusted_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_toa_sw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_adjusted_shortwave_forcing_assuming_clear_sky Long name Methane_sw_adjusted_radiative_forcing_at_toa_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_toa_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_adjusted_shortwave_forcing Long name Methane_sw_adjusted_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_trop_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard tropopause_adjusted_longwave_forcing_assuming_clear_sky name Long name Methane_lw_adjusted_radiative_forcing_at_tropopause_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_trop_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_adjusted_longwave_forcing Long name Methane_lw_adjusted_radiative_forcing_at_tropopause

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_trop_sw_c 3° x 3° Monthly (2003 – 2017) W m−2 s CMIP Standard tropopause_adjusted_shortwave_forcing_assuming_clear_sky name Long name Methane_sw_adjusted_radiative_forcing_at_tropopause_assuming_clear_sk y

Variable Name Spatial Resolution Temporal Resolution Units arf_ch4_trop_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_adjusted_shortwave_forcing Long name Methane_sw_adjusted_radiative_forcing_at_tropopause

Documentation of CAMS Climate Forcing products, version 1.5 Page 22 of 64

Copernicus Atmosphere Monitoring Service

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_srf_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard surface_instantaneous_longwave_forcing_assuming_clear_sky name Long name Methane_lw_instantaneous_radiative_forcing_at_surface_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_srf_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_longwave_forcing Long name Methane_lw_instantaneous_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_srf_sw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard surface_instantaneous_shortwave_forcing_assuming_clear_sky name Long name Methane_sw_instantaneous_radiative_forcing_at_surface_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_srf_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_shortwave_forcing Long name Methane_sw_instantaneous_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_toa_lw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_longwave_forcing_assuming_clear_sky Long name Methane_lw_instantaneous_radiative_forcing_at_toa_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_toa_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_longwave_forcing Long name Methane_lw_instantaneous_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_toa_sw_cs 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard toa_instantaneous_shortwave_forcing_assuming_clear_sky name Long name Methane_sw_instantaneous_radiative_forcing_at_toa_assuming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_toa_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_forcing Long name Methane_sw_instantaneous_radiative_forcing_at_toa

Documentation of CAMS Climate Forcing products, version 1.5 Page 23 of 64

Copernicus Atmosphere Monitoring Service

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_trop_lw_c 3° x 3° Monthly (2003 – 2017) W m−2 s CMIP Standard tropopause_instantaneous_longwave_forcing_assuming_clear_sky name Long name Methane_lw_instantaneous_radiative_forcing_at_tropopause_assuming_clear _sky

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_trop_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_instantaneous_longwave_forcing Long name Methane_lw_instantaneous_radiative_forcing_at_tropopause

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_trop_sw_c 3° x 3° Monthly (2003 – 2017) W m−2 s CMIP Standard tropopause_instantaneous_shortwave_forcing_assuming_clear_sky name Long name Methane_sw_instantaneous_radiative_forcing_at_tropopause_assuming_clear _sky

Variable Name Spatial Resolution Temporal Resolution Units rf_ch4_trop_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_instantaneous_shortwave_forcing Long name Methane_sw_instantaneous_radiative_forcing_at_tropopause

Documentation of CAMS Climate Forcing products, version 1.5 Page 24 of 64

Copernicus Atmosphere Monitoring Service

3.2 Methods 3.2.1 Radiative transfer calculations

The radiative transfer model used is a standalone version of the ECMWF IFS ecRad model (Hogan and Bozzo, 2018), configured like in cycle 43r1. Gaseous optical properties are computed by RRTMG (Mlawer et al., 1997) while cloud and aerosol optical properties are computed by schemes developed at ECMWF. The LW and SW solvers are based on McICA (Pincus et al., 2003). Cloud overlap is assumed to be exponential-random. Scattering by clouds and aerosols in the LW spectrum is neglected. The calculations of radiative fluxes by the CAMS Climate Forcing radiative transfer code have been compared against observational estimates (Kato et al., 2013) and found to be accurate within a few percent (Table 3.1). Fluxes with sizeable aerosol contributions, such as surface and clear-sky fluxes, are less accurate because pre-industrial aerosols were used in those benchmark tests.

Table 3.1 – Comparison of globally- and annually-averaged radiative fluxes computed by the standalone ecRad radiative transfer code against observational estimates by Kato et al. (2013).

Radiative flux Kato et al. (2013) ECMWF IFS radiative Difference (W m−2) transfer calculations (W m−2, %) (W m−2) TOA incoming solar 340 341.4 1.5 (0.5%)

All sky

TOA outgoing SW 99 to 100 93.9 5.1 (5.1%)

TOA outgoing LW 237 to 240 239.1 0.0 (0.0%)

Surf SW downward 187 197.0 10.0 (5.3%)

Surf SW upward 23 to 24 25.8 1.8 (7.5%)

Surf LW downward 342 to 344 339.6 2.4 (0.7)

Surf LW upward 398 395.3 2.7 (0.7)

Clear sky (cloud free)

TOA outgoing SW 53 51.0 2.0 (3.8%)

TOA outgoing LW 264 to 266 264.0 0.0 (0.0%)

Surf SW downward 242 to 243 248.6 6.6 (2.7%)

Surf SW upward 29 to 30 31.0 1.0 (3.3%)

Documentation of CAMS Climate Forcing products, version 1.5 Page 25 of 64

Copernicus Atmosphere Monitoring Service

Surf LW downward 314 313.6 0.4 (0.1%)

Surf LW upward 397 to 398 394.9 2.1 (0.5%)

Radiative fluxes are calculated at 61 model half-levels but for RF purposes, three levels are retained: surface, TOA, and tropopause. The tropopause is currently taken from the thermal definition set by the World Meteorological Organization, based on lapse rate (see section 4.2.2).

Adjustment of radiative fluxes to account for changes in stratospheric temperatures uses the fixed- dynamical heating method (Ramaswamy et al. 2001). The convergence criterion is that the sum of the SW+LW radiative fluxes at the tropopause equals that at the top of the atmosphere.

Methane RF is given in the LW and SW parts of the spectrum, although it is now known that ecRad RRTMG – and most other radiative transfer codes – are unable to properly handle methane absorption bands in the SW part of the spectrum. Therefore, the CAMS products underestimate methane RF in the SW spectrum. The SW contribution can be as large as 15% (Etminan et al., 2016). Improving the estimates involve a complex redefinition of wavelength intervals in ecRad, beyond the scope of the CAMS Climate Forcing project.

3.2.2 Inputs to the radiative transfer code

The radiative transfer code is run on distributions of atmospheric variables simulated by the CAMS Global Reanalysis (Table 3.2) and taken from ECMWF MARS and corresponding to data available at https://ads.atmosphere.copernicus.eu/cdsapp#!/dataset/cams-global-reanalysis- eac4?tab=overview. The distributions are used as the mean of 4 time steps (0Z, 6Z, 12Z, and 18Z) for the reanalysis dated 0Z daily. The distributions are used at the degraded horizontal resolution of 3.0° by 3.0°, down from the original 80-km resolution, to reduce computational cost. That decrease in resolution causes negligible (third decimal place) changes in globally-averaged RF.

Table 3.2 – Variables taken from the CAMS Global Reanalysis and used as input to the radiation scheme. Variable name MARS Parameter Number Levels Fraction of Cloud Cover 248.128 60 levels Forecast Albedo 243.128 Surface only GEMS ozone 203.210 60 levels Logarithm of surface pressure 152.128 Surface only Specific cloud ice water content 247.128 60 levels Specific cloud liquid water content 246.128 60 levels Skin temperature 235.128 Surface only Snow depth 141.128 Surface only Specific humidity 133.128 60 levels

Documentation of CAMS Climate Forcing products, version 1.5 Page 26 of 64

Copernicus Atmosphere Monitoring Service

Pressure on 61 half levels is calculated from the logarithm of the surface pressure following the definition and coefficients of the 60-level IFS configuration, as described at http://www.ecmwf.int/en/forecasts/documentation-and-support/60-model-levels. The radiative transfer code also requires temperature on 61 half levels, which are computed from temperature on 60 model levels using pressure-weighted linear interpolation following the equation:

푃 − 푃1 푃 − 푃1 ℎ 푓 1 1 ℎ 푓 2 2 푇ℎ 푃ℎ = (1 − 2 1) 푃푓 푇푓 + ( 2 1) 푃푓 푇푓 푃푓 − 푃푓 푃푓 − 푃푓

1 1 where 푇ℎ and 푃ℎ are temperature and pressure at the half-level, 푇푓 and 푃푓 are the pressure and 2 2 temperature for the model level above, and 푇푓 and 푃푓 are the pressure and temperature for the level below. Temperature at the top-most half-level is obtained by linear extrapolation.

LW surface emissivity is computed as done in the ECMWF IFS based on simulated snow and sand covers. Its value is computed by averaging the emissivity of four surface tiles in proportion to their coverage of each gridbox. Surface emissivities used in that calculation are listed in Table 3.3. Outside the window region, the value for sea is used.

Table 3.3 – Values of LW surface emissivity used in radiative transfer calculations. Surface type LW emissivity Land 0.96 Sand 0.93 Sea 0.99 Snow 0.98

RF is integrated diurnally by integrating over 6 solar zenith angles, computed as a function of local latitude and day of the year and symmetrically distributed around local noon.

Daily-averaged concentrations of carbon dioxide and methane are taken from the data-assimilated, three-dimensional distributions obtained by CAMS Greenhouse Gases Fluxes (Chevallier et al. (2005) and Bergamashi et al. (2013) for carbon dioxide and methane, respectively, with updates to both documented at atmosphere.copernicus.eu). The inversion product versions used are v18r2 for carbon dioxide and v17r1 for methane. Figure 3.2 shows time series of globally, monthly, total- column averages of carbon dioxide and methane concentrations. The annually averaged carbon dioxide concentration in 2017 was 404 ppm, up 8% from 374 ppm in 2003. For methane, the concentration for year 2017 was 1804 ppb, up 4.3% from 1730 ppb in 2003. Figure 3.2 also shows equivalent time series for background surface measurements by the NOAA Earth System Research Laboratory (downloaded from https://www.esrl.noaa.gov/gmd/ccgg/trends/global.html#global_data) for carbon dioxide and by the Advanced Global Atmospheric Gases Experiment (AGAGE, downloaded from https://agage.mit.edu/data/agage-data) for methane. Surface measurements are generally higher than the column averages, especially for methane that decreases with height by oxidation. Nitrous oxide is set to its preindustrial mixing ratio of 270 ppb (Myhre et al., 2013a).

Documentation of CAMS Climate Forcing products, version 1.5 Page 27 of 64

Copernicus Atmosphere Monitoring Service

Figure 3.2 – Time series of globally and monthly averaged concentrations of (top) carbon dioxide, in ppm, and (bottom) methane, in ppb, over the period 2003-2017. Bold lines show mass-weighted total column averages for the CAMS Greenhouse Flux Inversion products. Thin lines show background surface measurements from NOAA’s Earth System Research Laboratory for carbon dioxide and the Advanced Global Atmospheric Gases Experiment for methane, respectively.

