Content uploaded by Rohi Muthyala
Author content
All content in this area was uploaded by Rohi Muthyala on Apr 26, 2018
Content may be subject to copyright.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1036
*For correspondence. (e-mail: gbala@iisc.ac.in)
Regional scale analysis of climate extremes in
an SRM geoengineering simulation, Part 2:
temperature extremes
Rohi Muthyala1,2, Govindasamy Bala1,* and Aditya Nalam1
1Centre for Atmospheric and Oceanic Sciences, Indian Institute of Science, Bengaluru 560 012, India
2Present address: Department of Geography, Rutgers, The State University of New Jersey, Piscataway, NJ 08854, USA
In this study, we examine the statistics of temperature
extremes in a model simulation of solar radiation
management (SRM) geoengineering. We consider both
intensity and frequency-based extreme indices for
temperature. The analysis is performed over both
large-scale domains as well as regional scales (22
Giorgi land regions). We find that temperature
extremes are substantially reduced in geoengineering
simulation: the magnitude of change is much smaller
than that occur in a simulation with elevated atmos-
pheric CO2 alone. Large increase (~10–20 K) in the
lower tails (0.1 percentile) of Tmin and Tmax in the
northern hemisphere extra-tropics that are simulated
under doubling of CO2 are reduced in geoengineering
simulation, but significant increase (~4–7 K) persist
over high-latitude land regions. Frequency of temper-
ature extremes is largely offset over land regions in
geoengineered climate. We infer that SRM schemes
are likely to reduce temperature extremes and the
associated impacts on a global scale. However, we note
that a comprehensive assessment of moral, social,
ethical, legal, technological, economic, political and
governance issues is required for using SRM methods
to counter the impacts of climate change.
Keywords: Extreme events, geoengineering, regional
analysis, solar radiation constant.
INCREASED greenhouse gas (GHG) emissions induce a
warmer climate across the globe. This warming is associ-
ated with changes in several temperature extreme indices
that have been observed and are expected to continue in
the future. Several previous studies have shown that solar
radiation management (SRM) geoengineering can offset
the global mean surface warming caused by increase in
GHGs1–10. This article is Part 2 of our two-part study on
climate extremes under geoengineering. Part 1 discussed
changes in precipitation extremes11 and this Part 2 dis-
cusses changes in temperature extremes.
The trend of changes in temperature extremes is similar
to that of temperature means in many parts of the world12.
Changes in indices based on daily minimum temperature
are found to be more pronounced than changes in indices
based on daily maximum temperature13. The shifts toward
warmer temperatures of cold extremes are generally
larger than the corresponding shifts of warm extremes in
high-latitude regions12. In tropical and subtropical
regions, warm extremes shift toward warmer tempera-
tures faster than cold extremes.
Only a few studies in the past have investigated the
statistics of temperature extremes under SRM geoengi-
neering. Using daily model output the frequency of
temperature extreme events such as coldest night, warm-
est day and a few duration indices was analysed14. The
study showed that the climate extremes under geoengi-
neering are not just smaller than 4XCO2 conditions, but
they also differ significantly from those under pre-
industrial conditions. It was also found that geoengineer-
ing is more effective in reducing changes in temperature
extremes compared to precipitation extremes and more
effective in reducing changes in precipitation extremes
than means, but less effective in reducing changes in
temperature extremes compared to means14. Another
study analysed climate extremes for two SRM schemes15 –
stratospheric sulphate injection and marine cloud bright-
ening. In both climate engineering scenarios, extreme
temperature changes were similar to mean temperature
changes over much of the globe, except over the northern
hemisphere high latitudes. The increase in frequency of
temperature extremes was not completely alleviated in
both geoengineering scenarios.
In this article, we perform an extensive assessment of
the temperature extremes using 16 indices (12 for intensity
and 4 for frequency) and their projected changes in ge-
oengineered climate. We analyse a few percentile
indices (upper and lower tails) to account for the respec-
tive climatologies of different regions. Further, we quan-
tify the changes in extremes over 22 Giorgi land regions
and several large domains.
