The response of stratospheric water vapor to climate change driven by different forcing agents

. We investigate the response of stratospheric water vapor (SWV) to different forcing agents within the Precipitation Driver and Response Model Intercomparison Project (PDRMIP) framework. For each model and forcing agent, we break the SWV response into a slow response, which is coupled to surface temperature changes, and a fast SWV, which is the direct response to external forcing, but without any mediation from the surface temperature. Our results show that, for most climate perturbations, the slow SWV response dominates the fast response. The slow SWV response exhibits a similar 10 sensitivity to surface temperature across all climate perturbations. Specifically, the sensitivity is 0.35 ppmv K -1 in the tropical lower stratosphere (TLS), 2.1 ppmv K -1 in the northern hemispheric lowermost stratosphere (LMS), and 0.97 ppmv K -1 in the southern hemispheric LMS. The fast SWV response only dominates the slow SWV response when the forcing agent radiatively heats the cold point region — for example, black carbon, which directly heats the atmosphere by absorbing solar radiation. The fast SWV response in the TLS is primarily controlled by the fast adjustment of cold point temperature across 15 all climate perturbations. This control becomes weaker at higher altitudes and at higher latitudes below 150 hPa.


Introduction
Stratospheric water vapor plays an important role in global climate change. It is an important greenhouse gas (GHG), which affects the Earth's radiative budget (Forster and Shine, 2002;Solomon et al., 2010), and it also plays an important role in stratospheric ozone chemistry (Solomon et al., 1986;Dvortsov and Solomon, 2001). 20 SWV in the overworld (above 380-K isentropic surface) (e.g. Hoskins, 1991) and SWV in the extratropical lowermost stratosphere (LMS, between the extratropical tropopause and the 380-K isentropic surface) (e.g. Holton et al., 1995) are distinguished according to different mechanisms that control them. Overworld SWV is primarily controlled by the temperatures in the tropical tropopause layer (TTL) as air is transported through it (e.g. Mote et al., 1996;Fueglistaler et al., 2009) and by production from oxidation of methane (e.g. Brasseur and Solomon, 2005). The LMS SWV is controlled by 25 three major sources, including the transport of overworld air by the downward branch of Brewer-Dobson circulation, adiabatic quasi-horizontal transport from the tropical upper troposphere, and diabatic cross-tropopause transport due to deep convection (Dessler et al., 1995;Holton et al., 1995;Plumb, 2002;Gettelman et al., 2011). https://doi.org/10.5194/acp-2020-495 Preprint. Discussion started: 29 June 2020 c Author(s) 2020. CC BY 4.0 License.
The response of SWV to climate change can be partitioned into two components: the fast response and slow response. The addition of a radiatively active constituent to the atmosphere can influence the atmosphere even before the surface 30 temperature changes, leading to changes in SWV. This is often referred to as an "adjustment" to the forcing, and is generally considered part of the external forcing (e.g. Sherwood et al., 2015). We will refer to this as the "fast response" of SWV to the forcing. The slow response is the component in the SWV change that is coupled to changes of the surface temperature, which occurs on longer time scales. This slow response means that SWV could be an important positive feedback to global warming (Forster and Shine, 2002;Dessler et al., 2013;Huang et al., 2016;Banerjee et al., 2019). Banerjee et al. (2019) 35 have shown that, when CO2 is abruptly quadrupled, the change in SWV mainly consists of the slow response and that the fast response is less important.
