Updated Global Warming Potentials and Radiative Efficiencies of Halocarbons and Other Weak Atmospheric Absorbers

Abstract Human activity has led to increased atmospheric concentrations of many gases, including halocarbons, and may lead to emissions of many more gases. Many of these gases are, on a per molecule basis, powerful greenhouse gases, although at present‐day concentrations their climate effect is in the so‐called weak limit (i.e., their effect scales linearly with concentration). We published a comprehensive review of the radiative efficiencies (RE) and global warming potentials (GWP) for around 200 such compounds in 2013 (Hodnebrog et al., 2013, https://doi.org/10.1002/rog.20013). Here we present updated RE and GWP values for compounds where experimental infrared absorption spectra are available. Updated numbers are based on a revised “Pinnock curve”, which gives RE as a function of wave number, and now also accounts for stratospheric temperature adjustment (Shine & Myhre, 2020, https://doi.org/10.1029/2019MS001951). Further updates include the implementation of around 500 absorption spectra additional to those in the 2013 review and new atmospheric lifetimes from the literature (mainly from WMO (2019)). In total, values for 60 of the compounds previously assessed are based on additional absorption spectra, and 42 compounds have REs which differ by >10% from our previous assessment. New RE calculations are presented for more than 400 compounds in addition to the previously assessed compounds, and GWP calculations are presented for a total of around 250 compounds. Present‐day radiative forcing due to halocarbons and other weak absorbers is 0.38 [0.33–0.43] W m−2, compared to 0.36 [0.32–0.40] W m−2 in IPCC AR5 (Myhre et al., 2013, https://doi.org/10.1017/CBO9781107415324.018), which is about 18% of the current CO2 forcing.

Plain Language Summary Human activity has led to increased atmospheric concentrations of many gases, including halocarbons (used, e.g., in refrigeration and air conditioning), and may lead to emissions of many other gases. While some halocarbons, such as chlorofluorocarbons (CFCs), are known to deplete stratospheric ozone, they are also powerful greenhouse gases contributing to radiative forcing (the net change in the energy balance of the Earth system) and hence climate change. We find that the present-day contribution from halocarbons and related compounds to radiative forcing is about 18% of the forcing due to increased concentrations of CO 2 . By using established methods and available laboratory measurements of absorption of infrared radiation for each gas, we quantify the radiative efficiency (i.e., a compound's strength as a greenhouse gas) for a total of around 600 compounds. For around 250 compounds we provide so-called global warming potentials (GWP), which are used to compare the climate impact of emissions of different gases and are commonly used to inform policy decisions. Results presented here can be used to derive values for emission metrics other than GWP. The present work is the most comprehensive review of the radiative efficiency and GWP of halocarbons and other weak absorbers performed to date.
Protocol from 1987 and its amendments and adjustments, halocarbons still make an important contribution to radiative forcing of climate change because many have long atmospheric lifetimes. Furthermore, the concentrations of some replacement compounds, such as hydrochlorofluorocarbons (HCFCs) and hydrofluorocarbons (HFCs), are rising. More specifically, Figure 1 (WMO/GAW, 2019) shows that HCFC-22 has recently become the second most abundant compound (of the greenhouse gases with only anthropogenic sources) after CFC-12. HFC-134a has, in only 20 years, increased from very low abundance to become the fourth most abundant halocarbon. Emissions of HFCs, perfluorocarbons, SF 6 , and NF 3 are included in the United Nations Framework Convention on Climate Change (UNFCCC). Controls on emissions of HFCs, in addition to CFCs and HCFCs, are included in the 2016 Kigali Agreement to the Montreal Protocol (see discussion in Kochanov et al., 2019).
Differences in the intensity and wavelength of infrared (IR) absorption bands lead to distinct radiative forcing efficiencies of various gases. Radiative efficiency (RE) is a measure of the radiative forcing for a unit change in the atmospheric concentration of a gas, and for halocarbons and related compounds is usually reported in units of W m −2 ppb −1 . To provide policy makers with guidance on the relative effectiveness of actions limiting the emissions of different gases, metrics have been developed to place the impact of emissions of different gases on a common scale. The most widely used metric is the global warming potential (GWP) with a 100-year time horizon (hereafter GWP(100)), which is based on the time-integrated radiative forcing due to a pulse emission of a unit mass of gas, normalized by the reference gas CO 2 and was introduced in the first assessment report of the Intergovernmental Panel on Climate Change (IPCC, 1990) (see section 2.5).
In 2013 we reviewed the literature data and provided a comprehensive and self-consistent set of new calculations of REs and GWPs for halocarbons and related compounds (Hodnebrog et al., 2013, hereafter referred to as H2013). Unlike the major greenhouse gases, current atmospheric concentrations of these compounds are low enough for the forcing to scale almost linearly with abundance, and we will therefore refer to these compounds as weak atmospheric absorbers. Adopting a common method for calculating REs and GWPs provides a more consistent approach to comparing metrics between different compounds than if these metrics are taken from studies that used different methodologies. Our results were incorporated by the IPCC into the fifth assessment report (AR5)  and, as a result, they are now used in national and international agreements. The UNFCCC adopted AR5 values for reporting emissions under the Paris Agreement and the U.S. Environmental Protection Agency (EPA) uses GWP values from AR5 in its reports. To ensure that climate policy decisions are based on the latest scientific data, it is important to periodically review and update the assessments. Additional infrared absorption spectra and refinements in estimations of the atmospheric lifetimes of halocarbons and other compounds have become available since our last review. Specifically, we have considered and included absorption spectra given as supporting information to published papers, and from the HITRAN2016 (Kochanov et al., 2019) and PNNL (Sharpe et al., 2004)  databases. Atmospheric lifetimes have recently been updated in WMO (2019) and these estimates have been used here. The provision of GWP(100) values in this paper, and in H2013, should not be seen as an endorsement of that metric, as the choice of metric depends on the policy context ; the RE and lifetime values presented here can be used to derive values for alternative emission metrics.
We have updated and extended our previous assessment of REs and GWPs for halocarbons and other weak atmospheric absorbers. Updates are based on new absorption spectra for 60 compounds considered in our previous review, the latest estimates of atmospheric lifetimes, and an update to the RE calculation method. The review has been extended to include around 440 additional compounds to bring the total number of compounds considered to more than 600. Included are several isomeric species which have identical empirical formulae but are structurally and spectrally distinct. Therefore, there is no need to consider isomeric compounds together within the context of this review. The radiative forcing contributions of the 40 most abundant halocarbons and related compounds in the atmosphere are estimated. The present work is the most comprehensive review of the radiative efficiencies and GWPs of halogenated compounds performed to date.

