Introduction

Soil moisture plays a vital role in the climate system. The exchange of water and heat between land and atmosphere depends on the amount of soil moisture. Soil moisture impacts the surface energy balance and the hydrological cycle by directly affecting evaporation. Two major feedback loops are described in the literature: soil moisture and temperature, and soil moisture and precipitation1. The soil moisture-temperature feedback works as follows. When the surface air temperature rises, more water evaporates from the soil, reducing the soil moisture content. At some point, a lack of soil moisture limits evaporation and the associated evaporative cooling. As a result, the ground heats, and the sensible heat flux increases, leading to higher surface air temperatures. Reduced evaporation and higher air temperatures lead to reduced cloud cover and enhanced solar radiation at the surface. This further amplifies the drying and the surface warming. The soil moisture-precipitation feedback works as follows. Precipitation increases soil moisture. Wetter soils lead to more evaporation and moistening of the planetary boundary layer, favouring cloud formation. More clouds lead to enhanced precipitation. The drying and wetting of the soil takes time and introduces a memory into the climate system. Therefore, knowledge of the soil moisture state can lead to skillful forecasts on sub-seasonal to seasonal time scales2.

Previous studies have shown the significance of soil moisture in promoting and maintaining extreme events such as heat waves and droughts, both as local and remote drivers. Locally, there are various pathways in which dry soils trigger and amplify heat waves. These processes and mechanisms are mainly land-atmosphere coupling3,4, evapotranspiration5, feedback mechanisms6, and surface energy balance2,7,8. Miralles et al.4 demonstrated that the mega-heatwave in Europe in the summers of 2003 and 2010 would not have been possible without the soil desiccation conditions, emphasizing the land-atmosphere coupling. When soil drying intensifies diurnal warm air entrainment, it leads to the accumulation of atmospheric heat that persists over time4. Usually, the evapotranspiration process from soil and plants regulates the land surface temperature as it has a cooling effect. However, when the soil moisture becomes limited, the evapotranspiration rate reduces, which induces heat wave conditions, if prolonged, even drought conditions9. Soil moisture-temperature and soil moisture-precipitation are two dominating feedback mechanisms; for soil moisture-temperature feedback, surface drying is induced by reduced soil moisture, enhancing surface heating during heat waves. The heated surface again dries the soil moisture below6. Various studies, including a generalized framework7 and both regional2 and global climate models8, show that the partitioning of land surface energy is strongly affected by soil moisture as it affects the balance between sensible and latent heat fluxes; the dry soil contributes to increased sensible heat flux and, thus, persistent heat waves.

Additionally, apart from the role of soil moisture in heat waves, circulation and atmospheric dynamics play a crucial role. In the recent ExtremeX global climate model experiment study10 where the different multi-model experiments covered free atmospheric and land interaction, atmospheric nudging, and soil moisture prescription, the results from recent heatwave case studies demonstrated that atmospheric circulation patterns and soil moisture anomalies play key roles in extreme heat occurrences. While soil moisture effects are essential for the US Great Plains, tropics and monsoon regions, the effects of the circulation anomaly patterns dominate in the rest of the extratropic and high-latitude areas.

Furthermore, a critical aspect of extreme events study is atmospheric dynamics. In middle latitudes, large-scale atmospheric circulation, namely jet streams, affects the day-to-day weather systems and governs the land surface conditions. Previous studies showed that high-amplitude quasi-stationary Rossby waves can trigger extreme events11,12,13. Both reanalysis and model studies have illustrated the importance of specific Rossby wave patterns, especially amplified circumglobal wave numbers 5 and 7 in boreal summer, in promoting stagnating weather systems and, therefore, are also responsible for global simultaneous extremes14,15. Regionally, the waviness of jets can impact the numbers and locations of weather extremes at mid-latitudes16. A “recurrent synoptic-scale Rossby wave pattern” (RRWP) was identified based on reanalysis data that showed it to be responsible for persistent hot spells17. An overview of waveguide theory and its association with surface extreme weather was summarised in a recent study18.

Linking soil moisture to upper-level atmospheric circulation, a few studies have looked at the soil moisture-atmosphere interaction in one way: dry soils affect large-scale circulation in the Northern Hemisphere13,19 and Southern Hemisphere20,21. For example, during mega heat waves, dry soils can influence local circulation4, and dry soils can even affect remote circulation22. Di Capua et al.22 showed that the dry soil in western Russia enhances flood risks in Pakistan via a Rossby wave response. Gloege et al.23 also looked into spring but focused on the case study of the 2020 Siberian Heatwave.