Model, experiments and methodology
As discussed in Part 1, the model used for this study is
the National Center for Atmospheric Research (NCAR)
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1037
Table 1. Set of temperature extreme indices analysed in this study. These indices are recommended by the Expert Team of Climate Change
Detection and Indices, except those marked with an asterisk
Label Index Index definition Units
TNn Coldest daily Tmin Annual minimum value of daily minimum temperature TN K
TNx Warmest daily Tmin Annual maximum value of daily minimum temperature TN K
TXn Coldest daily Tmax Annual minimum value of daily maximum temperature TX K
TXx Warmest daily Tmax Annual maximum value of daily maximum temperature TX K
0.1TN* Lower extreme Tmin Intensity of daily minimum temperature events which do not exceed
0.1th percentile threshold K
99.9TN* Upper extreme Tmin Intensity of daily minimum temperature events which exceed 99.9th percentile threshold K
0.1TX* Lower extreme Tmax Intensity of daily maximum temperature events which do not exceed 0.1 th percentile threshold K
99.9TX* Upper extreme Tmax Intensity of daily maximum temperature events which exceed 99.9th percentile threshold K
TN10p Cold nights Let TNth 10 be the 10th percentile of TN in 1XCO2 simulation. The percentage of days %
in a year with TN < TNth 10
TN90p Warm nights Let TNth 90 be the 90th percentile of TN in 1XCO2 simulation. The percentage of days %
in a year with TN > TNth 90
TX10p Cold days Let TXth10 be the 10th percentile of TX in 1XCO2 simulation. The percentage of days in a %
year with TX < TXth 10
TX90p Warm days Let TXth 90 be the 90th percentile of TX in 1XCO2 simulation. The percentage of days in a %
year with TX > TXth 90
FD Frost days Number of days when TN < 0C days
ID Ice days Number of days when TX < 0C days
SU Summer days Number of days when TX > 25C days
TR Tropical nights Number of days when TN > 20C days
Community Earth System Model, version 1 (CESM1)16.
Three experiments have been performed: (i) a pre-
industrial control simulation ‘1XCO2’, (ii) ‘2XCO2’ with
doubled atmospheric CO2 concentration and (iii) ‘Geo-
Engg’ with doubled atmospheric CO2 concentration and
the solar constant reduced. A detailed explanation of the
model used and the experiments performed are provided
in Part 1 of this study11.
The model-simulated temperature indices were
evaluated using the daily data from the National Center
for Environmental Prediction–Department of Energy
(NCEP–DOE) Reanalysis 2 (http://www.esrl.noaa.gov/
psd/). This is an improved version of NCEP 1 that fixed
errors and updated parameterization of physical processes.
State-of-the-art analysis/forecast system was used to per-
form data assimilation with past data from 1979. The hori-
zontal resolution of the data was 2.0 2.0.
Here, we consider a subset of temperature extreme
indices available in EIA (ETCCDI Indices Archive). We
quantify the temperature extreme events in terms of both
intensity and frequency. The control simulation (1XCO2)
thresholds are used as reference thresholds in estimating
indices instead of a base observational threshold. Four
new temperature indices – 0.1TN, 99.9TN, 0.1TX and
99.9TX – have been added to our set of extreme tempera-
ture indices. The first index, i.e. 0.1TN represents the
0.1th percentile threshold value of daily minimum
temperature; 99.9TN is the intensity of daily minimum
temperature events which exceed 99.9th percentile thre-
shold. Similarly, 0.1TX and 99.9TX are defined for daily
maximum temperature. The selected indices (Table 1)
give a comprehensive overview of changes in tempera-
ture and precipitation extremes in both 2XCO2 and
geoengineering scenarios13,17. Regional extreme value
statistics was performed for various selected regions
(Supplementary Table 1) and for Giorgi land regions
(Supplementary Table 2 and Figure 1)18. Spatial statisti-
cal analysis was also performed to estimate their uncer-
tainties at local scale. Estimates for land and ocean
regions were also performed.