For the moment, the radiative transfer calculations for carbon dioxide and methane RF use fixed values for the effective radius of cloud liquid droplets and ice crystals, at 10 and 50 μm, respectively. Optically active gases are set to their PI values, see section 3.2.3.

Future improvement: Radiative transfer calculations for version 2 products will use consistent representations of aerosols and clouds for all forcing agents.

The 3D radiative transfer capabilities of the code are not used, and there are no plans to use them in the future.

Documentation of CAMS Climate Forcing products, version 1.5 Page 28 of 64

Copernicus Atmosphere Monitoring Service

3.2.3 Pre-industrial atmospheric concentrations

The three-dimensional distributions of carbon dioxide and methane derived for present-day (PD) strongly benefit from data assimilation of surface measurements and satellite retrievals, which partly offset the biases of the chemistry model. That, however, creates the difficulty that estimating PI concentrations by running the chemistry model with PI emissions would be biased with respect to the data-assimilated, present-day distributions. Instead, daily PI mixing ratios of carbon dioxide and methane are scaled from daily CAMS Greenhouse Gas Flux mixing ratios in each grid box and at each model level following:

퐴푅5 ⟨[푋]푃퐼,푠푢푟푓푎푐푒⟩ [푋]푃퐼 = [푋]푃퐷 ∙ , ⟨[푋]푃퐷,푠푢푟푓푎푐푒⟩ where [푋] denotes the mixing ratio of carbon dioxide or methane, and angle brackets denote annual averaging. All variables are taken from the CAMS Greenhouse Gas Flux inversions, except for 퐴푅5 PI surface mixing ratios, ⟨[푋]푃퐼,푠푢푟푓푎푐푒⟩, which come from footnote a of Table 8.2 of Myhre et al. (2013a), 278 ppm for carbon dioxide and 772 ppb for methane. The scaling factors are calculated at the surface because this is the level where PI concentrations are given in Myhre et al. (2013a): the whole profile is scaled like the surface level, which is justified by the relatively well-mixed nature of both gases. By construction, the scaled PI distribution has the same global, annual average value at the surface as given in Myhre et al. (2013a), but inherits the horizontal, vertical, and temporal variabilities of the PD distribution. Using this scaling method replicates the PD amplitude of the seasonal cycle of carbon dioxide and methane concentrations. For carbon dioxide, there is a suggestion, from modelling studies, that the amplitude of the seasonal cycle may have increased since PI (Lindsay et al., 2014). Replicating the PD amplitude would therefore cause a small underestimate of the forcing.

3.3 Uncertainties

See Section 8.

4. Tropospheric and stratospheric ozone 4.1 Product specifications

4.1.1 Tropospheric ozone

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_srf_lw_ 3° x 3° Monthly (2003 – 2017) W m−2 cs CMIP Standard surface_adjusted_longwave_forcing_assuming_clear_sky name

Documentation of CAMS Climate Forcing products, version 1.5 Page 29 of 64

Copernicus Atmosphere Monitoring Service

Long name Tropospheric_ozone_lw_adjusted_radiative_forcing_at_surface_assuming_cl ear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_srf_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_longwave_forcing Long name Tropospheric_ozone_lw_adjusted_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_srf_sw_ 3° x 3° Monthly (2003 – 2017) W m−2 cs CMIP Standard surface_adjusted_shortwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_sw_adjusted_radiative_forcing_at_surface_assuming_cl ear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_srf_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_adjusted_shortwave_forcing Long name Tropospheric_ozone_sw_adjusted_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_toa_lw_ 3° x 3° Monthly (2003 – 2017) W m−2 cs CMIP Standard toa_adjusted_longwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_lw_adjusted_radiative_forcing_at_toa_assuming_clear _sky

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_toa_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_adjusted_longwave_forcing Long name Tropospheric_ozone_lw_adjusted_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_toa_sw_ 3° x 3° Monthly (2003 – 2017) W m−2 cs CMIP Standard toa_adjusted_shortwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_sw_adjusted_radiative_forcing_at_toa_assuming_clea r_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_toa_sw 3° x 3° Monthly (2003 – 2017) W m−2

Documentation of CAMS Climate Forcing products, version 1.5 Page 30 of 64

Copernicus Atmosphere Monitoring Service

CMIP Standard name toa_adjusted_shortwave_forcing Long name Tropospheric_ozone_sw_adjusted_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_trop_lw 3° x 3° Monthly (2003 – 2017) W m−2 _cs CMIP Standard tropopause_adjusted_longwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_lw_adjusted_radiative_forcing_at_tropopause_assumin g_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_trop_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_adjusted_longwave_forcing Long name Tropospheric_ozone_lw_adjusted_radiative_forcing_at_tropopause

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_trop_s 3° x 3° Monthly (2003 – 2017) W m−2 w_cs CMIP Standard tropopause_adjusted_shortwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_sw_adjusted_radiative_forcing_at_tropopause_assumi ng_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units arf_tropo3_trop_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_adjusted_shortwave_forcing Long name Tropospheric_ozone_sw_adjusted_radiative_forcing_at_tropopause

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_srf_lw 3° x 3° Monthly (2003 – 2017) W m−2 _cs CMIP Standard surface_instantaneous_longwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_lw_instantaneous_radiative_forcing_at_surface_assuming _clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_srf_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_longwave_forcing Long name Tropospheric_ozone_lw_instantaneous_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_srf_sw 3° x 3° Monthly (2003 – 2017) W m−2

Documentation of CAMS Climate Forcing products, version 1.5 Page 31 of 64

Copernicus Atmosphere Monitoring Service

_cs CMIP Standard surface_instantaneous_shortwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_sw_instantaneous_radiative_forcing_at_surface_assumin g_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_srf_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_shortwave_forcing Long name Tropospheric_ozone_sw_instantaneous_radiative_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_toa_lw_ 3° x 3° Monthly (2003 – 2017) W m−2 cs CMIP Standard toa_instantaneous_longwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_lw_instantaneous_radiative_forcing_at_toa_assuming_cl ear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_toa_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_longwave_forcing Long name Tropospheric_ozone_lw_instantaneous_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_toa_sw 3° x 3° Monthly (2003 – 2017) W m−2 _cs CMIP Standard toa_instantaneous_shortwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_sw_instantaneous_radiative_forcing_at_toa_assuming_c lear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_toa_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_forcing Long name Tropospheric_ozone_sw_instantaneous_radiative_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_trop_l 3° x 3° Monthly (2003 – 2017) W m−2 w_cs CMIP Standard tropopause_instantaneous_longwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_lw_instantaneous_radiative_forcing_at_tropopause_assu ming_clear_sky

Documentation of CAMS Climate Forcing products, version 1.5 Page 32 of 64

Copernicus Atmosphere Monitoring Service

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_trop_lw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_instantaneous_longwave_forcing Long name Tropospheric_ozone_lw_instantaneous_radiative_forcing_at_tropopaus e

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_trop_s 3° x 3° Monthly (2003 – 2017) W m−2 w_cs CMIP Standard tropopause_instantaneous_shortwave_forcing_assuming_clear_sky name Long name Tropospheric_ozone_sw_instantaneous_radiative_forcing_at_tropopause_ass uming_clear_sky

Variable Name Spatial Resolution Temporal Resolution Units rf_tropo3_trop_sw 3° x 3° Monthly (2003 – 2017) W m−2 CMIP Standard name tropopause_instantaneous_shortwave_forcing Long name Tropospheric_ozone_sw_instantaneous_radiative_forcing_at_tropopause

4.2 Methods 4.2.1 Radiative transfer calculations

The radiative transfer calculations made for computing ozone RF are identical to those described in section 3.2.1 for carbon dioxide and methane RF.

4.2.2 Inputs to the radiative transfer code

Calculations are made in a similar way to those for carbon dioxide and methane, as described in section 3.2.2. In version 1.5 products, the ozone distributions are taken from the CAMS Global Reanalysis (variable “GEMS ozone” in Table 3.2). The tropopause is identified according to its thermal definition, adopted by the World Meteorological Organization (WMO), where the tropopause is the lowest altitude at which lapse rate drops to 2 K km−1.

4.2.3 Pre-industrial atmospheric concentrations

Pre-industrial ozone conditions are derived from three-dimensional, monthly-averaged distributions from the CMIP6 dataset derived from CCMI models with representations of stratosphere- troposphere chemistry. CMIP6 ozone files used by CAMS Climate Forcing were obtained from https://pcmdi.llnl.gov/search/input4mips/ . The CMIP6 dataset is given as volume mixing ratios and are converted to mass-mixing ratios by multiplying by 1.66.

To produce pre-industrial distributions that are consistent with the CAMS Global Reanalysis, the reanalysed ozone distribution for each day is multiplied by the ratio of pre-industrial to present-day

Documentation of CAMS Climate Forcing products, version 1.5 Page 33 of 64

Copernicus Atmosphere Monitoring Service

ozone as given by the CMIP6 dataset. The year being processed is taken to represent present-day conditions. The year 1850 is taken to represent pre-industrial conditions and it is assumed here that ozone distributions have not changed between 1750 and 1850. That assumption is reasonable, although contributions from wildfires and anthropogenic activities linked to the start of the industrial revolution may have introduced small variations.

Future improvement: Pre-industrial ozone distributions will be based on several simulations of the pre-industrial atmosphere using the ECMWF IFS.

4.3 Uncertainties

See Section 8.