The study from Koster et al.24 supported the positive feedback loop idea that drying various interior regions in North America leads to atmospheric circulation changes that reinforce the heating on the continental interior. In their experiments, drying soil moisture simulations were performed with a stationary wave model (SWM) and an atmospheric general circulation model (AGCM). In the SWM experiments, the model was forced to use idealized diabatic heating at a few locations in the continental United States; whereas in the AGCM experiments, soil moisture dryness was imposed in several places in the US interior. Despite the different models used and locations chosen, the atmospheric circulation anomaly showed similar patterns: a high-pressure anomaly in West-Central North America and a low-pressure anomaly to the east. The proposed positive feedback loop suggests that drying in soil moisture imposed in the US interior triggers a specific atmospheric circulation pattern, which warms up the central US with increased surface sensible heating that reinforces the original atmospheric circulation anomalies24. The need for large ensembles (96–192) was also emphasized for detecting the spatial patterns of the induced stream function, precipitation, and 2-m temperature anomalies in their results.

In this paper, we build upon this earlier work and try to understand the role of soil moisture-atmosphere interactions on the hemispheric-scale boreal summer circulation. We cover the analysis of the summer season, mainly using the month of July as an example to understand the soil moisture conditions and, hence, their implications for surface variables. Furthermore, we analyze jet stream patterns and planetary wave changes induced by different soil moisture experiments.

This is the first study to examine the role of two-way soil moisture interaction between land and atmosphere on the climatological summer circulation employing large ensemble global climate model experiments.

This paper aims to answer the following research questions:

  • How does the two-way interaction between soil moisture and atmosphere dynamics influence the climatological mean state of the jet stream and planetary wave in boreal summer?

  • What is the circulation’s response to variability in soil moisture?

Results

Jet stream response to soil moisture-atmosphere interaction

First, we focus on the question of whether the two-way interaction between soil moisture and the atmosphere influences the climatological state of the jet streams. To do so, we compare the July climatology of the experiment with interactive soil moisture (SMI) and without (SME). Figure 1 shows the climatology for thermodynamic variables for SMI in the left column. The impact of interactive soil moisture is in the right column, as it is the difference in the climatology between SMI and SME. Dynamical variables are shown in Fig. 2. Stippled areas outline the regions where the difference is significant at 99% confidence (determined from bootstrapping 1000 times of random samples from 800 July months). The soil moisture climatology difference between SMI and SME is not shown in Fig. 1, as the prescribed soil moisture in SME comes from the SMI soil moisture climatology. By design, the difference is zero.

Fig. 1: Difference plots between SMI and SME surface thermodynamic variables for ensemble climatology means in July for simulated years 2009–2016.
Fig. 1: Difference plots between SMI and SME surface thermodynamic variables for ensemble climatology means in July for simulated years 2009–2016.
Full size image

The left column is the SMI climatology mean, and the right column is the difference between SMI and SME. a, b Near-surface temperature [k], c, d surface sensible heat flux [W m−2], e, f surface latent heat flux [W m−2], g, h net short-wave radiation [W m−2]. Stippled areas show significant (α = 0.01, two-sided) differences between SMI and SME climatology means obtained from 1000 bootstraps based on 800 ensemble monthly means.

Fig. 2: Difference plots between SMI and SME atmospheric dynamic variables for ensemble climatology means in July for simulated years 2009–2016.
Fig. 2: Difference plots between SMI and SME atmospheric dynamic variables for ensemble climatology means in July for simulated years 2009–2016.
Full size image

The left column is the SMI climatology means, and the right column is the difference between SMI and SME. a, b Azonal plots for geopotential height at 500 hPa [m], c, d meridional wind at 250 hPa [m s−1], e, f zonal wind at 250 hPa [m], g, h stream function at 250 hPa [m2 s−1], i, j precipitation pr [mm]. Stippled areas show significant (α = 0.01, two-sided) differences between SMI and SME climatology means obtained from 1000 bootstraps based on 800 ensemble monthly means.

The mean surface air temperature is generally much higher with interactive soil moisture (Fig. 1b). The largest impact (0.5–1.5 k) occurs in east Siberia, Russia, Europe, and central to east North America. In these areas, we see enhanced net short-wave radiation (Fig. 1h) and increased stream function values at 250 hPa (Fig. 2h). These changes suggest that in these areas, on average soil moisture more often limits evaporation in SMI, leading to reduced evaporative cooling of the surface, increased surface temperature and sensible heat flux, fewer clouds, more solar radiation and a warmer air column beneath 250 hPa leading to the positive stream function anomalies aloft. Increased sensible heat flux due to increased available energy (Fig. 1d). In the areas of increased radiation, we find significantly reduced precipitation (Fig. 2j).