Methodology for estimating all temperature extreme
indices is not similar. We used three methods of aggre-
gating individual events to create samples. The first
method aggregates the events for the whole time period
(over the entire 10-year period in this study) for each grid
point. Then temperature extremes are estimated at each
grid point based on their respective index definition and
statistical analysis is performed over the spatial domain
of interest. The second method aggregates the individual
events on the annual timescale at each grid point to create
the sample; then climate extremes are estimated from
these samples and the mean of these extremes over the
10-year time period is calculated at each grid point. Then
statistical analysis is performed over the selected spatial
domains. The second method is suitable for estimating
the annual indices (FD, TR, TNn, TXx, etc.), while the
first method is suitable for the other indices. The third
method, suitable for estimating zonal means of extremes,
aggregates individual events over the 10-year period to
estimate climate extreme events at each grid point and
then averages along each latitude circle.
Results
In this article, we discuss the changes in temperature
extremes in a doubled CO2 (2XCO2) and geoengineered
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1038
Figure 1. (Top) Spatial pattern of annual mean surface temperature for NCEP II (reanalysis data for 2004–2013; top left
panel) and 1XCO2 (control simulation during 91–100 years; top right panel). (Bottom panel) Difference between model
simulation and reanalysis data.
climate (Geo-Engg) relative to the 1XCO2 case. The
changes in precipitation and temperature means in a ge-
oengineered climate have been discussed in several pre-
vious studies1,2,8–10,19–22. In the 2XCO2 case, we found
that the change in mean surface temperature was 4.1 K
and mean precipitation was 0.24 mm/day (7.9%). How-
ever, in the Geo-Engg case the change in global mean
temperature reduced to –0.07 K and precipitation to
–0.08 mm/day (–2.8%).
Evaluation of model-simulated temperature extremes
We found that the model-simulated surface mean temper-
ature showed a spatial pattern similar to that in NCEP II
reanalysis data (for the period 2004–2013; correlation co-
efficient of 0.99) with a global mean difference of
–0.08 K (Figure 1). The model underestimated the sur-
face temperature over extra-tropical land regions, but
overestimated the same in the ocean areas of the southern
hemisphere. Figure 2 shows a comparison of annual max-
imum and minimum temperature. The pattern is similar
for the daily minimum temperature extremes with a cor-
relation coefficient of 0.97 and the mean bias is small
(Figure 2). However, daily maximum temperature ex-
tremes show similar pattern (correlation coefficient of
0.95) in the 1XCO2 case and reanalysis data with an
overestimation over most of the land regions (mean bi-
as = 5.5 K). Hence, the minimum temperature extremes
are slightly underestimated and the maximum tempera-
ture extremes are overestimated. The extreme tempera-
ture frequency indices were also evaluated (Supplemen-
tary Figures 2 and 3). We found that the frequency
indices showed similar pattern (correlation coefficient of
FD was 0.98, TR was 0.92, SU was 0.95) to reanalysis
data. Tropical nights were slightly underestimated
(Supplementary Figure 4) (mean bias 18 days) and
summer days were slightly overestimated over tropical
land regions (Supplementary Figure 3) (mean bias ~32
days). The daily minimum indices were slightly underes-
timated over high-latitude land regions and daily maxi-
mum indices overestimated over mid- and low-latitude
land regions (Figure 2), thereby overestimating the
diurnal range by a large bias (Supplementary Figure 3)
(mean bias ~8 K) and also with a smaller correlation co-
efficient of 0.46. Overall, we found the spatial pattern in
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1039
Figure 2. Spatial pattern of TNn (annual minimum daily minimum temperature; top panels) and TXx (annual maximum
daily maximum temperature; bottom panels; description of the indices is given in Table 1) for NCEP II (reanalysis data
for 2004–13; left panels) and 1XCO2 (control simulation for 91–100 years; right panels).
model-simulated extremes, annual temperature maximum
and minimum to be in good agreement with NCEP II
reanalysis. However, there were large regional biases in
daily maximum temperature extreme indices, but when
we compare two model simulations, it is likely that the
biases will cancel out.
Changes in intensity of temperature extremes
We analysed the intensity of temperature extremes using
the four absolute indices: annual maximum and minimum
of Tmax and Tmin, i.e. TNn, TNx, TXn and TXx (Table 1).
