Impacts of late-spring North Eurasian soil moisture variation on summer rainfall anomalies in Northern East Asia

This paper reports findings from a diagnostic and modeling analysis that investigates the impacts of the boreal late-spring soil moisture anomalies over North Eurasia on the summer rainfall over northern East Asia (NEA). The soil moisture in May from the Kara-Laptev Sea coasts to Central Siberian Plateau is found to be negatively correlated with the summer rainfall from Mongolia to Northeast China. The atmospheric circulation anomalies associated with the anomalously dry soil are characterized by a pressure dipole with the high-pressure center located over North Eurasia and low-pressure center over NEA, where an anomalous water vapor convergence occurs, favoring rainfall formation. Diagnoses and modeling experiments demonstrate that the effects of the spring lower soil moisture over North Eurasia may persist into the following summer, then increase the low-level air temperature at higher latitudes through modulating local surface latent and sensible heat fluxes, and effectively reduce the meridional temperature gradient north of NEA. The weakened temperature gradient could induce the decreased vertical shear of zonal wind and generate an anomalously cyclonic center over NEA by affecting the baroclinicity around 60° N, associated with a favorable condition of local synoptic activity to increase rainfall. The above relationships and mechanisms are vice versa for the prior wetter North Eurasian soil and decreased NEA rainfall. These findings suggest that the soil moisture anomalies at higher latitudes may act as a new precursor providing an additional predictability source for better predicting the summer rainfall in NEA.


Introduction
As an important part of surface hydrological cycle, soil moisture plays a vital role in global land-atmosphere interactions (Seneviratne et al. 2010). By influencing numerous physical processes in land-atmosphere interactions, soil moisture has considerable impacts on climate elements, such as precipitation and temperature. Similar to ocean, soil moisture has the character of "memory" (Koster and Suarez 2001;Shinoda 2001;Ruosteenoja et al. 2018). Previous studies have indicated that the soil moisture signals in middle and high latitudes of the Northern Hemisphere could last 1-3 months and have a lagged effect on atmospheric circulation anomalies, which will further influence climate elements (Shukla and Mintz 1982;Dirmeyer et al. 2009). As a result of its memory effect, soil moisture as a precursor is instructive for short-term climate prediction.
The climatic effects of soil moisture have been extensively studied during the past few decades (e.g., Yeh et al. 1984;Chahine 1992;Dai et al. 1999; Koster et al. 2003;Seneviratne et al. 2006;Sato and Nakamura 2019). Based on highly controlled numerical experiments by different climate models, Koster et al. (2004) proposed that there are specific "hot spots" with strong land-atmosphere couplings, where soil moisture anomalies have substantial impacts on precipitation through influencing evaporation and other surface energy fluxes. In recent years, studies have shown that better precipitation simulations and predictions may be expected with the improved accuracy of initial soil moisture, which could exert distinct influences on precipitation change (Shi et al. 2021). Wu and Zhang (2013), based on Weather Research and Forecasting (WRF) model simulations, explored the role of soil moisture playing in the formation of flood and drought over East China in the summers of 1998 and 1999, and showed that soil moisture substantially contributes to both of droughts and floods over northern East China through its memory. These results could help improve the skill of seasonal precipitation prediction.
However, owing to the deficiencies of reliable large-spatial and long-term measurement of soil moisture, many studies have had to rely on model simulations to investigate the relationship between soil moisture and climate variability (Wu and Dickinson 2004;Kim and Hong 2007). Therefore, further diagnoses based on increasing amounts of the new observations as well as the latest reanalysis products are still needed. Moreover, previous studies have mainly focused on the local climatic effects of soil moisture anomalies and rarely on the potential remote impacts. Xue et al. (2018) demonstrated that the upstream land surface, by influencing heat fluxes, may affect the subsequent downstream droughts/ floods in East Asia. This study further demonstrated that such effects are not only constrained within local areas but also extended downstream.
