Article https://doi.org/10.

1038/s41467-023-40641-y

Soil moisture–atmosphere coupling

accelerates global warming

Received: 27 December 2022 Liang Qiao1,5, Zhiyan Zuo 1,5 , Renhe Zhang 1
, Shilong Piao 2
,
Dong Xiao3 & Kaiwen Zhang1,4
Accepted: 2 August 2023

Soil moisture–atmosphere coupling (SA) amplifies greenhouse gas-driven

Check for updates global warming via changes in surface heat balance. The Scenario Model
Intercomparison Project projects an acceleration in SA-driven warming due to
1234567890():,;

the ‘warmer climate – drier soil’ feedback, which continuously warms the globe
1234567890():,;

and thereby exerts an acceleration effect on global warming. The projection

shows that SA-driven warming exceeds 0.5 °C over extratropical landmasses
by the end of the 21st Century. The likelihood of extreme high temperatures
will additionally increase by about 10% over the entire globe (excluding Ant-
arctica) and more than 30% over large parts of North America and Europe
under the high-emission scenario. This demonstrates the high sensitivity of SA
to climate change, in which SA can exceed the natural range of climate varia-
bility and play a non-linear warming component role on the globe.

Since the onset of the Industrial Revolution, anthropogenic emission experiment in Coupled Model Intercomparison Project phase 6
of greenhouse gases (GHG) has caused Earth’s climate to warm mea- (CMIP6) — to isolate the climatic impact of SA during the extra-tropical
surably. Nonetheless, the rates of warming vary greatly across the summer under different warming scenarios. Experiments under the
globe1–3. Previous researches have revealed that regional variability is same emission scenario are driven by the same forcing agents (i.e., sea
due in part to land–atmosphere coupling4–7, whereby differences in surface temperature, sea ice, and CO2 concentrations). In the Land
land surface conditions (e.g., soil moisture, snowpack, and vegetation Feedback Model Intercomparison Project with prescribed Land Con-
cover) modulate mass fluxes and energy to the atmosphere8–16 and, ditions experiment (LFMIP-pdLC) from LS3MIP, the soil moisture is
furthermore, influence weather patterns and climate anomalies asso- fixed to its climatological state for the period 1980–2014 which is
ciated with GHG forcing. For instance, soil moisture–atmosphere derived from historical global climate model output (Supplementary
coupling (SA) is linked to increased intensity and frequency of high- Fig. 1). Then the SA effect can be isolated from the relative differences
temperature extremes and heat waves via the impact of atmospheric between fully coupled experiments (historical, the SSP1-2.6, and SSP5-
warming on soil moisture in some regions17–22. However, a coherent 8.5 experiments) and LFMIP-pdLC experiment. In this study, we con-
picture of the extent to which global warming (including warming rate sider three time horizon periods:1995–2014, 2040–2059, and
and heat extremes) will be impacted by SA and its time evolution under 2080–2099 to represent modern, mid-term, and long-term future
different emission scenarios remains unclear. conditions, respectively.
In order to analyze the SA effect on global warming, we employed
six global climate models (CESM2, CMCC-ESM2, EC-Earth3, IPSL- Results
CM6A-LR, MIROC6, and MPI-ESM1-2-LR) — the Land Surface, Snow and Under each emission scenario investigated here, SA amplifies global
Soil Moisture Model Intercomparison Project (LS3MIP)23, the Scenario warming over much of Earth’s land surface (Fig. 1). Even under the
Model Intercomparison Project (ScenarioMIP)24, and the historical most stringent GHG mitigation pathway, the SA-induced warming is

1
Department of Atmospheric and Oceanic Sciences/Institute of Atmospheric Sciences, Fudan University, Shanghai, China. 2Sino-French Institute for
Earth System Science, College of Urban and Environmental Sciences, Peking University, Beijing, China. 3Key laboratory of Cites’ Mitigation and
Adaptation to Climate Change in Shanghai, China Meteorological Administration, Shanghai, China. 4State Key Laboratory of Severe Weather, Chinese
Academy of Meteorological Sciences, Beijing, China. 5These authors contributed equally: Liang Qiao, Zhiyan Zuo. e-mail: [email protected];
[email protected]

Nature Communications | (2023)14:4908 1

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Article https://doi.org/10.1038/s41467-023-40641-y

a SSP1-2.6 (mid-term) b SSP5-8.5 (mid-term)

c SSP1-2.6 (long-term) d SSP5-8.5 (long-term)

Temperature (℃)

e Historical (modern) f Regions comparison

3 MMM modern
Mid-term Long-term MMM SSP1-2.6
Temperature ℃

2 MMM SSP5-8.5
CESM2
CMCC-ESM2
1 EC-Earth3
IPSL-CM6A-LR
0 MIROC6
MPI-ESM1-2-LR
GLO EUR NA GLO EUR NA
Temperature (℃) Region

