Direct forcing of stratospheric sulfate
The underlying physical processes behind the injection of SO2 into
the atmosphere have been widely studied thanks to the various explosive
volcanic eruptions of the 20th century. For instance, in the year following
the Mount Pinatubo eruption of June 1991, when 7–10 Tg S were
injected into the stratosphere , a sharp
reduction in the TOA net radiative flux was observed
(∼ 2.5 Wm-2) , as well as a significant
drop in global surface temperatures of about 0.5 K
. This was calculated as a monthly mean for September 1992,
compared to pre-Pinatubo levels. However, more recent results with detrended
analyses have shown that the Pinatubo volcanic impact on
surface temperatures was probably overestimated by about a factor of 2, with
a cooling estimate of 0.14 and 0.32 K, globally and over land,
respectively.
These effects can be explained by SO2 oxidation into SO4
followed by the formation of H2O–H2SO4 supercooled liquid
droplets, which create an optically thick cloud that reflects part of the
incoming solar radiation. This results in a surface cooling and a local
stratospheric warming. The stratospheric warming is due to changes in
diabatic heating rates produced by aerosol absorption of solar near-infrared
and planetary radiation and by the ozone absorption of the additional UV
radiation scattered by the volcanic aerosols .
When considering the effects of the proposed injection of sulfur into the
atmosphere, however, a series of factors must be taken into account,
complicating the analogy between this kind of geoengineering experiments and
volcanic eruptions. Obviously, the amount of sulfur and the height and
latitude at which it is injected in a geoengineering experiment all play a
prominent role in its related effects. Some recent papers, such as
and , analyzed a series of
geoengineering experiments accounting for the different factors previously
mentioned. Their results show that the relation between injected SO2
and the resulting sulfate mass burden is nonlinear, with larger injection
rates producing a lower efficiency of SG. This is due to the fact that
injections of larger amounts of SO2 lead to the formation of larger
aerosol particles by gas condensation, which are rapidly removed from the
stratosphere by gravitational settling (see Fig. , with
calculated vertical profiles of the aerosol effective radius).
Aside from the reduction in the aerosol lifetime, the size of the produced
aerosol particles also influences the amount of scattered radiation, because
the sulfate scattering efficiency peaks at a particle radius of around
140 nm and decreases as aerosols become larger .
The highest burden to injection ratio is achieved for stratospheric
injections between 30∘ N and 30∘ S ,
because gas condensation and particle coagulation are both reduced with
SO2 injection spanning over a broader latitude. The altitude also
plays a significant role in determining the aerosol lifetime, due to a faster
sedimentation removal in the upper troposphere (UT) when the sulfur injection
is localized closer to the tropical tropopause layer (TTL)
.
Annual averaged vertical profiles of aerosol effective radius
(µm) in the tropical stratosphere
(25∘ S–25∘ N), with increasing geoengineering injection of
SO2 (see legend). The heavy dashed line indicates the mean tropical
tropopause. Profiles are calculated in the University of L'Aquila
Chemistry–Climate Model (ULAQ-CCM), which includes explicit gas–particle
conversion and aerosol microphysics .
As shown in , the injection of
5 Tg SO2 yr-1 produces, according to the models
used in the experiment G4, a net TOA radiative forcing (RF) of -1.54,
-1.27, -1.31 and -0.73 Wm-2 for the University of L'Aquila
Chemistry–Climate Model (ULAQ-CCM), NASA Goddard Earth Observing System
Chemistry–Climate Model (GEOSCCM), Goddard Institute for Space Science Model E2-R
(GISS-E2-R) and Model for Interdisciplinary Research on Climate Earth System
Model coupled with an Atmospheric Chemistry model (MIROC-ESM-CHEM), respectively (see
for model description and details). The different results
are mainly dependent on the (calculated, or imposed in one case) different
aerosol optical depth (AOD) and size distribution among models. It should
also be considered that, in general, even with the same AOD distribution,
models may produce different radiative responses depending on the adopted
radiation scheme . Other RF values are available from the
literature, for a variety of conditions of sulfur injection (amount and
altitude, mainly). With a linear scaling to
5 Tg SO2 yr-1 (in the case of different injection
values), we get the following values: -1.13 Wm-2
; -1.17 Wm-2 ;
-1.53 Wm-2 ; -1.4 and
-1.0 Wm-2 ; and -0.55 Wm-2
. In two cases, the forcing value was reported as the
surface shortwave (SW) RF : it has been
converted to a net TOA RF by scaling the SW surface value with a factor
(25-8)/20, where 25, 20 and 8 are the approximate factors to derive TOA SW,
surface SW and TOA adjusted longwave (LW) RFs from the stratospheric AOD.