5. Speciated aerosol optical depths

5.1 Product specifications

Variable Name Spatial Resolution Temporal Resolution Units abs550anth 1° x 1° Daily (2003 – 2017) 1 CMIP Standard atmosphere_absorption_optical_thickness_due_to_ambient_aerosol_particles name Long name anthropogenic_aerosol_absorption_optical_depth_at_550_nm

Variable Name Spatial Resolution Temporal Resolution Units abs550dust 1° x 1° Daily (2003 – 2017) 1 CMIP Standard atmosphere_absorption_optical_thickness_due_to_ambient_aerosol_particles name Long name mineral_dust_aerosol_absorption_optical_depth_at_550nm

Variable Name Spatial Resolution Temporal Resolution Units abs550landnat 1° x 1° Daily (2003 – 2017) 1 CMIP Standard atmosphere_absorption_optical_thickness_due_to_ambient_aerosol_particles name Long name non-dust_land_natural_aerosol_absorption_optical_depth_at_550nm

Variable Name Spatial Resolution Temporal Resolution Units abs550marine 1° x 1° Daily (2003 – 2017) 1 CMIP Standard atmosphere_absorption_optical_thickness_due_to_ambient_aerosol_particles name Long name marine_aerosol_absorption_optical_depth_at_550_nm

Documentation of CAMS Climate Forcing products, version 1.5 Page 34 of 64

Copernicus Atmosphere Monitoring Service

Variable Name Spatial Resolution Temporal Resolution Units od550aer 1° x 1° Daily (2003 – 2017) 1 CMIP Standard name atmosphere_optical_thickness_due_to_ambient_aerosol_particles Long name total_aerosol_optical_depth_at_550_nm

Variable Name Spatial Resolution Temporal Resolution Units od550anth 1° x 1° Daily (2003 – 2017) 1 CMIP Standard name atmosphere_optical_thickness_due_to_ambient_aerosol_particles Long name anthropogenic_aerosol_optical_depth_at_550_nm

Variable Name Spatial Resolution Temporal Resolution Units od550dust 1° x 1° Daily (2003 – 2017) 1 CMIP Standard name atmosphere_optical_thickness_due_to_ambient_aerosol_particles Long name mineral_dust_aerosol_optical_depth_at_550nm

Variable Name Spatial Resolution Temporal Resolution Units od550landnat 1° x 1° Daily (2003 – 2017) 1 CMIP Standard name atmosphere_optical_thickness_due_to_ambient_aerosol_particles Long name non-dust_land_natural_aerosol_optical_depth_at_550nm

Variable Name Spatial Resolution Temporal Resolution Units od550marine 1° x 1° Daily (2003 – 2017) 1 CMIP Standard name atmosphere_optical_thickness_due_to_ambient_aerosol_particles Long name marine_aerosol_optical_depth_at_550_nm

5.2 Methods 5.2.1 Identification of aerosol origin

To obtain aerosol RF, it is necessary to distinguish between aerosols of natural origin and aerosols of anthropogenic origin. The ECMWF IFS does not keep track of the aerosol origin mainly to keep computational cost reasonable but also because aerosol origin is not always given in emission inventories, because the same aerosol particle may be an internal mixture with anthropogenic and natural contributions, and because data assimilation cannot constrain natural and anthropogenic aerosols independently. Instead the aerosol origin is obtained using the algorithm described by Bellouin et al. (2013) where aerosol size is used as a proxy for aerosol origin. The algorithm identifies four aerosol origins: anthropogenic, mineral dust, marine, and land-based fine-mode natural aerosol. That last origin mostly includes biogenic aerosols. The reader is referred to section 3 of Bellouin et al. (2013) for details of the algorithm. The present documentation describes two updates made to the algorithm since the publication of Bellouin et al. (2013).

The first update is the replacement of continental-wide anthropogenic fractions used over land surfaces by a fully gridded dataset that includes seasonal variations. Over land, identification of component AODs starts with removing the contribution of mineral dust aerosols from total AOD.

Documentation of CAMS Climate Forcing products, version 1.5 Page 35 of 64

Copernicus Atmosphere Monitoring Service

The remaining non-dust AOD, τnon-dust, is then distributed between anthropogenic and fine-mode natural components, noted τanth and τfine-natural, respectively, following:

τanth = fanth . τnon-dust τfine-natural = (1 – fanth) . τnon-dust where fanth is the anthropogenic fraction of the non-dust AOD. In version 1.5 products, fanth is given by monthly distributions on a 1x1-degree grid. This new dataset derives from an analysis of AeroCom 2 numerical models (Kinne et al., 2013). Its annual average is shown in Figure 5.1. Anthropogenic fractions show a North-South gradient, as expected from the location of population and industrial activities. Anthropogenic fractions are larger than 0.8 over most industrialised regions of North America, Europe, and Asia. The largest fractions are located over China, where more than 90% of non-dust AOD is attributed to anthropogenic aerosols. In the southern hemisphere, anthropogenic fractions are typically smaller than 0.7 on an annual average. In terms of seasonality, anthropogenic fractions remain larger than 0.7 throughout the year in the northern hemisphere, with a peak in winter when energy consumption is high. In the southern hemisphere, seasonality is driven by biomass-burning aerosols, which are considered purely anthropogenic in the CAMS Climate Forcing products. Anthropogenic fractions therefore peak in late summer in South America and southern Africa.

Figure 5.1 - Annually-averaged anthropogenic fraction of non-dust aerosol optical depth over land.

The second change concerns the fine-mode fraction (FMF) of marine AOD at 0.55 μm, which gives the fraction of marine AOD that is exerted by particles with radii smaller than 0.5 μm. In Bellouin et al. (2013), that fraction was set to a fixed value of 0.3. In version 1.5 products, that fraction is determined by a gridded dataset that includes monthly variations. The dataset is obtained by applying the method of Yu et al. (2009) to daily MODIS Collection 6 aerosol retrievals of AOD and FMF. First, the marine aerosol background is isolated by selecting only ocean-based scenes where total AOD at 0.55 μm is between 0.03 and 0.10. Then, an AOD-weighted averaged FMF is computed. The analysis has been applied to retrievals from MODIS instruments on both the Terra (dataset covering 2001—2015) and Aqua (dataset covering 2003—2015) platforms. Both instruments yield very similar marine FMF distributions, and the distributions used in version 1.5 product are the

Documentation of CAMS Climate Forcing products, version 1.5 Page 36 of 64

Copernicus Atmosphere Monitoring Service

multi-annual monthly averages of the two instruments. Figure 5.2 shows the marine FMF derived from MODIS/Terra for the months of January and July. It suggests that marine FMF varies over a wide range of values. Regions of high wind speeds, around 40-50° in both hemispheres, are associated with large FMFs indicating that the marine aerosol size distribution includes a sizeable fraction of smaller particles there. There are indications of contamination by fine-mode anthropogenic and mineral dust aerosols in coastal areas, but the impact on speciated AODs is small because the aerosol identification algorithm uses broad FMF categories rather than absolute values.

Figure 5.2 – Fine-mode fraction of marine aerosol optical depth at 0.55 μm as derived from MODIS/Terra Collection 6 aerosol retrievals for the months of January (left) and July (right).

In fact, the impact of using monthly-varying distributions instead of a global, annual marine FMF is small. Anthropogenic AOD decreases slightly in the roaring forties in the Southern Ocean but tends to increase slightly in the Northern Atlantic and Pacific oceans. On a global average, the change in anthropogenic AOD due to the improved specification of marine FMF is +0.001 (+1.4%). 5.2.2 Inputs to the algorithm

The list of input distributions from the CAMS Global Reanalysis used by the aerosol origin identification algorithm is given in Table 5.1.

Table 5.1 – Variables taken from the CAMS Global Reanalysis and used as input to the aerosol origin identification algorithm. Variable name MARS Parameter Number Levels 10 metre U wind component Surface only 10 metre V wind component Surface only Black Carbon Aerosol Optical Depth at 550nm 210.211 Column Dust Aerosol Optical Depth at 550nm 210.209 Column Land-sea mask Surface only Organic Matter Aerosol Optical Depth at 550nm 210.210 Column Sea Salt Aerosol Optical Depth at 550nm 210.208 Column Sulphate Aerosol Optical Depth at 550nm 210.212 Column Total Aerosol Optical Depth at 550nm 210.207 Column

Documentation of CAMS Climate Forcing products, version 1.5 Page 37 of 64

Copernicus Atmosphere Monitoring Service

5.3 Uncertainties

See Section 8.

6. Aerosol-radiation interactions

6.1 Product specifications

Variable Name Spatial Resolution Temporal Resolution Units re_ari_anth_srf_sw_cs 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_shortwave_radiative_effect_assuming_clear_sky Long name anthropogenic_sw_direct_effect_at_surface

Variable Name Spatial Resolution Temporal Resolution Units re_ari_anth_toa_sw_cs 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_radiative_effect_assuming_clear_sky Long name anthropogenic_sw_direct_effect_at_toa

Variable Name Spatial Resolution Temporal Resolution Units re_ari_dust_srf_sw_cs 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_shortwave_radiative_effect_assuming_clear_sky Long name mineral_dust_sw_direct_effect_at_surface

Variable Name Spatial Resolution Temporal Resolution Units re_ari_dust_toa_sw_cs 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_radiative_effect_assuming_clear_sky Long name mineral_dust_sw_direct_effect_at_toa

Variable Name Spatial Resolution Temporal Resolution Units re_ari_landnat_srf_sw_c 1° x 1° Daily (2003 – 2017) W m−2 s CMIP Standard name surface_instantaneous_shortwave_radiative_effect_assuming_clear_sk y Long name non-dust_land_natural_aerosol_sw_direct_effect_at_surface

Variable Name Spatial Resolution Temporal Resolution Units re_ari_landnat_toa_sw_cs 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_radiative_effect_assuming_clear_sky Long name non-dust_land_natural_aerosol_sw_direct_effect_at_toa

Variable Name Spatial Resolution Temporal Resolution Units re_ari_marine_srf_sw_cs 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_shortwave_radiative_effect_assuming_clear_sky

Documentation of CAMS Climate Forcing products, version 1.5 Page 38 of 64

Copernicus Atmosphere Monitoring Service

Long name sea-salt_sw_direct_effect_at_surface

Variable Name Spatial Resolution Temporal Resolution Units reari_marine_toa_sw_cs 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_radiative_effect_assuming_clear_sky Long name sea-salt_sw_direct_effect_at_toa

Variable Name Spatial Resolution Temporal Resolution Units rf_ari_srf_sw_cs 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name surface_instantaneous_shortwave_forcing_assuming_clear_sky Long name anthropogenic_sw_direct_forcing_at_surface

Variable Name Spatial Resolution Temporal Resolution Units rf_ari_toa_sw_cs 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_forcing_assuming_clear_sky Long name anthropogenic_sw_direct_forcing_at_toa

Variable Name Spatial Resolution Temporal Resolution Units rfari_toa_sw 1° x 1° Daily (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_forcing Long name anthropogenic_sw_direct_forcing_at_toa_allsky

6.2 Methods

Radiative effect and forcing of aerosol-radiation interactions are computed by radiative transfer calculations that combine the speciated AODs derived in Section 5 with prescriptions of aerosol size distribution and single-scattering albedo. The methods are as described in Section 4 of Bellouin et al. (2013) with one exception: the prescription of single-scattering has been updated from a few, continental-wide numbers to gridded monthly climatologies. This updated dataset introduces two major improvements compared to Bellouin et al. (2013). First, the new dataset provides the monthly cycle of fine-mode absorption. Second, the data set is provided on a finer, 1x1-degree grid. The method used to produce the dataset is described in Kinne et al. (2013). First, distributions of fine-mode extinction and absorption AODs are obtained from a selection of global aerosol numerical models that participated in the AeroCom 1 simulations using a common set of aerosol and precursor emissions for present-day conditions (Kinne et al. 2006). To include an observational constraint, those modelled distributions are then merged with retrievals of aerosol SSA for the period 1996–2011 at more than 300 AERONET sites. The merging is based on a subjective assessment of the quality of the measurements at each of the AERONET sites used, along with their ability to represent aerosols in a wider region around the site location. The main impact of merging observed SSAs is to make aerosols in Africa and South Asia more absorbing than numerical models predicted. The distribution of annual-averaged aerosol SSA is shown in Figure 6.1. The dataset represents the local maximum of absorption over California and the change in absorption as biomass-burning aerosols during transport, which is visible over Africa. Over Asia, Europe, and South America, absorption is also larger near source regions, with less absorption elsewhere.