Absorption Cross Sections
In addition to the experimental spectra included in H2013 we have included, either in the main or supporting information, all IR absorption spectra available from the HITRAN2016 (Gordon et al., 2017;Kochanov et al., 2019) and PNNL (Sharpe et al., 2004) databases. The vast majority of spectra from PNNL are also available in HITRAN2016 and we have only included data from one of the databases to avoid overlap. The main sources of experimental infrared absorption cross sections in H2013 were the Ford Motor Company (e.g., Sihra et al., 2001), the Spectroscopy and Warming potentials of Atmospheric Greenhouse Gases project (Ballard et al., 2000b;Highwood & Shine, 2000), HITRAN-2008(Rothman et al., 2009 and GEISA-2009(Jacquinet-Husson et al., 2011 databases, and data provided by authors of published papers (e.g., Imasu et al., 1995). Several of the spectra used in H2013 were provided in the supporting information and later included in the HITRAN2016 and GEISA-2015(Jacquinet-Husson et al., 2016 databases. Many publications now make available their measured absorption cross sections as supporting information. Since spectra provided as supporting information are typically not in a standardized data format and need to be converted, we could only carry out RE calculations for a limited number of these supporting information spectra, and we have prioritized the 40 most atmospherically abundant compounds. For other studies the reported integrated absorption cross section and RE value, if available, are listed (Tables S1-S20).
As in H2013, each of the available spectra has been evaluated and if several spectra from the same laboratory group exist, we only use the latest published spectrum. For example, spectra from Sihra et al. (2001) supersede those from Pinnock et al. (1995) and Christidis et al. (1997) due to improvements in the methodology of the Ford laboratory measurements. When more than one spectrum was available from a source, the spectrum that was recorded nearest room temperature and atmospheric pressure was used (see section 2.2 for a discussion of the temperature dependence of cross sections). The choices of spectra to be used in RE  (Massie et al., 1991;McDaniel et al., 1991). (Sharpe et al., 2004). (c) SF 6 : T ¼ 180 K, p ¼ 75 Torr; T ¼ 225 K, p ¼ 78 Torr (referred to as Varanasi, private communication, 2000, in HITRAN). calculations have been explained for each group of compounds in the supporting information (Texts S1-S20).
In contrast to H2013, we only consider experimental absorption cross sections that are measured in a laboratory. As a result, 44 of the compounds included in H2013 have been omitted here because experimental spectra are not available, while nine of the compounds that only had calculated spectra in H2013 have been updated with RE values based on experimental spectra. Calculated IR spectra have been published for a vast number of compounds (e.g., Davila et al., 2017;Papanastasiou et al., 2018), with some studies including thousands of compounds (Betowski et al., 2016;Kazakov et al., 2012;McLinden et al., 2014) but these have a considerably larger uncertainty than experimental spectra (see Table 1 of H2013).

Temperature Dependence of Cross Sections
Although absorption cross sections are temperature dependent, integrated absorption cross sections show little dependence on temperature. The origin of the temperature dependence of absorption cross sections is the strong dependence of rotational states on temperature. Consequently, spectral bands are generally broader and have a lower peak intensity when observed at higher temperatures. This effect is illustrated in Figure 2 for a range of compound types (CFC, halon and sulfur-containing species), temperature range and pressure. The effect is noticeable even for the 20 K temperature difference illustrated in Figure 2 for CBrClF 2 . These small changes in band structure have a negligible effect on calculated REs, and hence GWPs.
However, when molecules exist in two or more distinct conformational forms, the possibility of significant temperature dependence of the integrated cross section exists (Godin et al., 2019). For example, the absorption spectra for CFC-114 reported by McDaniel et al. (1991) indicate that there are bands within the spectrum that show relatively strong positive temperature dependence, bands that show a weak negative temperature dependence, and bands that are not temperature dependent. These observations can be rationalized in terms of the temperature dependence of the populations of the two different conformers of CFC-114. However, the integrated cross sections of most molecules show little temperature dependence, and for consistency, we have used spectra obtained at ambient temperatures, where the experimental uncertainties are typically smallest.

Radiative Efficiency
In H2013, a common method was used to calculate the RE for most gases. This employed the "Pinnock curve" (Pinnock et al., 1995) where the RE as a function of wave number was calculated for a weak absorber absorbing equally at all wave numbers. Multiplying this curve by the absorption cross section of a given gas yields its RE. In H2013 the Pinnock curve was updated (Figure 3, blue line), most notably by increasing its spectral resolution from 10 to 1 cm −1 using the Oslo Line-By-Line (OLBL) radiative transfer model run at 0.02 cm −1 resolution (note that there was a typo in the caption of Figure 6 in H2013, wrongly stating a resolution of 0.2 cm −1 ); the updated calculations also used more refined atmospheric profiles of temperature, cloudiness and greenhouse gas concentrations. For instance, the atmospheric representation was expanded from one global mean profile to two profiles, one for the tropics and one for the extratropics, and the inclusion of refined cloud profiles led to weaker RE in the 800-1,200 cm −1 region (see sections 2.3 and 3.3.1 of H2013 for details). The Pinnock et al. (1995) method, and the H2013 update, yield the instantaneous RE (i.e., the radiative efficiency in the absence of stratospheric temperature adjustment). Since the RE, which . Instantaneous radiative forcing (IRF) efficiency (for a 0-1 ppb increase in mixing ratio) per unit cross section compared between the previous (Hodnebrog et al., 2013) and updated (Shine & Myhre, 2020) results from the Oslo line-by-line (OLBL) radiative transfer model run at 0.02 cm −1 spectral resolution. Also shown is the new radiative forcing (RF) efficiency where the effect of stratospheric temperature adjustment per unit cross section, based on 10 cm −1 narrow band model (NBM) simulations (Shine & Myhre, 2020), have been used to modify the OLBL curve. The curves have been averaged to 10 cm −1 spectral resolution in the plot, to improve readability, but RE calculations in this paper have been made using a 1 cm −1 version of the RF efficiency curve (as provided in the supporting information of Shine and Myhre, 2020).

10.1029/2019RG000691
Reviews of Geophysics includes this adjustment, provides a more accurate representation of a gas's impact on surface temperature, H2013 incorporated a correction to account for this. For most gases, the instantaneous RE was simply increased by 10%. For several gases  the correction was explicitly calculated using OLBL, either because of the absolute importance of that gas or because, in the case of HFC-41, it was known that the RE is less than its instantaneous value. However, this approach was somewhat ad hoc and may not have been applicable to all gases. Shine and Myhre (2020) have incorporated stratospheric temperature adjustment into the Pinnock curve for the first time, by calculating the impact of absorption by a gas at a given wave number on stratospheric temperatures ( Figure 3, red vs. purple line). The calculation of this adjustment is computationally intensive, as the RE due to absorption by a gas at a given wave number occurs not only at that wave number (as in the case of instantaneous RE) but now depends on the emission by gases (mostly CO 2 , H 2 O, and O 3 ) at all other wave numbers. Because of this, Shine and Myhre (2020) calculated the effect of adjustment using a narrow-band (10 cm −1 ) radiation code, and applied this to updated instantaneous RE calculations using OLBL (which included an improved representation of the water vapor continuum and some changes to the representation of clouds). The new method reproduced detailed calculations for a range of gases (including HFC-41 and CFC-11) to better than 1.5%. Although more complicated in its derivation, it is no more complicated than the original Pinnock method in its application. This new method (which also requires the use of the lifetime correction described in section 2.4) is applied to all gases here and hence improves the relative consistency of derived REs.