Previous studies have shown that climate models, which are able to accurately reproduce observed interannual variations in SWV (Dessler et al., 2013;Smalley et al., 2017), robustly project a positive long-term trend in overworld SWV at entry level with a warming climate due to increasing GHGs (Gettelman et al., 2010;Dessler et al., 2013;Smalley et al., 2017). This is 40 mainly due to a warmer tropopause (Thuburn and Craig, 2002;Gettelman et al., 2010;Lin et al., 2017;Smalley et al., 2017;Xia et al., 2019), which is controlled, to some extent at least, by the warming surface (Gettelman et al., 2010;Shu et al., 2011;Dessler et al., 2013;Huang et al., 2016;Revell et al., 2016;Lin et al., 2017;Smalley et al., 2017;Banerjee et al., 2019). Dessler et al. (2016) suggested that increases in convective injection into the stratosphere due to a warming climate may also be contributing to the trend in entry SWV. In the LMS, the climate models show larger increases in SWV (Dessler 45 et al., 2013;Huang et al., 2016;Banerjee et al., 2019). It is not known how SWV responses to different forcing agents.
The goal of this study is to investigate the response of both overworld and LMS SWV to forcing agents with different physical properties. We will explicitly investigate the fast and slow responses in SWV and compare them. We will also investigate how SWV responds to surface temperature change when the climate is forced by different forcing agents.

The PDRMIP set-up
In this paper, we analyze nine models from the Precipitation Driver and Response Model Intercomparison Project (PDRMIP) (Samset et al., 2016;Myhre et al., 2017;Tang et al., 2018Tang et al., , 2019. These are Coupled Model Inter-comparison Project phase 5 (CMIP5) era models (Table 1) and each performed a baseline and multiple climate perturbation experiments (Table 1).
In the perturbation experiments, perturbations on a global scale are applied abruptly at the beginning of the model 55 simulation. The five core experiments include a doubling of CO2 concentration (2xCO2), a tripling of CH4 concentration (3xCH4), a 2% increase in solar irradiance (2%Solar), an increase of present-day black carbon concentration or emission by https://doi.org/10.5194/acp-2020-495 Preprint. Discussion started: 29 June 2020 c Author(s) 2020. CC BY 4.0 License. factor of 10 (10xBC), and an increase of present-day SO4 concentration or emission by factor of 5 (5xSO4). In addition to the five core experiments, a subset of models also performed additional perturbation experiments: an increase in CFC-11 concentration from 535 ppt to 5 ppb (hereafter, 10xCFC-11), an increase in CFC-12 concentration from 653.45 ppt to 5 ppb 60 (hereafter,, an increase in N2O concentration from 316 ppb to 1 ppm (hereafter, 3xN2O), an increase tropospheric O3 concentration used in MacIntosh et al. (2016) by factor of 5 (5xO3), and an increase of present-day black carbon with shorter lifetime by factor of 10 (10xBCSLT). Table 1 provides details about the models and the perturbations each one simulated.
The perturbations in GHGs and solar irradiance are relative to the models' baseline simulations, in which the concentration 65 of the GHGs and solar irradiance are either at present-day levels or pre-industrial levels. The perturbations in the aerosols depend on whether it is possible to prescribe aerosol concentrations in the models. For models that are able to prescribe aerosol concentrations, the aerosol perturbations are based on a multi-model mean baseline aerosol concentration in 2000 obtained from the AeroCom Phase II initiative (Myhre et al., 2013). For those that are only able to produce aerosols through emissions, the perturbation is applied by increasing the emissions by the factors listed above. The 10xBCSLT experiment is 70 performed only by models that are able to prescribe aerosol concentrations.
Each perturbation experiment is performed in two configurations: a fixed sea surface temperatures simulation ("fixed SST") and a fully coupled (slab ocean for CAM4 only) simulation. The fixed SST simulations use the SST climatology at either present-day level or pre-industrial level. The fixed SST simulations are at least 15 years and the coupled simulations are at least 100 years. 75

Fast response and slow response
When available, SWV mixing ratio is obtained directly from the specific humidity output by each model simulation. For the models that do not output specific humidity (CAM5, GISS-E2-R, and MIROC-SPRINTARS), we calculate specific humidity by multiplying the models' relative humidity by the saturation mixing ratio with respect to ice calculated using model temperature and pressure. 80 We define ∆SWV, the change in SWV mixing ratio in response to a particular perturbation, to be the difference between SWV in the perturbed coupled run and that in the baseline coupled run. As discussed above, the ∆SWV can then be broken down into the two components: the fast response (∆SWVfast) and slow response (∆SWVslow). We compute results in the  Shu et al., 2011;Smalley et al., 2017). But recent studies report that the climate is most sensitive to changes in water vapor in the LMS (Solomon et al., 2010;Dessler et al., 2013;Banerjee et al., 2019), so we also investigate that region.