We also used four percentile indices: 0.1TN, 99.9TN,
0.1TX and 99.9TX (Table 1). The probability density
function (PDF) of surface temperature and threshold for
extreme temperature events would be different at every
grid point, and hence these indices were estimated at each
grid point. This avoids errors in analysing the extremes
due to nonuniformity in temperature ranges on a regional
scale.
In the 2XCO2 case, we found a large increase of 10–
15 K in daily minimum temperature indices (TNn and
TNx) and 4–6 K in daily maximum temperature indices
(TXn and TXx) in the high latitudes (Supplementary Fig-
ure 4). However, in the Geo-Engg case, this increase was
largely offset in daily maximum indices, but residual
warming (~5 K) persisted in daily minimum indices over
mid- and high-latitude land regions in the northern hemi-
sphere. From the spatial pattern of changes in 0.1TN
(lower extreme Tmin) and 99.9TX (upper extreme Tmax),
we found that in the 2XCO2 case there was global mean
increase of 5.1 K in 0.1TN and 3.6 K in 99.9TX
(Supplementary Figure 5), while the surface mean tem-
perature increase was 4.1 K (Supplementary Figure 6).
We found that there was a large increase of ~20 K in
0.1TN over some extra-tropical oceanic regions and around
10 K in 99.9TX over northern extra-tropical oceanic region
(Supplementary Figure 5). In the Geo-Engg case, the tem-
perature extremes were brought close to the 1XCO2 case on
a global scale (Figures 3 a and 4 a). However, on the
regional scale, we found that whereas the medians of
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1040
Figure 3. Change in intensity of temperature extremes in the 2XCO2 (red) and Geo-Engg (green) cases relative to the 1XCO2 case, represented
using two indices TNn (coldest daily Tmin; top panels) and TXx (warmest daily Tmax; bottom panels; descriptions of the indices are given in Table 1)
for (a) large domains and (b) Giorgi land regions.
Figure 4. Change in intensity of temperature extremes in the 2XCO2 (red) and Geo-Engg (green) cases relative to the 1XCO2 case, represented
using the two indices 0.1TN (lower extreme Tmin; top panels) and 99.9TX (upper extreme Tmax; bottom panels; description of the indices is given in
Table 1) for (a) large domains and (b) Giorgi land regions.
changes were nearly zero in 0.1TN and 99.9TX, extremes
still existed over a wide range on a local scale for most of
the land regions (Figures 3 b and 4b).
The changes in other tails of daily minimum and max-
imum temperature indices (TNx, TXn, 99.9TN and
0.1TX) followed similar pattern when compared to the
indices TXx, TNn, 99.9TX and 0.1TN (Supplementary
Figures 7 and 8). Both intensity-based indices TNn, TXx
and 0.1TN, 99.9TX followed similar pattern, but the
magnitude of changes in conventional absolute indices
(TNn and TXx) was larger than the percentile indices
(0.1TN and 99.9TX) in the Geo-Engg case (compare the
right panels of Supplementary Figures 4 and 5). In sum-
mary, upper extreme Tmax in the Geo-Engg case was
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1041
Figure 5. Change in frequency of temperature extremes in the 2XCO2 (red) and Geo-Engg (green) cases relative to the 1XCO2
case, represented using the two indices TN10p (cold nights; top panels) and TX90p (warm days; bottom panels; description of the indices is given
in Table 1) for (a) large domains and (b) Giorgi land regions.
closer to the 1XCO2 case and changes in lower
extreme Tmin remained slightly positive on the regional
scale.