Fig. 1 | Soil moisture-atmosphere coupling (SA) impact on surface air tem- 40–60°N, 20–50°E). Black dots in a–e denote that the sign of the change is con-
perature (tas). a, b represent spatial distributions of the SA influence on tas (°C, sistent with the sign of multi-model mean (MMM) in at least five of the six CMIP6
calculated the tas changes by subtracting fixed soil moisture experiment from fully models. f is SA impact on global (GLO, excluding Antarctica), NA, and EUR tas under
coupled experiment under low-emission (SSP1-2.6) and high-emission (SSP5-8.5) low- and high-emission scenarios for both mid-term and long-term future periods.
scenarios for the mid-term future (2040–2059)). c, d represent same as for a, b but Gray, blue, and red bars represent the MMM values for modern conditions, the low-
corresponding to the long-term future (2080–2099). e same as for a, but corre- emission (SSP1-2.6), and high-emission (SSP5-8.5) pathways, respectively. Indivi-
sponding to the modern period (1995–2014). Black boxes in a–e represent central dual models are represented by different types of points.
North America (NA: 28–55°N, 88–110°W) and central and eastern Europe (EUR:

considerably greater than that experienced currently (Fig. 1a, c). Cen- term and long-term future projections is 0.67 °C ± 0.66 °C and
ters of SA-amplified warming occur primarily within the mid-latitudes 0.71 °C ± 0.57 °C, respectively. The area over which SA-induced
of Northern Hemisphere and the subtropical regions of Southern warming exceeds 0.5 °C (1.0 °C) would increase by nearly 30%
Hemisphere, respectively. The most extreme warming occurs over (>200%) by the end of the 21st Century compared with mid-term
central North America (NA: 28–55°N, 88–110°W) and central and projections (Supplementary Figs. 2–3).
eastern Europe (EUR: 40–60°N, 20–50°E) in modern period Our findings exhibit a clear uptrend in SA-induced warming in the
(0.89 ± 0.53 °C and 0.56 ± 0.41 °C), which is generally consistent with mid-latitude Northern Hemisphere under the very high-emission
Berg et al4. The warming will be up to 2.4 °C in both regions in the long- pathway, with comparatively weak trends under the lower-emission
term future under the very high-emission scenario (Fig. 1d). Under low- scenario (Fig. 2). For the very high-emission pathway, the positive
emission scenario, SA-induced warming varies little between the mid- trends caused by SA over EUR (0.17 °C ± 0.08 °C per decade) and NA
term and long-term future projections; under the very high-emission (0.16 °C ± 0.09 °C per decade) explain 18.5 ± 9.7% and 18.8 ± 10.0%,
pathway, however, warming is significantly more intense for long-term respectively, of the overall warming rate in each region (Supplemen-
than for mid-term periods, particularly in EUR and NA. Specifically, tary Table 1). Clearly, the uptrend of SA-induced warming associated
global land-surface air temperature (excluding Antarctica) is projected with very high greenhouse gases emission would accelerate the speed
to rise by 0.68 °C ± 0.38 °C owing to SA-induced warming by the end of of global warming, and the magnitude of acceleration would increase
this century, which is measurably higher than the mid-term projection with time. Therefore, we posit that, unless we take early action to
(0.44 °C ± 0.28 °C; Fig. 1f). In NA and EUR, the difference between mid- reduce emission, SA- and GHG-induced warming would become

Nature Communications | (2023)14:4908 2

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Article https://doi.org/10.1038/s41467-023-40641-y