We use the fixed SST simulations to get DSWVfast, the change in SWV before the surface temperature changes. DSWVfast is 90 the difference between the SWV mixing ratio averaged over the last 10 years in the fixed SST run with the forcing perturbation and the SWV mixing ratio averaged over the last 10 years in the fixed SST run with the baseline atmosphere.
We calculate DSWVslow as DSWV minus DSWVfast. To estimate the time series of DSWVslow, we use annual mean DSWV over the entire coupled run period (at least 100 years) minus the ten-year average DSWVfast. To estimate equilibrium DSWVslow, we use a regression method similar to the methodology introduced by Gregory et al. (2004). The basic concept is 95 that we regress the annual mean global average net downward radiative flux (R) at the top of atmosphere (TOA) against the annual mean DSWV averaged at TLS, NH LMS, or SH LMS. The equilibrium DSWV is where the linear fit intercepts at R=0. Then we simply subtract DSWVfast from the equilibrium DSWV to estimate equilibrium DSWVslow.
These regressions can be very noisy and yield highly uncertain parameters, particularly for perturbations with relatively small amounts of radiative forcing and warming. To account for this, we first fit the R and DSWV time series using an 100 exponential function ( ( ) = + 1 • ,-//0 + 2 • ,-//2 ), and then do the regression using the fitted time series. For fully coupled models, we constrain 1 to be within the range of 4±2 years and 2 to be within the range of 250±70 years; for CAM4, in which the atmosphere is coupled to a slab ocean, we constrain 1 to be within the range of 4±2 years. We then compute the best fit of all parameters. The ranges for the time constants are based on previous estimations of climate system time scales (Geoffroy et al., 2013). We estimate the DSWV-intercept at R=0 by regressing the fitted R and DSWV data over 105 the last 30 years, since the relation between R and DSWV is not necessarily linear over the entire 100-year period. The slow and fast responses of other variables, such as global average surface temperatures and cold point temperatures are computed using the same method.
We tested this method in a climate model that nearly reaches the equilibrium climate state. We analysed runs of the fully ppmv in the NH LMS, and 9.69 ppmv in the SH LMS. We expect this to be close to equilibrium ∆SWV because the trend in global average surface temperature over the last 500 years of the 4xCO2 run is 0.02 K per century. We use the regression 115 method to estimate the equilibrium DSWV using MPI-ESM1.1 water vapor mixing ratio time series over the first 100 years and obtain estimates of 4.38 ppmv in the TLS, 20.01 ppmv in the NH LMS, and 9.07 ppmv in the SH LMS; these yield https://doi.org/10.5194/acp-2020-495 Preprint. Discussion started: 29 June 2020 c Author(s) 2020. CC BY 4.0 License. differences of 0.22 ppmv in the TLS, 2.39 ppmv in the NH LMS, and 0.62 ppmv in the SH LMS. Thus, our method underestimates the true equilibrium value by 5% in the TLS, 11% in the NH LMS, and 6% in the SH LMS.
Uncertainty for slow and fast responses of different quantities shown in this paper are obtained from Monte Carlo samples as 120 follows: For each perturbation, we randomly sample with replacement 100,000 times for each model that performed that perturbation and from these samples compute the 2.5%-97.5% percentiles.