Atmospheric Lifetimes and Lifetime Correction
The atmospheric lifetime of a compound is required for calculations of GWPs and Global Temperature-change Potentials (GTPs) (see section 2.5). The RE value obtained from the method described in section 2.3 assumes the compound is well-mixed in the atmosphere. Most of the compounds included in this study have a nonuniform vertical and horizontal distribution in the atmosphere, and the lifetime can be used to correct for that. Here we use the method presented in H2013 (their section 3.3.4), where two approximations are given depending on the primary loss mechanism of the compound. One approximation is used for compounds primarily being lost through photolysis in the stratosphere: the fractional correction f to the RE of f(τ) ¼ 1 − 0.1826τ −0.3339 is applicable for lifetimes τ of 10 < τ < 10 4 years. Another approximation is used for compounds primarily lost through reaction with OH in the troposphere: , and is applicable for 10 −4 < τ < 10 4 years. The lifetime corrections for very short-lived compounds should be treated as particularly approximate, as the correction depends on where the emissions take place. Excepted from these approximations are CFC-11, CFC-12, and Halon-1211 because explicit LBL calculations were made in H2013 (see their section 3.3.3) to derive factors to account for non-uniform mixing. The derived factors were 0.927, 0.970, and 0.937, respectively, and are used here in the RE calculations for these compounds. These factors are less than one, despite being quite long-lived compounds, because of stratospheric loss due to photolysis.
The recent WMO (2019) report gives the most up-to-date and complete overview of atmospheric lifetimes of halocarbons and related compounds, and we rely on these estimates. Explanations and sources for the lifetime estimates in WMO (2019) are given for each compound in their Chapter 1.2 and Table A-1. For some compounds that do not have a lifetime estimate in WMO (2019), lifetime estimates have been taken from previous literature and sometimes as an average across different estimates if more studies exist (see Tables S1-S20 for references to lifetime estimates). For several compounds, we are not aware of any estimates of lifetimes; for these we only present REs assuming a constant horizontal and vertical distribution in the atmosphere, and no estimates of GWPs can be given.

Description of Metrics
The most widely used emission metric in climate policy is the GWP. It was introduced by IPCC (1990) where values for three time horizons (20, 100, and 500 years) were given. The GWP values were updated in following assessment reports. GWP has been widely adopted in climate policies, and the Kyoto Protocol adopted GWPs for a time horizon of 100 years as its metric for implementing a multigas approach. At UNFCCC COP24 it was decided to use GWP(100) for reporting national emissions to the Paris Agreement, while 10.1029/2019RG000691

Reviews of Geophysics
parties may in addition use other metrics (e.g., global temperature change potential) to report supporting information on aggregate emissions and removals of greenhouse gases, expressed in CO 2 equivalents (UNFCCC, 2019).
The GWP is based on the time-integrated radiative forcing due to a pulse emission of a unit mass of a gas. It can be given as an absolute GWP for gas i (AGWP i ) (usually in W m −2 kg −1 year) or as a dimensionless value by dividing the AGWP i by the AGWP of a reference gas, normally CO 2 . Thus, the GWP for gas i over a time horizon of H years is defined as IPCC has usually presented GWPs for a time horizon (H) of 20, 100, and 500 years (although IPCC AR5  only gave GWPs for 20 and 100 years). We use updated lifetimes and RE values presented in section 3 to calculate GWPs for 20, 100, and 500 years as in H2013.
The models used to calculate the impulse response function for CO 2 (Joos et al., 2013) include climate-carbon cycle feedbacks, but usually no feedbacks are included for the non-CO 2 gases when metrics are calculated. IPCC AR5  included this feedback tentatively in the metric values (see their Table 8.7 and supporting information Table 8.SM.16), which increased the GWP(100) values by 10-20%. Gasser et al. (2017) found that accounting for climate-carbon feedback increases the emission metrics of non-CO 2 species but, in most cases, less than indicated in AR5. They also found that when the feedback is removed for both the reference and target gas, the relative metric values are generally only modestly different compared to when the feedback is included in both (absolute metric values change more markedly); in the case of GWP(100) the differences are less than 1%. As pointed out by Gasser et al. (2017), including or excluding the climate-carbon feedback ultimately depends on the user's goal, but consistency should be ensured in either case. To resolve the consistency issue, we have excluded the climate-carbon feedback also for CO 2 by using the impulse response function for CO 2 based on the Gasser et al. (2017) simple Earth system model (see their Appendix C); their model shows very good agreement with Joos et al. (2013) when the climate-carbon feedback is included. Our documentation of input data and presentation of calculations allow for the inclusion of the climate-carbon feedback to our results in further studies or applications, both for CO 2 and the non-CO 2 compounds.
Changes to the parameters in AGWP CO2 impact all GWP values, and the GWP(100) values presented in section 3 are about 14% higher than if the old AGWP CO2 from AR5 or H2013 had been used. This is due to two changes: (i) The impulse response function for CO 2 is updated as explained above and (ii) the RE of CO 2 is updated using 409.8 ppm for 2019 (Butler & Montzka, 2020) and the simplified expression for CO 2 RF presented in Etminan et al. (2016), which is an update of the formula from Myhre et al. (1998)    and H2013 was about 14% higher, mainly because we updated the impulse response function (accounts for about 8% of the 14% change) and because of a higher atmospheric concentration of CO 2 which lowers its RE (accounts for~5%), and slightly because of the new formula from Etminan et al. (2016) (accounts for~1%). Accounting for all these changes, but including the climate-carbon feedback for CO 2 , as has been done in much of the prior literature, would give AGWP CO2 values which are 3%, 8%, and 13% higher for 20, 100, and 500 year time horizons, respectively.
It is worth highlighting that the impact of increasing CO 2 mixing ratios on GWP values is the net result of two opposing effects. First, many CO 2 absorption features are saturated, or close to saturation, and hence the RE of CO 2 decreases as its mixing ratio increases. Second, the fraction of CO 2 remaining in the atmosphere (measured by the impulse response function) increases with CO 2 mixing ratio (see Figure 8.31 in 10.1029/2019RG000691 Myhre et al., 2013). The first effect decreases AGWP CO2 while the second effect increases AGWP CO2 . Hence, GWP calculations for optically thin gases which are defined as AGWP X /AGWP CO2 will change with CO 2 mixing ratio.