From these RF values available in the literature, we may derive a mean value
of -1.16 ± 0.33 Wm-2.
Changes in circulation and its feedback
While on the one hand these results show that SG leads to the desired effect
of (at least partially) offsetting the positive RF of increasing well-mixed
greenhouse gases (WMGHGs), on the other hand they show that SG effects, such
as the lower-stratospheric warming, must be carefully studied.
Enhanced lower-stratospheric diabatic heating rates after major explosive
volcanic eruptions and the consequent temperature increase were well
documented both in observations and through
modeling experiments . The tropical
lower-stratospheric warming induces a significant increase of westerly winds
from the thermal wind equation, with peaks at midlatitudes in the
mid-stratosphere. These dynamical changes tend to increase the amplitude of
planetary waves in the stratosphere and to enhance the tropical upwelling in
the rising branch of the Brewer–Dobson circulation (BDC)
.
The effects of the aerosol heating rates on the quasi-biennial oscillation
(QBO) under geoengineering conditions have been analyzed in the
aforementioned study by using GEOSCCM, which includes an
internally generated QBO. Four different experiments were designed, using
5 Tg SO2 yr-1 for the first two and
2.5 Tg SO2 yr-1 for the others, injected at
different altitudes (16–25 and 22–25 km; both at the Equator and
0∘ longitude in a single lat–lon box). They found that SG perturbs
the QBO by prolonging the westerly phase in the 20–50 hPa layer with an
increasing stratospheric SO4 mass burden (ranging from
1.5 Tg S for the 16–25 km injection of
2.5 Tg SO2 yr-1 to 4.7 Tg S for the
22–25 km injection of 5 Tg SO2 yr-1).
also mention a perturbation of the QBO in SG simulations
performed with the ECHAM-HAM model. This was an
ensemble of simulations with variable SO2 injection
(1–100 Tg S yr-1), altitude and latitude of injection
(60 and 30 hPa; Eq-2.8∘ N; 5∘ S–5∘ N;
30∘ S–30∘ N; all in a single longitude box centered at
122.3∘ E). Their simulations include explicit aerosol microphysics,
so that the effects of the perturbed QBO on the aerosol size distribution are
taken into account. They found that an injection of about
8 Tg S yr-1 would cause a slowing of the QBO oscillation
with a constant QBO westerly phase in the lower stratosphere with overlaying
easterlies, consistent with the findings by . The overall
conclusion of both these studies is that a stratospheric sulfur injection
could dramatically alter the QBO periodicity, up to producing a permanent
westerly phase in the lower stratosphere, thus reducing the meridional
transport efficiency .
The SO4 stratospheric lifetime in the simulations included in
was approximately 1.2 and 1.8 years for sulfur injection
in the altitude layers 16–25 and 22–25 km, respectively. However,
it is interesting to note that the sulfate lifetime is systematically longer
in the 5 Tg SO2 yr-1 case with respect to the
2.5 Tg SO2 yr-1 injection case (∼ 1.9 years vs.
∼ 1.7 years with injection in the 22–25 km layer and
∼ 1.25 years vs. ∼ 1.2 years with injection in the 16–25 km
layer). The higher heating rates produced by the aerosol in the
5 Tg SO2 yr-1 case are responsible for a stronger modification
of the stratospheric circulation, resulting in the QBO changes and increased
tropical upwelling, and hence a better confinement of the particles in the
tropical pipe . This reduces the amount of
aerosol that may be transported downwards across the extratropical
tropopause in the lower branch of the BDC. A compact summary of all these
feedback mechanisms is presented in Fig. (superimposed to the
calculated sulfate mass density anomaly due to an injection of
5 Tg SO2 yr-1).
The prolonging of the aerosol lifetime found by , however,
could be canceled if the microphysical effects of the QBO-dependent sulfur
confinement in the tropical pipe were taken into account. In the simulations
by using the ECHAM-HAM model, which includes a
representation of aerosol microphysics, the enhanced aerosol tropical
confinement under conditions of a locked QBO westerly phase in the lower
stratosphere decreases the SG aerosol lifetime; this is because the tighter
tropical confinement of the aerosol also leads to larger particles and
therefore a more efficient gravitational settling (U. Niemeier, personal
communications, 2016) (see Fig. b).