Documentation of CAMS Climate Forcing products, version 1.5 Page 39 of 64

Copernicus Atmosphere Monitoring Service

Figure 6.1 – Annually-averaged distribution of single-scattering albedo at 0.55 μm used to characterize absorption of anthropogenic aerosols.

It is worth noting that the SSA distribution characterises absorption of fine-mode aerosols but is used to provide the absorption of anthropogenic aerosols, which is not fully consistent. The inconsistency is however mitigated by two factors. First, fine-mode aerosols are the main proxy for anthropogenic aerosols in the MACC algorithm that identifies aerosol origin, and their distributions are broadly similar. Second, regions where natural aerosols such as marine and mineral dust may contaminate the fine-mode AOD often correspond to minima in anthropogenic AOD.

Like in Bellouin et al. (2013), REari and RFari are estimated in clear-sky (cloud-free sky) then scaled by the complement of the cloud fraction in each gridbox to represent all-sky conditions, thus assuming that cloudy-sky aerosol-radiation interactions are zero. However, experimental products that include estimates of cloudy-sky RF exist but are based on a simplified account of cloud albedo, which limits their usefulness. For 2003, globally-averaged above-cloud anthropogenic and mineral dust AODs weighted by cloud fraction, are 0.005 and 0.003, respectively, or 8% of their clear-sky counterparts. Above-cloud marine and fine-mode natural AODs are negligible. Those above-cloud aerosols exert a positive REari because of their absorbing nature and the high reflectance of clouds. REari commonly reach +5 to +10 Wm−2 during the biomass-burning season that lasts from late August to October over the south-eastern Atlantic stratocumulus deck. That would translate into a cloudy-sky anthropogenic RFari of +0.01 Wm−2. Neglecting above-cloud aerosols therefore introduces a small uncertainty on the global average but leads to larger errors regionally and seasonally.

Future improvement: Products will use the ECMWF IFS radiation code and will therefore properly account for impact of clouds on all-sky radiative forcing of aerosol-radiation interactions.

Documentation of CAMS Climate Forcing products, version 1.5 Page 40 of 64

Copernicus Atmosphere Monitoring Service

6.2.1 Inputs to the algorithm

The list of input distributions from the CAMS Global Reanalysis used by the radiative transfer calculations to obtain REari and RFari is given in Table 6.1.

Table 6.1 – Variables taken from the CAMS Global Reanalysis and used to compute aerosol-radiation interactions. Variable name Levels Near IR albedo for diffuse radiation Surface only Near IR albedo for direct radiation Surface only Total cloud cover Column UV visible albedo for diffuse radiation Surface only UV visible albedo for direct radiation Surface only

6.2.2 Pre-industrial aerosol optical depths

Version 1.5 products are not defined with respect to PI conditions. Rather, RFari is defined with respect to PD aerosols, which is a different reference to PI so a correction is required. That correction factor is taken from Bellouin et al. (2013) and is equal to 0.8, i.e. RFari defined with respect to PI is 80% of RFari defined with respect to PD aerosols. The correction factor has been applied to variables rf_ari_toa_sw_cs, rf_ari_srf_sw_cs, and rf_ari_toa_sw.

Future improvement: Pre-industrial aerosol optical depth distributions will be based on several simulations of the pre-industrial atmosphere using the ECMWF IFS.

6.3 Uncertainties

See Section 8.

7. Aerosol-cloud interactions 7.1 Product specifications

Variable Name Spatial Resolution Temporal Resolution Units rfaci_toa_sw 1° x 1° (from 60S to 60N) Monthly (2003 – 2017) W m−2 CMIP Standard name toa_instantaneous_shortwave_forcing Long name anthropogenic_sw_first_indirect_forcing_at_toa_allsky

Documentation of CAMS Climate Forcing products, version 1.5 Page 41 of 64

Copernicus Atmosphere Monitoring Service

7.2 Methods

The aerosol-cloud RF estimation algorithm is the same as that used in Bellouin et al. (2013). It is based on satellite-derived cloud susceptibilities to aerosol changes, which are given seasonally and regionally. Those susceptibilities are applied to low-level (warm) clouds only.

7.2.1 Inputs to the algorithm

The list of input distributions from the CAMS Global Reanalysis used by the radiative transfer calculations to obtain RFaci is given in Table 7.1.

Table 7.1 – Variables taken from the CAMS Global Reanalysis and used to compute aerosol-cloud interactions. Variable name Levels Total column water Column Total column water vapour Column Total cloud cover Column Low cloud cover Column

7.2.2 Pre-industrial reference state

Like for RFari (section 6.2.2), version 1.5 products of RFaci are not defined with respect to PI conditions. Rather, RFaci is defined with respect to PD aerosols, which is a different reference to PI so a correction is required. That correction factor is assumed to be the same as for RFari at 0.8 (Bellouin et al. 2013). So RFaci defined with respect to PI is 80% of RFaci defined with respect to PD aerosols. The ratio has been applied to variable rf_aci_toa_sw.

Future improvement: Pre-industrial aerosol optical depth distributions will be based on several simulations of the pre-industrial atmosphere using the ECMWF IFS.

7.3 Uncertainties

See Section 8.

Documentation of CAMS Climate Forcing products, version 1.5 Page 42 of 64

Copernicus Atmosphere Monitoring Service

8. Uncertainties

Uncertainty can be structural or parametric in nature. The structural uncertainty relates to methodological choices in the characterisation of the radiative forcing. It is influenced by the atmospheric time step used in evaluating the radiative forcing (Colman et al., 2001), the effect of any climatological averaging (Mülmenstädt et al., 2019) and for instantaneous or stratospheric- adjusted forcing, the definition of tropopause (Collins et al., 2006).

Parametric uncertainty relates to choices about how certain physical processes are parameterised. As a full line-by-line radiative transfer simulation is too expensive for operational use in climate reanalysis or general circulation models, the SW and LW parts of the spectrum are divided into a small number of bands which exhibit similar scattering and absorption properties. This parameterisation error can be significant (Pincus et al., 2015). Different radiative transfer solvers divide the bands in different ways, and the choice of radiative transfer code contributes structural uncertainty (as there are methodological differences in how the radiative transfer equation is solved) in addition to parametric uncertainty. Parametric uncertainty is also present from the choices of how aerosol scattering and absorption processes are represented in models (Carslaw et al., 2013).

8.1 Data and methods

Uncertainty analysis in this document is performed using version 0.9.36 of ecRad (Hogan and Bozzo, 2018). ecRad is a two-stream broad-band radiation code designed for efficiency use in numerical weather prediction (NWP) and general circulation models (GCMs). The gas parameterisation is taken from the Rapid Radiative Transfer Model for GCMs (RRTMG) (Mlawer et al., 1997; Morcrette et al., 2008), the cloud solver from the the SPeedy Algorithm for Radiative TrAnsfer through CloUd Sides (SPARTACUS) (Hogan et al., 2018) and aerosol optical properties are taken from CAMS (Bozzo et al., 2017).

All experiments are performed using the CAMS Reanalysis (Inness et al., 2019) dataset for the year 2003. The starting point is the CAMS reanalysis data. The CAMS reanalysis was produced using ECMWF’s Integrated Forecast System (IFS) at CY43R1. The version of IFS used has 60 hybrid sigma/pressure levels in the vertical, with the top level at 0.1 hPa. The underlying IFS model has a time step of 30 minutes, with output analyses and forecasts produced every 3 hours. The analysis fields are “instantaneous”, although in reality are representative of the model time step. Reanalysis and forecast fields are obtained from the model output at 3-hourly resolution and radiative forcing calculations performed offline.

Reanalysis fields used are atmospheric temperature, specific humidity, cloud fraction, cloud ice water content, cloud liquid water content, ozone mass mixing ratio, skin temperature, logarithm of surface pressure and forecast albedo. The forecast albedo includes the effect of snow and ice cover on land. Greenhouse gas concentrations for CO2, CH4, N2O, CFC11, CFC12, HCFC22, and CCl4 from 2003 and 1850 are taken from the Representative Concentration Pathways (RCP) Historical dataset (Meinshausen et al., 2011). Although these forcings do not comprise the totality of

Documentation of CAMS Climate Forcing products, version 1.5 Page 43 of 64

Copernicus Atmosphere Monitoring Service

anthropogenically-emitted greenhouse gases, a large majority of the GHG forcing is included from these species. The radiation code also requires temperatures on level boundaries. These half-level temperatures are estimated by interpolating the full-level temperatures in log pressure.

8.2 Uncertainty from methodological choices

8.2.1 Timestepping and averaging

Uncertainty relating to timestepping comes from both the resolution of the climatology (the effect of averaging or sampling frequency of the input data), as well as the frequency of the radiation calls. In the IFS, full radiation calls are only made every 3 model hours, with reduced radiation calls made on intermediate model timesteps (30 model minutes). Only calling the full radiation scheme is in common with other reanalyses and GCMs, in which radiation is one of the more expensive processes to simulate. The reduced radiation calls on intermediate time steps allow for a better representation of TOA flux as the error by taking radiation calls at discrete time steps is minimised, particularly in the SW. As reanalysis data from CAMS is only available every 3 hours, there will be some sampling error from our offline radiation calls that cannot be quantified, because there is no access to the intermediate 30-minute climate states. On the converse, running the radiation code even every 3 hours can be computationally expensive, and it prompts the question whether either the time frequency of radiation calls could be reduced, or the climatology averaged, or both.

To investigate the option of using reduced frequency climatologies, reanalysis data is prepared as both daily and monthly means, alongside 3-hour instantaneous data. Using time-averaged climatologies as input to the radiative transfer code may cause biases, but the purpose of this investigation is to determine whether these biases are mild (and worth the increased computational efficiency) or severe.

Reducing the frequency of changes in climatology is only worthwhile computationally if the number of radiation calls is also reduced. In the SW this requires an appropriate choice of solar zenith angle. For daily and monthly climatologies, 3 representative zenith angles per day or month are selected. For monthly climatologies, 10 representative zenith angles are also investigated (this is not repeated for the daily climatology as the number of SW calls, 10 per day, would exceed the number of calls for a 3-hourly climate; 8 per day). The zenith angles and their weighting are selected based on Gaussian quadrature (Li, 2017).

For the actual 3-hourly climatologies, another important consideration is the solar zenith angle for SW radiation calls. Using an instantaneous zenith angle for 3-hourly radiation calls could create longitude-dependent biases. One example is high latitudes in the winter on the frontier of polar night: if the daylight period is less than 3 hours, using an instantaneous zenith angle every 3 hours may not be frequent enough to coincide with a time where that grid cell is sunlit.