The slow stratospheric water vapor response
We show equilibrium DSWVslow and its percentage contribution to the total equilibrium DSWV in Figure  In all cases, we calculate this by differencing the average of the last 10 years of the fixed SST run with the perturbed atmosphere from the same quantity in the fixed SST run with the baseline atmosphere. The equilibrium global averaged surface temperature changes (DTs), estimated using the regression method described in Section 2.2 and normalized by ERF, are plotted in Fig. 2b. The ensemble average DTs/ERF shows general agreement across different perturbations. This quantity is the inverse of the feedback parameter l (e.g. Dessler and Zelinka, 2015), so Fig. 2b implies that the climate sensitivity to 135 these different perturbations is similar (Richardson et al., 2019). We also list the quantities for each model and perturbation in Table S1.
In each region, the magnitude of ensemble average DSWVslow/ERF shows general agreement for different perturbations. The magnitudes of DSWVslow/ERF in the LMS tend to be larger than those in the TLS . This is consistent with previous studies, which showed that the long-term trend in SWV over the century in climate models is largest near the LMS 140 tropopause (Dessler et al., 2013;Huang et al., 2016;Banerjee et al., 2019). This reflects different transport pathways into the LMS, including the downward transport by the Brewer-Dobson circulation, quasi-horizontal isentropic mixing from tropical troposphere, and convective influence (Dessler et al., 1995;Holton et al., 1995;Plumb, 2002;Gettelman et al., 2011).
In the LMS, the ensemble average DSWVslow/DSWV ratio is close to 100% for many perturbations (Figs. 1e-f). In the TLS, the ensemble average DSWVslow/DSWV ratio is generally above 50%, with a few exceptions. We will discuss this in detail in 145 Section 3.3.
We note that inter-model variability in DSWVslow/ERF and DSWVslow is generally consistent for different perturbations. For example, HadGEM3 produces larger responses than the rest of the ensemble for most perturbations (Figs. 1a-c, Table S1), likely connected to larger surface warming per ERF than the rest of the ensemble (Fig. 2b). GISS-E2-R and MIROC-SPRINTARS have DSWVslow/ERF and DSWVslow values generally below the rest of the ensemble (Figs. 1a-c, Table S1), 150 likely connected to smaller surface temperature changes per ERF (Fig. 2b).

The slow stratospheric water vapor response and the surface temperature change
Our results show that, in most climate perturbations analyzed in this study, the equilibrium response of water vapor in both the TLS and the LMS is dominated by ΔSWVslow, which is the component mediated by surface temperature change. To directly quantify how SWV responds to surface temperature across a range of different climate change mechanisms, we 155 linearly regress the time series of annual mean ΔSWVslow over the entire period of the coupled simulations (at least 100 years) against the time series of annual mean global averaged surface temperature change (ΔTs). This is similar to the analysis of Banerjee et al. (2019), who did this for quadrupled CO2 perturbation, but we do this for multiple perturbations.
The scatter plot for each perturbation and model is shown in supplement ( Figures S1-3). For most perturbations and models, the ΔSWVslow time series in both the TLS and the LMS is positively correlated with the ΔTs time series, supporting the 160 hypothesis that the surface temperature change contributes to the long-term trend in SWV for most cases. Figure 3 shows the slopes of the regression for all perturbations and models. The ensemble average and uncertainty of the slopes are obtained from Monte Carlo samples: For each model and perturbation, we first randomly sample the slope 100,000 times, assuming a Gaussian distribution. Then, for each perturbation, we sample from the slope distributions with replacement 100,000 times for each model that performed that perturbation and from these samples compute the ensemble 165 mean and 2.5%-97.5% percentiles.
In both the TLS and LMS, the slopes from different perturbations show general agreement (Fig. 3). In the TLS, the ensemble and perturbation average slope is 0.35 ppmv K -1 with a 95% confidence interval of 0.28-0.44 ppmv K -1 (Fig. 3a). The LMS ΔSWVslow time series has stronger correlations with the ΔTs time series (Figures S1-3) and produces larger sensitivities . Specifically, the ensemble and perturbation average slope is 2.1 ppmv K -1 in the NH, and is 0.97 ppmv K -1 in 170 the SH, with 95% confidence intervals of 1.82-2.39 ppmv K -1 and 0.79-1.15 ppmv K -1 , respectively. The larger LMS SWV sensitivity reflects a different mix of transport pathways into the LMS compared to the TLS (Dessler et al., 1995;Holton et al., 1995;Plumb, 2002;Gettelman et al., 2011). Our results are similar to those of Dessler et al. (2013) and Smalley et al. (2017) despite the fact that they used 500-hPa temperature as their regressor.