Many of the previously cited studies have focused on specific aspects of
formation, transport and removal of stratospheric aerosols under
geoengineering conditions. As noted above, significant feedback mechanisms
exist among the magnitude and location of SO2 injection, aerosol
microphysics, background stratospheric dynamics, aerosol-induced surface
cooling and stratospheric heating rates, and induced changes in the
stratospheric circulation and stratosphere–troposphere exchange. This means
that a significant improvement on the knowledge of direct and indirect
effects of SG may be obtained through model experiments designed in such a
way that all these aspects are explicitly considered and interacting with
each other. One important limitation of many of the above-cited studies is
the use of atmosphere-only models forced by prescribed sea surface
temperatures (SSTs), so that an explicit interaction of geoengineering
aerosols with surface ocean is not considered. A missing explicit aerosol
microphysics is another limitation for some of these studies: in this case,
the increased gas–particle conversion cannot feed back on the aerosol size
distribution shape and finally on the particle sedimentation rate and aerosol
optical properties for the radiative transfer calculations.
(a) Annually and zonally averaged sulfate mass density
anomalies (µgm-3), due to a geoengineering injection of
5 Tg SO2 yr-1, with respect to a RCP4.5
background atmosphere. The aerosol mass density distribution is calculated in
the Goddard Earth Observing System Chemistry Climate Model (GEOSCCM), with SG
treated as described in . Arrows superimposed to the
aerosol distribution indicate the main transport pathways of the aerosol
particles, as explained in (b). The white dashed line shows the mean
tropopause; the dash-dotted white lines highlight the stratospheric tropical
region. The sensitivity of each dynamical effect to the SO2 injection
is highlighted in (b), along with the physical mechanisms driving
the perturbation and the net effect on sulfate lifetime and optical
depth.
Indirect radiative forcing
In the following sections we shall summarize the indirect changes caused by
the SG-induced stratospheric warming and surface cooling. This section
answers the question of whether any of these indirect effects could significantly
counteract or enhance the primary goal of SG of counteracting the positive RF
from WMGHGs.
Ozone
Early studies of the potential impact of SG on stratospheric ozone are those
of and .
focus on polar ozone and estimate that SG could favor stratospheric ozone
destruction and delay the recovery of the Antarctic ozone hole by
30–70 years. In addition, this ozone depletion produces a significant
increase of erythemal surface UV, up to 5 % at mid- and high latitudes
and 10 % over Antarctica . The polar ozone depletion is
favored by enhanced NOx removal via heterogeneous chemical
reactions on the surface of stratospheric sulfate aerosols, as in the case of
major volcanic eruptions taking place with high atmospheric levels of
chlorine and bromine species .
and analyze the SG impact in chemical
ozone loss rates and find that the chemical ozone changes are significantly
impacted by the strong reduction of the NOx cycle, due to the
efficient NOx-to-HNO3 conversion on the surface of
sulfate aerosols. The NOx depletion, in turn, favors an increase of
HOx, Clx and Brx loss rates: the net effect on
the ozone column will then be time-dependent and regulated by the amount of
halogen species in the lower stratosphere. have calculated a
global ozone reduction of 4.5 % (i.e., ∼ 13 DU), for an
injection of 10 Tg SO2 yr-1 and assuming halogen concentrations
appropriate for the year 2000.
have run the GeoMIP G4 experiment from 2020 to 2070:
despite the constant stratospheric aerosol loading, the magnitude of the
geoengineering aerosol-induced ozone depletion is found to decrease in time,
due to the decreasing atmospheric concentration of chlorine and bromine
species. Two of the models used in this study (ULAQ-CCM and MIROC-ESM-CHEM)
even show a global ozone increase starting from about 2050, when the
NOx-driven chemical ozone increase is no longer over-balanced by
the HOx-, Clx- and Brx-driven ozone loss.