Reviews of Geophysics
An alternative, the GTP was introduced by Shine et al. (2005). It uses the change in global mean temperature following a pulse emission for a chosen point in time as the impact parameter. While GWP is a metric integrated over time, the GTP is based on the temperature change per unit emissions for a selected year, t after the pulse emission. As for the GWP, the impact of CO 2 is normally used as reference: where AGTP (K kg −1 ) is the absolute GTP. The GTP uses the same input as for GWP but in addition includes a temperature response function that represents the thermal inertia of the climate system. AR5 presented values for both GWP and GTP. Here we follow the method used by AR5  and H2013 for calculating GTPs, except that the impulse response function and RE for CO 2 are updated as explained above and the climate response parameters are updated from Boucher and Reddy (2008) to Geoffroy et al. (2013) (as given in Appendix C of Gasser et al., 2017), which are based on an ensemble of models from the Coupled Model Intercomparison Project phase 5 (CMIP5) (Taylor et al., 2011) and involve a lower climate sensitivity (0.88 compared to 1.1 K (W m −2 ) −1 in Boucher and Reddy, 2008 There continues to be a vigorous debate about the applicability of different emission metrics (e.g., Myhre et al., 2013); metric choice depends on the particular policy context in which they are applied, and the degree to which continuity of choice is important in that context (e.g., Allen et al., 2018;Cain et al., 2019;Rogelj & Schleussner, 2019). A specific development has been the suggested use of metrics that compare one-off pulse emissions of long-lived gases (such as CO 2 ) with step-changes in emissions of short-lived species (e.g., gases with lifetimes less than a few decades), on the basis that this leads to a more informed comparison of their ultimate impact on temperature; such approaches can either adopt GWP values, but adapt their usage (Allen et al., 2016) or more directly compute the pulse-step equivalence (W. J. Collins et al., 2019). In the context of this review, the important point is that all such metrics require the same set of inputs (RE and lifetimes).
It is important to note that the RE and GWP(100) calculations presented here only include the direct effect, while indirect effects can be important for several compounds. Some compounds, and particularly CFCs and halons, influence radiative forcing indirectly through depletion of stratospheric ozone as shown in other work (e.g., Daniel et al., 1995;. The removal of organic compounds by reaction with OH in the troposphere acts as a source of ozone and prolongs the lifetime of methane, and this has been shown to be important for several hydrocarbons (W. J. Collins et al., 2002;Hodnebrog et al., 2018).

Uncertainties
An overview of estimated contributions to uncertainties associated with the radiative forcing of halocarbons was given in Table 1 of H2013. A total RE uncertainty of~13% was estimated for compounds with lifetimes longer than about 5 years, and~23% for compounds with lifetimes shorter than that. The much higher uncertainty for shorter-lived compounds is caused by the difficulty of estimating nonuniform horizontal and vertical distributions in the atmosphere, which in turn are dependent on the location of emissions (see section 2.4). Table 1 gives updated estimates of contributions to the total radiative forcing uncertainties. As in H2013, the uncertainty estimates are based on published literature and subjective judgment and we estimate the total uncertainty to be valid for a 5% to 95% (90%) confidence range. The total RF uncertainty, calculated using the root-sum-square (RSS) method, is~14% and 24% for compounds with lifetimes longer and shorter thañ 5 years, respectively. These total RF uncertainties are slightly higher than in H2013 and explanations are given below.
One issue with the use of laboratory data is that it does not always cover the entire spectral range for which radiative forcing is important (see, e.g., Figure 3). For example, the PNNL measurements mostly cover the 600-6,500 cm −1 wave number range, and so their use would neglect any absorption (and 10.1029/2019RG000691

Reviews of Geophysics
hence forcing) at lower wave numbers, although in general it extends to much higher wave numbers than those in other data sets.
The uncertainty due to lack of spectral data at low wave numbers cannot be assessed for every gas in our analysis, but there is some evidence to indicate its typical size. Highwood and Shine (2000) computed the contribution of wave numbers less than 700 cm −1 to the RE for HFC-134a and found it contributed around 2% to the forcing. Bravo et al. (2010) presented an analysis of the RE due to a set of seven perfluorocarbons. They compared the RE calculated using ab initio methods for the wave number interval 0-2,500 cm −1 with calculations for the wave number interval 700-1,400 cm −1 , chosen because it coincided with the wave number range for their associated laboratory measurements. Most of the additional absorption was at wave numbers below 700 cm −1 . They found that the integrated absorption cross sections and REs for the narrow range were within 2% for the lighter PFCs, but this difference increased to 10% for heavier PFCs. Since many of the measured data sets (e.g., the PNNL data) use a broader wavelength range than 700-1,400 cm −1 , it is unlikely that our estimates are systematically in error by such a large amount. Nevertheless, we introduce an additional generic uncertainty to our estimates, which was not included in the analysis of H2013, of~3% due to neglected bands (Table 1); clearly this could be systematically investigated in future work, perhaps by including ab initio calculations outside the range of measured cross sections.
Another source of uncertainty not considered in H2013 is the contribution to RE from absorption of shortwave (SW), or solar, radiation in the near-infrared (3,000 to 14,000 cm −1 ). There has been renewed interest in the SW forcing due to methane (e.g., W. D. Collins et al., 2018;Etminan et al., 2016). Etminan et al. (2016) find the direct effect of methane's near-IR bands enhances its forcing by 6% but there is an additional 9% impact via the effect of this absorption on stratospheric temperatures (and hence on longwave forcing). This contrasts with the impact of the near-IR bands of CO 2 which cause a decrease of a few percent, because much of the additional forcing is at higher altitudes. The contribution of these near-IR bands to RE is further complicated by the fact that it depends strongly on the overlap between these bands and those of water vapor (Etminan et al., 2016), many of which are saturated for typical atmospheric paths, making generic statements difficult.
The potential impact of SW absorption is difficult to constrain for the diverse range of gases discussed here, without much more detailed study, not least because many of the experimental data sets do not extend to such high wave numbers (the PNNL data are a notable exception). For the heavier halogenated gases, the strongest fundamental and combination bands will generally be at lower wave numbers, at which SW absorption is less important (see, e.g., Bera et al., 2009). The lighter, more hydrogenated, gases, will have more significant absorption bands in the solar near-infrared but, on the other hand, these gases are likely to be much shorter-lived, so that their impact on stratospheric temperatures is likely to be lower. We introduce an additional uncertainty of~5% due to the potential effect of this shortwave absorption (Table 1).
Since H2013, surface emissivity has been included as a source of uncertainty together with surface temperature and atmospheric temperature, and consequently the estimated contribution to RF uncertainty has been increased from~3% to~5% (Table 1). The stratospheric temperature adjustment is now based on a much more sophisticated method compared to the generic 10% increase used in H2013 (see section 2.3), and we have lowered the uncertainty contribution for this term from~4% to~2%. The remaining sources of uncertainties and their estimated contributions given in Table 1 are unchanged, and we refer to H2013 for detailed explanations of each term.
Uncertainties in the atmospheric lifetime of the compounds are also important for metric calculations, and since H2013, SPARC (2013) have provided recommended lifetime values and uncertainties for a range of halocarbons. Their estimates are derived using atmospheric chemistry transport and inverse modeling, and analysis of atmospheric observations and laboratory measurements. Possible uncertainty ranges for most of the compounds in SPARC (2013) have been evaluated in Velders and Daniel (2014;their Table 1) and range from ±3% to ±33% (1 standard deviation), depending on the compound; they are typically in the range from ±15% to ±20% (or ±25% to ±33% when converted from 1 standard deviation to 5-95% (90%) confidence range). However, Velders and Daniel (2014) point out that the possible uncertainty range is likely an overestimation of the true uncertainty and the most likely range, given for some of the compounds, is substantially lower (±12% to ±20% when converted from 1 standard deviation to 5-95% (90%) confidence range).