The East Asian summer rainfall directly influences the economy and human wellbeing across the region. However, increasing the accuracy of seasonal prediction of summer rainfall in East Asia has always posed considerable challenges, particularly in its northern region (Ding and Chan 2005;Wang 2006;Zhou and Zou 2010). Some studies have shown that the summer precipitation in East China is influenced by a number of factors, including soil moisture, which is particularly important (Zuo and Zhang 2007;Zhang and Zuo 2011;Dai and Zuo 2010;Zhang et al. 2016a). However, the relationship between soil moisture and precipitation illustrated in these studies has generally been constrained to only local effects within East China.
North Eurasia in higher latitudes covers a vast area and the soil moisture there with seasonal scale of memory (Dirmeyer et al. 2009) may have impacts on the surrounding climate. Given the relatively low accuracy of summer rainfall prediction over northern East Asia (NEA) which is located in the middle latitudes near south of North Eurasia, the soil moisture variability over North Eurasia and its possible impacts on the summer rainfall in NEA are worthy of examination for a potential future application in the seasonal prediction. In this study, we aimed to investigate the linkage between the summer (June-August, JJA) precipitation anomalies in NEA and the precedent late-spring (May) soil moisture anomalies in North Eurasia through observational data diagnoses and numerical model experiments. The results may provide new insights on how to improve the prediction of summer precipitation in NEA.
The remainder of the paper will be organized as follows. Section 2 describes the data, methods and climate model used here, while Sect. 3 discusses the relation of late-spring soil moisture anomalies and summer rainfall anomalies in NEA. Section 4 presents the impacts of soil moisture on atmospheric circulation connected with rainfall. The design and results of numerical model experiments are given in Sect. 5. Finally, the summary and discussions are provided in Sect. 6.

Data
Due to the lack of reliable soil moisture gauge measurement datasets covering large-domain and long-term scales in situ observations, assimilation and reanalysis products are generally two kinds of alternative soil moisture datasets available for researches. Assimilated soil moisture data products such as the Global Land Data Assimilation System (GLDAS) use advanced land surface modeling and data assimilation techniques to provide combined satellite-and ground-based observational products for soil moisture researches (Rodell et al. 2004). In this study, two versions of the GLDAS (v2.0 and v2.1) data from the Goddard Earth Sciences Data and Information Services Center were combined to form the monthly mean near-surface (0-10 cm) soil moisture dataset on a 0.25° latitude × 0.25° longitude grid covering the period of 1980-2018. In addition, soil moisture on the first level (7 cm) from the European Centre for Medium-Range Weather Forecasts (ECMWF) reanalysis (ERA) Interim/ Land data (Balsamo et al. 2015), which includes the land reanalysis product using the latest ECMWF land surface model with numbers of parameterization improvements in land surface scheme, was used as a reference to make our conclusions more robust. Correspondingly, the surface heat fluxes data was also taken from the GLDAS and ERA-Interim/Land. Unless stated otherwise, the related figures and indices displayed in this paper were calculated using the GLDAS. The monthly mean precipitation data used here was extracted from the CPC (Climate Prediction Center) Merged Analysis of Precipitation (CMAP, Xie and Arkin 1997). The daily and monthly atmospheric circulation field data (e.g., sea level pressure (SLP), geopotential heights, zonal and meridional wind) were obtained from the National Center for Environmental Prediction and Department of Energy Reanalysis 2 (NCEP-DOE R2, Kanamitsu et al. 2002).

Methods
To demonstrate the dominant coupled modes between the NEA summer rainfall and the preceding soil moisture in North Eurasia, we used the singular value decomposition (SVD) analysis (also known as maximum covariance analysis) (Wallace et al. 1992;Von Storch and Zwiers 1999). Other commonly used statistical analysis techniques and metrics such as the Pearson correlation and linear regression were also adopted in this study. All the data were linearly detrended and removed annual cycle to focus on climate anomalies. The climate state ranges from 1980 to 2018, which is the period focused on in this study. Furthermore, the Eady growth rate (Eady 1949) was calculated using daily data to represent the baroclinic instability.