Model simulations in showed that SG produces changes in
stratospheric ozone due to a series of concurring factors, i.e., perturbation
of photolysis rates because of the increased AOD, enhanced heterogeneous
chemistry, and modifications of atmospheric dynamics. The models used in the
G4 experiment show significant changes in the ozone profile, with a decrease
in the tropical column between 100 and 30 hPa in the tropics, for the
combined effects of enhanced upwelling and losses in the chemical cycles.
Above that layer, ozone was found to increase because of the reduction of
NOx via enhanced heterogeneous chemistry. Combined with similar
changes in the extratropics, which are largely produced by modifications in
the chemical processes, an average total change SG-induced perturbation of
-2.8 ± 3.0 DU is calculated in the global mean ozone column,
considering decadal averages from 2020 to 2070 for the four models that ran the
G4 experiment (ULAQ-CCM, GEOSCCM, GISS-E2-R and MIROC-ESM-CHEM) and for the
two models that ran the G3 experiment (ULAQ-CCM and GISS-E2-R). In terms of
RF this produces a rather small negative result, on the order of
-0.04 Wm-2 : RF = -0.045 ± 0.035 Wm-2,
with the same decadal averages used for the global mean ozone column change.
Stratospheric water vapor
SG is expected to increase stratospheric water vapor concentration by warming
the TTL. In the stratosphere, the water vapor concentration is regulated by
the TTL temperature , combined with methane oxidation. The
higher the TTL temperatures, the more water vapor is able to enter the
stratosphere. However, when considering the behavior of the TTL in a
geoengineering scenario, we must consider two overlapping effects: an
upper-tropospheric cooling caused by the aerosol scattering, which cools the
surface and stabilizes the troposphere (thus reducing convective heating),
and a lower-stratospheric warming caused by the infrared absorption by the
aerosol particles. The amount of water vapor predicted in the stratosphere
will thus depend on how the models represent these processes
.
Water vapor contributes to global warming, since it works as a GHG both in
the troposphere and in the stratosphere .
Following the definition of radiative forcing, i.e., the net radiative flux
change at the tropopause with fixed tropospheric temperatures and adjusted
stratospheric temperatures, only stratospheric water vapor changes concur to
the determination of the RF associated with any considered anthropogenic
perturbation, SG in the present case. gave an estimate of
the RF of the SG-induced increase in stratospheric water vapor. At
100 hPa in the tropics, three out of four models produce a warming ranging
from +0.16 to +0.58 K that leads to an increase in water vapor
mixing ratio from 0.08 to 0.35 ppmv. This in turn produces a net
positive RF = 0.055 ± 0.025 Wm-2, considering decadal
averages from 2020 to 2070 for the three of the four models that ran the G4
experiment (ULAQ-CCM, GEOSCCM and MIROC-ESM-CHEM). The fourth model
(GISS-E2-R), on the other hand, predicts a TTL cooling with a decreased
amount of stratospheric H2O and thus a negative RF. This is partly
due to an underestimated lower-stratospheric aerosol warming, originated by
an insufficient tropical confinement of the aerosol cloud.
Upper-tropospheric ice
Several studies have proposed mechanisms by which the SG would affect
upper-tropospheric cirrus clouds, reaching, however, contradictory
conclusions. found that SG directly provides ice nuclei
(IN) of a larger size with respect to those in the unperturbed atmosphere,
resulting in a rather small increase in cirrus cloud coverage.
, on the other hand, found that SG would decrease cirrus
cloud coverage because of changes in temperature, vertical velocity and water
vapor updraft. The aerosol driven surface cooling, coupled with the
lower-stratospheric warming, stabilizes the atmosphere due to a decreased
vertical temperature gradient, thus reducing the available turbulent kinetic
energy and the vertical updraft . This results
in a decrease of the upper-tropospheric ice crystal formation, which in turn
produces a less efficient trapping of the planetary longwave radiation and a
reduction of the net atmospheric greenhouse effect. Figure presents
a compact summary of the dynamical perturbations induced by SG and relevant
for the ice particle formation via homogeneous freezing. Lower vertical
velocities force a decrease in ice crystal number concentration due to the
decreasing water vapor transport from below, with consequent lower
supersaturation. The temperature dependence is inverse, because lower
temperatures allow for more ice crystals, due to the slower depositional
growth and the higher nucleation rate .