Updated Spectra, REs, and GWPs for the Most Abundant Halocarbons and Related Compounds
This section broadly follows the structure of section 4.1 in H2013, where absorption cross sections and radiative efficiency estimates in the literature were reviewed and new RE and GWP calculations were presented. However, we limit this section to only include studies and spectra that were not included in H2013, and only to the 40 most abundant halocarbons presented in Table 7 of Meinshausen et al. (2017) (see section 3.3 for other compounds). Also, only experimental spectra are used as a basis for our calculations here, unlike H2013 which included RE and GWP calculations for some compounds where only calculated spectra existed. In cases where spectra have been measured at different temperatures, we have used the spectra closest to room temperature (see section 2.2 for a discussion of temperature dependence of cross sections). All REs are given for all-sky and with stratospheric temperature adjustment included (see section 2.3). The lifetime correction method from H2013, to account for a nonhomogeneous vertical and horizontal distribution in the atmosphere, has been applied to the calculated REs (see section 2.4). Table 2 lists absorption cross sections that are new since H2013 and Tables S1-S6 in the supporting information list all (to the best of our knowledge) absorption cross sections and reported RE values from the literature. Tables S1-S6 also include calculations using the Pinnock curve from H2013 for easier identification of possible changes in RE that are due to the updated Pinnock curve from Shine and Myhre (2020). We have followed the International Union of Pure and Applied Chemistry, IUPAC, naming scheme and included the unique Chemical Abstract Service Registry Number, CASRN, for each compound listed in the tables.   (100) values, are given in the supporting information.

Chlorofluorocarbons
Since H2013, new spectra have been included for the five most-abundant CFCs, but the RE remains unchanged for four of the compounds (Tables 2 and 3). CFC-115 now has a much larger RE than in H2013 (0.25 compared to 0.20 W m −2 ppb −1 ) due to the addition of spectra from the PNNL database (Sharpe et al., 2004). In H2013, and in two out of four previous studies (Jain et al., 2000;Myhre & Stordal, 1997), the CFC-115 spectrum used is that from McDaniel et al. (1991), which has an integrated absorption cross-section of 1.21 × 10 −16 cm 2 molecule −1 cm −1 and gives an RE of 0.20 W m −2 ppb −1 in our calculations (Table S1) (1991) and PNNL absorption spectra shows that the locations and relative strength of the main absorption bands are similar, but that the overall magnitude of the bands are higher in the PNNL spectrum (not shown). Due to the large difference between the two spectra, we have also inspected the PNNL spectra measured at different temperatures (278 and 323 K), and these have similar integrated absorption cross sections and yield similar RE values as the 296 K PNNL spectrum (Table S1), and so give no indication of error in the 296 K PNNL spectra. A fourth source for CFC-115 spectra is Fisher et al. (1990) who report an integrated absorption cross section of 1.74 × 10 −16 cm 2 molecule −1 cm −1 , which is higher than McDaniel et al. (1991) and lower than (but nearer to) PNNL. Reasons for the large difference between the spectra remain unknown. We have calculated our new RE value of 0.25 W m −2 ppb −1 by averaging the RE values based on the three available spectra Sharpe et al., 2004;Totterdill et al., 2016).
The stratospheric temperature adjustment for the CFCs ranges from 9% to 12% increase of the instantaneous RE, and the generic 10% increase used in H2013 was a relatively good approximation for these compounds ( Figure 4). (Note that the 10% assumption was not used for CFC-11 and CFC-12 in H2013.) The atmospheric lifetimes of the five CFCs have been updated based on WMO (2019) since H2013, most notably for CFC-11 (52 vs. 45 years in H2013) and CFC-115 (540 vs. 1,020 years in H2013). A combination of updated lifetimes, REs, and the AGWP CO2 leads to higher GWP(100) values for all five CFCs (Table 3 and Figure 5).

Hydrochlorofluorocarbons
Six new spectra have been included for the three most-abundant HCFCs in this category, but their REs are unchanged when rounded to two decimals (Tables 2 and 3). The updated AGWP CO2 , and slightly longer lifetimes for two of the compounds (HCFC-141b and HCFC-142b), contribute to higher GWP(100) (Tables 3 and S2 and Figure 5).

Hydrofluorocarbons
Since H2013, spectra have been added to eight of the 11 most-abundant HFC compounds (Table 2) and in most cases this led to little or no change in the RE (Table 3). For HFC-23, the two new spectra (Harrison, 2013;Sharpe et al., 2004) each have higher integrated absorption cross sections than the two spectra used in H2013 (Table S3); this leads to a higher RE for this compound (0.19 compared to 0.17 W m −2 ppb −1 in H2013). Another contributing factor is the stratospheric temperature adjustment. The RE is now 13% higher than the instantaneous RE for HFC-23 (Figure 4), while in H2013 a generic 10% increase was used. In fact, all 11 HFC compounds have stratospheric temperature adjustments larger than 10% and most of them around 13%.
For HFC-43-10mee, the H2013 RE value of 0.42 W m −2 ppb −1 was not calculated using new spectra but was based on the RE given in the fourth assessment report (AR4) (Forster et al., 2007), which was again based on personal communication with D. A. Fisher in IPCC (1994). Recently, Le Bris et al. (2018) measured the absorption cross section and calculated a much lower RE of 0.36 W m −2 ppb −1 for HFC-43-10mee when using the method in H2013 (Table S3). They also showed that the RE calculated with their spectrum agreed very well with that calculated from the PNNL spectrum. Here, we have used the spectra from both Le Bris 10.1029/2019RG000691  Updated GWP (100) values are higher for all HFCs (Table 3 and Figure 5), and this is due to a combination of updated AGWP CO2 , higher RE values for several compounds (HFC-43-10mee is a notable exception), and longer lifetimes for all compounds except HFC-227ea and HFC-236fa.

