Theor Appl Climatol https://doi.org/10.1007/s00704-017-2298-z

ORIGINAL PAPER

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over northwestern China and western Mongolia in July Yang Zhou 1 & Ben Yang 2 & Yong Zhao 3 & Jing Jiang 2 & Anning Huang 2 & Mengke La 4

Received: 16 February 2017 / Accepted: 11 October 2017 # Springer-Verlag GmbH Austria 2017

Abstract A significantly negative interannual relationship is identified between the ground surface temperature (GTS) over the Tibetan Plateau (TP) and the rainfall over northwestern China and western Mongolia (NWC-WM) through analyzing the Chinese weather station data, GPCP precipitation, and ERA-Interim reanalysis in July during 1980–2012. This relationship is verified by the model sensitivity experiments carried out by using RegCM4.1 during 1982–2011. The positive/ negative GTS forcing of three different magnitudes is added in two key regions over the TP in RegCM4.1. One of the key regions covers the central and eastern TP (denoted as TPC). The other covers the northern and north slope of the TP (denoted as TPN). The model results suggest that when the GTS anomalies in either of the two key regions are negative (positive), the rainfall anomalies over NWC-WM are positive

(negative), which is consistent with observations. Furthermore, rainfall anomalies over NWC-WM are more sensitive to the GTS anomalies over the TPN region than those over the southern TP. The model results also reveal that the negative (positive) GTS anomalies over region TPN mainly cause the decrease (increase) of the latent heat release related to rainfall (surface sensible heat) and descent (ascent) over the TPN region but ascent (descent) to the north of the TP between 40° and 50° N. In addition, the specific humidity between 40° and 50° N is increased (decreased). Therefore, the increase (decrease) in specific humidity and the ascent (descent) between 40° and 50° N cause the increase (decrease) in the rainfall over NWC-WM.

1 Introduction Key points • TP GTS negatively correlates with rainfall over northwestern China and western Mongolia. • Negative relationship is found in sensitivity experiments. • Vertical motion anomaly causes rainfall changes. * Yong Zhao [email protected]

1

Collaborative Innovation Center on Forecast and Evaluation of Meteorological Disasters/Key Laboratory of Meteorological Disaster, Ministry of Education, Nanjing University of Information Science & Technology (NUIST), Nanjing 210044, China

2

School of Atmospheric Sciences, Nanjing University, Nanjing 210032, China

3

School of Atmospheric Sciences, Chengdu University of Information Technology, No. 24 First Stage of Xuefu Road, Chengdu, Sichuan 610225, China

4

Research Academy of Environmental Planning & Design, Co. Ltd, Nanjing University, Nanjing 210093, China

Northwestern China and western Mongolia (NWC-WM) is located in the middle latitude and to the north of the Tibetan Plateau (TP). The mechanical and thermal effects of the TP are generally responsible for the typically dry climate over NWCWM (He et al. 1987; Manabe and Broccoli 1990; Duan and Wu 2005; Liu et al. 2007). In summer, the rainfall over NWCWM is mainly associated with the eastward-moving disturbances along the westerly jet stream (Barlow et al. 2005; Schiemann et al. 2008, 2009). Rainy season firstly starts over western Xinjiang in mid-May and ends over central Inner Mongolia in late August, lasting about 40–60 days, with most of the rainfall occurring in July and providing about 30–60% of the total annual rainfall (Ding and Wang 2008). However, the annual rainfall amount is small over northwestern China (Zhou et al. 2015) and western Mongolia. Drought is one of the most catastrophic climate events that affect the lives of local residents (Yang et al. 2011; Jiang et al. 2013). Therefore, July rainfall is an important water supply for