Numerical model
The Community Earth System Model version 1.2.2 (CESM 1.2.2), a state-of-the-art coupled climate model, with interactive components of the atmosphere, land, ocean and sea ice, was used to detect the effects of preceding soil moisture anomalies in this study. The Version 4.0 of Community Land Model (CLM 4.0) is the land component of CESM 1.2.2, which mainly includes the processes of surface energy transportation, hydrology and biogeochemical cycles (Lawrence et al. 2011 3 Relationship between the NEA summer rainfall and preceding soil moisture in North Eurasia

The statistical relationship
Severe flood and drought events occurred over Northeast China in the summers of 2013 and 2015, respectively, and caused substantial economic losses (Yang and He 2013;Wang et al. 2016). We found that the summer rainfall anomalies in these two cases are well accompanied by the precedent late-spring soil moisture anomalies in North Eurasia. As shown in Fig. 1, the negative soil moisture anomalies at higher latitudes (i.e. North Eurasia) occur along with the positive anomalies relating to the summer rainfall in NEA, and vice versa. To confirm whether or not these negative correlations are only rare cases, we applied the SVD at a longer time scale ranging from 1980 to 2018. Figure 2 presents the results of SVD analysis of the soil moisture anomalies in May over North Eurasia and precipitation anomalies in summer over NEA. The first SVD (SVD1) modes of the precipitation anomalies in summer and soil moisture anomalies in May account for 66.2% of their covariance. Both of the two variables depict the same sign of homogeneous correlations in their respective areas (Fig. 2a, c), demonstrating that the late-spring soil moisture and summer precipitation are characterized by consistent variations. Figure 2b, d show that the patterns of SVD1 heterogeneous correlations are similar to those of homogeneous correlations, indicating a close connection between these two fields. Specifically, for the North Eurasian soil moisture anomalies in May, the SVD1 spatial pattern shows significantly negative correlations covering the coastal areas of Kara-Laptev Sea and Central Siberian Plateau. The spatial pattern of summer precipitation anomalies features pronounced positive correlations from Mongolia to Northeast China, the sign of which is opposite to that of soil moisture. The SVD1 modes calculated by the ERA-Interim/Land (figures not shown) resemble those of the GLDAS; however, the explained variance is relatively lower (35.6%). Figure 2e presents the two extension coefficient time series corresponding to the SVD1 modes with apparent interannual and interdecadal variability are significantly correlated (r = 0.71, significant at the 95% confidence level of Student's t test). Note that the products of SVD pattern values and their corresponding time series in each year are proportional to the actual anomalies of corresponding variables for the year. When both of the time series of precipitation and soil moisture are positive, the soil moisture in North Eurasia will be negatively anomalous and the precipitation over NEA will be positively anomalous. This indicates that the precedent dryer soil moisture in North Eurasia is significantly related to the increased summer precipitation in NEA, and vice versa. To more precisely describe the variations of the late-spring soil moisture and summer rainfall, we selected the key regions of soil moisture (62° N-74° N, 90° E-125° E) and precipitation the blue and red lines denoting their 5-year running means. Red and blue boxes in the maps denote the key regions of soil moisture and precipitation, respectively. Values that are statistically significant at the 90 and 95% confidence levels are indicated by green and purple dots, respectively (43° N-51° N, 107° E-130° E), and defined the time series of averaged soil moisture anomalies in May and averaged precipitation anomalies in JJA over their respective boxes as the soil moisture and precipitation indices, respectively. The correlation coefficients between the two new indices and corresponding SVD1 time series are − 0.81 and 0.70 for soil moisture and precipitation, respectively, suggesting that the indices defined by box-average can realistically represent the primary coupled and individual variations of two variables. More importantly, the correlation coefficient between the indices of soil moisture and precipitation is − 0.72 (significant at the 95% confidence level, Fig. 2f), which is almost the same as the minus correlation coefficient between the SVD1 time series. This further indicates that the summer rainfall anomalies in NEA are significantly negatively correlated with the precedent soil moisture anomalies in late spring over North Eurasia. Furthermore, these two indices remain highly correlated (r = − 0.78, significant at the 0.05 level) after the removal of 5-year running mean to eliminate the influence of interdecadal background, which demonstrates that these two variables are highly correlated not only at interdecadal scale, but also at interannual scale. Hence, the indices used in the following analysis are all removed interdecadal signal.