Chlorocarbons and Hydrochlorocarbons
Nine new spectra have been added for the five most-abundant compounds since H2013 (Table 2). Wallington et al. (2016) made new measurements of the absorption spectra of the chloromethanes CH 3 Cl, CH 2 Cl 2 , CHCl 3 , and CCl 4 , and provided recommended spectra for these compounds by combining existing and new experimental data. We have used their recommended spectra to calculate REs for all four chloromethanes (see Text S4 for an explanation of the choice of spectra). The resulting RE for CCl 4 of 0.17 W m −2 ppb −1 is unchanged since H2013 (Tables 3 and S4), where the spectrum from Nemtchinov and Varanasi (2003) was used. The RE of CHCl 3 is lower than in H2013 (0.07 vs. 0.08 W m −2 ppb −1 ), where the spectrum from Vander Auwera (2000) was used. For CH 3 Cl and CH 2 Cl 2 , new RE calculations were not carried out for H2013 but retained from IPCC AR4 (Forster et al., 2007). Our calculations using the Wallington et al. (2016) spectrum show that the RE value of 0.03 W m −2 ppb −1 for CH 2 Cl 2 is unchanged since H2013 when rounded to two decimal places. The RE of CH 3 Cl is now 0.005 W m −2 ppb −1 , which is lower than the 0.01 W m −2 ppb −1 value in H2013 (which originated from AR4), but in excellent agreement with the original instantaneous RE value of 0.005 W m −2 ppb −1 from Grossman et al. (1997).
For CH 3 CCl 3 , we added the spectrum from the HITRAN 2016 database, which was again adopted from the PNNL database, and calculate a lower RE value compared to H2013 (0.06 vs. 0.07 W m −2 ppb −1 ) (Tables 2 and 3). The addition of the new spectrum did not change the RE, but the updated Pinnock curve and particularly the method to account for stratospheric temperature adjustment (see section 2.3) led to the lower value (Table S4). For all five compounds in this group, the stratospheric temperature adjustment is lower than the generic 10% increase used in H2013, and ranges from 0% change to a 9% increase of the instantaneous RE ( Figure 4). GWP(100) values are lower for CH 3 Cl, and higher for the remaining four compounds (Table 3 and Figure 5). Lifetime updates for four of the compounds contribute to the changes in GWP 100-year values.

Bromocarbons, Hydrobromocarbons, and Halons
Since H2013, absorption spectra from the HITRAN 2016 and PNNL databases have been included in the RE calculations for each of the four most-abundant compounds (Table 2). Changes in RE since H2013 are negligible (<5%) for the three halons, while CH 3 Br shows a lower RE (0.004 vs. 0.005 W m −2 ppb −1 ) (Table 3), mainly because the stratospheric temperature adjustment is lower (~4%) compared to the generic 10% increase used in H2013 (Figure 4). Since H2013, lifetimes are longer for Halon-1301 and Halon-2402 while GWP (100) values are higher for all four compounds (Tables 3 and S5 and Figure 5).  10.1029/2019RG000691

Fully Fluorinated Species
For 9 of the 12 most-abundant compounds, spectra have been added from the HITRAN 2016 database (where spectra were again adopted from the PNNL database) since H2013 (Table 2). Still, the calculated RE values for all these compounds are relatively similar to those reported in H2013 (Table 3). Sulfuryl fluoride shows the largest change of around 5%, mainly due to a slightly higher integrated absorption cross

10.1029/2019RG000691
Reviews of Geophysics section in the new PNNL spectrum compared to that of Andersen et al. (2009), which was used in H2013 (Table S6). This is in turn partly because the PNNL spectrum also includes a weak absorption band around 550 cm −1 (not shown).
For NF 3 , two new spectra have been added since H2013 and the calculated RE value is now based on three different spectra (Robson et al., 2006;Sharpe et al., 2004;Totterdill et al., 2016) (Tables 2, 3, and S6). The RE value of 0.20 W m −2 ppb −1 is the same as in H2013, but the RE of 0.25 W m −2 ppb −1 presented in Totterdill et al. (2016) is substantially higher (>20%). Totterdill et al. (2016) attribute the differences to a higher integrated absorption cross section compared to Robson et al. (2006) (which was used to calculate the RE value in H2013 and AR5), but our RE calculation differs by less than 5% when using spectra from each of the two studies separately (Table S6) so this is only part of the reason. Other potential reasons include differences between the radiative transfer models, treatment of clouds, and stratospheric temperature adjustment.
The RE of SF 6 has had a relatively wide range in reported literature values from 0.49 W m −2 ppb −1 (Jain et al., 2000) to 0.68 W m −2 ppb −1 (H. Zhang et al., 2011) (Table S6). Since H2013, Kovács et al. (2017) have made new measurements of the SF 6 absorption spectrum and used a LBL model to calculate a RE value of 0.59 W m −2 ppb −1 . Their spectrum is not included here, but their RE value is close to our calculated RE value of 0.57 W m −2 ppb −1 using spectra from the HITRAN and PNNL databases; this value was also presented in H2013 and used in AR5.
The stratospheric temperature adjustment for the fully fluorinated species ranges from 8% to 13% increase of the instantaneous RE (Figure 4). For most of these compounds, the generic 10% increase used in H2013 was compounds. The red line shows the 10% assumption used in H2013 for nearly all compounds (note that the 10% assumption was not used for CFC-11, CFC-12, and PFC-14).

10.1029/2019RG000691
Reviews of Geophysics a relatively good approximation for stratospheric temperature adjustment (note that the 10% assumption was not used for PFC-14 in H2013).
GWP(100) values are higher than in H2013 for all compounds in this category (Tables 3 and S6 and Figure 5), mainly due to the updated AGWP CO2 . The only lifetime change since H2013 is for NF 3 , which has a longer lifetime of 569 years compared to the value of 500 years that was used earlier. While we have adopted atmospheric lifetimes from WMO (2019), we note that two recent studies have calculated substantially shorter lifetimes for SF 6 than the widely used estimate of 3,200 years (Ravishankara et al., 1993). If the shorter SF 6 lifetimes of 1,278 [1,120-1,475] years (Kovács et al., 2017) or 850 [580-1,400] years (Ray et al., 2017) would have been used instead of 3,200 years, our GWP(100) value of 26,700 would not have been significantly affected (by less than 5%), but a shorter lifetime could be important for metric calculations using time horizons of several hundred years.  Table 4 for details). In the RF calculation, we use the preindustrial concentrations recommended by Meinshausen et al. (2017); these are nonzero for CH 3 Cl, CHCl 3 , CH 2 Cl 2 , CH 3 Br, and PFC-14/CF 4 , and assumed to be zero for the remaining compounds (see Table 4 footnote).