a Hist + future (SSP1-2.6) intensity will rise by >1.5 °C globally and by as much as 8.0 °C over
North America and Europe. Under the most stringent emission-
mitigation scenario, the relationship between SA and the intensity and
probability of extreme high-temperature would weaken (Supplemen-
tary Fig. 4), suggesting that reductions in GHG emission would actively
reduce the severity of such events. We note that this feature is present
in all the models considered here (Supplementary Figs. 5–6). The
extreme high-temperature due to SA in India is diverse among differ-
ent models though the SA-driven warming is significant, which may be
associated with the large uncertainty of Indian summer monsoon
precipitation25.
GHG-driven warming is projected to dry the soil column26
(Fig. 4a, b and Supplementary Fig. 7), thereby reducing evapo-
transpiration and allowing the ground surface to receive more solar
b Hist + future (SSP5-8.5) shortwave radiation (longwave radiation is much weaker) through a
reduction in cloud cover (Fig. 5a, b and Supplementary Fig. 8).
Meanwhile, decreasing evapotranspiration could increase sensible
heat flux from the land surface to the low-level atmosphere via
decreasing the latent heat flux (Fig. 5c, d). The increasing sensible
heat flux caused by the joint enhanced shortwave radiation and
reduced latent heat flux trigger the nonlinear warming under severe
GHG emission. Those phenomena are the strongest in the Northern
mid-latitude, especially in EUR and NA, and Southern subtropical,
where decreasing soil moisture can significantly change surface
energy partitioning from latent heat flux to more sensible heat flux
via decreasing evapotranspiration. On the contrary, there is a slight
cooling effect caused by SA in a few regions, such as Sahara and
Arabian Peninsula. The evapotranspiration change over those
Fig. 2 | Temporal evolution of soil moisture-atmosphere coupling (SA) driven regions is negligible, while the negative cloud cover and positive
surface air temperature (tas). Temporal evolution of tas variability due to SA shortwave radiation anomalies are generally significant, which sug-
under (a) the low-emission scenario (SSP1-2.6) and (b) the high-emission scenario gests that the local evapotranspiration associated with soil moisture
(SSP5-8.5). The time series is performed with Lanczos low-pass filtering to remove is not the primary factor dominating the surface energy balance. The
the interannual variability in 1980–2099. The total number of weights is 11 and the impact of SA on surface energy over those regions may be due to the
cut-off frequency of the low-pass filter is 1/11. For both periods, “Hist” denotes the non-local effect and related to the large-scale circulation16,27.
period 1985–2014, and “future” is the period 2015–2094. The blue line is central Under the very high-emission scenario, progressively drying soil
North America (NA: 28–55°N, 88–110°W), the black line is central and eastern
column leads to an acceleration of the decline in evapotranspiration
Europe (EUR: 40–60°N, 20–50°E), the red line is mid-latitudes of Northern Hemi-
(Fig. 4b, d), with the result of increased positive radiative budgets and
sphere (30–60°N, 180°W–180°E), and the orange line is subtropical regions of the
thereby the acceleration of the amplified-warming, particularly over
Southern Hemisphere (20–40°S, 180°W–180°E). Shading represents the uncer-
tainty among models.
NA and EUR. Moreover, the correlation coefficient between evapo-
transpiration and soil moisture will increase under a drying soil back-
ground, which eventually aggravates the reduction of
closely coupled, resulting in a positive feedback that may hasten the evapotranspiration caused by soil drying over time (Fig. 4e–h). These
approach of distinct climate range. Should the most stringent emission characteristics are not only shown in the surface soil moisture which is
pathway be adopted, our results suggest SA-induced warming will most directly related to evapotranspiration, but also in the root zone
weaken significantly (Fig. 2a). soil moisture (Supplementary Fig. 9). The enhanced sensitivity of
Although SA contributes <10% to the likelihood of extreme high- evapotranspiration to soil drying leads to the increase of SA-induced
temperature under modern conditions, this influence is likely to be non-linear warming under very high GHG emission background. The
strengthened over northern China and northernmost South America in non-linear increase of SA-induced warming, combined with the GHG-
the mid-term future (Fig. 3a). By the end of this century, the role of SA warming, will make global warming to act like a snowball, corre-
in the frequency and intensity of extreme high-temperature events sponding to SA and other climate factors approaching a novel and
globally is projected to rise significantly, consistent with the uptrend of unpredictable statement.
SA-induced warming, and could approach 20% in the mid–high-lati- Furthermore, we decompose the SA-induced changes in surface
tude Northern Hemisphere and subtropical Southern Hemisphere air temperature into the following radiative forcing terms: surface
(Fig. 3a). Over NA and EUR hotspots, SA would cause a rightward shift albedo, evapotranspiration, shortwave transmissivity, air emissivity,
and pronounced flattening of the surface air temperature probability aerodynamic resistance, and residual terms28 (see Methods). The
distribution function, representing an overall increase in the prob- primary driver of SA-induced warming is positive radiative forcing
ability of extreme high-temperature over both regions (NA: +52.5%; arising from the changes in evapotranspiration term and shortwave
EUR: +30.8%; Fig. 3c, d). Viewed another way, in the absence of SA- transmissivity term outlined above (Supplementary Figs. 10 and 11),
induced warming, the probability of the extreme high-temperature which is the consequence of the decrease in evapotranspiration on
under the very high-emission pathway decreases by almost a third over shortwave radiation and sensible heat flux. Generally, the shortwave
NA and by a quarter over EUR by the end of the 21st Century. transmissivity term is larger than the evapotranspiration term for the
In addition to frequency, the SA effect might also influence the global land in the modern period, while the effects of the two terms
intensity of extreme high-temperature; this process would impact become equivalent in the future, especially in very high-emission
North China and parts of South America in mid-term projections and scenario. Over EUR and NA in the long-term future, the combined
globally in long-term projections (Fig. 3b). Our results indicate that the positive radiation (sum of the above five terms) dominated by

Nature Communications | (2023)14:4908 3

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Article https://doi.org/10.1038/s41467-023-40641-y

a Probability of extreme high-temperature by SA b Intensity of extreme high-temperature by SA

Modern Modern

Mid-term future Mid-term future

Long-term future Long-term future

Percent (%) Temperature (℃)

c Probability distribution function of tas in EUR

0.2
Fixed soil moisture
Modern Mid-term Fully coupled Long-term
0.15
Probability density

0.1

0.05

0
10 15 20 25 30 10 15 20 25 30 10 15 20 25 30 35
Temperature ℃

d Probability distribution function of tas in NA

Fixed soil moisture
Modern Mid-term Long-term
0.2 Fully coupled
Probability density

0.1

0
15 20 25 30 35 15 20 25 30 35 15 20 25 30 35
Temperature ℃
Fig. 3 | Changes in the spatial distribution of 90th percentile of surface air term future periods (see Methods). b is same as a but for the intensity (°C) of
temperature (tas) caused by soil moisture-atmosphere coupling (SA) and the extreme high-temperature events attributed to SA. c, d represent tas probability
probability distribution function of tas in the MPI-ESM1-2-LR model under the distribution functions under the high-emission scenario for central and eastern
high-emission scenario. a is spatial distribution of the change in probability of Europe (EUR: 40–60°N, 20–50°E) and central North America (NA: 28–55°N,
extreme high-temperature (%) caused by SA for the modern, mid-term, and long- 88–110°W) during the modern, mid-term, and long-term future periods.