(a) Schematic profile changes of upper
troposphere–lower stratosphere temperature (K) and UT vertical
velocity (cms-1) in the tropics, due to a geoengineering
injection of 5 Tg SO2 yr-1. The perturbation
scheme is based on the findings of and
. The dash-dotted black lines indicate the
region of ice particle formation (up to the mean tropopause). The sensitivity
of each thermal-dynamical effect to the SO2 injection is highlighted
in (b), along with the physical mechanisms driving the perturbation
and the net effect on UT ice optical depth.
Ice formation via homogeneous freezing
As clearly demonstrated in a number of papers focusing on the physical
processes taking place in the upper troposphere
, the formation of ice particles may take
place via heterogeneous and homogeneous freezing mechanisms. Airborne
measurements by reported typical concentrations of newly
formed ice crystals on the order of 0.3 cm-3 in a young cirrus
cloud at T=220 K in the upper troposphere of Northern Hemisphere
midlatitudes, in agreement with the model estimate of
based on the assumption of ice particle formation via homogeneous freezing.
The homogeneous freezing mechanism normally dominates in the upper
troposphere and involves water vapor freezing over liquid supercooled
particles (as sulfate aerosols or sulfate coated aerosols) when the ice
supersaturation ratio exceeds ∼ 1.5. In a SG perturbed atmosphere, more
sulfate aerosols are available in the upper troposphere with respect to
unperturbed background conditions thanks to extratropical downwelling and
gravitational settling from the lower stratosphere. However, the background
number density of sulfate aerosols in the upper troposphere is normally
already much larger than the number of ice particles that can form
. This means that the SG-driven increase of IN number
density has basically no effect on the population of ice particles, but we
may expect some impact on the ice particle size due to the larger size of IN
made available by SG. This is the main conclusion of , who
note that the more large geoengineered particles exist (of typical sizes
close to 0.5 µm), the less particles have to struggle against the
Kelvin effect and more droplets may grow to larger sizes. This study analyzes
in detail the direct SG impact on IN, as a complementary effect with respect
to the dynamical indirect effect investigated by . The
main conclusion of is that the microphysical impact on
cirrus clouds from geoengineered stratospheric sulfate aerosols is not an
important side effect. They estimate a resulting midlatitude average RF in
the range of +0.02 to -0.04 Wm-2, depending on upwelling
velocities and geoengineering scenario.This is consistent with the
conclusions by , who found that the effect of a perturbed
aerosol size distribution on the ice particle population formed via
homogeneous freezing is of secondary importance. It should be considered,
however, that the estimates from are based on box model
simulations and radiative transfer model calculations, and do not consider
the dynamical impact and the feedback to microphysics.
Ice formation via heterogeneous freezing
The other possible pathway for ice crystal formation is through heterogeneous
freezing, which requires solid nuclei as mineral dust or black carbon. In
this case, when the ice supersaturation ratio exceeds approximately 1.1,
heterogeneous freezing may start ; sulfate aerosols do
not act as potential IN in this case. and, indirectly,
have demonstrated that only the indirect dynamical
perturbation induced by SG may be capable of significantly perturbing the
number density of upper-tropospheric ice particles, with decreased vertical
velocities due to the enhanced atmospheric stabilization. As noted in
, the idea proposed in some studies that volcanic
eruptions may lead to enhanced ice crystal number concentrations was indeed
confirmed by ISCCP lidar measurements , whereas modeling
studies found only a weak aerosol effect even in the case of large
perturbations . However, it should be noted
that in the case of explosive volcanic eruptions (contrary to SG) there are
also solid ash particles injected in the lower stratosphere that will settle
down below the tropopause (although with a rather short lifetime for the
mass-dominant coarse mode), thus potentially contributing to some increase of
the upper-tropospheric IN population actually available for heterogeneous
freezing. have shown that mineral dust particles can
play an important role in cirrus cloud formation, because their ice active
fraction may be rather large (> 10 % for a supersaturation ratio close
to the homogeneous freezing threshold). However, this is not the case for the
proposed SG, where the homogeneous freezing mechanism actually dominates.