Reviews of Geophysics
When using the same RE values as in AR5 (from H2013), we see that the change from 2011 to 2014/2019 concentrations has led to a decrease in radiative forcing of CFCs (Figure 6). At the same time, concentrations of the CFC replacement compounds HCFCs and HFCs have increased and this leads to stronger RF for these compound groups, most notably for HFCs with a 83% increase in the RF. In total, RF due to increasing concentrations of HCFCs and HFCs more than outweighs the decrease in RF due to declining concentrations of CFCs.  Figure 6), the main reason for the RF increase can be attributed to increased concentrations (yellow vs. green bars in Figure 6). Table 4 shows that the main contributors to the~4 mW m −2 increase in RF are the updated RE values for CFC-12, HCFC-22, and HFC-134a. Chloromethane has the largest relative change in RF (and RE) with a 53% decrease. While its atmospheric concentration is the highest among the compounds, its high abundance is mainly due to natural sources (WMO, 2019) and its influence on anthropogenic RF is therefore much smaller than would otherwise be expected. Here we have assumed a pre-industrial value of 457 ppt from Meinshausen et al. (2017) who used a simple budget equation for its derivation, and it should be noted that this number is associated with uncertainties due to a lack of observations. Table 4 further shows that CFC-115 and HFC-43-10mee, respectively, have the second and third largest relative RF change due to new REs. While the new REs of methyl chloride and HFC-43-10mee are lower compared to H2013, the RE of CFC-115 is higher (see section 3.1).
The RF of 0.38 W m −2 due to halocarbons and other weak atmospheric absorbers can be put into context by comparison with the RF due to increased CO 2 concentrations. When using the simplified formula from Etminan et al. (2016), and assuming preindustrial (1750) and 2019 CO 2 concentrations of 278 ppm  and 409.8 ppm (Butler & Montzka, 2020), respectively (and of 270 ppb and 331.9 ppb, respectively, for N 2 O), the present-day RF due to CO 2 is 2.09 W m −2 . Thus, the RF due to halocarbons and other weak absorbers is 18% of the RF due to increased CO 2 concentrations.

Updated Spectra, REs, and GWPs for Other Weak Atmospheric Absorbers
This section has a similar structure to section 3.1 but presents and discusses lifetimes, REs, and GWP (100) values for compounds other than the 40 most abundant halocarbons and related compounds. Table 5 shows results for the compound groups included in our previous review (H2013), and brief discussions of these results are given in sections 3.3.1-3.3.7 below. Tables S7-S13 in the supporting information provide information on how the RE numbers were derived and list previously published absorption cross sections and reported RE values from the literature. In addition to the compound groups included in H2013, we have made RE calculations for a number of other compounds, mainly based on absorption spectra from the HITRAN 2016 (Kochanov et al., 2019) and PNNL (Sharpe et al., 2004) databases. These results are presented in Tables S14-S20 and a brief discussion of these results is given in section 3.3.8 below.

Hydrochlorofluorocarbons
Absorption spectra from the PNNL database were added for four HCFCs in H2013 ; the new RE values are in good agreement (difference of 5% or less) with the H2013 values (Tables 5 and S8). The spectrum from Gierczak et al. (2014) has been added for HCFO-1233zd(E), but the increased RE value since H2013 is mainly due to the longer lifetime and is therefore less influenced by the lifetime-correction factor.
HCFC-133a has been identified in the atmosphere recently . RE and GWP(100) values are given for this compound, based on absorption spectra from three sources (Etminan et al., 2014;McGillen et al., 2015;Sharpe et al., 2004). Calculations have been made for six additional HCFCs since H2013 and most notable is the relatively long-lived HCFC-124a (lifetime 17 years) with RE and GWP(100) values of 0.25 W m −2 ppb −1 and 2,170, respectively. In addition to the compounds listed here, Papanastasiou  (100) values are updated to include the effect of changes in lifetime and the AGWP of CO 2 . For most of the HFCs, changes in RE values since H2013 are minor and mostly reflect changes in the method to account for stratospheric temperature adjustment, which, for the HFCs, is generally higher than the 10% assumption used in H2013 (Figure 4). Additional factors include the revised Pinnock et al. curve (see section 2.3) and, particularly for the short-lived compounds, changes in lifetime which influence the factor to account for nonuniform distribution in the atmosphere (see section 2.4). Since H2013, RE values have been added for 14 compounds, of which four compounds are based on calculations using absorption spectra from PNNL.

Chlorocarbons and Hydrochlorocarbons
Among the 33 compounds in this category, only one (1,2-dichloroethane) was included in H2013 (Table 5). For all these compounds, RE calculations are based on additional absorption spectra from the PNNL database, and for 1,2-dichloroethane the RE value of 0.01 W m −2 ppb −1 is the same as in H2013 (when rounded to two decimals) (Table S10). Atmospheric lifetimes for most of the compounds in Table 5 are not available and the RE values listed are most likely upper limits since a uniform distribution in the atmosphere is assumed.

Bromocarbons, Hydrobromocarbons, and Halons
Six compounds in this category were also included in H2013 and for two of these (dibromomethane and halon-1202), new spectra from PNNL have been included in the RE calculations (Tables 5 and S11). The RE values remain unchanged for all six compounds (when rounded to two decimals). RE calculations have been made for 10 compounds in addition to those presented in H2013 and these are based on nine absorption spectra from PNNL and one from HITRAN2016.

Fully Fluorinated Species
Among the 10 compounds that were included in H2013, three of the four compounds with a very long lifetime (800 years or more) have new RE values that are less than 5% different from H2013 (Tables 5 and  S12). Octafluorocyclopentene was estimated to have a RE of 0.08 W m −2 ppb −1 in H2013, based on literature instantaneous RE values which were increased by 10% to crudely account for stratospheric temperature adjustment, and further adjusted by applying a lifetime correction. Since H2013 the lifetime of octafluorocyclopentene has been revised upward from 31 days to 1.1 years (WMO, 2019). We do not have the absorption spectrum and our recommended RE of 0.25 W m −2 ppb −1 is from N Zhang et al. (2017) who use a lifetime of 0.715 years. For the five remaining compounds with lifetimes of 31 days or less the new RE values are the same (after rounding) as in H2013 (see Table S12 for RE values with more significant figures).
Eight new compounds have been added since H2013 and one of the new compounds, heptacosafluorotributylamine/PFTBA, was recently observed in the Arctic (Schlabach et al., 2018). Its absorption spectrum has been measured in three recent studies (Bernard et al., 2018;Godin et al., 2016;Hong et al., 2013) and using the spectra from Hong et al. (2013) and Godin et al. (2016) we calculate a RE value of 0.91 W m −2 ppb −1 . Bernard et al. (2018) measured spectra for three other perfluoroamines and report large RE values also for these compounds (in the range 0.61-0.95 W m −2 ppb −1 ). Lifetimes are estimated to be more than 1,000 years (WMO, 2019) and therefore these compounds are potent greenhouse gases. Two of the compounds added are based on RE values from the literature and RE for the remaining two compounds have been calculated using absorption spectra from PNNL. Heptafluoroisobutyronitrile is a potential replacement for sulfur hexafluoride and its atmospheric chemistry has been studied by Blazquez et al. (2017) and Andersen et al. (2017). The RE value of 0.25 W m −2 ppb −1 in Table 5 is an average of the REs from the two studies.