Nature Communications | (2023)14:4908 4

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Article https://doi.org/10.1038/s41467-023-40641-y

a Soil moisture SSP1-2.6 b Soil moisture SSP5-8.5

c Evapotranspiration SSP1-2.6 d Evapotranspiration SSP5-8.5

e Northern mid-latitude f Southern subtropical

Mod=0.53* Mod=0.79**
Mid=0.58** Mid=0.68**
Long=0.76** Long=0.62**

g EUR h NA
Mod=0.67** Mod=0.56*
Mid=0.60** Mid=0.53*
Long=0.69** Long=0.77**

evapotranspiration term and shortwave transmissivity term will and 25.3 ± 12.6 W m−2), they are smaller than half of those in high-
reach 33.5 ± 14.1 and 32.8 ± 16.7 W m−2 under high-emission scenario. emission scenario.
Compared with the modern period (15.4 ± 9.9 and 22.1 ± 9.2 W m−2),
the combined radiation will increase by 117.5 ± 276.5% and Discussion
48.4 ± 53.3% over EUR and NA. However, under low-emission sce- Output from CMIP6’s LS3MIP and ScenarioMIP experiments projects
nario, although the combined radiation will also increase (23.1 ± 5.9 that very high GHG emission will result in soil drying and reduced

Nature Communications | (2023)14:4908 5

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Article https://doi.org/10.1038/s41467-023-40641-y

Fig. 4 | Soil moisture-atmosphere coupling (SA) mediated relationship between month) due to SA under high-emission scenario over northern middle latitudes:
soil moisture and evapotranspiration. a, b Represent temporal evolution of soil 30–60°N, 180°W–180°E, southern subtropical: 20–40°S, 180°W–180°E, central and
moisture (kg m−2) due to SA under low- and high-emission scenarios. c, d same as eastern Europe (EUR): 40–60°N, 20–50°E, and central North America (NA):
a, b, but for evapotranspiration (mm per month). In a–d, the former (1985–2014) 28–55°N, 88–110°W. Numbers given in the upper left are correlation coefficients
denotes modern conditions and the latter (2015–2094) denotes future conditions. between evapotranspiration and soil moisture for the modern (Mod), mid-term
The time series is performed with Lanczos low-pass filtering to remove the inter- (Mid), and long-term future (Long) periods. Two stars indicate that the correlation
annual variability in 1980–2099, the total number of weights is 11 and the cut-off coefficient passes the 99% significance test, and one star indicates that the corre-
frequency of the low-pass filter is 1/11. Shading is uncertainty among models. e–h lation coefficient passes the 95% significance test. Black, blue, and red lines are the
Represent Scatter between soil moisture (kg m−2) and evapotranspiration (mm per fitting lines in the modern, mid-term, and long-term future periods.