The atmospheric circulation related to the preceding soil moisture anomalies
We further explored whether the late-spring soil moisture anomalies are linked to atmospheric circulation and further impact the middle-latitude rainfall variation over NEA in the following summer. Figure 3 presents the patterns of SLP and geopotential height anomalies in JJA regressed against − 1 × soil moisture index (henceforth the minus soil moisture index). In summer, the statistically significant below-normal SLP, which is conducive to forming positive rainfall anomalies, can be detected around the key region of precipitation in NEA, extending from Mongolia to Northeast China ( Fig. 3a). At the same time, the area from the coasts of Kara-Laptev Sea to Central Siberian Plateau, including the key region of soil moisture, is covered by the significantly positive SLP anomalies. Similarly, the geopotential height anomalies at 850 hPa ( Fig. 3b) also show a dipole over Eurasia that highly resembles the pattern presented in SLP. The similar pattern emerges to the heights of 500 and 200 hPa (Fig. 3c, d) except that the ranges of anomalous geopotential height centers show some shrinkage and westward-moving. Therefore, the summer atmospheric circulation anomalies over NEA are clearly related to the earlier soil moisture in late spring over North Eurasia and further likely affect the summer precipitation. Specifically, the negative North Eurasian soil moisture anomalies intimately correspond to the dipole pattern of "positive in the north and negative in the south" in atmospheric pressure and geopotential height anomalies. Correspondingly, the key region in NEA where the summer precipitation is highly correlated with the North Eurasian soil moisture is occupied by subnormal pressure from lower to upper levels. This quasi-barotropic anomaly structure favors the formation of positively anomalous summer rainfall in this area. On the contrary, the positive soil moisture anomalies at higher latitudes are supposed to be related with the positive pressure anomalies and deficit of rainfall over NEA. Figure 4a, b show the anomalies of wind at 850 hPa and vertically integrated water vapor fluxes in summer regressed against the minus soil moisture index, respectively. Both of the two variables show that there is a significantly anomalously cyclonic circulation pattern in NEA, which is conducive to the atmospheric convergence and upward motion favoring the increase of rainfall. There are two significant water vapor transport channels to NEA: one is from South China Sea, and the other is from the ocean east of Japan.
Storm track described by the variance of 2-8-days bandpassed filtered (synoptic-scale) meridional wind at 850 hPa could represent activity of synoptic eddies. Figure 4c shows the JJA storm track regressed against the minus soil moisture index. It can be seen that NEA is featured with positive values especially in Northeast China, which demonstrates that the background circulation anomalies (i.e., zonal wind, high/ low pressure) in NEA related to the anomalously dry soil are beneficial to the generation of synoptic eddies. By further regressing the summer Eady growth rate at 850 hPa against the minus soil moisture index (Fig. 4d), we could find that the positive Eady growth rates are more significant over the area south of the Baikal. It suggests that the background Values that are statistically significant at the 90 and 95% confidence levels are indicated by green and purple dots, respectively circulation anomalies related with the negative soil moisture anomalies will lead to the strengthened baroclinic instability over the key region of precipitation in NEA. Therefore, the strengthened baroclinic instability will allow the synoptic eddies to be more active over NEA (Hoskins and Valdes 1990). Previous studies have shown that synoptic eddies, such as the Northeast-China Cold Vortex at mid-high latitudes of East Asia in summertime, are closely responsible for heavy rain over the northeastern and northern China (He et al. 2006;Zhou et al. 2020).