Other Compounds
In addition to the seven compound groups listed in Tables 3 and 5, RE values for the following eight compound groups not considered in H2013 are presented in the supporting information: hydrocarbons (Table S14), alcohols, ethers, and other oxygenated hydrocarbons (Table S15), iodocarbons and hydroiodocarbons (Table S16), nitriles, amines and other nitrogenated hydrocarbons (Table S17), sulfur-containing compounds (Table S18), silicon-containing compounds (Table S19), and other compounds (Table S20). In contrast to the compounds presented in Tables S1-S13, the previous literature has not been reviewed for these compounds. Rather, RE values presented in Tables S14-S20 are for the most part purely from calculations based on available absorption spectra from the HITRAN2016 and PNNL databases. It is also important to note that the RE values assume, with a few exceptions, constant horizontal and vertical atmospheric distribution and are thus regarded as upper estimates.

Summary and Conclusions
We present a comprehensive assessment of the radiative efficiencies and GWPs for a large number of halocarbons and other weak atmospheric absorbers. The present work is an update of our review in H2013 where a consistent method for calculating RE was used for all compounds. A major advantage of using a common method for calculating REs is that the RE and GWP values for different compounds can be directly compared. This method has now been updated, and best estimate RE values have been calculated based on approximately 700 experimental absorption cross sections, versus 200 in H2013. The majority of the new spectra have been obtained from the HITRAN2016 and PNNL databases which were not included in our previous review.
Best estimate REs based on experimental spectra are now provided for more than 600 compounds compared to 168 compounds in H2013 (221 compounds when including REs based on calculated spectra). Most of the REs are based on our calculations, while some are based on published values. Figure 7 shows a comparison of our updated RE values with those presented in H2013 (and used in IPCC AR5; Myhre et al., 2013) for the 177 compounds included in both studies. For compounds with RE > 0.5 W m −2 ppb −1 , changes are less than 5%. For compounds with RE < 0.5 W m −2 ppb −1 , 61 compounds have RE values which differ by more than 5% from H2013, and 42 differ by more than 10%.
We have adopted recommended atmospheric lifetimes from the literature and, when available, calculated GWP values. In the main part of the paper (Tables 3 and 5) we have chosen to show only GWPs for a 100-year time horizon in addition to the lifetimes and RE values. However, many metrics exist, and the choice of metric and time horizon depends on the context in which they are used (see section 2.5). Table  16 of H2013 presented GWP values for 20, 100, and 500 year time horizons in addition to GTPs for 20, 50, and 100 year time horizons for a selection of compounds. Table 6 shows updated numbers for these compounds, and (A)GWP and (A)GTP values for the same time horizons are given for all compounds in the supporting information.
In principle, and as noted by H2013, it would be desirable to calculate the Effective Radiative Forcing (ERF)  which includes rapid adjustments beyond stratospheric temperature; ERF better represents the ultimate impact of a gas on surface temperature. It remains impracticable to calculate ERF for the gases discussed here (see Shine and Myhre, 2020 for discussion) because ERF requires computationally expensive calculations using general circulation models. The radiation codes in these models do not have sufficient spectral resolution to properly represent differences between the many halocarbons presented here, and the model's unforced variability would be much larger than the RF at their current, or likely future, concentrations. Although excellent progress has been made in understanding the generic nature of rapid adjustments and intermodel differences for many climate forcing agents (e.g., Smith et al., 2018), this has not yet extended to the halocarbons in a way that would allow a reliable generic correction to be made to the RFs calculated here. Figure 7. Comparison of radiative efficiencies (W m −2 ppb −1 ) calculated in this study (lifetime-corrected adjusted cloudy-sky) and from H2013 for (a) all compounds and (b)-(d) zoomed in using different scales for the RE. Black dots represent compounds where the RE in this study is less than 5% different from H2013, while yellow and red dots represent compounds where the REs are significantly different (>5% and >10%, respectively). Red dots have been labeled and represent compounds where the RE calculated here is more than 10% different from H2013.
One interesting potential consequence of revisions in the GWP(100) of halocarbons is the impact on existing legislation. For example, the European Union's legislation on the usage of fluorinated greenhouse gases (EUR-Lex, 2014), in part, puts restrictions on usages of gases based on their GWP(100); it places dates on prohibition of marketing certain equipment which uses products with GWP(100) values exceeding 150, 1,500, and 2,500. Within that legislation, the GWP(100) values are clearly defined as those mostly originating from AR4 (Forster et al., 2007), but it does not appear to account for the uncertainties inherent in those GWP (100) values. Some of the updates presented here would push gases that were on one side of these GWP (100) boundaries to the other side. For example, HFC-152a has now breached the 150 boundary (172 compared to 138), and HFC-134a has breached the 1,500 boundary (1,600 compared to 1,300). Hence, future updates to legislation would either have to stick to using outdated values, amend the boundaries between allowed and prohibited fluorinated gases, or else decide that some gases that were previously accepted for certain usages, are no longer so. Similarly, Japan's "Act on Rational Use and Proper Management of Fluorocarbons" has target values for GWP(100) (at values of 100, 150, 750, and 1,500) for different products (MEGJ, 2016) as does Canada's "Ozone-depleting Substances and Halocarbon Alternatives Regulations" (MJGC, 2019), with various products limited at different values of GWP(100) (150, 750, 1,400, 1,500, and 2,200). Table 7 gives an overview of the six (out of the 40 most abundant) compounds that enter a new policy category due to our updated GWP(100) values. All six compounds show higher GWP(100) than in H2013, and updates to lifetimes and the AGWP CO2 (100) explain most of the increase for these compounds. A considerable part of the GWP(100) increase due to updated AGWP CO2 (100) (around 5% out of the 14%) arises  Note. See section 4 for a discussion of, and references to, the policies that are referred to here, and section 3.1 for discussion of changes in lifetime and RE estimates. The rightmost column shows the contribution of change in GWP(100) due to the different factors: new AGWP CO2 , new lifetime estimate (τ), new instantaneous RF "Pinnock curve" (IRF curve), new method to account for stratospheric temperature adjustment (STA), and addition of absorption spectra.
solely because of increasing CO 2 concentrations and thereby reduced radiative efficiency of CO 2 since H2013 (see section 2.5). Hence, it can be anticipated that continued accumulation of CO 2 in the atmosphere will lead to changes in GWP values for weak atmospheric absorbers also in future updates.
Finally, we have combined our new updated RE values with present-day atmospheric concentrations of halocarbons to determine their radiative forcing. We find that the most abundant halocarbons cause a present-day RF of 0.38 [0.33 to 0.43] W m −2 , compared to 0.36 [0.32 to 0.40] W m −2 in AR5  and this is almost 20% of the preindustrial (1750) to present-day (2019) RF due to CO 2 . Most of the increase in halocarbon forcing since AR5 can be attributed to increasing concentrations of CFC replacement compounds (HCFCs and HFCs) which more than outweigh the forcing due to decreasing concentrations of CFCs. However, the stronger halocarbon RF is also a consequence of updated RE values, which are slightly higher compared to AR5 for some of the most abundant compounds.