The mean position and strength of the jet streams are affected by interactive soil moisture. (Fig. 2e, f). In Fig. 2e, the Atlantic and subtropical jets are visible, as shown by the anomalies in their climatological positions. Over North America and Eurasia, the jet stream is weaker and slightly northward shifted with interactive soil moisture. The polar front jet above the northern coast of Eurasia is enhanced and above the North Atlantic there is a clear northward shift of the jet stream leading to a more south-west to north-east tilt. Stationary waves are also affected. A circumglobal wave response is visible in the meridional wind response at 250 hPa as well as in the azonal part of geopotential height at 500 hPa (Fig. 2b, d). The positive geopotential height changes match as expected the positive stream function changes over Canada, East Europe, and Northeast Asia. The stationary wave response has a clear circumglobal wave number 5 signature.

The monsoon is strengthened in SMI runs, as shown in Fig. 2j. There is more precipitation in Indian monsoon belt regions in SMI than in SME runs. This corresponds well with the larger latent heat fluxes in SMI (Fig. 1f).

Surface temperature and jet stream response to soil moisture variability

In order to assess the linearity of the atmospheric response to soil moisture, we assess how the dry and wet composites differ from the mean atmospheric July state of the whole ensemble. Figure 3 shows the results for an Eastern European region (EEU) indicated by the red square for the left response to dry conditions and the right response to wet conditions. The first row shows the difference in soil moisture (Fig. 3a, b). In the EEU region, the soil is indeed drier than the mean in the dry composite and wetter in the wet composite. This also applies to most regions outside of EEU. Furthermore, the surface temperature and the circulation response to the variability in the soil moisture can be analyzed from the SMI experiments. Since the soil moisture is not prescribed in SMI experiments, each ensemble has a different soil moisture state, and the whole ensemble preserves the variability in soil moisture; the surface temperature response to soil moisture deficits, therefore, can be compared with periods of soil moisture abundance. The composites of soil moisture (SM), near-surface temperature (tas, Fig. 3c, d), and upper-level meridional wind (v250, Fig. 3e, f) anomalies were constructed based on the Eastern European (EEU) soil moisture index time series to investigate the asymmetric response of temperature relating to soil moisture variability. To exclude the potential influence of prescribed different sea surface temperature states on the model’s near-surface temperature and atmospheric circulations, 100 ensemble members are compared for the selected specific year, in this case, year 2011.

Fig. 3: Composites for difference maps between EEU dry/wet soil and ensemble climatology means in July for the simulated year 2011.
Fig. 3: Composites for difference maps between EEU dry/wet soil and ensemble climatology means in July for the simulated year 2011.
Full size image

The left column is based on the driest 20% soil moisture, and the right column is the wettest 20% soil moisture. a, b Dry and wet 20% soil moisture difference to climatology [kg m−2], c, d near surface temperature [K], e, f meridional wind at 250 hPa [m]. EEU is the region in the red box.

Figure 3 shows the driest (SM_dry20) and wettest (SM_wet20) 20% quartiles relative to the climatological mean (ensemble mean) for variables soil moisture, near-surface temperature, and upper-level meridional wind. Under the dry soil conditions, pronounced soil moisture deficits emerge over the EEU region, stretching zonally across the selected domain (Fig. 3a). These anomalies coincide with near-surface warming for at least 1.5 K in the central EEU region (Fig. 3c). These local warming anomalies are dynamically linked to a negative v250 anomaly extending northeast of EEU (Fig. 3e), indicating a strengthened upper-level trough and enhanced southward flow. The resulting cold-air advection aloft promotes subsidence and clear-sky conditions downstream, which reinforces surface warming through soil moisture-temperature feedbacks. However, on the contrary, the wet soil anomalies (SM_wet20 – SM_mean) are associated with cooling (Fig. 3d) and a positive v250 response (central and southern EEU region; Fig. 3f), suggesting a ridge pattern and weakened meridional temperature gradients. Overall, the anomalies for both near-surface temperature and upper-level meridional wind, with dry EEU soil conditions, the response is stronger and more pronounced compared to the wet soil conditions. Similar analyses and composites for the other five regions25 (Fig. S4), West & Central Europe (WCE), East Siberia (ESB), Central North America (CNA), North South Africa (NSA), and Central Africa (CAF), are shown in Fig. S5. Figure S5 shows that the near-surface temperature warming anomalies for dry soil conditioned composites for EEU, ESB, and CAF regions are stronger than the cooling anomalies; the signals are also more pronounced in the upper-level meridional winds v250. For the WCE, CNA and NSA regions, the temperature anomalies have similar magnitudes, with NSA exhibiting a greater atmospheric response compared to the wet soil states. However, the circulation responses to dry anomalies are not always stronger than the responses to wet anomalies, for example, over region CNA (Fig. S5).