Therefore, the statistically significant relationship between the anomalies of summer precipitation in NEA and late-spring soil moisture over North Eurasia is linked with the dipole of anomalous pressure, cyclonic/anticyclonic circulation and related water vapor transport. In the following section, we will discuss the possible mechanisms by which the preceding soil moisture anomalies can affect the summer circulation and rainfall in NEA.

The impacts of soil moisture on atmospheric circulation connected with rainfall
Soil moisture could affect precipitation through its local and nonlocal effects (Seneviratne et al. 2010;Li and Xue 2014;Li et al. 2016). In the case of local effects, the increase of soil moisture lead to strengthened evapotranspiration, including vegetation transpiration and surface evaporation, and thus an increased supply of water vapor into atmosphere, which is conducive to the increase of precipitation at local scale (Koster et al. 2003). For the nonlocal effects, the change of surface temperature in continental monsoon region caused by the soil moisture variation would affect thermal contrast between land and ocean, and further modulate large-scale monsoonal circulation associated the precipitation anomalies at larger scale (Zhang et al. 2016b). At these two different scales, the soil moisture anomalies strongly affect atmosphere through changing surface heat fluxes. Specifically, the wet soil leads to higher surface latent heat flux through evapotranspiration at the expense of surface sensible heat flux, which further affects the atmosphere above (Dirmeyer and Brubaker 1999;Koster et al. 2003;Xue et al. 2004Xue et al. , 2006. Thus, it is necessary to further investigate whether the summer atmospheric anomalies discussed above are resulted from the changes in thermal conditions caused by the effects of soil moisture on surface heat fluxes. Figure 5a presents the North Eurasian soil moisture anomalies in JJA regressed against the minus soil moisture index of May. The area from the West Siberian Plain to Central Siberian Plateau, including the key soil moisture region, is featured with the significantly negative soil moisture anomalies, which validates that the late-spring soil moisture signals could persist into following summer. Indeed, the soil moisture indices of May and JJA are highly autocorrelated (r = 0.89, significant at the 95% confidence level), indicating that the memory is of seasonal scale. Figure 5b shows the North Eurasian JJA air temperature anomalies at 2 m height regressed against the minus soil moisture index. North Eurasia is distinguished by the positively anomalous temperature in summer, especially the key soil moisture region, illustrating that the negative soil moisture anomalies in May over North Eurasia are associated with the increased lower-level air temperature at high latitudes in summer, and vice versa. The distributions of surface heat fluxes anomalies regressed against the minus soil moisture index are shown in Fig. 6. North Eurasia is characterized by the significantly negative latent heat flux anomalies in summer (Fig. 6a). This suggests that the persistent abnormally dry soil moisture signal from May to JJA could result in the negative anomalies of latent heat fluxes over the same key soil moisture region in summer. The patterns of sensible heat flux anomalies also show clear features. Figure 6b shows that the significantly positive sensible heat flux anomalies nearly appear over the whole of North Eurasia. The sign of sensible heat flux anomalies is opposite to that of latent heat flux anomalies, which is consistent with the findings of previous studies (Koster et al. 2006;Seneviratne et al. 2010). The positive sensible heat flux anomalies will directly increase the air temperature at lower atmosphere layers above the key soil moisture region, as shown by Fig. 5b and the air temperature in latitude-height section along 105° E in Fig. 6c. The increased air temperature in North Eurasia will lead to the decrease of meridional temperature gradient between middle and high latitudes. Simultaneously, the cooling caused by the rainfall increase above the key precipitation region further intensifies the decreased temperature gradient.