We show that the relation between ΔSWVslow and ∆Ts time series can be extended to the entire stratosphere (Figs. 4a). We 175 re-gridded the zonal mean ΔSWVslow from all models and perturbations onto the same pressure-latitude grid (10 hPa above 100 hPa and 50 hPa below 100 hPa, 4 degrees latitude) and regress the ΔSWVslow time series at each grid point against global average ∆Ts time series. The ensemble and perturbation average slope of the linear fit at each grid point is shown in  Fig. S4). Since the vertical gradient of water vapor is large, we plot the percentage change of mixing ratio per degree K relative to the baseline. Lapse rate tropopause, the lowest level 180 where the lapse rate decreases to 2 K km -1 , also plotted, is obtained using the atmospheric temperatures from the baseline coupled run and ensemble averaged.
We clearly see the larger sensitivity of ΔSWVslow to ΔTs in the LMS than in the overworld. In the LMS, the slope has a hemispheric asymmetry, with larger values in the NH. This is consistent with previous studies, which showed that isentropic transport brings more tropospheric water vapor to the NH than the SH (Pan et al., 1997(Pan et al., , 2000Dethof et al., 1999Dethof et al., , 2000185 Ploeger et al., 2013). In addition, convective moistening may be more important to the NH due to more land in the NH and, consequently, more convection (Dessler and Sherwood, 2004;Smith et al., 2017;Ueyama et al., 2018;Wang et al., 2019).
We see large values in the tropical upper troposphere, which is the main part of the tropospheric water vapor feedback. The sensitivity declines as one ascends through the TTL. Once above the TTL, the sensitivity in the overworld is relatively uniform with altitude. 190

The fast stratospheric water vapor response
Figure 1 also shows the DSWVfast normalized by the ERF (Figs. 1g-i) as well as its contribution to total equilibrium DSWV (Figs. 1j-l). As discussed previously, DSWVfast is the change in SWV due to the perturbation, but before the surface temperature has responded. For most perturbations, especially in the LMS, ΔSWVfast/ERF is smaller than DSWVslow/ERF, with a magnitude of a few tenths of a ppmv×(Wm -2 ) -1 . 195 For 2xCO2, the near-zero TLS ΔSWVfast/ERF is the result of cancellation between cooling by a strengthening Brewer-Dobson circulation and increased local radiative heating (Lin et al., 2017). Some other GHG forcing agents, however, produce larger TLS ΔSWVfast/ERF and contributions in the TLS. The ensemble average DSWVfast from 10xCFC-12 and 10xCFC-11 contribute about half of the total ΔSWV, respectively (Fig. 1j). This is a consequence of halocarbons producing more TTL warming per Wm -2 by efficiently absorbing upwelling longwave radiation from the troposphere in the 200 atmospheric window (Jain et al., 2000). Fig. 5 shows the fast temperature response per unit ERF due to different perturbations and it shows heating in the TTL for both 10xCFC-12 and 10xCFC-11.
Increases of tropospheric O3 (5xO3) reduce the shortwave radiation absorption by stratospheric O3, which cools the stratosphere (Fig. 5). Meanwhile, the O3 in the upper troposphere absorbs short wave radiation and heats the TTL (Fig. 5), which results in larger TLS DSWVfast/ERF magnitude than DSWVslow/ERF and larger contributions to total equilibrium 205 DSWV (77%) (Figs. 1g and 1j). Tropospheric O3 is also transported to the LMS region, which heats the LMS by absorbing short wave radiation and results in larger LMS ΔSWVfast/ERF magnitude than DSWVslow/ERF (Figs. 1h-j). We note that our conclusion on 5xO3 is based on only one model, MIROC-SPRINTARS.