Changes in frequency of temperature extremes
We analysed the frequency of temperature extremes using
the eight indices: cold nights (TN10p), warm nights
(TN90p), cold days (TX10p), warm days (TX90p), frost
days (FD), ice days (ID), summer days (SU) and tropical
nights (TR) (Table 1). These indices were estimated at
each grid point annually and then averaged over the
10-year period (91–100 years). The indices TN10p,
TN90p, TX10p and TX90p were estimated for the
2XCO2 and Geo-Engg cases relative to the 1XCO2 case
(Supplementary Figure 2). In Figure 5 (see also Supple-
mentary Figures 9 and 10), a value of 10 corresponds to
no change in extremes (relative to the 1XCO2 case). In
the 2XCO2 case, 10th percentile of minimum and maxi-
mum temperature of the 1XCO2 case reduced to about
0.2th percentile on a global mean basis, thereby resulting
in decrease in occurrence of cold days and nights (TN10p
and TX10p) (Figure 5 a; also see Supplementary Figure 9
and 10 a). However, in the Geo-Engg case, the cold days
(TX10p) and nights (TN10p) increased to ~15%
(Supplementary Figure 9). On a regional scale, the fre-
quency of temperature extremes in the Geo-Engg case
was nearly similar to that in the 1XCO2 case (Figure 5 b
and also see Supplementary Figure 10 b). Similarly, in the
2XCO2 case, the occurrence of warm days and nights
(TN90p and TX90p) increased to ~60% from 10% in the
1XCO2 case. However, in the Geo-Engg case, we found
that the frequency of temperature extremes was close to the
1XCO2 case over land regions. Over the ocean areas, the
lower extreme indices (TN10p and TX10p) were slightly
larger than those in the 1XCO2 case and upper extreme
indices (TN90p and TX90p) were slightly less than those
in the 1XCO2 case. Overall, there was a shift in the PDF
of temperature to the left: SRM geoengineering slightly
reduced the warm temperature extremes and increased the
cold extremes relative to the 1XCO2 case. For example,
TN90p decreased to 5.5% and TN10p increased to 14.7%,
indicating that the number of warm nights decreased and
that of cold nights increased in the Geo-Engg case.
The spatial pattern of FD and SU showed that the for-
mer reduced by 16 days per annum on a global mean
basis and the latter increased by 42 days per annum in the
2XCO2 case (Supplementary Figure 11). Regionally,
there were large differences: large reduction of 150–200
days in the number of frost days was found in the high
latitudes and large increase of 200–250 days in the num-
ber of summer days was found in subtropical oceanic re-
gions. In contrast, we simulated large reduction in the
frequency of temperature extremes in the Geo-Engg case
when compared to the 2XCO2 case, which were brought
close to the 1XCO2 case globally (Supplementary Figure
11). Over large domains, the changes in medians of both
FD and SU were close to zero in the Geo-Engg case
(Figure 6 a). On the regional scale, we found that in the
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1042
Figure 6. Change in frequency of temperature extremes in the 2XCO2 (red) and Geo-Engg (green) cases relative to the 1XCO2 case, represented
using the two indices FD (frost days; top panels) and SU (summer days; bottom panels; description of the indices is given in Table 1) for (a) large
domains and (b) Giorgi land regions.
Geo-Engg case the frequency of temperature extremes (FD
and SU) reduced to a large extent when compared to the
2XCO2 case (Figure 6 b). The changes in medians were
close to zero and simultaneously the spatial variability
(length of whiskers) is also reduced to a large extent. The
changes in ID and TR followed patterns similar to those FD
and SU (Supplementary Figures 11 and 12).
Changes in zonal mean intensity and frequency of
temperature extremes
The zonal mean intensity of temperature extremes (0.1TN
and 99.9TX) showed that in the 2XCO2 case the zonal
mean intensity of temperature extremes was 3–5 K
larger than those in the 1XCO2 case (Figure 7 a and b).
However, the zonal mean of frequency of FD was 10–20
days and 40–50 days less than that in the 1XCO2 case
over the southern hemisphere high latitudes and northern
hemisphere mid- and high-latitudes respectively (Figure
7 c). We also found that the zonal mean frequency of SU
was 50–100 days larger than that in the 1XCO2 case over
the tropics and subtropics (Figure 7 d). In contrast,
geoengineered climate showed zonal mean of both inten-
sity and frequency of temperature extremes (0.1TN,
99.9TX, FD and SU) similar to that in the 1XCO2 case,
except SU over the southern tropics where we found a
slight reduction. Overall, we found that the zonal mean
temperature extremes were brought closer to pre-
industrial climatic conditions by geoengineering. Abso-
lute temperature indices (TNn and TXx) also showed
similar results to 0.1TN and 99.9TX in all the simulations
(Supplementary Figure 13).