According to the thermal wind principle, the decreased meridional temperature gradient between the key regions of soil moisture and precipitation will weaken the vertical shear of zonal wind and strengthen the anomalous easterlies around 60° N at lower and higher levels (Fig. 7a, b). The anomalous easterlies will cause the weakened storm track at 300 hPa around 60° N, especially in the north of NEA (Fig. 7c). The weakened storm track or baroclinicity at higher levels is beneficial to the formation of cyclonic circulation in the south of the anomalous storm track and an anticyclone in the north of it (Fig. 4a) due to the vorticity fluxes transport (Lau and Holopainen 1984). This distribution of wind circulation acts to generate the dipole pattern of SLP and geopotential height over Eurasia.
To summarize, by affecting the surface heat fluxes, the soil moisture anomalies over North Eurasia will influence the meridional temperature gradient by changing lower-layer b zonal wind at 300 hPa (unit: m/s) and c storm track at 300 hPa, derived from regressions against the minus soil moisture index. Blue and red boxes in the maps denote the key regions of soil moisture and precipitation. Values that are statistically significant at the 90 and 95% confidence levels are indicated by green and purple dots, respectively air temperature. Then the anomalous zonal wind and storm track around 60° N caused by the variation of meridional temperature gradient will finally affect the atmospheric circulation related to the Eurasian summer precipitation.

Model experiments
In this section, we assessed the influences of precedent soil moisture anomalies in North Eurasia on the atmospheric circulation related to the NEA summer precipitation and investigated the associated physical processes establishing such connections, using model experiments conducted with the CESM 1.2.2. Three sets of experiments forced by prescribed climatological monthly-mean SST and sea-ice concentration were performed. The control run (exp_CTL) was integrated for 50 model years and the first 10 years were discarded for model spin-up. The high-soil moisture experiment (exp_ HSM) and low-soil moisture experiment (exp_LSM) were two sets of sensitivity experiments used for detecting the influences of soil moisture anomalies in North Eurasia. The surface soil moisture anomalies at each grid point over the soil moisture key region SM anom were calculated by equation where SM max are the maxima of soil moisture at each grid point over the key region of the 40-year control run; and SM min are the minimum values. SM anom were added and subtracted onto the outputs over the key region of exp_CTL, which were regarded as the initial conditions of exp_HSM and exp_LSM, respectively. For the initial conditions of exp_HSM, if the values of soil moisture were larger than the maxima of the 40 model years in exp_CTL, then they were assigned as the maxima. Similarly in the initial conditions of exp_LSM, zeroes were assigned to the values less than zero. Each sensitivity experiment consisting of 40 ensemble members was integrated from 1 May to 31 August. The model responses to the soil moisture anomalies were defined as the ensemble mean differences between exp_LSM and exp_HSM. Figure 8 displays the ensemble mean differences of summer soil moisture, precipitation and atmospheric circulation between exp_LSM and exp_HSM. As we can see from Fig. 8a, b, the dryer soil in North Eurasia could persist from May to summer, and NEA is featured with the positive rainfall anomalies. Significantly negative SLP anomalies covering the area ranging from Mongolia to Northeast China are favorable for the positive rainfall anomalies (Fig. 8c). At the same time, North Eurasia, which includes the eastern part of the key soil moisture region, is characterized by higher SLP. Pattern at the height field of 850 hPa (Fig. 8d) similar with that of SLP shows the dipole pattern of "positive in the north SM anom = SM max − SM min , and negative in the south" except that the positive anomalies are less significant. The wind at 850 hPa (Fig. 8e) matches well with the height field. A clearly anomalous anticyclone appears above the key soil moisture region. On the south of the anticyclone, there is a significantly cyclonic shear circulation above NEA that will enhance convergence and upward motion at lower level and further promote the formation of positive rainfall anomalies. The ensemble mean difference of vertically integrated water vapor fluxes (Fig. 8f) shows that NEA is a zone of significant convergence. The pattern of divergence of vertically integrated water vapor fluxes clearly shows that the regions south of the Lake Baikal ranging from Mongolia to Northeast China are characterized by significantly negative