evapotranspiration, thereby forcing more heat into the atmosphere via low-emission scenario), Supplementary Fig. 5 (CMCC-ESM2), and Sup-
enhanced downward shortwave radiation and sensible heat flux. Such plementary Fig. 6 (IPSL-CM6A-LR). And the results showed that they had
SA conditions will serve to further amplify the GHG-driven warming. similar characteristics. In the six global climate models, CESM2, CMCC-
Under the worst (highest) emission scenario, the amplification due to ESM2, IPSL-CM6A-LR, and MPI-ESM1-2-LR include simulated dynamic
SA is projected to increase over time owing to the uptrend evapo- vegetation, and EC-Earth3 and MIROC6 models do not include simu-
transpiration rate associated with drying soil, which follows an accel- lated dynamic vegetation. The surface soil moisture depth is 10 cm, and
erating amplified-warming. Such acceleration in SA-warming will make the root zone soil moisture depth is 100 cm.
extreme high-temperature events both more frequent and more The LFMIP-pdLC experiment in LS3MIP, SSP1-2.6,and SSP5-8.5
severe, particularly over North America and Europe. The implication of experiments in ScenarioMIP, and the historical experiment are ana-
these findings suggests that mitigation efforts corresponding to lyzed. The LFMIP-pdLC experiment in LS3MIP is used to assess the
acceleration of SA-driven warming must be implemented at an early impact of land-atmosphere coupling caused by soil moisture on
stage to minimize the risk of climate shock. Policymakers must also weather and climate through fixing the soil moisture as their clima-
consider maintaining ecosystem stability as a tool for sustaining soil tological state (the annual mean cycle for the period 1980–2014
moisture within appropriate limits, thereby avoiding the worst impacts derived from historical global climate model output). Then the SA
of elevated SA, especially in North America and Europe. Here we effect can be isolated through the relative differences between fully
emphasize on the local effect of SA on surface air temperature while coupled experiments (historical, the SSP1-2.6, and SSP5-8.5 experi-
the latest research shows that SA can also affect the large-scale ments) and fixed soil moisture experiment (LFMIP-pdLC
atmospheric circulation under the global warming background16. In experiment)23. The ScenarioMIP reflects future climate (2015–2099)
this view, the joint contribution of local and non-local SA effect on under high and low Shared Socioeconomic Pathways (SSP)24. The SSP1-
global land warming needs to be further investigated. Finally, the 2.6 experiment and r1i1p1f1 member in the LFMIP-pdLC experiment
results may have some projection uncertainty considering the limita- corresponds to the same low-emission scenarios (2015–2099), and the
tions of the parameterization in SA29–31 and scenario uncertainty due to SSP5-8.5 experiment and r1i1p1f2 member in the LFMIP-pdLC experi-
the lack of knowledge of future radiative forcing32. Nevertheless, the ment corresponds to the same high-emission scenarios (2015–2099).
models we used in this research have high reliability in the historical
simulation of soil moisture33, and we reduced the uncertainty by multi- Extreme high-temperature
model mean. Meanwhile, different single models have similar results The spatial distribution of extreme high-temperature (Fig. 3a, b, Sup-
for the acceleration in SA-driven warming in the future projection. So, plementary Fig. 4a, b, Supplementary Fig. 5a, b, and Supplementary
the conclusion that SA amplifies greenhouse gas-driven global warm- Fig. 6a, b) is determined by calculating the 90th percentile of surface
ing is relatively reliable and robust. air temperature in each grid point. And for the probability distribution
function of surface air temperature over NA and EUR (Fig. 3c, d, Sup-
Methods plementary Fig. 4c, d, Supplementary Fig. 5c, d, and Supplementary
CMIP6 models Fig. 6c, d), regional average of surface air temperature is calculated
Six global climate models in CMIP6 (CESM2, CMCC-ESM2, EC-Earth3, first, and then the probability distribution function of the regional
IPSL-CM6A-LR, MIROC6, and MPI-ESM1-2-LR) are used to analyze the average is calculated. In order to analyze the extreme high-
impact of SA on surface air temperature. Extra-tropical summer is temperature changes driven by SA, the 90th percentile of surface air
defined as the June–August (JJA) average in Northern Hemisphere and temperature in LFMIP-pdLC experiment is used as the threshold
the December–February (DJF) average in Southern Hemisphere. Since temperature in three time periods (modern, mid-term, and long-term)
there have been only six CMIP6 LS3MIP models published so far, we separately. The probability and intensity differences of surface air
only use these six models in this article. The multi-model mean is used temperature in coupled experiment (historical, the SSP1-2.6, and SSP5-
to remove uncertainties arising from model differences. For the 8.5 experiments) relative to the threshold temperature in LFMIP-pdLC
majority of our analyses, we incorporated monthly data derived from all experiment are taken as the SA effect across the three time horizon
six models. Due to the lack of latent heat flux in MIROC6 model, the periods (modern, mid-term, and long-term).
multi-mode mean of latent heat flux is the result of the other five
models. The surface air temperature probability distribution function Contribution to the warming trend
and extreme high-temperature are calculated from daily data for each We calculated the warming trend (°C per decade, 2015-2099) of total,
model (no multi-model mean) separately. Because the amount of GHG, and SA effect on ground surface air temperature under high-
monthly data is not enough compared with daily data, we are concerned emission scenario (SSP5-8.5) over the globe (excluding Antarctica),
that the probability distribution function characteristics cannot be northern middle latitudes (Northern: 30–60°N, 180°W–180°E),
accurately captured. Meanwhile, the multi-model mean will mask the southern subtropical latitudes (Southern: 20–40°S, 180°W–180°E),
intensity distribution of extreme high-temperatures when the number Europe (EUR: 40–60°N, 20–50°E), and North America (NA: 28–55°N,
of models is little. So, we used daily data to calculate the probability 88–110°W) in future projections (Supplementary Table 1). (1) The total
distribution function and extreme high-temperature probability for contribution to the warming trend is obtained by calculating the trend
each model separately in Fig. 3 (MPI-ESM1-2-LR model under high- of the surface air temperature in fully coupled experiment under high-
emission scenario), Supplementary Fig. 4 (MPI-ESM1-2-LR model under emission scenario (SSP5-8.5), because the time series of this

Nature Communications | (2023)14:4908 6

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Article https://doi.org/10.1038/s41467-023-40641-y

a Total cloud cover b Surface received shortwave

Modern Modern

Mid-term Mid-term

Long-term Long-term

Percent (%) Shortwave radiation (W m-2)

c Latent heat flux d Sensible heat flux

Modern Modern

Mid-term Mid-term

Long-term Long-term

Heat flux (W m-2) Heat flux (W m-2)

Fig. 5 | Spatial distributions of land surface meteorological elements caused by shortwave radiation (W m−2), c is latent heat flux (W m−2), and d is sensible heat flux
soil moisture-atmosphere coupling (SA) in modern (1995–2014), mid-term (W m−2). Black dots signify agreement between the sign of change and the multi-
future (2040–2059), and long-term future (2080–2099) periods under the very model mean in at least five of the six (or four of five) CMIP6 models.
high-emission scenario (SSP5-8.5). a is total cloud cover (%), b is surface-received

Nature Communications | (2023)14:4908 7

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Article https://doi.org/10.1038/s41467-023-40641-y

experiment included both GHG and SA effects; (2) the contribution of resistance. Because in water-restricted areas such as arid and semi-arid
SA to the warming trend is obtained by calculating the trend of the regions, the change of latent heat flux is more determined by soil
difference of surface air temperature between fully coupled experi- moisture, so latent heat flux is not written as a function of the change
ment (SSP5-8.5) and fixed soil moisture experiment (LFMIP-pdLC) of T a .
under high-emission scenario, because the time series of this differ- Using Eqs. (2)–(8), the surface energy balance equation can be
ence between the two experiments considered as SA effect; (3) the expanded as following:
contribution of GHG to the warming trend is obtained by calculating

the trend of the surface air temperature in fixed soil moisture experi-   Ts  Ta
Sτ ð1  α Þ + εs σ εa T 4a  T 4s = λE + ρC d ð9Þ
ment under high-emission scenario (LFMIP-pdLC), because the time ra
series of this experiment included GHG effect and excluded SA effects.
Assuming that S, λ, ρ, C d , σ and εs are independent of T s , it can be
Decomposition of SA-driven changes in surface air further differentiated the Eq. (9) with respect to T s :
temperature28