Documentation of CAMS Climate Forcing products, version 1.5 Page 44 of 64

Copernicus Atmosphere Monitoring Service

Longitude bias can be reduced by using an effective zenith angle for the 3-hour time step, defined as the mean of the cosine of the solar zenith over the daylight period. This corresponds to method A3 of Blanc and Wald (2016):

푡2 ∫ 휃푆(푢) 퐼 휋(푢) 푑푢 푡1 휃푠< 휃eff = 2 푠 푡2 휋( ) ∫ 퐼휃 < 푢 푑푢 푡1 푠 2 where θs is the instantaneous solar zenith angle, I is the indicator function for daylight (1 if θs < π/2, eff 0 otherwise), t1 and t2 are the start and end of the 3-hour period and θs is the effective solar zenith angle. This daylight-weighted solar zenith angle definition is used for 3-hourly radiation calls. Furthermore, the effective zenith angle is multiplied by the fraction of time that each grid point is sunlit over the 3-hour period.

An experiment in which the climatology is varied every 21 hours (i.e. every 7th model timestep) is performed. This experiment preserves the instantaneous nature of the underlying reanalysis data while reducing the number of radiation calls by not sampling every timestep. A 21-hour sampling frequency is chosen to preserve the diurnal as well as seasonal radiation cycle, as recommended in partial radiative perturbation studies (Colman et al., 2001).

The approximations introduced by using a 3-hourly effective zenith angle are compared by using the same underlying reanalysis data with a 1-hourly effective zenith angle. At periods of 1 hour or less, the effective and instantaneous zenith angles are very similar in most grid points. Table 8.1 summarizes the 9 timestepping and climatological averaging experiments undertaken.

Table 8.1 – Time stepping and climatological averaging experiments

Label Reanalysis data Solar zenith angle Radiation calls/yr SW LW Total 3hr 3-hourly instantaneous 3-hour effective 2920 2920 5840 3hr_1hrzen 3-hourly instantaneous 1-hour effective 8760 2920 11680 3hr_21hr 3-hourly instantaneous, every 3-hour effective, every 7th 418 418 836 7th model timestep model timestep day_3hrzen daily mean 3-hour effective 2920 365 3285 day_3gzen daily mean 3 representative Gaussian 1095 365 1460 mon_1hrzen monthly mean 1-hour effective 8760 12 8772 mon_3hrzen monthly mean 3-hour effective 2920 12 2932 mon_10gzen monthly mean 10 representative Gaussian 120 12 132 mon_3gzen monthly mean 3 representative Gaussian 36 12 48

Top-of-atmosphere flux imbalance

Documentation of CAMS Climate Forcing products, version 1.5 Page 45 of 64

Copernicus Atmosphere Monitoring Service

The results from the timestepping experiment are shown in Figure 8.1, and root-mean-squared errors (RMSE) for the simulated data versus observations from the Clouds and the Earth’s Radiant Energy System, Energy Balanced and Filled dataset (CERES EBAF TOA Ed4.0) are given in Table 8.2. The CERES data assumes a nominal TOA height of 20 km, which corresponds to somewhere between the 3rd and 4th from top model level in the CAMS Reanalysis data. As this is well above the cloud layer the fluxes are not significantly different to those at the top level of the model.

Figure 8.1 – (a) TOA SW downwelling; (b) TOA SW upwelling; (c) TOA LW upwelling; (d) TOA net downwelling radiation from ecRad using 2003 CAMS Reanalysis data for the nine timestepping experiments described in Table 8.1. The black line shows the radiation fluxes for CERES.

Figure 8.1a shows that there is little error introduced in TOA downwelling SW radiation by using representative Gaussian zenith angles or averaged climatologies. The latter result should be expected as there is no atmosphere dependence on downwelling SW radiation at the TOA. Furthermore, the simulated incoming solar irradiance agrees very well to CERES. Figure 8.1b shows that accuracy in the SW upwelling TOA radiation is compromised by using climatological averaging. Monthly averaging is less accurate than daily averaging, whereas 3-hourly instantaneous climatologies agree reasonably well with observations. Figure 8.1c shows the corresponding fluxes for LW outgoing radiation. Again, 3-hourly instantaneous climatologies perform better than daily,

Documentation of CAMS Climate Forcing products, version 1.5 Page 46 of 64

Copernicus Atmosphere Monitoring Service

which in turn perform better than monthly. Agreement with observations is less good with the 3- hourly instantaneous fluxes in the LW than in the SW. Figure 8.1d shows net TOA radiation. Again, 3-hour instantanous climatologies agree better with observations than daily means, which are in turn better than monthly means. Biases with mean climatologies add rather than cancel, as upwelling radiation is underestimated in both the LW and the SW for daily and monthly means.

Figure 8.1 and Table 8.2 also show that the effect of climatological averaging dominates over the frequency of SW radiation calls. For 3-hourly experiments, there is no discernible loss of accuracy by only using every 7th model timestep. If using averaged climatological data is unavoidable (due to only having monthly mean diagnostics available from GCM output, for example), using 3 representative Gaussian zenith angles results in little loss of accuracy in either the LW or SW over sub-daily radiation calls, and increasing the number of representative zenith angles to 10 has no benefit. It should be re-iterated that monthly mean climatologies from reanalysis or GCMs are not recommended for use in offline radiation simulations.

Table 8.2 – Root-mean-square error (RMSE) of monthly top-of-atmosphere (TOA) radiation compared to CERES-EBAF for 2003. Experiment SW TOA RMSE LW TOA RMSE Net TOA RMSE 3hr 1.07 1.9 1.79 3hr_1hrzen 1.02 1.9 2.48 3hr_21hr 1.18 1.91 1.74 day_3gzen 3.78 4.52 8.23 day_3hrzen 2.77 4.52 7.18 mon_10gzen 11.25 10.33 21.55 mon_1hrzen 11.26 10.33 21.57 mon_3gzen 11.24 10.33 21.54 mon_3hrzen 10.34 10.33 20.65

Figure 8.2 shows the corresponding clear-sky radiation fluxes. It can be inferred that the inaccuracies in the climatological averaging are mostly due to clouds as these SW (Figure 8.2b) and LW (Figure 8.2c) biases are much reduced for clear-sky scenes (Figure 8.2a is identical to Figure 8.1a). A physical explanation is that averaging clouds (cloud fraction, cloud ice content and cloud water content) over several timesteps produces layers of thin cloud which have very different optical properties to more realistic, instantaneous cloud scenes, which are likely to be more individually optically thick clouds interspersed with significant patches of clear sky. Cloud transmission is a non-linear function of optical depth and hence cloud water path (Lacis and Hansen, 1974) and in the SW cloud radiative effect also depends on the effective solar zenith angle (Cronin, 2014). This makes SW all-sky fluxes unreliable with averaged climatologies.

Documentation of CAMS Climate Forcing products, version 1.5 Page 47 of 64

Copernicus Atmosphere Monitoring Service

SW clear-sky upwelling radiation shows a slight improvement for 3-hourly zenith angle calls than for representative zenith angles, and for 1-hourly radiation calls. Clear-sky outgoing LW radiation modelled compared to CERES improves slightly with increasing timestep, suggesting a secondary role for water vapour and atmospheric temperature. However, clear-sky upwelling LW radiation is underestimated by about 7 to 9 W m−2 in all cases compared to CERES. It is not clear why this is the case. One possibility could be differences in the definitions of “clear sky” between CERES observations (which may include observations of very thin cloud) and the model’s theoretical clear- sky radiation (which has clouds removed by construction). Other possibilities may be errors in the reanalysis values of albedo and temperature, and the assumed surface emissivity of 0.98.

Figure 8.2 – As for Figure 8.1, but for clear-sky fluxes.

The recommended offline radiation call configuration is the 3hr_21hr approach. This results in the lowest RMSE of all experiments compared to CERES (probably fortuitously, as there is no reason why it should be more accurate than the 3hr experiment) and has the third-lowest number of radiation calls per model year. The two experiments with fewer calls are mon_10gzen and mon_3gzen, which are grossly inaccurate.

Radiative forcing at top-of-atmosphere and tropopause

Documentation of CAMS Climate Forcing products, version 1.5 Page 48 of 64

Copernicus Atmosphere Monitoring Service

Instantaneous RF is estimated by comparing net fluxes at the tropopause and at the TOA for 2003 and 1850 all-sky conditions. A simplified definition of the tropopause is employed for this comparison, defined as the 29th model level in the CAMS Reanalysis, the level closest to 200 hPa. Alternative tropopause assumptions are investigated in section 8.2.3.

The 1850 atmosphere is created first by adjusting GHG concentrations to 1850 levels from the Historical RCP (Meinshausen et al., 2011). Mass-mixing ratios of ozone and aerosol species are prescribed using a gridded pre-industrial to present-day ratio. The same year-2003 climatologies for the atmosphere (temperature, water vapour and cloud variables) are prescribed, in the 1850 experiment, which measures instantaneous RF rather than ERF.

As no observations are possible using these 1850 atmospheric composition values, the RF calculated in the 3hr experiment is assumed to be closest to the truth, given the better agreement to CERES TOA fluxes than the daily- or monthly-averaged reanalysis data.

Figure 8.3 shows the results for the 3hr, day_3hrzen and mon_3hrzen experiments. Corresponding time stepping experiments for different solar zenith time steps give almost identical results. SW instantaneous RF is deficient when using averaged climatology, with mon_3hrzen disagreeing on sign with 3hr. The errors introduced in the LW by climatological averaging are relatively smaller, amounting to about 6% at the tropopause and 10% at the TOA for mon_3hrzen compared to 3hr. LW forcing dominates, and the errors oppose in sign, so the net climatological averaging effect is significant but somewhat mitigated. This suggests that unlike for TOA fluxes, averaged climatologies can still provide a useful first-order estimate of the net forcing. The error in net IRF is 0.21 W m−2 at the tropopause for day_3hrzen (and day_3gzen, not shown) compared to 3hr. This is used as our 5 to 95% uncertainty range in the CAMS Reanalysis RF product, which is calculated using a day_3gzen methodology.

Documentation of CAMS Climate Forcing products, version 1.5 Page 49 of 64

Copernicus Atmosphere Monitoring Service

Figure 8.3 – Year 2003 instantaneous radiative forcing, in W m−2, at the tropopause and top of atmosphere for 3-hourly solar zenith angle timesteps for 3-hourly, daily and monthly climatologies.

8.2.2 Resolution of reanalysis data

CAMS RF products are provided at a global grid of 3°x3°, partially for the purposes of computational efficiency. To determine whether this coarse grid introduces additional error, the 2003 TOA fluxes were analysed using the 3hr_21hr methodology at an increased resolution of 0.75°x0.75°. Only minor differences are found in the TOA radiative fluxes: −0.02 W m−2 in the SW and +0.07 W m−2 in the LW, resulting in a +0.05 W m−2 net difference. As the pre-industrial ratios of ozone and aerosol precursors are not available on this higher-resolution grid, instantaneous RF cannot be calculated using the finer grid. But taking the difference in TOA (or tropopause) fluxes is likely to result in smaller errors than the absolute TOA difference. On the other hand, 0.75° is still around 80 km in mid-latitudes, far below the resolution of satellites, and it is possible that a still-higher resolution would provide a different result. Taking these opposing effects into account, resolution error is assessed to be 0.05 W m−2 at a 5 to 95% confidence interval.