divergence anomalies (Fig. 8g), favoring the positive rainfall anomalies. Figure 9a shows that the latent heat fluxes over North Eurasia in summer simulated by exp_LSM are significantly lower than that of exp_HSM. Contrary to the latent heat fluxes, the responses of sensible heat fluxes show that the positive anomalies cover North Eurasia (Fig. 9b). As a result of the directly heating from sensible heat fluxes, the air temperature at lower atmosphere levels in North Eurasia gets increased and the meridional temperature gradient between middle and high latitudes over Eurasia becomes decreased (Fig. 9c). This leads to the weakened zonal wind in the Lake Baikal between two key regions (Fig. 9d). This anomalous easterly flow will generate the cyclonic shearing circulation above NEA and then the pressure dipole spanning East Asia. Clearly, the results of the sensitivity experiments by the CESM 1.2.2 basically validate the main conclusions of observational diagnoses that the summer circulation anomalies related to the rainfall anomalies in NEA can be influenced by the late-spring soil moisture anomalies over North Eurasia.

Summary and discussions
In this study, we investigated the relation between the latespring soil moisture over North Eurasia and the summer rainfall over NEA based on observations and numerical model experiments. The SVD1 modes of NEA precipitation in summer and North Eurasian soil moisture in May with 66.2% of their explained variance show that both of the fields depict the same sign of heterogeneous correlations in their respective areas, which is very similar to the patterns of homogeneous correlations. In particular, the soil moisture anomalies over North Eurasia in May show a significantly negative correlation with the following summer rainfall anomalies over the regions from Mongolia to Northeast China. Furthermore, the two extension coefficient time series corresponding to SVD1, which reveal nearly consistent interannual and interdecadal variations with each other, are highly correlated with the coefficient of 0.71. The late-spring soil moisture index and summer precipitation index calculated by the method of box-average, are also negatively correlated (r = − 0.72). We further found that the late-spring negative soil moisture anomalies over North Eurasia are associated with the dipole of SLP and geopotential height pattern over the whole of East Asia in summer: the key region of soil moisture is occupied by positive pressure anomalies, and the key region of precipitation is covered by negative ones that favor the formation of positive rainfall anomalies. Pattern at the height field of 850 hPa (Fig. 8d) similar with that of SLP shows the dipole pattern of "positive in the north and negative in the south" except that the positive anomalies are less significant. The water vapor fluxes show two significant water vapor transport channels to the key region related to precipitation. At the same time, the background circulation anomalies related to the abnormal soil moisture will also influence baroclinic instability and synoptic activity, then affect the distribution of rainfall anomalies. Hence, the related conditions of circulation lead by the high-latitude abnormally dryer (wetter) soil will enhance (decrease) the precipitation over NEA.
Upon further analysis, we identified the mechanism by which the North Eurasian soil moisture anomalies can generate the appropriate atmospheric circulation that affects the middle-latitude precipitation. We found that the abnormal soil moisture signal in May will persist into the following summer and consistently influence the surface heat fluxes. In particular, the dryer soil in North Eurasia will lead to the negative latent heat flux anomalies and positive sensible heat flux anomalies. Correspondingly, the increased air temperature in North Eurasia will narrow the meridional temperature gradient between middle and high latitudes and weaken the vertical shear of zonal wind, then promote the anomalous easterlies at around 60° N. The storm track north of NEA are weakened by the anomalous easterlies and then generates the cyclonic circulation over NEA through vorticity transport according to Lau and Holopainen (1984). By strengthening the storm track (baroclinicity) and eddy activity at 45° N, this cyclonic circulation finally increase the summer rainfall in NEA. With the increase of precipitation in NEA, the cooling effect of rainfall further decreases the meridional temperature gradient and promotes this cyclic process summarized in Fig. 10. The influences of the wetter soil in North Eurasia are vice versa. Thus, changing the thermal conditions by affecting the heat fluxes is a vital pathway by which the soil moisture influences the atmospheric circulation.