The 3xCH4 also includes multiple models that produce larger TLS ΔSWVfast/ERF magnitudes and contributions than DSWVslow/ERF. Figure 5 shows TTL heating for 3xCH4. The TTL heating could be due to the shortwave absorption by CH4, 210 which is explicitly treated in models including CAM5, CanESM2, MPI-ESM, and MIROC-SPRINTARS (Smith et al., 2018). These models are also the ones that produce TLS ΔSWVfast contributions larger than 50% in 3xCH4 (Figs. 1g and 1j).
ΔSWVfast from 10xBC dominates total equilibrium DSWV in the TLS, with ensemble average contribution of 84%. The magnitude of the ensemble average ΔSWVfast/ERF from 10xBC is also larger than any other perturbations in each region.
This occurs because the 10xBC strongly absorbs shortwave radiation, causing large heating of the tropopause region in both 215 the tropics and extra-tropics. Figure 5 shows the 10xBC gives by far the most warming per unit ERF.
The 10xBC DSWVfast/ERF in the NH and SH LMS contributes to about 50% of the total equilibrium DSWV, with smaller magnitudes in the SH . This is because the total amount of black carbon is smaller in the SH (Myhre et al., 2017), since black carbon is a combustion product and is predominantly emitted over the NH continents (Ramanathan and Carmichael, 2008). The 10xBCSLT DSWVfast also contributes about 50% of the total 10xBCSLT DSWV. The 220 10xBCSLT does not produce as strong a DSWVfast/ERF as 10xBC, since the reduction in BC lifetime leads to less BC in the TTL and therefore less heating per unit ERF.
We quantify control of TLS ΔSWVfast by the fast TTL temperature adjustments across a range of different climate perturbations by regressing the TLS ΔSWVfast against the fast response of the cold point temperature (ΔTCPfast). To estimate ΔTCPfast in the models, we first find the minimum temperature in the profile at each grid point in the fixed SST runs (no 225 interpolation is done, we simply find the minimum temperature on the output model levels). These minimum temperatures are then averaged between 30°N -30°S to yield TCPfast in each run. ∆TCPfast is the difference between TCPfast in the perturbed model run minus that in the baseline runs.
We find that TLS ΔSWVfast is strongly correlated with ΔTCPfast across all perturbations and models (Fig. 6a), with a slope of 0.52 ppmv K -1 and a 95% confidence interval of 0.43 to 0.61 ppmv K -1 . Randel and Park (2019) pointed out that the slope 230 from the Clausius-Clapeyron relationship evaluated near the tropical tropopause is close to this value, about 0.5 ppmv K -1 .
We also separately plot the slopes between ΔSWVfast and ΔTCPfast for each perturbation (Figs. 6d-f). For the perturbations that have more than five participating models, including 2xCO2, 3xCH4, 2%Solar, 10xBC, 5xSO4, and 10xCFC-12, we https://doi.org/10.5194/acp-2020-495 Preprint. Discussion started: 29 June 2020 c Author(s) 2020. CC BY 4.0 License. calculate the linear regression between ΔSWVfast and ΔTCPfast from the models and show the slopes and 95% confidence intervals. For the perturbations that have fewer participating models, including 10xCFC11, 3xN2O, 5xO3, and 10xBCSLT, 235 we plot the ratio ΔSWVfast/ΔTCPfast and show only the ensemble averages. The slopes produced by different perturbations show general agreement (Fig. 6d). The larger uncertainty in the slopes produced by 2%Solar and 10xCFC-12 occurs because both the ΔTCPfast and ΔSWVfast produced by different models are similar and therefore the slope of the linear regression is uncertain. Overall, we find that the fast response of TTL temperature is a good predictor for the TLS ΔSWVfast across a range of different climate mechanisms and across multiple models. 240 For the LMS ΔSWVfast, the ΔTCPfast does not show a control as strong as that in the TLS (Figs. 6b-c) due to the fact that TTL temperatures are only one factor that influences the LMS. In addition, the regression between ΔSWVfast and ΔTCPfast across all perturbations at each grid point in the pressure-latitude domain shows that the slope (% K -1 ) follows the transport pattern of the BDC (Fig. 4b). The slope is large in the tropical overworld stratosphere and become weaker as one moves poleward and downward in the extra-tropics below 150 hPa. The value is lower in the LMS, again consistent with the fact 245 that water vapor in the LMS is controlled by several processes, not just TTL cold-point temperature. Clearly, more work on this is warranted.