8.2.3 Tropopause definition

The IPCC Fifth Assessment Report (Myhre et al., 2013a) recommends using instantaneous RF calculated at the climatological tropopause, unless it does not differ from TOA instantaneous RF in which case TOA instantaneous RF is acceptable. It is seen from Figure 8.3 that TOA instantaneous RF differs significantly from tropopause instantaneous RF. This is in large part due to stratospheric

Documentation of CAMS Climate Forcing products, version 1.5 Page 50 of 64

Copernicus Atmosphere Monitoring Service

temperature adjustment resulting from the increasing carbon dioxide concentrations between 1850 and 2003 (Manabe and Wetherald, 1975) but may also be driven by differing tropospheric adjustments to varying forcing species that can also affect the TOA energy balance (Smith et al., 2018). Stratospheric adjustments are taken into account in the CAMS RF product, using a fixed- dynamical heating (FDH) approximation to iteratively allow stratospheric temperatures to readjust (Ramanathan and Dickinson, 1979; Fels et al., 1980). The number of additional radiation calls needed to produce an FDH estimate of stratospherically-adjusted radiative forcing precludes us from performing this for the uncertainty analysis. In common with the instantaneous RF at the tropopause, calculating RF requires definition of the tropopause. Experiment 3hr_21hr is used as a basis to investigate the uncertainty in the tropopause definition for instantaneous RF.

As in section 8.2.1 the base calculation is performed on level 29 of the CAMS reanalysis output. Alternative definitions used here are:

• the 200 hPa level, calculated by interpolating ecRad-calculated fluxes on model levels in logarithm of pressure. This level is used as a proxy for the tropopause from GCM results in forcing calculations by Collins et al. (2006); • a linearly-varying tropopause, from 100 hPa at the equator to 300 hPa at the poles, as used by Soden et al. (2008); • 100 hPa from the equator to 39° N/S where it drops abruptly to 189 hPa, and is then linear in latitude to 300 hPa at the poles, as used by Hansen et al. (1997); • the World Meteorological Organization (WMO) definition of the lowest altitude at which lapse rate drops to 2 K km−1. Occasionally, this is undefined in the Himalayas or Antarctic regions, and in these cases, tropopause is set to level 39, which is approximately 400 hPa in these regions. The number of grid points in which this occurs is small and is unlikely to affect the global calculation; • the CAMS model-defined tropopause, based on the WMO definition, but calculated daily for the CAMS74 forcing product.

Results are presented in Table 8.3. The WMO definition gives the largest net IRF at 2.56 W m−2, whereas the level 29 definition is somewhat smaller at 2.33 W m−2. The CAMS definition of the tropopause results in a net IRF of 2.46 W m−2 that is near the centre of this range. In determining the tropopause level uncertainty, equal weight is assigned to the WMO, CAMS, Soden et al. (2008) and Hansen et al. (1997) definitions. A weighting of 0.5 is assigned to the level 29 and 200 hPa definitions, as the former approximates the latter by construction. Even though CAMS use the WMO definition, they are calculated differently: the CAMS tropopause is calculated online and reported as daily averages, whereas the WMO tropopause is calculated offline from instantaneous 3-hour fields. These estimates are therefore treated as independent, with the reported differences in IRF justifying this choice. Using these weights, the uncertainty for the choice of tropopause level is 0.15 W m−2, which is the 5 to 95% confidence interval of these estimates taking into account weighting.

Documentation of CAMS Climate Forcing products, version 1.5 Page 51 of 64

Copernicus Atmosphere Monitoring Service

Table 8.3 – Shortwave, longwave and net instantaneous radiative forcings, in W m−2, calculated with different tropopause definitions. Definition SW LW Net Level 29 -0.55 2.88 2.33 200 hPa -0.56 2.88 2.31 Hansen 1997 -0.46 2.98 2.52 Soden 2008 -0.52 2.92 2.40 WMO -0.44 3.01 2.57 CAMS -0.50 2.97 2.46

8.2.4 Radiation code

Parameterisation error is also introduced by the reduction of both the solar and thermal radiation parts of the electromagnetic spectrum into a small number of bands. This reduction is required to facilitate rapid run time of radiation schemes in GCM and reanalysis schemes, as solving a full line- by-line radiative transfer problem is too computationally expensive. Parameterisation uncertainty also arises from the treatment of scattering and absorption of gases. Structural uncertainty in radiative transfer arises from the choices of approximations and numerical methods used in the actual solving of the radiative transfer equation. Further uncertainty is introduced by use of a two- stream radiative transfer model, which is standard in most GCMs as well as in ecRad, again for reasons of efficiency. This component of uncertainty is not quantified here.

Instantaneous RF calculated by ecRad is compared against the Suite Of Community Radiative Transfer codes based on Edwards and Slingo (SOCRATES), as configured in the UK Met Office’s GA3.1 configuration (Manners et al, 2015) optimized for use in the HadGEM3 family of GCMs. In this configuration, SOCRATES uses a Delta-Eddington two-stream solver with 6 SW and 9 LW radiation bands. In comparison, ecRad uses 16 bands in the LW and 14 in the SW.

Owing to the differences to how aerosols are specified between the ecRad and SOCRATES interfaces, comparisons are performed in aerosol-free cases. All-sky and clear-sky cases are compared between ecRad and SOCRATES, but it should also be noted that methodological differences between the two codes, including the scattering and absorption profiles of cloud droplets, and treatment of cloud overlap, may preclude a fair comparison of all-sky cases.

For the instantaneous RF calculations, full-year, 3hr_21hr calculations with 2003 CAMS reanalysis are again used but with GHGs set to 1850 levels in the 1850 simulation. The simulations are run only with the GHGs common to both codes (CO2, CH4, N2O, CFC11, CFC12 and HCFC22). A global effective radius of 10 micron is set for liquid water clouds and 50 micron for ice clouds.

Documentation of CAMS Climate Forcing products, version 1.5 Page 52 of 64

Copernicus Atmosphere Monitoring Service

The net GHG-only tropopause (level 29) instantaneous RF is 2.71 W m−2 in ecRad and 2.97 W m−2 in SOCRATES, whereas clear-sky IRF is 3.17 W m−2 in ecRad and 3.44 W m−2 in SOCRATES. SOCRATES therefore calculates a somewhat stronger IRF of around 0.27 W m−2, which is not reduced by the inclusion of clouds.

One further comparison against a quasi line-by-line calculation in the libRadtran implementation of DISORT (Mayer and Kylling 2005; Mayer et al. 2018) is performed for a global reference profile. The reanalysis data from 21 March 2003 at 15:00 is selected, for clear-sky conditions only. This comparison is a “quasi” rather than “true” line-by-line calculation because the Representative Wavelength parameterisation (REPTRAN; Gasteiger et al. 2014) is used to provide solar and thermal fluxes at a resolution of 15 cm−1.

This clear-sky comparison against the reference profile for 21 March at 15:00 results in an instantaneous RF of 2.85 W m−2 in libRadtran, 3.13 W m−2 in ecRad and 3.34 W m−2 in SOCRATES. The error due to radiation parameterisation is estimated to be 0.33 W m−2 at the 5 to 95% level from these three estimates. The three estimates are treated as equally plausible. A comparison from only three radiation codes is somewhat unsatisfactory. A radiation code inter-comparison is planned for several reference atmospheric profiles as part of RFMIP (Pincus et al., 2016) which should better quantify uncertainties in GCM radiation codes.

8.3 Uncertainty from aerosol optical properties and climatology

In addition to the methodological choice uncertainty discussed in section 8.2, there is uncertainty from the base climate state unrelated to any climatological averaging. Meteorological reanalysis is not perfect since limited and spatially incomplete observations are used to drive an atmospheric model (Dee et al., 2011). Additionally, the SW, and to a lesser extent LW, transmission and reflectivity of the atmosphere is heavily dependent on aerosol optical properties, which are not well constrained from observations (Carslaw et al., 2013).

A 240-member perturbed parameter ensemble is built by sampling 24 input variables using a Latin hypercube approach (Lee et al., 2011) according to assumed prior distributions (Table 8.4). For each sample set, a 2003 and 1850 simulation are performed, using the 2003 reanalysis data as before. Prior distributions of each parameter are informed from literature ranges and other modelling studies. Tropopause instantaneous RF is calculated on level 29, and a 3hr_21hr timestepping methodology is used.

In many cases the prior distributions in Table 8.4 are not the same as those used in referenced studies, but these references have been taken into account along with known information about the default parameter combinations used in ecRad, which produce a 2003 instantaneous RF estimate that is well within the expected range (see section 8.2.1). For example, the geometric standard deviation of the sulphate size distribution is modified from the prior used in Lee at al. (2013) of 1.2—1.8 to account for the fact that the IFS by default uses a small size distribution mean radius of 35 nm with a larger geometric standard deviation of 2.0. The prior for mean sulphate size

Documentation of CAMS Climate Forcing products, version 1.5 Page 53 of 64

Copernicus Atmosphere Monitoring Service

distribution used in the PPE admits values that are mostly larger than 35 nm, so the geometric standard deviation is reduced to compensate.

Table 8.4 – Variables perturbed and their ranges for use in the 240-member perturbed parameter ensemble. Variable How Scaling or Range Distribution Basis of prior perturbed absolute Mean of sulphate size CDNC Absolute 30 to 100 Uniform Asmi et al. distribution namelist nm (2011) Geometric standard CDNC Absolute 1.5 to 2.0 Uniform Lee et al. deviation of sulphate namelist (2013) size distribution Mean of OC size CDNC Absolute 30 to 100 Uniform Asmi et al. distribution namelist nm (2011) Geometric standard CDNC Absolute 1.5 to 2.0 Uniform Lee et al. deviation of OC size namelist (2013) distribution Mean of BC size CDNC Absolute 10 to 80 Uniform Asmi et al. distribution namelist nm (1) (2011) Geometric standard CDNC Absolute 1.5 to 2.0 Uniform Lee et al. deviation of BC namelist (2013) Mean of sea salt size CDNC Absolute 100 to 200 Uniform Dubovik et al. distribution (fine mode) namelist nm (2002) Geometric standard CDNC Absolute 1.2 to 1.8 Uniform Lee et al. deviation of sea salt size namelist (2013) distribution (fine mode) Mass mixing ratio of Atmospheric Scaling ⅓ to 3 log-uniform Myhre et al. hydrophilic BC profile (2013b) Mass mixing ratio of Atmospheric Scaling ⅓ to 3 log-uniform Myhre et al. sulphate profile (2013b) Mass mixing ratio of sea Atmospheric Scaling ⅓ to 3 log-uniform Lee et al. spray profile (2013) Cloud updraft speed CDNC Absolute 0.1 to 1.2 Uniform Regayre et al. (covering all cloud types) namelist m s-1 (2011) Cloud fraction, specific Atmospheric Scaling 0.9 to 1.1 Uniform Bellouin et al. cloud liquid content and profile (2013) specific cloud ice content Scattering coefficient of Aerosol Absolute 0.10 to Uniform Bond et al. BC optical 0.28 at 550 (2013) properties nm Absorption coefficient of Aerosol Absolute 4.4 to 18.6 Uniform Myhre et al. BC optical m2 g-1 at (2013b) properties 550 nm Scattering coefficient of Aerosol Absolute 0.887 to Uniform Feng et al.