The results of the ensemble mean differences between the exp_LSM and exp_HSM are basically consistent with those of observational diagnoses. We should notice that although the differences pass the significant test, the values of the response in model as compared to the observed values are relatively small. But overall, the North Eurasian abnormal soil moisture persisting from May to summer could influence the distribution of surface temperature and then affect the atmospheric circulation related with precipitation in NEA, though the high-pressure systems are less significant and the the source of water vapour is different in model. By comparing the exp_CTL outputs and observations (figure not shown), we found that the underestimation of pressure in exp_CTL over North Eurasia, especially in the area around the Lake Baikal and the north, could account for the smaller and less significant positive anomaly. In the aspect of winds, the relatively weaker southeast current blowing form the Bay of Bengal and South China Sea to NEA in model simulation might cause the difference of water vapor flux. But overall the observational diagnoses and model sensitivity experiments both show that the preceding soil moisture anomalies over North Eurasia play a vital role in affecting the largearea atmospheric circulation and further influence the summer precipitation in NEA.
However, it is still not clear how the synoptic eddy activity could contribute to the seasonal-scale precipitation in NEA and related atmospheric circulation though we have observed some results that the background steering flow and baroclinity changes induce variations of synoptic storm/ eddy activity. This question is definitely important to deeply understand formation dynamics of the NEA cyclonic anomaly circulation as excited by prior external factors such as soil moisture, and worthy of further studies on the next step in terms of the scale interaction paradigm between highfrequency synoptic eddy with a lifetime of 2-8 days and low-frequency flow with a lifetime of seasonal scale (e.g., Ren et al. 2011Ren et al. , 2014. The prediction of summer rainfall over East Asia has been a major issue in short-range climate prediction. Moreover, the forecasting accuracy of summer precipitation in NEA, especially in Northeast China, is relatively poor and has even been contrary to the actual situation in recent years (Sun et al. 2017;Zhao et al. 2020). In this study, from the perspective of land factors, we found observational evidence of a linkage between the summer precipitation anomalies over NEA and the late-spring soil moisture anomalies over higher-latitude region and confirmed these by performing numerical model experiments. The results provide a base Fig. 8 The simulated ensemble-mean difference patterns in summer between the exp_LSM and exp_HSM in a soil moisture (unit: kg/m 2 ); b precipitation (unit: mm/day); c SLP (unit: Pa); d 850-hPa height (unit: gpm); e wind vector at 850-hPa (unit: m/s); f vertically integrated water vapor fluxes (unit: kg/m/s) and g divergences of the vertically integrated water vapor fluxes (unit: kg/s). Blue and red boxes in the maps denote the key regions of soil moisture and precipitation. Values that are statistically significant at the 90% confidence level are indicated by green dots ◂ Fig. 9 The simulated ensemble-mean difference patterns in summer between the exp_LSM and exp_HSM in a, b surface latent and sensible heat fluxes (units: W/m 2 ), respectively; c air temperature at 850 hPa (unit: °C); and d zonal wind at 850 hPa (unit: m/s). Blue and red boxes in the maps denote the key regions of soil moisture and precipitation. Values that are statistically significant at the 90% confidence level are indicated by green dots Fig. 10 Schematic diagram depicting the physical processes responsible for the influence of the late-spring soil moisture over North Eurasia on the subsequent summer precipitation over NEA for improving the short range climate forecasts of summer rainfall over East Asia. However, the mid-high-latitude precipitation is usually affected by multiple local and remote factors, and the circulation related to precipitation behaves chaotically. Exploration and research are needed to further quantify the amount of rainfall that can be explained by land factors such as soil moisture and surface temperature.