Y. Zhou et al.

NWC-WM, and the causes for its interannual changes need to be explored. Previous studies have pointed out that the TP can significantly influence the summer rainfall over Asia (Yanai and Wu 2006; Song et al. 2010; Wang et al. 2014; Hu and Duan 2015). The mechanical and thermal effects of the TP strongly impact the transition of the Indian monsoon rainfall through influencing tropospheric circulation (Sato and Kimura 2007). The heating of the TP can modulate about 30% of the total monsoon rainfall over India during the early and late summer monsoon season (Rajagopalan and Molnar 2013). In summer, the sensible heat on the TP can amplify the intensity of the precipitation over southern China (Li et al. 2014) and affect the precipitation in the northern boundary zone of the East Asian monsoon (Wang and Li 2011). Zhang et al. (2012) have found that latent heat is one of the principal factors affecting the TP vortex, which can cause the heavy rainfall over eastern China. The spring and summer snow cover on the TP, which is an important factor that affects the heat source over the TP, is associated with the summer monsoon rainfall over eastern China (Wu and Qian 2003; Qian et al. 2003; Si and Ding 2013; Liu et al. 2014; Zhu et al. 2015). Furthermore, the decrease trend of the TP heat source can induce the weakening of the East Asian summer monsoon (Duan and Wu 2009; Xu et al. 2013). A comprehensive overview of the effects of the TP on Asian climate by Yanai and Wu (2006) had shown that the mechanical effects of the TP include splitting winter westerly, forcing out atmospheric waves, and affecting the formation of mid-latitude mean troughs and ridges. The TP is also an important heat source to the atmosphere, and the sensible and latent heats over the TP are the major heat contributors to the atmosphere in spring and summer, respectively (Yanai and Tomita 1998; Wu et al. 2007). TP heating produces a shallow cyclonic circulation over the TP surface, and deep anticyclonic circulation at the upper levels above the TP (Liu et al. 2001, 2012; Yanai and Wu 2006; Song et al. 2010; Pan et al. 2013). Therefore, ascent occurs over the TP, and descent occurs over the regions surrounding the TP (He et al. 1987). The heating on the TP can affect the onset of the Asia summer monsoon and modulate the South Asian high (He et al. 1987; Yanai and Wu 2006; Park et al. 2012; Liu et al. 2013). TP heating can be influenced by the vegetation variation, snow depth/cover, and precipitation over the TP (Zhang et al. 2012; Duan et al. 2013; Chen et al. 2015). The South Asian high is an important atmospheric system that influences the rainfall over Asia in summer (Ren et al. 2015; Wei et al. 2014, 2015; Choi et al. 2016). Overall, the TP is an important topography and heat source. The TP can significantly influence summer rainfall and modulate atmospheric circulation over India and eastern Asia. However, less attention has been paid on the effects of the TP on the rainfall over NWC-WM, though previous studies

have found that the dry climate over those regions is associated with the influences of the TP (He et al. 1987; Manabe and Broccoli 1990; Duan and Wu 2005; Liu et al. 2007). The effects of the TP on the annual rainfall changes over NWCWM in mid-summer are still not clear. Therefore, in the present study, the role played by the TP ground surface temperature in affecting the interannual variation of the July rainfall over NWC-WM is examined by analyzing both observations and the results from numerical model sensitivity experiments. The structure of this paper is as follows. Section 2 introduces data. Section 3 provides the relationship between the surface temperature on the TP and the July rainfall over NWC-WM in the observations, as well as the relevant atmospheric circulation. To further support the findings in Section 3, numerical model sensitivity experiments are conducted, and the model experiments and their results are presented in Section 4. Section 5 provides a discussion of the model results. Finally, a summary is presented in Section 6.