Documentation of CAMS Climate Forcing products, version 1.5 Page 54 of 64

Copernicus Atmosphere Monitoring Service

OC optical 0.947 at (2013) properties 550 nm and 75% RH Absorption coefficient of Aerosol Absolute 2.5 to 12.6 Uniform Feng et al., OC optical m2 g-1 at (2013), Myhre properties 550 nm et al. (2013b) Temperature (vertical Atmospheric Absolute ± 1 K Uniform Dee et al. profile) profile (2011) Specific humidity Atmospheric Scaling 0.8 to 1.2 Uniform Dee et al. profile (2011) Forecast/surface albedo Atmospheric Absolute ± 0.02 Uniform Maclaurin et profile al. (2016) O3 concentration Atmospheric Scaling 0.5 to 1.5 Uniform Myhre et al. profile (2013a) (2) (3) CH4 concentration Atmospheric Scaling 2003: Normal Myhre et al. profile 0.9986 to (2013a) 1.0014

1850: 0.9684 to 1.0316 (3) CO2 concentration Atmospheric Scaling 2003: Normal Myhre et al., profile 0.9996 to (2013a) 1.0004

1850: 0.9930 to 1.0070 (3) N2O concentration Atmospheric Scaling 2003: Normal Myhre et al. profile 0.9997 to (2013a) 1.0003

1850: 0.9745 to 1.0254 Notes: 1. Assumed to be lower than OC. 2. O3 forcing presumed to scale linearly with O3 concentration. 3. CH4, CO2 and N2O use the same relative uncertainty compared to the best estimate concentrations for 1850 and 2003 simulations.

The distribution of tropopause instantaneous RF in the 240-member PPE is shown in Figure 8.4. The distribution of forcings is positively skewed and well-represented by a lognormal distribution (red curve in Figure 8.4). This contrasts with the anthropogenic forcing assessment in the IPCC AR5

Documentation of CAMS Climate Forcing products, version 1.5 Page 55 of 64

Copernicus Atmosphere Monitoring Service

which shows a mild negative skew (Myhre et al., 2013a), mostly due to the influence of the asymmetric uncertainty in AR5-assessed aerosol forcing. It should be noted however that the two different methods of arriving at distributions of radiative forcing are not equivalent.

The mean (5-95%) instantaneous RF from the 240-member ensemble is 2.44 (1.67 to 3.42) W m−2, which is slightly above the 2.33 W m−2 arising from using default ecRad parameters (section 8.2.1). The mean (5-95%) instantaneous RF from the lognormal curve fit is 2.44 (1.67 to 3.40) W m−2. Due to the good agreement between the sample and distribution fit, the mean and uncertainty range from the lognormal curve fit to the PPE is used in our overall uncertainty assessment for computational ease.

Figure 8.4 – Probability density function resulting from the CAMS Climate Forcing Perturbed Parameter Ensemble. A lognormal fit to the distribution is shown in red.

8.4 Combined uncertainty

Each individual source of uncertainty from sections 8.2 and 8.3 is combined to produce an overall uncertainty estimate (Table 8.5). To produce the combined uncertainty, each individual source of uncertainty is assumed to be uncorrelated. 106 Monte Carlo samples were drawn from each corresponding distribution in Table 8.5. This approach is taken as it is not straightforward to add non-symmetric uncertainties in quadrature. For the Gaussian distributions the uncertainty is applied as an addition to the lognormal PPE distribution with mean zero and 5 to 95% uncertainty range as given in Table 8.5.

Documentation of CAMS Climate Forcing products, version 1.5 Page 56 of 64

Copernicus Atmosphere Monitoring Service

Table 8.5 – Combined parametric and structural uncertainty in net tropopause instantaneous radiative forcing for 2003. Source of forcing error Uncertainty or forcing Distribution estimate (W m−2) Grid resolution ± 0.05 Gaussian Tropopause definition ± 0.15 Gaussian Radiative transfer parameterisation ± 0.33 Gaussian Timestepping (CAMS day_3gzen versus ± 0.21 Gaussian 3hr_21hr) Parametric: atmospheric reanalysis and 2.44 (1.67 to 3.40) Lognormal aerosol optical properties Total 2.44 (1.55 to 3.48)

Our combined uncertainty in instantaneous RF represents a range of 64 to 143% of the mean. This range is used to evaluate the RF uncertainty of the CAMS RF product, assuming that the uncertainty range calculated for the instantaneous RF in 2003 applies also to RF in future years.

The instantaneous radiative forcing in 2003 of 2.44 W m−2 is larger than the anthropogenic component of effective radiative forcing in 2003 of 1.98 W m−2 evaluated in AR5 (1.92 W m−2 relative to 1850) (see Annex II of Myhre et al. (2013a)), although well within the AR5 uncertainty range, which was calculated for the year 2011. However, if CAMS’ positive instantaneous RF is dominated by carbon dioxide RF, the ERF to that forcing is about 10-15% lower than the instantaneous RF calculated at the tropopause (Chung and Soden, 2015; Forster et al., 2016) so CAMS’ ERF would be correspondingly lower and more in line with the IPCC best estimate, in the range of 2.1 to 2.2 W m−2. Other forcing agents produce many different adjustment mechanisms in the stratosphere, troposphere and land surface (Smith et al., 2018), which would need to be fully taken into account in order to evaluate ERF. It is not possible to evaluate ERF in this study due to the requirement for a full pre-industrial atmospheric climatology to exist for 1850.

In summary, this assessment of uncertainty in CAMS RF products combines a perturbed parameter ensemble, where aerosol optical properties and atmospheric state variables are varied within their prescribed uncertainty ranges, and a structural uncertainty from climatological averaging, selection of radiation code, tropopause definition and grid spacing. Uncertainty is dominated by aerosol optical property parameterisation and reanalysis uncertainty. Future work will perform a variance- based sensitivity analysis on the perturbed parameter ensemble to determine which components of the PPE contribute most to the variance in instantaneous RF.

On a technical point of view, the best trade-off between computational efficiency and accuracy when calculating instantaneous RF from CAMS Reanalysis data is to use 3-hourly instantaneous reanalysis every 21 model hours. Longer timesteps may be permissible (27 or 45 hours) if the diurnal and seasonal variation is still adequately sampled, but this has not been investigated. There is very little error introduced in the 21-hour radiation call methodology compared to running the radiation code every 3 hours. The CAMS RF products use daily-mean climatological data using three representative solar zenith angles in the SW. The error introduced in this approximation is about 0.2 W m−2 for total instantaneous RF.

Documentation of CAMS Climate Forcing products, version 1.5 Page 57 of 64

Copernicus Atmosphere Monitoring Service

9. List of acronyms

Acronym Meaning AGAGE Advanced Global Atmospheric Gases Experiment AOD Aerosol Optical Depth CALIOP Cloud-Aerosol Lidar with Orthogonal Polarization CALIPSO Cloud-Aerosol Lidar and Infrared Pathfinder Satellite Observation CAMS Copernicus Atmosphere Monitoring Service CCCM CERES CALIPSO Cloudsat MODIS CCN Cloud Condensation Nuclei CDNC Cloud Droplet Number Concentration CERES Clouds and the Earth's Radiant Energy System CMIP6 Coupled Model Intercomparison Project, phase 6 DISORT Discrete Ordinates Radiative Transfer ECHAM ECMWF Hamburg (the climate model of Max Institute for Meteorology in Hamburg) ECMWF European Centre for Medium-range Weather Forecasts ESRL Earth System Research Laboratory ERF Effective Radiative Forcing FMF Fine-Mode Fraction IFS Integrated Forecast System IPCC AR5 Intergovernmental Panel on Climate Change Fifth Assessment Report LW Long-wave part of the electromagnetic spectrum MACC Monitoring Atmospheric Composition and Climate MODIS Moderate Resolution Imaging Spectroradiometer NOAA National Oceanic and Atmospheric Administration PD Present Day PDF Probability Density Function PI Pre-Industrial PPE Perturbed Parameter Ensemble RCP Representative Concentration Pathway RE Radiative Effect RF Radiative Forcing RFaci Radiative Forcing of aerosol-cloud interactions RFari Radiative Forcing of aerosol-radiation interactions RMSE Root Mean Square Error RRTMG Rapid Radiative Transfer Model for General Circulation Models SW Short-wave part of the electromagnetic spectrum TOA Top Of Atmosphere

Documentation of CAMS Climate Forcing products, version 1.5 Page 58 of 64

Copernicus Atmosphere Monitoring Service

10. References

Asmi, A., et al. Number size distributions and seasonality of submicron particles in Europe 2008– 2009, Atmos. Chem. Phys., 11, 5505-5538, doi :10.5194/acp-11-5505-2011, 2011.

Bellouin, N., et al. Estimates of aerosol radiative forcing from the MACC re-analysis. Atmos. Chem. Phys., 13, 2045-2062, doi:10.5194/acp-13-2045-2013, 2013.

Benedetti, A., et al. Aerosol analysis and forecast in the European Centre for Medium-Range Weather Forecasts Integrated Forecast System: 2. Data assimilation, J. Geophys. Res., 114, D13205, doi:10.1029/2008JD011115, 2009.

Bergamaschi, P., et al.: Atmospheric CH4 in the first decade of the 21st century: Inverse modeling analysis using SCIAMACHY satellite retrievals and NOAA surface measurements, J. Geophys. Res. Atmos., 118, 7350– 7369, doi:10.1002/jgrd.50480, 2013.

Blanc, P. and L. Wald. On the effective solar zenith and azimuth angles to use with measurements of hourly irradiation. Adv. Sci. Res. 13, pp.1-6, doi:10.5194/asr-13-1-2016, 2016.

Bond, T.C., et al. Bounding the role of black carbon in the climate system: A scientific assessment. J. Geophys. Res., 118, no. 11, 5380-5552, doi:10.1002/jgrd.50171, 2013.

Boucher, O., et al. Clouds and Aerosols in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change 571–658 (eds Stocker, T F. et al.) (Cambridge, 2013).

Bozzo, A., Remy, S., Benedetti, A., Flemming, F., Bechtold, P., Rodwell, M.J. and Morcrette, J.-J. Implementation of a CAMS-based aerosol climatology in the IFS. ECMWF, 2017.

Carslaw, K.S., et al. Large contribution of natural aerosols to uncertainty in indirect forcing. Nature, 503(7474), doi: 10.1038/nature12674, 2013.

Chevallier, F., Fisher, M., Peylin, P., Serrar, S., Bousquet, P., Bréon, F.-M., Chédin, A., and Ciais, P.: Inferring CO2 sources and sinks from satellite observations: Method and application to TOVS data, J. Geophys. Res., 110, D24309, https://doi.org/10.1029/2005JD006390, 2005.