Discussion

In our study, we compared the July ensemble climatological mean fields for surface thermodynamically controlled variables and upper-level atmospheric circulation variables between large-ensemble simulations with different soil moisture and land-atmosphere interactions. Notably, panel Fig. 1b patterns identified similar regions that have seen the strongest warming over recent decades by the study26 in Fig. 1a–c, suggesting a potential role of soil moisture in amplifying the warming trends. Additionally, the observed linear trend from 1979 to 2020 for near-surface temperature anomalies during JJA seasonal mean during selected extreme events27 also coincides with our hotspots in Fig. 1b, such as east Siberia, Russia, Europe, and northern Canada.

When two-way land-atmosphere interaction is turned on (e.g. SMI versus SME), we observe a poleward migration of the jet and a more pronounced Arctic front jet state. Additionally, we found an anomalous wave number 5 circumglobal wave pattern (Fig. 2b, d). This implies that the two-way dynamic interaction between land (soil moisture) and atmosphere affects the position and waves in the jet stream. This anomalous circumglobal wave number 5 pattern and/or its correspondence with surface heat extremes were also found in a recent study by ref. 27. The wave amplitudes over land are strengthened by 24% in the SMI experiments compared to the SME experiments, based on July ensemble means of meridional wind (v250) (Fig. S4), suggesting that the inclusion of two-way land-atmosphere interactions enhances wave amplitudes. Overall, the SMI experiments are warmer than SME runs, suggesting the two-way interaction also introduces more extreme events. As seen in Fig. S3, from the 90th percentile of near-surface temperature comparisons between SMI and SME, we see more heat extremes in Central North America.

A possible explanation of the northward jet shift is that the atmospheric response to local surface energy flux changes is not always in situ, as changes in soil moisture conditions locally induce both local and remote changes in forcings. As the soil dries out, initially, latent heating is reduced, and sensible heating is increased; but later, the atmosphere starts to respond. There will be cloud formation and precipitation processes, then divergence in the air and thus wind changes. As a result, the jets shift their positions. The persistent anomalies in soil moisture can also act as forcing for Circumglobal Wave Trains (CGWTs), specifically amplifying the wave number 5 pattern in the mid-latitudes. This results from the influence of soil moisture on surface heat flux, which further modulates atmospheric baroclinicity and wave activity19.

A few other studies have highlighted the emerging climate signals, based on decades of reanalysis data and modelling data, such as the poleward shift of jet streams in the Northern Hemisphere and attributed that to global warming-induced changes in atmospheric circulation, notably enhanced upper-tropospheric baroclinicity and alterations in stratospheric winds that impact tropospheric eddies and momentum flux28,29,30.

One special aspect of our soil moisture prescription experiments is that the soil moisture prescription is applied at every grid point, every soil level, and every time step. This separates our simulations from previous studies, which used location-specific soil moisture prescription experiments19,24. Our analyses thereby shed light on the global effect of soil moisture-atmosphere interactions on large-scale circulation, but also make it more challenging to interpret in terms of physics.

In addition, from our analysis, the response from drying vs. wetting soil is asymmetric. This is due to the effect of soil moisture depletion on near-surface temperature6 since, as long as there is enough water in the soil (Figs. 3b and S3 wet regime), it is an energy-limited regime, and sensible heat fluxes would be small. However, warming has more effect when the soil water content is entirely evaporated in SMI runs (Figs. 3a and S3 dry regime). For the EEU region, Fig. 3c shows a stronger local temperature warming effect and a strengthened circumglobal wave response (Fig. 3e) compared to the cooling (Fig. 3d) and upper-level circulation (Fig. 3f). In the areas of the Northern Hemisphere mid-latitudes and land-dominated regions, such as EEU and ESB, the near-surface warming effect is stronger than the cooling effect (Fig. S5, columns 3 and 4).

In our analysis, we selected both the driest and wettest 20% from the 100 SMI ensemble members for the year 2011. The magnitudes of surface temperature warming and upper-level meridional wind anomalies are stronger for dry soil conditioned members than for wet soil conditioned ensembles, for the EEU region. The spatial distributions of these anomalies differ for the two sets of composites. The wave pattern causes a dry anomaly over Europe, and this soil moisture deficit (potentially leading to droughts) feeds back into the wave again. This implies that the variability in soil moisture influences surface temperature, the spatial pattern of such temperature anomalies, and, furthermore, circumglobal wave patterns. Thus, one must consider this asymmetrical effect when studying soil moisture and its relationship to surface energy balance and fluxes, as well as its influence on upper-level atmospheric jets.