 
We use the surface energy balance to decompose the surface air 1 4 ρC d T s  T a
ΔT s = SτΔα  λΔE + Sð1  αÞΔτ + εS σT a Δεa + Δr a
temperature (Ta) changes caused by SA. The changes in Ta are mainly fs 2
ra
driven by the land surface temperature (Ts) and atmospheric circula- ρC d =r a + 4εS σεa T 3a
tion (ΔT cir
a ), where Ts interacts with Ta through radiative and non-
+ ΔT a
ρC d =r a + 4εS σT 3s
radiative fluxes (ΔT rad
a ). So, the ΔT a can be expressed as:
ð10Þ
ΔT a = ΔT rad
a + ΔT cir
a
ð1Þ
where f s is an energy redistribution factor:
In order to calculate the change of Ts caused by SA, we start with
the formula of land surface energy balance: f s = ρC d =r a + 4εS σT 3s ð11Þ

1
Sn + Ln = λE + H + G ð2Þ f s is the Ts sensitivity to 1 W m−2 radiative forcing.
On the right side of Eq. (10), the first term is the Ts change caused
where Sn and Ln are the net short-wave and long-wave radiation at the by radiative and thermodynamic forcings associated with SA-driven
surface. λ, E, and H are the vaporization latent heat, evapotranspira- changes in surface albedo, evapotranspiration, short-wave transmis-


tion, and sensible heat flux. G is the ground heat flux, which magnitude sivity, air emissivity, and aerodynamic resistance (ΔT rad = 1=f ðSτΔ
s
is relatively small and can be ignored in seasonal and longer timescales. ρC ðT T Þ
Equation (2) can be simplified as: α  λΔE + Sð1  α ÞΔτ + εS σT 4a Δεa + d r 2s a Δr a Þ). The second term
a
ρC d =r a + 4εS σεa T 3a
( fs
ΔT a ) quantifies the coupling strength between T a and
Sn + Ln = λE + H ð3Þ
T s . It means that T s varies with T a caused by SA-driven change in air
ρC d =r a + 4εS σεa T 3a
Sn can be expressed as: advection (ΔT cir
s = fs ΔT cir
a ). Meanwhile, it also suggests
that the T s change further drives a change in T a through the surface
Sn = Sτ ð1  αÞ ð4Þ heating rate change (ΔT rad fs
ΔT rad
a = ρC d =r a + 4εS σεa T 3a s ). Through equations

(1) and (10), Ta changes caused by SA can be decomposed into the

where S is the solar radiation flux at atmosphere top, τ is the atmo-
following forms:
spheric short-wave transmissivity, α is the surface albedo.
According to the Stephan–Boltzmann law, the downward long- 

1 ρC d T s  T a
wave radiation at the land surface can be calculated roughly as: ΔT a = SτΔα  λΔE + Sð1  α ÞΔτ + εS σT 4a Δεa + Δr a + ΔT cir
a
f r 2a
ð12Þ
L# = εa σT 4a ð5Þ
1
Where f = ρC d =r a + 4εS σεa T 3a ,
f is the T a sensitivity to 1 W m−2
where εa is atmospheric air emissivity, σ is the Stephan–Boltzmann radiative forcing.
constant (5.67 × 10−8 Wm−2 K−4). The change in surface air temperature (ΔT a ) caused by SA is
The upward long-wave radiation is expressed as: decomposed into radiative forcing terms by Eq. (12)28 and incorpor-
ating surface albedo (SτΔα), evapotranspiration (λΔE), shortwave


L" = 1  εS εa σT 4a + εS σT 4S ð6Þ transmissivity (Sð1  α ÞΔτ), air emissivity (εS σT 4a Δεa ), aerodynamic
ρC ðT T Þ
resistance ( d r 2s a Δr a ), and residual term (ΔT cira ). In the decom-
a
where εS is the land-surface emissivity, and it can be treated as a position of surface air temperature, we employed four modes of
constant of 0.95, since the εS varies very little over different land multi-model mean (CESM2, CMCC-ESM2, EC-Earth3, and IPSL-CM6A-
covers, around 0.95. Therefore, the net long-wave radiation at the land LR) in our calculations. The remaining two models (MIROC6 and MPI-
surface is expressed as: ESM1-2-LR) were omitted because they lack the relevant radiation
  input data.
Ln = L#  L" = εs σ εa T 4a  T 4s ð7Þ
Data availability
The sensible heat flux is expressed as: All data used in this study are freely available online. The CMIP6 model

simulations are from https://esgf-node.llnl.gov/search/cmip6/.
ρC d T s  T a
H= ð8Þ
ra Code availability
Analysis and figure generation were performed using NCL and
where ρ is the air density (1.21 kg m−3), C d is the air specific heat at MATLAB. The code and scripts of the five figures in the paper are
constant pressure (1013 J kg−1 K−1), and r a is the aerodynamic available from Zenodo: https://doi.org/10.5281/zenodo.7928584.