Chung, E.-S. and B.J. Soden. An assessment of methods for computing radiative forcing in climate models. Environ. Res. Lett, 10, 074004, doi:10.1088/1748-9326/10/7/074004, 2015.

Collins, W.D., et al. Radiative forcing by well-mixed greenhouse gases: Estimates from climate models in the Intergovernmental Panel on Climate Change (IPCC) Fourth Assessment Report (AR4). J. Geophys. Res., 111(D14), doi:10.1029/2005JD006713, 2006.

Colman, R., et al. Climate feedbacks in a general circulation model incorporating prognostic clouds. Clim. Dyn. 18(1), 103—122, doi:10.1007/s003820100162, 2001.

Documentation of CAMS Climate Forcing products, version 1.5 Page 59 of 64

Copernicus Atmosphere Monitoring Service

Cronin, T.W. On the Choice of Average Solar Zenith Angle. J. Atmos. Sci., 71(8), 2994—3003, doi:10.1175/JAS-D-13-0392.1, 2014.

Dee, D. P., et al. The ERA‐Interim reanalysis: configuration and performance of the data assimilation system. Q.J.R. Meteorol. Soc., 137: 553-597. doi:10.1002/qj.828, 2011.

Dubovik, O., et al. Variability of Absorption and Optical Properties of Key Aerosol Types Observed in Worldwide Locations. J. Atmos. Sci., 59, 590–608, doi:10.1175/1520- 0469(2002)059<0590:VOAAOP>2.0.CO;2, 2002.

Etminan, M., et al. Radiative forcing of carbon dioxide, methane, and nitrous oxide: A significant revision of the methane radiative forcing, Geophys. Res. Lett., 43, doi:10.1002/2016GL071930, 2016.

Fels, S.B., et al. Stratospheric Sensitivity to Perturbations in Ozone and Carbon Dioxide: Radiative and Dynamical Response. J. Atmos. Sci. 37(10), 2265—2297, doi:10.1175/1520- 0469(1980)037<2265:SSTPIO>2.0.CO;2, 1980.

Feng, Y., et al. Brown carbon: a significant atmospheric absorber of solar radiation? Atmos. Chem. Phys., 13, 8607-8621, doi:10.5194/acp-13-8607-2013, 2013.

Gasteiger, J., et al. Representative wavelengths absorption parameterization applied to satellite channels and spectral bands, J. Quant. Spectrosc. Rad. Transfer, 148, 99—115, doi:10.1016/j.jqsrt.2014.06.024, 2014.

Hansen, J., et al. Radiative forcing and climate response. J. Geophys. Res., 102, 6831—6864, doi:10.1029/96JD03436, 1997.

Hogan, R. J., and A. Bozzo. A flexible and efficient radiation scheme for the ECMWF model. J. Advances in Modeling Earth Systems, 10, 1990–2008, doi:10.1029/2018MS001364, 2018.

Inness, A., Ades, M., Agustí-Panareda, A., Barré, J., Benedictow, A., Blechschmidt, A.-M., Dominguez, J. J., Engelen, R., Eskes, H., Flemming, J., Huijnen, V., Jones, L., Kipling, Z., Massart, S., Parrington, M., Peuch, V.-H., Razinger, M., Remy, S., Schulz, M., and Suttie, M.: The CAMS reanalysis of atmospheric composition, Atmos. Chem. Phys., 19, 3515–3556, doi:10.5194/acp-19-3515-2019, 2019.

Kato, S., et al. Surface Irradiances Consistent with CERES-Derived Top-of-Atmosphere Shortwave and Longwave Irradiances. J. Clim., 26, 2719—2740, doi:10.1175/JCLI-D-12-00436.1, 2013.

Kato, S., et al. Variable descriptions of the A-train integrated CALIPSO, CloudSat, CERES, and MODIS merged product (CCCM or C3M). Technical report, Atmospheric Science Data Center (ASDC) at NASA Langley Research Center, 2014.

Key, J. and A. J. Schweiger. Tools for atmospheric radiative transfer: Streamer and FluxNet. Computers & Geosciences, 24, 5, 443—451, 1988.

Documentation of CAMS Climate Forcing products, version 1.5 Page 60 of 64

Copernicus Atmosphere Monitoring Service

Kinne, S. et al. An AeroCom initial assessment – optical properties in aerosol component modules of global models, Atmos. Chem. Phys., 6, 1815-1834, doi:10.5194/acp-6-1815-2006, 2006.

Kinne, S., et al. MAC-v1: A new global aerosol climatology for climate studies. Journal of Advances in Modeling Earth Systems, 5, 704-740, doi:10.1002/jame.20035, 2013.

Lacis, A.A., and J.E. Hansen. A parameterization for the absorption of solar radiation in the Earth's atmosphere. J. Atmos. Sci., 31, 118-133, doi:10.1175/1520-0469(1974)031<0118:APFTAO>2.0.CO;2, 1974

Lee, L. A., et al. Emulation of a complex global aerosol model to quantify sensitivity to uncertain parameters, Atmos. Chem. Phys., 11, 12253-12273, doi:10.5194/acp-11-12253-2011, 2011.

Lee, L. A., et al. The magnitude and causes of uncertainty in global model simulations of cloud condensation nuclei, Atmos. Chem. Phys., 13, 8879-8914, doi:10.5194/acp-13-8879-2013, 2013.

Li, J. Comments on “On the Choice of Average Solar Zenith Angle”. J. Atmos. Sci., 74, 1669–1676, doi:10.1175/JAS-D-16-0185.1, 2017.

Lindsay, K., G.B. Bonan, S.C. Doney, F.M. Hoffman, D.M. Lawrence, M.C. Long, N.M. Mahowald, J. Keith Moore, J.T. Randerson, and P.E. Thornton: Preindustrial-Control and Twentieth-Century Carbon Cycle Experiments with the Earth System Model CESM1(BGC). J. Climate, 27, 8981–9005, (2014).

Maclaurin, G., et al. Development of a MODIS-Derived Surface Albedo Data Set: An Improved Model Input for Processing the NSRDB. National Renewable Energy Laboratory Technical Report NREL/TP- 6A20-67306, 23pp, 2016.

Manabe, S. and R.T. Wetherald. The Effects of Doubling the CO2 Concentration on the climate of a General Circulation Model. J. Atmos. Sci., 32, 3–15, doi :10.1175/1520- 0469(1975)032<0003:TEODTC>2.0.CO;2, 1975.

Mayer, B. and Kylling, A. Technical note: The libRadtran software package for radiative transfer calculations - description and examples of use, Atmos. Chem. Phys., 5, 1855-1877, doi :10.5194/acp- 5-1855-2005, 2005.

Meinshausen, M., et al. The RCP greenhouse gas concentrations and their extensions from 1765 to 2300. Climatic Change, 109:213, doi:10.1007/s10584-011-0156-z, 2011.

Mlawer, E.J. et al. Radiative transfer for inhomogeneous atmospheres: RRTM, a validated correlated-k model for the longwave. J. Geophys. Res., 102, doi:10.1029/97JD00237, 1997.

Morcrette, J., et al. Impact of a New Radiation Package, McRad, in the ECMWF Integrated Forecasting System. Mon. Wea. Rev., 136, 4773–4798, doi:10.1175/2008MWR2363.1, 2008.

Documentation of CAMS Climate Forcing products, version 1.5 Page 61 of 64

Copernicus Atmosphere Monitoring Service

Mülmenstädt, J., et al. Separating radiative forcing by aerosol–cloud interactions and fast cloud adjustments in the ECHAM-HAMMOZ aerosol–climate model using the method of partial radiative perturbations, Atmos. Chem. Phys. Discuss., doi:10.5194/acp-2018-1304, in review, 2019.

Myhre, G., et al. Anthropogenic and Natural Radiative Forcing in Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate Change 659–740 (eds Stocker, T F. et al.) (Cambridge), 2013a.

Myhre, G., et al. Radiative forcing of the direct aerosol effect from AeroCom Phase II simulations. Atmos. Chem. Phys., 13, 1853-1877, doi:10.5194/acp-13-1853-2013, 2013b.

Pincus, R., et al. A fast, flexible, approximate technique for computing radiative transfer in inhomogeneous cloud fields. J. Geophys. Res., 108, 4376, doi:10.1029/2002JD003322, D13, 2003.

Pincus, R., et al. Radiative flux and forcing parameterization error in aerosol‐free clear skies, Geophys. Res. Lett., 42, 5485—5492, doi:10.1002/2015GL064291, 2015.

Pincus, R., et al. The Radiative Forcing Model Intercomparison Project (RFMIP): experimental protocol for CMIP6, Geosci. Model Dev., 9, 3447-3460, doi:10.5194/gmd-9-3447-2016, 2016.

Ramanathan, V. and R.E. Dickinson. The Role of Stratospheric Ozone in the Zonal and Seasonal Radiative Energy Balance of the Earth-Troposphere System. J. Atmos. Sci., 36, 1084–1104, doi:10.1175/1520-0469(1979)036<1084:TROSOI>2.0.CO;2, 1979.

Ramaswamy, V., Boucher, O., Haigh, J., Hauglustaine, D., Haywood, J., Myhre, G., et al. Radiative forcing of climate change. In Climate change 2001: The physical science basis. Contribution of working group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change 6 (pp. 349–417). Cambridge, United Kingdom and New York, NY, USA: Cambridge University Press, 2001.

Reddy, M. S., et al. Estimates of global multicomponent aerosol optical depth and direct radiative perturbation in the Laboratoire de Météorologie Dynamique general circulation model. J. Geophys. Res., 110 (D10S16). doi:10.1029/2004JD004757, 2005.

Regayre, L. A., et al. Aerosol and physical atmosphere model parameters are both important sources of uncertainty in aerosol ERF. Atmos. Chem. Phys., 18, 9975-10006, doi:10.5194/acp-18- 9975-2018, 2018.

Smith, C. J., et al. Understanding rapid adjustments to diverse forcing agents. Geophys. Res. Lett., 45, 12,023–12,031, doi:10.1029/2018GL079826, 2018.

Soden, B.J., et al. Quantifying Climate Feedbacks Using Radiative Kernels. J. Climate, 21, 3504–3520, doi :10.1175/2007JCLI2110.1, 2008.

Yu, H., et al. Variability of marine aerosol fine-mode fraction and estimates of anthropogenic aerosol component over cloud-free oceans from MODIS. J. Geophys. Res., 114, D10206, doi:10.1029/2008JD010648, 2009.

Documentation of CAMS Climate Forcing products, version 1.5 Page 62 of 64

Copernicus Atmosphere Monitoring Service

11. User support and contacts

For questions regarding the CAMS Climate Forcing products or this documentation, it is recommended to raise a query through the CAMS user support system at https://atmosphere.copernicus.eu/help-and-support.

Documentation of CAMS Climate Forcing products, version 1.5 Page 63 of 64

Copernicus Atmosphere Monitoring Service

ECMWF - Shinfield Park, Reading RG2 9AX, UK

Contact: [email protected] atmosphere.copernicus.eu copernicus.eu ecmwf.int