Recent studies by have quantified the
direct radiative effects produced by seeding upper-tropospheric cirrus ice
clouds with large IN. Although this is not directly related to our specific
discussion on SG side effects, it can be considered as indirect evidence of
the importance of correctly understanding the balance between the complex
microphysical processes regulating the formation and growth of
upper-tropospheric ice particles.
RF estimates from cirrus ice thinning
We may conclude that the assumption of limiting our discussion to the
indirect dynamical effect is a robust one and based on a sound physical
basis. have calculated a LW TOA
RF = -0.51 Wm-2 for cloud adjustment due to optically
thinner cirrus, under a SG injection of
5 Tg SO2 yr-1. However, we should keep in mind
that some degree of uncertainty remains for the processes regulating the
potential direct perturbation of upper-tropospheric ice crystals through
changes in the size distribution of sulfate aerosols acting as IN. In
addition, as noted by the authors themselves, one limitation of the study by
is that sea surface temperatures were prescribed. The
SG-induced cooling of the surface would on the one hand enhance the
atmospheric stabilization and then further reduce the vertical updraft and
cirrus ice optical depth (see Fig. ), but on the other hand it would
contribute to cooling the whole troposphere, thus favoring additional ice
crystal formation (see Fig. ). Although
suggest that in principle it would be important to redo the simulations with
a mixed-layer ocean, on the other hand they conclude that the overall
difference in the GCM response would be small in terms of UT ice anomalies.
As shown in for the atmospheric stabilization
resulting from tropospheric aerosols by non-explosive volcanoes, the combined
effect of the aerosol-induced tropospheric decrease in temperature and
updraft velocities produces a net global reduction of ice optical thickness
in the upper troposphere of 1.0×10-3 at λ=0.55 µm, which then causes a radiative forcing of
-0.08 Wm-2. This corresponds to an aerosol optical depth
increase of 5.3×10-3 and an average surface cooling of
0.07 K. The same ULAQ-CCM module for ice crystal formation via
homogeneous freezing has been applied to the SG case with stratospheric
injection of 5 Tg SO2 yr-1, obtaining a globally
averaged LW TOA RF = -0.45 Wm-2 due to optically thinner
cirrus, consistent with the findings of . A
corresponding net TOA RF = -0.30 Wm-2 was calculated in
all sky conditions, with the SW RF = 0.15 Wm-2 (i.e.,
34 % of the absolute LW RF). If this same SW / LW RF fraction is
applied, a net TOA RF = -0.34 Wm-2 is obtained for
, for the cloud adjustment due to optically thinner
cirrus.
Methane
Another indirect effect of SG is a lifetime modification for many long-lived
species. Among these species CH4 is particularly important, due to
its sensitivity to OH abundance and its impact on tropospheric chemistry. A
CH4 lifetime increase takes place for three main reasons
, all connected with a decrease in OH concentration, which
represents the main sink for methane: (a) the surface cooling directly
lessens the amount of water vapor in the troposphere, which in turn
diminishes the OH concentration. (b) A decrease in tropospheric UV occurs in
the tropics because of the stratospheric aerosols. This reduces the
production of O(1D), which in turns decreases the amount of OH produced by
the reaction O(1D) + H2O. (c) The increase of aerosol surface
area density (SAD) enhances heterogeneous chemistry in the mid- to upper
troposphere, reducing the amount of NOx and O3 production
and thus of OH production. The increased aerosol SAD causes significant ozone
depletion in the stratosphere, which results in an increase of UV radiation
able to reach the surface. However, this effect is overbalanced by the direct
scattering of solar radiation, so that the net amount of tropospheric UV is
reduced (except over the polar latitudes) . The high-latitude
UV increase has little effect over the methane lifetime, which is mostly
influenced from OH changes in the tropics.
In addition, it should be noted that the stratospheric aerosol heating rates
produce a strengthening of the BDC, where more stratospheric air is
transported from the stratosphere to the upper-troposphere extratropics.
Since the concentration of methane is lower in the stratosphere than in the
troposphere, this strengthening of the BDC leads to a CH4 decrease in
the upper troposphere. All these effects together produce a longer lifetime
of CH4 that is estimated by the ULAQ-CCM to increase from 8 years for
RCP4.5 to 9 years for SG with injection of
5 Tg SO2 yr-1. According to the model, such a
lifetime increase is estimated to produce a positive TOA
RF = +0.11 ± 0.04 Wm-2 , as an
average from year 2020 to 2090.