Summary
It is of great interest for the climate community to understand how SWV changes when the climate changes since SWV plays an important role in the Earth's radiative budget and stratospheric ozone chemistry (Solomon et al., 1986(Solomon et al., , 2010250 Dvortsov and Solomon, 2001;Forster and Shine, 2002). In this study, we investigate the response of stratospheric water vapor (SWV) to a range of different climate forcing mechanisms using a multi-model and multiple forcing agent framework.
To better understand the SWV response (DSWV), we partition it into two parts: the slow response (DSWVslow) and the fast response (DSWVfast). The DSWVfast is the change in response to a perturbation on short time scales, before the surface temperature has responded. DSWVslow occurs on longer time scales and is coupled to the surface temperature change. Our results show that, for most perturbations, ∆SWV in the tropical lower stratosphere (TLS) and in the lowermost stratosphere 260 (LMS) (200 hPa, 50°N-90°N and 50°S-90°S) is dominated by DSWVslow (Fig. 1).
Analysis of DSWVslow shows that a warming surface increases SWV (Figures S1-3). Furthermore, the response of SWV to the surface temperature change has a similar sensitivity across different climate perturbations in both the overworld https://doi.org/10.5194/acp-2020-495 Preprint. Discussion started: 29 June 2020 c Author(s) 2020. CC BY 4.0 License. stratosphere and the lowermost stratosphere (Figs. 3 and 4a). Specifically, the ensemble and perturbation average slope is 0.35 ppmv K -1 in the TLS, 2.1 ppmv K -1 in the northern hemispheric (NH) LMS, and 0.97 ppmv K -1 in the southern 265 hemispheric (SH) LMS (Fig. 3).
The fast response of SWV from most perturbations are weak compared to the slow response and therefore plays a smaller role in ∆SWV (Fig. 1). However, for forcing agents that directly heat tropopause levels (Fig. 5), ΔSWVfast makes a larger contribution to ∆SWV. In particular, when climate is perturbed by 10xBC, the ΔSWVfast dominates the ΔSWVslow and has larger magnitude than any other perturbed simulations in both the TLS and LMS. This occurs because black carbon absorbs 275 shortwave radiation in the atmosphere and directly heats the temperatures at tropopause levels. Other forcing agents also heat the tropopause levels and increase ΔSWVfast through absorption of shortwave radiation or longwave radiation at the atmospheric window range (3xCH4, 5xO3, 10xBCSLT, 10xCFC-12, 10xCFC-11), but are not as strong as 10xBC.
The TLS DSWVfast is controlled by the fast response of the cold point temperature across different climate change mechanisms (Fig. 6), with a slope of 0.52 ppmv K -1 , which is consistent with the Clausius-Clapeyron relationship evaluated 280 near the tropical tropopause (Randel and Park, 2019). The control of cold point temperature fast response over ΔSWVfast is stronger in the tropical overworld and becomes weaker at higher latitudes in the LMS below 150 hPa (Fig. 4b).
Data availability: The PDRMIP data can be downloaded from this website: https://cicero.oslo.no/en/PDRMIP.

Competing interests. The authors declare that they have no conflict of interest.
Author contribution: Xun Wang performed analyses and wrote the original draft. Andrew E. Dessler provided the 285 conceptualization, guidance, and editing.