2 Data The data during 1980–2012 used in the present study is as follows. The 6-hourly ERA-Interim reanalysis include geopotential height, air temperature, specific and relative humidity, horizontal/vertical winds, surface pressure, and skin temperature. Skin temperature is used as the ground surface temperature (GTS). These reanalyses are with the horizontal resolution of 2.5° × 2.5°. Except that surface pressure and skin temperature are at the ground surface, all the reanalysis are at 17 pressure levels from 1000 to 10 hPa. The ERA-Interim reanalysis are obtained from the European Centre for Medium-Range Forecasts (http://www.ecmwf.int). The monthly precipitation during 1980–2012 is with the horizontal resolution of 2.5° × 2.5° for the Global Precipitation Climatology Project (GPCP) v2.2 combined precipitation. The GPCP precipitation is provided by the US Earth System Research Laboratory/Physical Sciences Division (ESRL/PSD, http://www.esrl.noaa.gov/psd). The US National Oceanic and Atmospheric Administration (NOAA) optimum interpolation sea surface temperature (OISST) v2 at monthly intervals with the horizontal resolution of 1° × 1° during 1982–2011 is also obtained from the ESRL/ PSD and used to drive the regional climate model. The 6hourly ERA-Interim reanalysis are averaged during July for analysis. Global 30 arc-second elevation dataset (GTOPO30) is obtained from the U.S. Geological Survey (https://eros. usgs.gov/elevation-products). In addition to the ERA-Interim reanalysis and GPCP precipitation, the daily precipitation and ground surface temperature (GTS) of the 824 weather stations in China during 1980– 2012 are supplied by the China Meteorology Administration. Before using the station data, a simple quality control is

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over...

conducted. When the number of the days with the missing data of either precipitation or GTS is greater than four during July in any year of 1980–2012 in a station, this station is abandoned. Therefore, 777 stations are left for analysis (Fig. 1a), and the stations with the elevation above 2.5 km over the TP are shown in Fig. 1b. After that, with the missing records removed, daily precipitation and GTS are averaged during July. In Fig. 1b, the geographic names used in this study are also presented, and the northwestern Chinese provinces of Xinjiang, Gansu, and Inner Mongolia are denoted as XJ, GS, and IM, respectively.

3 Relationship between GTS over the TP and rainfall over NWC-WM 3.1 Relationship between GTS and rainfall Singular value decomposition (SVD, Bretherton et al. 1992) is used to explore the interannual relationship between the GTS on the TP and the rainfall over NWC-WM in July during 1980–2012. Before performing the SVD, the trend and mean of the time series of a variable (e.g., GTS and rainfall) in July during 1980–2012 are removed. The time series is standardized through being divided by its standard deviation. For the weather station data, SVD is carried out to firstly obtain the expansion coefficients. The expansion coefficients of GTS (rainfall) are correlated to the rainfall (GTS) that is interpolated to 1° × 1° horizontal grids to get the left (right) heterogeneous correlation maps. The Cressman (1959) objective analysis is used as the interpolation method for rainfall. The inverse-distance algorithm, in which the topography (GTOPO30) is considered, is used to interpolate GTS (Willmott and Matsuura 1995; Pan et al. 2004). Moreover, GTS data is used only from the stations shown in Fig. 1b, which are the stations with the elevation above 2.5 km over the TP. For ERA-Interim GTS and GPCP precipitation, the SVD is also performed with the same procedure to obtained left and right heterogeneous correlation maps but without interpolation. Figure 2 presents the heterogeneous correlation maps of the first SVD patterns between the station rainfall and GTS (Fig. 2a, b) and between the GPCP precipitation and ERAInterim GTS (Fig. 2c, d) in July during 1980–2012. In Fig. 2, shaded areas show the regions where the SVD is performed, and a marker on a grid indicates that the correlation coefficient is significant at the 10% level according to Student’s t test. The percentages of the covariance that is explained by the first SVD patterns are about 41% and 46% for station data (Fig. 2a, b) and reanalysis (Fig. 2c, d), respectively. The correlation coefficients between the expansion coefficients of rainfall and GTS are 0.79 and 0.74 for the station data and reanalysis, respectively. The first SVD patterns generally represent the relationship between the GTS over the TP and