To what extent may SG balance WMGHG RF?
Here we discuss how the estimated net RF from direct and indirect effects of
SG may be compared with the positive RF associated with increasing WMGHGs.
The current IPCC scenarios for the next century will produce by 2100 a RF
relative to 2011 of 0.3 Wm-2 (RCP2.6), 2.2 Wm-2
(RCP4.5), 3.7 Wm-2 (RCP6.0) and 6.2 Wm-2 (RCP8.5)
. In the subsequent discussion, we choose not to
consider the most optimistic, but probably not realistic, scenario RCP2.6
with a sharp RF reduction already before 2100.
A total estimate of the net RF from SG must take into account the wide range
of factors discussed in the previous sections. Here we would like to
highlight that the relationship between the SO2 amount and the
subsequent AOD is nonlinear, as larger amounts of SO2 will produce
larger aerosol particles and the aerosol scattering efficiency decreases.
Furthermore, the gravitational settling becomes faster with increasing
particle size, therefore reducing the stratospheric aerosol lifetime.
Summary of direct and indirect SG TOA RF per component (see
Sects. 2.1 and 2.2) (global mean values).
As highlighted in Sect. 2.1, another factor that may change the aerosol
lifetime is the prolonged QBO westerly phase caused by SG .
As shown by for explosive volcanic eruptions, a QBO with
dominant easterly shear leads to a longer lifetime for the volcanic aerosol,
due to a greater isolation of the tropical pipe. This helps confine the
aerosols in an area where downward transport is not present. In a similar
way, the extension of the lower-stratospheric QBO westerly phase simulated by
leads to a longer aerosol lifetime. This result, however,
could be partly canceled or even overcompensated if the microphysical effects
of the QBO-dependent sulfur confinement in the tropical pipe were taken into
account. found that a locked QBO westerly phase globally
produces a net decrease of the SG aerosol lifetime, because the tropical
isolation leads to larger particles and subsequently to a more efficient
gravitational settling.
Figure summarizes the RF breakdown per component, including
direct and indirect effects of SG, as discussed in Sects. 2.1 and 2.2 and
based on published estimates. Aside from the direct effect of sulfate aerosol
scattering, we see that the changes in UT ice particle formation and size may
produce a significant negative RF, due to the thermal-dynamically induced
thinning of cirrus clouds formed via homogeneous freezing. The indirect
effects related to SG-induced changes in GHG concentrations (CH4,
O3, stratospheric H2O) are approximately 1 order of
magnitude smaller, so that we may assume that they are globally negligible
with respect to the direct effect of SG aerosols and their indirect impact on
ice cloudiness.
Considering the results in Fig. , we find that the sum of all
direct and indirect RFs of SG with an injection of
5 Tg SO2 yr-1 accounts for
-1.4 ± 0.5 Wm-2, which means a compensation of the
projected positive RF in 2100 relative to 2011 by 64, 38, and 23 % for
the IPCC “realistic” scenarios RCP4.5, RCP6.0 and RCP8.5, respectively. The
November 2015 Paris Agreement aims to strengthen the global response to the
threat of climate change by keeping a global temperature rise this century
well below 2 ∘C above pre-industrial levels and to pursue efforts to
limit the temperature increase even further to 1.5 ∘C. According to
the , the best estimate of the total anthropogenic RF relative
to 1750 is 2.29 Wm-2 in 2011, and the increase in global mean
surface temperature over the period 1880–2012 is 0.85 ∘C. This
means that the 2100 RF relative to 2011 projected in the three RCPs (2.2, 3.7
and 6.2 Wm-2, respectively) could not allow reaching the Paris
Agreement target of a maximum temperature increase of ∼ 0.6 ∘C
up to ∼ 1.1 ∘C in the period 2011 to 2100. In the hypothesis
of SG implementation with injection of
5 Tg SO2 yr-1 during the 21st century, the Paris
Agreement target could likely be reached with the previously estimated
SG RF = -1.4 ± 0.5 Wm-2. This could only happen in
the case of simultaneous WMGHG emissions regulated under scenario RCP4.5 or
(barely) under scenario RCP6.0 (assuming a climate sensitivity of
0.5 KWm-2).