Comparison of changes in temperature means and
extremes
We used changes in 99.9TX (daily maximum extreme)
and 0.1TN (daily minimum extreme) to represent the
temperature extremes for comparison with surface mean
temperature changes (Figure 8). In the 2XCO2 case, we
found that the global changes in 0.1TN were larger than
those in the mean, whereas changes in 99.9TX were
smaller than those in the mean relative to the 1XCO2
case. For other regions (tropics and subtropics), the
changes in the means and extremes did not differ signifi-
cantly. The range of changes in means was around 2.5–
5.5 K whereas it was 2.5–8 K for the changes in daily
minimum extreme 0.1TN (Figure 8 a), with the largest
increase simulated for the extra-tropics. On a regional
scale (Figure 8 b), the range increased to 2–13 K. Our re-
sults are qualitatively consistent with previous studies13.
However, in the Geo-Engg case the extreme temperatures
were not only offset but were slightly less than in the
1XCO2 case. We found a reduction of 0.5–1 K in daily
maximum extreme over the tropics and subtropics with a
maximum reduction of 1 K over tropical land region
(Figure 8 a). Overall, we found that though there were
regional disparities in the geoengineering simulations,
temperature means and extremes were close to the
1XCO2 case. Absolute temperature indices (TNn and
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1043
Figure 7. Zonal mean of (a, b) intensity of extreme temperature (0.1TN and 99.9TX) and (c, d) frequency of extreme temperature (FD and SU)
for 10-year analysis period (91–100 years period in our simulations) calculated for the control ( 1XCO2; blue), doubled CO2 (2XCO2; red) and ge-
oengineering (Geo-Engg; green) simulations. Grey bars represent the range of extremes in ten 5-years segments of the last 50-years data of the con-
trol simulation (1XCO2).
TXx) also showed similar results to 0.1TN and 99.9TX in
all the simulations (Supplementary Figure 14).
Discussion and conclusion
Similar to Part 1 where the precipitation extremes were
studied, here we have analysed the temperature extremes
in a doubled CO2 climate with and without geoengineer-
ing and compared them with a control simulation using a
subset of temperature indices available in EIA and some
new temperature indices appropriately defined for this
study (Table 1). We have analysed four intensity-based
and eight frequency-based temperature extreme indices,
and discuss the changes in these indices upon CO2 dou-
bling and SRM geoengineering.
In the 2XCO2 case, we simulated an increase in global
mean temperature of ~4.1 K, daily minimum temperature
(0.1TN) of ~5.1 K and daily maximum temperature
(99.9TX) of ~3.6 K relative to the 1XCO2 case. How-
ever, on a regional scale we simulated large changes in
temperature extremes of up to ~20 K mainly in the extra-
tropical regions. Temperature extremes were reduced
considerably in the Geo-Engg case compared to the
2XCO2 case, with departures from the 1XCO2 case
smaller than those in the 2XCO2 case. Though the
temperature extremes in the Geo-Engg case were brought
close to those in the 1XCO2 case, they were not uniformly
reduced over the globe. The change in intensity of
temperature extremes persisted over high latitudes in the
northern hemisphere. On a regional scale, upper extremes
were brought close to the 1XCO2 case, while the residual
changes in lower extremes remained.
The frequency of cold nights (TN10p) and cold days
(TX10p) decreased, whereas that of warm nights (TN90p)
and warm days (TN90p) increased by up to ~50% in the
2XCO2 case relative to the 1XCO2 case. Similarly, in the
2XCO2 case, we simulated a reduction in the number of
frost days (16 days per year decrease) and increase in the
number of summer days (42 days per year increase) rela-
tive to the 1XCO2 case. In the Geo-Engg case, there was
a leftward shift in the PDF of temperature resulting in a
small reduction in the warm temperature extremes
(TN90p) and increase in the cold temperature extremes
(TN10p) relative to the 1XCO2 case. We found that
changes in FD and SU in the 2XCO2 case relative to the
1XCO2 case were offset in the Geo-Engg case to a large
extent. We also simulated a reduction in the medians in
the Geo-Engg case when compared to the 1XCO2 case.