Nature Communications | (2023)14:4908 8

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Article https://doi.org/10.1038/s41467-023-40641-y

References temperature feedbacks. Geophys. Res. Lett. 44, 1511–1519, https://

1. IPCC. Climate Change 2013: The Physical Science Basis. Contribu- doi.org/10.1002/2016gl071235 (2017).
tion of Working Group I to the Fifth Assessment Report of the Inter- 20. Fischer, E. M., Seneviratne, S. I., Lüthi, D. & Schär, C. Contribution of
governmental Panel on Climate Change [Stocker, T.F., D. Qin, G.-K. land-atmosphere coupling to recent European summer heat waves.
Plattner, M. Tignor, S.K. Allen, J. Boschung, A. Nauels, Y. Xia, V. Bex Geophys. Res. Lett. 34, https://doi.org/10.1029/2006gl029068
and P.M. Midgley (eds.)]. (Cambridge University Press: Cam- (2007).
bridge, 2013). 21. Seo, E. et al. Impact of soil moisture initialization on boreal summer
2. Ji, F., Wu, Z., Huang, J. & Chassignet, E. P. Evolution of land surface subseasonal forecasts: mid-latitude surface air temperature and
air temperature trend. Nat. Clim. Change 4, 462–466, https://doi. heat wave events. Clim. Dyn. 52, 1695–1709, https://doi.org/10.
org/10.1038/nclimate2223 (2014). 1007/s00382-018-4221-4 (2018).
3. Jones, G. S., Stott, P. A. & Christidis, N. Attribution of observed 22. Zhang, P. et al. Abrupt shift to hotter and drier climate over inner
historical near-surface temperature variations to anthropogenic East Asia beyond the tipping point. Science 370, 1095–1099,
and natural causes using CMIP5 simulations. J. Geophys. Res.: Atm. https://doi.org/10.1126/science.abb3368 (2020).
118, 4001–4024, https://doi.org/10.1002/jgrd.50239 (2013). 23. Bart van den Hurk, B. et al. LS3MIP (v1.0) contribution to CMIP6: the
4. Berg, A. et al. Impact of Soil Moisture–Atmosphere Interactions on Land Surface, Snow and Soil moisture Model Intercomparison
Surface Temperature Distribution. J. Clim. 27, 7976–7993, https:// Project – aims, setup and expected outcome. Geoscientific Model
doi.org/10.1175/jcli-d-13-00591.1 (2014). Dev. 9, 2809–2832, https://doi.org/10.5194/gmd-9-2809-2016
5. Seneviratne, S. I., Luthi, D., Litschi, M. & Schar, C. Land-atmosphere (2016).
coupling and climate change in Europe. Nature 443, 205–209, 24. O’Neill, B. C. et al. The Scenario Model Intercomparison Project
https://doi.org/10.1038/nature05095 (2006). (ScenarioMIP) for CMIP6. Geoscientific Model Dev. 9, 3461–3482,
6. Zhang, J., Wu, L. & Dong, W. Land-atmosphere coupling and sum- https://doi.org/10.5194/gmd-9-3461-2016 (2016).
mer climate variability over East Asia. J. Geophys. Res. 116, https:// 25. Prasanna, V. Assessment of South Asian Summer Monsoon Simu-
doi.org/10.1029/2010jd014714 (2011). lation in CMIP5-Coupled Climate Models During the Historical
7. Koster, R. D. et al. Regions of Strong Coupling Between Soil Period (1850–2005). Pure Appl. Geophys. 173, 1379–1402, https://
Moisture and Precipitation. Science 305, 1138–1140, https://doi. doi.org/10.1007/s00024-015-1126-6 (2015).
org/10.1126/science.1100217 (2004). 26. Qiao, L., Zuo, Z., Xiao, D. & Bu, L. Detection, Attribution, and Future
8. Williams, C. J. R., Allan, R. P. & Kniveton, D. R. Diagnosing atmosphere–- Response of Global Soil Moisture in Summer. Front. Earth Sci. 9,
land feedbacks in CMIP5 climate models. Environ. Res. Lett. 7, 044003, https://doi.org/10.3389/feart.2021.745185 (2021).
https://doi.org/10.1088/1748-9326/7/4/044003 (2012). 27. Berg, A., Lintner, B., Findell, K. & Giannini, A. Soil Moisture Influence
9. Betts, A. K., Ball, J. H., Beljaars, A. C. M., Miller, M. J. & Viterbo, P. A. on Seasonality and Large-Scale Circulation in Simulations of the
The land surface-atmosphere interaction: A review based on West African Monsoon. J. Clim. 30, 2295–2317, https://doi.org/10.
observational and global modeling perspectives. J. Geophys. Res. 1175/jcli-d-15-0877.1 (2017).
Atm. 101, 7209–7225, https://doi.org/10.1029/95jd02135 (1996). 28. Zeng, Z. et al. Climate mitigation from vegetation biophysical
10. Zhang, R. L. L. Z. Z. Variation of soil moisture over China and their feedbacks during the past three decades. Nat. Clim. Change 7,
influences on Chinese climate. Chin. J. Nat. 38, 313–319 (2016). 432–436, https://doi.org/10.1038/nclimate3299 (2017).
11. Lemordant, L. & Gentine, P. Vegetation Response to Rising CO2 29. Dirmeyer, P. A. Characteristics of the water cycle and
Impacts Extreme Temperatures. Geophys. Res. Lett. 46, 1383–1392, land–atmosphere interactions from a comprehensive reforecast
https://doi.org/10.1029/2018GL080238 (2019). and reanalysis data set: CFSv2. Clim. Dyn. 41, 1083–1097, https://
12. Alan Williams, C. Heat and drought extremes likely to stress eco- doi.org/10.1007/s00382-013-1866-x (2013).
system productivity equally or more in a warmer, CO2 rich future. 30. Roundy, J. K., Ferguson, C. R. & Wood, E. F. Impact of land-
Environ. Res. Lett. 9, 101002, https://doi.org/10.1088/1748-9326/9/ atmospheric coupling in CFSv2 on drought prediction. Clim. Dyn.
10/101002 (2014). 43, 421–434, https://doi.org/10.1007/s00382-013-1982-7 (2013).
13. Skinner, C. B., Poulsen, C. J. & Mankin, J. S. Amplification of heat 31. Santanello, J. A., Roundy, J. & Dirmeyer, P. A. Quantifying the
extremes by plant CO2 physiological forcing. Nat. Commun. 9, Land–Atmosphere Coupling Behavior in Modern Reanalysis Pro-
1094, https://doi.org/10.1038/s41467-018-03472-w (2018). ducts over the U.S. Southern Great Plains. J. Clim. 28, 5813–5829,
14. Li, L. et al. Responses of greenhouse gas fluxes to climate extremes https://doi.org/10.1175/jcli-d-14-00680.1 (2015).
in a semiarid grassland. Atmos. Environ. 142, 32–42, https://doi.org/ 32. Lehner, F. et al. Partitioning climate projection uncertainty with
10.1016/j.atmosenv.2016.07.039 (2016). multiple large ensembles and CMIP5/6. Earth Syst. Dyn. 11,
15. Clark, R. T., Brown, S. J. & Murphy, J. M. Modeling Northern Hemi- 491–508, https://doi.org/10.5194/esd-11-491-2020 (2020).
sphere Summer Heat Extreme Changes and Their Uncertainties 33. Qiao, L., Zuo, Z. & Xiao, D. Evaluation of Soil Moisture in CMIP6
Using a Physics Ensemble of Climate Sensitivity Experiments. J. Clim. Simulations. J. Clim. 35, 779–800, https://doi.org/10.1175/jcli-d-20-
19, 4418–4435, https://doi.org/10.1175/JCLI3877.1 (2006). 0827.1 (2022).
16. Zhou, S. et al. Diminishing seasonality of subtropical water avail-
ability in a warmer world dominated by soil moisture-atmosphere Acknowledgements
feedbacks. Nat. Commun. 13, 5756, https://doi.org/10.1038/ We acknowledge the World Climate Research Programme’s Working
s41467-022-33473-9 (2022). Group on Coupled Modelling, which is responsible for CMIP, and thank
17. Jaeger, E. B. & Seneviratne, S. I. Impact of soil moisture–atmosphere the various climate modeling groups who produced and made available
coupling on European climate extremes and trends in a regional their model output. Z.Z. acknowledges the National Key Research and
climate model. Clim. Dyn. 36, 1919–1939, https://doi.org/10.1007/ Development Program (Grant No. 2022YFF0801703) and the National
s00382-010-0780-8 (2010). Natural Science Foundation of China (41822503). D.X. acknowledges the
18. Mueller, B. & Seneviratne, S. I. Hot days induced by precipitation National Natural Science Foundation of China (42175053).
deficits at the global scale. Proc. Natl Acad. Sci. USA 109,
12398–12403, https://doi.org/10.1073/pnas.1204330109 (2012). Author contributions
19. Vogel, M. M. et al. Regional amplification of projected changes in Z.Z., L.Q., R.Z., and S.P. conceived of and designed the study, L.Q.
extreme temperatures strongly controlled by soil moisture‐ performed analyses, Z.Z. and L.Q. wrote the paper. D.X. and K. Z.