The EEU dry soil composites v250 anomaly patterns (Fig. 3e) indicate that local soil moisture forcing can contribute to the circumglobal wave-5 teleconnection pattern, which in turn modulates the amplitude and phase of the summer jet. This two-way interaction between land and atmosphere supports the view that Eurasian soil moisture deficits can produce quasi-stationary circumglobal waves, enhancing persistent ridges and downstream heat extremes. This barotropical reinforcement mechanism is consistent with previous findings13,14, which show that mid-latitude land-atmosphere interactions/feedbacks amplify wave persistence. Overall, the EEU composites demonstrate how local soil moisture conditions can influence hemispheric circulation and furthermore, the surface temperature extremes across Eurasia.

One strength in our approach and results, which echoed the study from Koster et al.24, is that the large number of ensemble members (in our case 100, and theirs 96–192) was carried out for detecting some specific spatial anomalous patterns in anomalous atmospheric circulation and near-surface temperature. However, at the same time, we have only conducted experiments with EC-Earth3; more models can be added to make more confident and general statements. Although a large ensemble size improves the sampling of internal climate variability, it does not compensate for the lack of model diversity, particularly with respect to differing representations of land-atmosphere interactions, surface fluxes, and soil moisture processes. As such, results may still be model-dependent.

In our study, we compared the large ensemble simulations’ climatological means for various variables. This allowed us to capture the signal of circulation change induced by land-atmosphere interaction. Despite the overall robustness of our results, we still lack confidence in explaining the physical process in detail. The changes we observe are a combination of local and remote effects, which are also outside this study’s scope.

Future work can investigate the amplified circumglobal wave characteristics and associated extreme events. It is also possible to investigate the relationship between the early spring soil moisture conditions in each experiment set and the corresponding summer extreme heat events, to analyze the impact on the seasonal predictability of summer heat waves. A further aspect would be to investigate land-atmosphere interaction extremes by considering the tails of the distributions of temperature and circulation, rather than focusing on the mean states.

In summary, our results show that the two-way interaction between land and atmosphere is critical in driving circulation changes, such as poleward shifts and strengthening of jet stream patterns. This land-atmosphere interaction amplifies the potential for extreme events, as shown by the 90th percentile temperature anomalies increasing when SME is compared to SMI (Fig. S1). Moreover, the wave amplitudes over land are strengthened by approximately 24% from SME to SMI experiments (Fig. S2), which implies the key role of land-atmosphere coupling in modulating atmospheric wave dynamics. This has important implications, especially for certain amplified waves that trigger and maintain more extremes in the mid-latitudes.

Methods

European Community Earth System Model version 3 (EC-Earth3)

The model output from the European Community Earth System Model (EC-Earth3)31 is used in this study. The atmosphere component from EC-Earth is based on the European Centre for Medium-Range Weather Forecasts (ECMWF) Integrated Forecasting System IFS cycle cy36r4. There are 91 vertical model levels; the horizontal model resolution is T255 (ca. 80 km). IFS also has a land component H-TESSEL land surface model32 that consists of four soil layers: 0–7, 7–28, 28–100, and 100–255 cm. Monthly sea surface temperature and sea ice are prescribed in the model, using the HadISST1 and NOAA OI2 data sets33. EC-Earth3 was run with model version 3.3.1 with CMIP6 scenario SSP3-7.0 for all the experiments.

Experiment set-up

This study includes two sets of ensemble experiments. An overview of the experiments is presented in Table 1. All experiments have the same prescribed SSTs, sea ice cover fraction, and land cover. All the experiments consist of 100 ensemble members for the period of 2009–2016. From these experiments, by comparing SMI and SME simulations, we gain insight into the effect of the two-way interaction between soil moisture and atmospheric dynamics on summertime Northern Hemisphere jet streams.

Table 1 EC-Earth3 model Experiments description
Full size table

Soil moisture data

All four soil layers were prescribed for SME experiments. The SME prescribed soil moisture was calculated based on all 100 SMI ensemble members’ soil moisture values. First, the median ensemble soil moisture values were computed for each day, then a 15-day running mean was applied, followed by a multi-year (2009–2016) daily mean calculation.

July climatology plots

Climatology plots are computed as the ensemble monthly mean from 100 ensemble members from 2009 to 2016 and over the month of July. All the ensemble members are initially output every 6 h, then climatological means are calculated based on these data to create monthly means. We restrict our study domain from 20S to 90N and 180W to 180E to cover the Northern Hemisphere and Southern tropics.