As discussed in Part I, there are several limitations to
this study, as we use idealized experiments to demon-
strate the effects of SRM geoengineering on climate
extremes. Some of the limitations are related to the use of
a single model and absence of feedbacks on longer time-
scales (deep ocean and carbon cycle feedbacks), as we
have used a slab ocean model. We simulated temperature
extremes that were close to the observations, but with
some biases on a regional scale. Although we used a
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1044
Figure 8. Change in temperature means and extremes (99.9TX and 0.1TN) over (a) large domains and (b) 22 Giorgi land regions
in the 2XCO2 (red bars) and Geo-Engg cases (green bar) relative to the 1XCO2 case.
single model with some limitations, our results are quali-
tatively in agreement with previous SRM geoengineering
studies using solar constant reduction that analysed
climate extremes from multiple models10,14.
In conclusion, we find that geoengineering has the po-
tential to ameliorate the impacts of climate change from
extreme events. However, there could be undesired side
effects such as ozone depletion for stratospheric aerosol
injections6 and several reasons for not considering ge-
oengineering to counter climate change23. Also, a com-
prehensive assessment of the moral, social, ethical, legal,
technological, economic, political and governance issues
related to geoengineering needs to performed before
implementation of any SRM geoengineering methods. In
the absence of a global consensus on these issues, reduc-
ing GHG emissions is likely the best strategy to tackle
climate change.
1. Govindasamy, B. and Caldeira, K., Geoengineering earth’s
radiation balance to mitigate CO2-induced climate change.
Geophys. Res. Lett., 2000, 27, 2141–2144; doi:10.1029/1999GL00
6086.
2. Govindasamy, B., Thompson, S., Duffy, P. B., Caldeira, K. and
Delire, C., Impact of geoengineering schemes on the terrestrial bi-
osphere. Geophys. Res. Lett., 2002, 29(22), 2061; doi:10.1029/
2002GL015911.
3. Govindasamy, B., Caldeira, K. and Duffy, P. B., Geoengineering
earth’s radiation balance to mitigate climate change from a
quadrupling of CO2. Global Planet Change, 2003, 37(1–2), 157–
168; doi:10.1016/S0921-8181(02)00195-9.
4. Matthews, H. D. and Caldeira, K., Transient climate–carbon simu-
lations of planetary geoengineering. Proc. Natl. Acad. Sci. USA,
2007, 104, 9949–9954; doi:10.1073/pnas.0700419104.
5. Robock, A., Oman, L. and Stenchikov, G. L., Regional climate
responses to geoengineering with tropical and Arctic SO2 injec-
tions. J. Geophys. Res., 2008, 113, D16101; doi:10.1029/
2008JD010050.
6. Rasch, P. J. et al., An overview of geoengineering of climate using
stratospheric sulphate aerosols. Philos. Trans. R. Soc. London,
Ser. A, 2008, 366, 4007–4037; doi:10.1098/rsta.2008.013.
7. Ricke, K. L., Granger Morgan, M. and Allen, M. R., Regional
climate response to solar-radiation management. Nature Geosci.,
2010, 3, 537–541; doi:10.1038/ngeo915.
RESEARCH ARTICLES
CURRENT SCIENCE, VOL. 114, NO. 5, 10 MARCH 2018
1045
8. Schmidt, H. et al., Solar irradiance reduction to counteract radia-
tive forcing from a quadrupling of CO2: climate responses simu-
lated by four earth system models. Earth Syst. Dyn., 2012, 3, 63–
78; doi:10.5194/esd-3-63-2012.
9. Kravitz, B. et al., Climate model response from the Geoengineer-
ing Model Intercomparison Project (GeoMIP). J. Geophys. Res.
Atmos., 2013, 118, 8320–8331; doi:10.1002/jgrd.50646.
10. Tilmes, S. et al., The hydrologic impact of Geoengineering in the
Geoengineering Model Intercomparison Project (GeoMIP). J. Ge-
ophys. Res. Atmos., 2013, 118(11036–11), 058; doi:10.1002/
jgrd.50868.