Nature Communications | (2023)14:4908 9

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Article https://doi.org/10.1038/s41467-023-40641-y

assisted in the framing and development of ideas. Z.Z., L.Q., R.Z., S.P., Publisher’s note Springer Nature remains neutral with regard to jur-
D.X., and K.Z. edited the paper. isdictional claims in published maps and institutional affiliations.

Competing interests Open Access This article is licensed under a Creative Commons
The authors declare no competing interests. Attribution 4.0 International License, which permits use, sharing,
adaptation, distribution and reproduction in any medium or format, as
Additional information long as you give appropriate credit to the original author(s) and the
Supplementary information The online version contains source, provide a link to the Creative Commons license, and indicate if
supplementary material available at changes were made. The images or other third party material in this
https://doi.org/10.1038/s41467-023-40641-y. article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not
Correspondence and requests for materials should be addressed to included in the article’s Creative Commons license and your intended
Zhiyan Zuo or Renhe Zhang. use is not permitted by statutory regulation or exceeds the permitted
use, you will need to obtain permission directly from the copyright
Peer review information Nature Communications thanks the anon- holder. To view a copy of this license, visit http://creativecommons.org/
ymous reviewer(s) for their contribution to the peer review of this work. A licenses/by/4.0/.
peer review file is available.
© The Author(s) 2023
Reprints and permissions information is available at
http://www.nature.com/reprints

Nature Communications | (2023)14:4908 10

Content courtesy of Springer Nature, terms of use apply. Rights reserved
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:

1. use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
2. use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
3. falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
4. use bots or other automated methods to access the content or redirect messages
5. override any security feature or exclusionary protocol; or
6. share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at

[email protected]