11. Muthyala, R., M., Bala, G. and Nalam, A., Regional scale analysis
of climate extremes in an SRM geoengineering simulation, Part 1:
precipitation extremes. Curr. Sci., 2018, 114(5), 1024–1035.
12. Kharin, V. V., Zwiers. F. W., Zhang, X. and Hegerl, G. C.,
Changes in temperature and precipitation extremes in the IPCC
ensemble of global coupled model simulations. J. Climate, 2007,
20, 1419–1444; doi:10.1175/JCLI4066.1.
13. Sillmann, J., Kharin, V. V., Zhang, X., Zwiers, F. W. and
Bronaugh, D., Climate extreme indices in the CMIP5 multi-model
ensemble: Part 2. Future climate projections. J. Geophys. Res.
Atmos., 2013, 118, 2473–2493; doi:10.1002/jgrd.50188.
14. Curry, C. L. et al., A multimodel examination of climate extremes
in an idealized geoengineering experiment. J. Geophys. Res.
Atmos., 2014, 119, 3900–3923; doi:10.1002/2013JD020648.
15. Aswathy, V. N., Boucher, O., Quaas, M., Niemier, U., Muri, H.
and Quaas, J., Climate extremes in multi-model simulations of
stratospheric aerosol and marine cloud brightening climate engi-
neering. Atmos. Chem. Phys. Discuss., 2014, 14, 32393–32425;
doi:10.5194/acpd-14-32393-2014.
16. Richard, B. N. et al., Description of the NCAR Community
Atmosphere Model (CAM 5.0). NCAR Technical Note
NCAR/TN-464 + STR, National Centre for Atmospheric
Research, Boulder, CO, USA, 2012.
17. Sillmann, J., Kharin, V. V., Zhang, X., Zwiers, F. W. and
Bronaugh, D., Climate extreme indices in the CMIP5 multi-model
ensemble: Part 1. Model evaluation in the present climate. J. Geo-
phys. Res. Atmos., 2013, 118, 1716–1733; doi:10.1002/jgrd.50203.
18. Giorgi, F. and Francisco, R., Uncertainties in regional climate
change prediction: a regional analysis of ensemble simulations
with the HADCM2 coupled AOGCM. Climate Dyn., 2000, 16,
169–182.
19. Bala, G., Duffy, P. B. and Taylor, K. E., Impact of geoengineering
schemes on the global hydrological cycle. Proc. Natl. Acad. Sci.
USA, 2008, 105, 7664–7669; doi:10.1073/pnas.0711648105.
20. Bala, G., Caldeira, K., Nemani, R., Cao, L., Ban-Weiss, G. and
Shin, H. J., Albedo enhancement of marine clouds to counteract
global warming: impacts on the hydrological cycle. Climate Dyn.,
2010, 37(5–6), 915–931; doi:10.1007/s00382-010-0868-1.
21. Bala, G. and Nag, B., Albedo enhancement over land to counteract
global warming: impacts on hydrological cycle. Climate Dyn.,
2012, 39, 1527–1542; doi:10.1007/s00382-011-1256-1.
22. Kalidindi, S., Bala, G., Modak, A. and Caldeira, K., Modeling of
solar radiation management: a comparison of simulations using
reduced solar constant and stratospheric sulphate aerosols. Climate
Dyn., 2015, 44, 2909–2925; doi:10.1007/s00382-014-2240-3.
23. Robock, A., 20 Reasons why geoengineering may be a bad idea.
Bull. Atmos. Sci., 2008, 64(2), 14–18, 59; doi:10.2968/064002006.
ACKNOWLEDGEMENTS. This study was funded by the Divecha
Centre for Climate Change, Indian Institute of Science (IISc), Benga-
luru and the Department of Science and Technology (DST), New Delhi
(grant DSTO1203). R.M. and A.N. thank IISc for providing the
scholarships. Computations were carried out at the Centre for Atmos-
pheric and Oceanic Sciences High Performance Computing facility at
IISc funded by Fund for Improvement of S&T Infrastructure (FIST),
DST and Divecha Centre for Climate Change.
Received 30 August 2017; accepted 22 October 2017
doi: 10.18520/cs/v114/i05/1036-1045