rainfall over northwestern China (Fig. 2), as well as the rainfall over western Mongolia and the southern TP. In addition, the second SVD patterns explain about 25% and 17% of the covariance for the station data and reanalysis, respectively. The second SVD patterns only reflect the relationship between the GTS and rainfall over the TP (figure not shown). Therefore, only the first SVD patterns are shown and analyzed. Now, we examine the first SVD patterns. Figure 2 shows the heterogeneous correlation maps for rainfall (left heterogeneous) and GTS (right heterogeneous). For both the station data and reanalysis, the GTS on the TP (Fig. 2b, d) significantly and negatively correlates to the rainfall (Fig. 2a, c) over northern and eastern XJ, northern GS, and western IM. The magnitudes of the correlation coefficients are about 0.3–0.6. In addition to the areas mentioned above, station rainfall also significantly correlates to the GTS over southern TP and southwestern XJ (Fig. 2a). The right heterogeneous correlation map for the GTS (Fig. 2b) has the largest correlation coefficient of about 0.6 at about 98° E and 35° N. Large correlations (≥ 0.4) appear generally over central and eastern TP. Figure 2c and d shows the heterogeneous correlation maps for the reanalysis. In Fig. 2c, besides of northern and eastern XJ, northern GS, and western IM, significant correlations between the rainfall and GTS on the TP are found to the west of the Lake Baikal, over western Mongolia, on the south slope of the TP, and over the southeastern TP. Large negative correlation coefficients (≤ − 0.4) are generally along the border between China and Mongolia. Over the southern part of the central TP, the correlation is not significant, which is different from the pattern obtained by using the station data. The right heterogeneous correlation maps for the GTS (Fig. 2d) are generally with the large correlation (≥ 0.4) over the northern TP and the north slope of the TP, and also exhibit some differences compared to the pattern obtained by using the station data. Overall, the relationship between the GTS on the TP and the rainfall over NWC-WM is significantly negative, except for southern XJ. This means that except for southern XJ, when the GTS anomalies on the TP are positive (negative), the rainfall over NWC-WM decreases (increases). In the present study, we mainly focus on the regions over northern XJ and GS, and western IM and Mongolia. Although there are some inconsistencies between the SVD patterns of the station data and reanalysis over the TP, the first SVD patterns between the station data and reanalysis are generally similar. The correlation coefficient of the expansion coefficients of the SVD for the GTS on the TP between the station data and reanalysis is 0.83. The inconsistency between SVD patterns in the station data and reanalysis is generally caused by the differences in the SVD regions and horizontal resolution. According to the SVD analysis between the ERAInterim GTS and GPCP rainfall over the regions that are generally the same as those of the station data (figure not shown), the first SVD patterns of the rainfall and GTS over the TP

Y. Zhou et al. Fig. 1 a Locations of the 777 weather stations (red markers) over China, and topography in the model area (shaded, in meters). The blue contour line shows the elevation of 3.0 km, indicating the location of the TP. b The key regions used in the sensitivity experiments (Section 4). Area surrounded by red (green) line is the key region for the experiment of TPN (TPC). XJ, GS, and IM denote the provinces of Xinjiang, Gansu, and Inner Mongolia in northwestern China, respectively. Orange markers are the locations of the weather stations with the elevation above 2.5 km over the TP

generally agree with those shown in Fig. 2a and b. In addition, the SVD analyses between the station GTS and GPCP precipitation and between the ERA-Interim GTS and station rainfall are also performed (figures not shown). The SVD patterns obtained are also consistent with those shown in Fig. 2. The reanalyses are adopted because of the lack of weather station over the western TP (Fig. 1b) and can present the relationship between the GTS on the TP and the rainfall outside of China (Fig. 2c). Previous studies (Wang and Zeng 2012; Song et al. 2016) have compared various reanalyses with the station observations over the TP and point out that the ERA-Interim can generally well represent the interannual variation of the GTS over the TP. Thus, we are not going to further explore the differences between the reanalysis and station data over the TP, which is out of the scope of this study. In general, both the

station data and reanalysis indicate a significantly negative relationship between the GTS on the TP and rainfall over northwestern China. Figure 3a and b shows the time series of the station GTS over the TP, and the rainfall over northwestern China for the station data and over NWC-WM for GPCP data in July during 1980–2012. In Fig. 3b, the time series have been standardized but not in Fig. 3a. Figure 2a and c has shown the regions for averaging, where the correlation between the station/GPCP rainfall and GTS is generally significant. GTS data on the TP is used only from the stations shown in Fig. 1b. The average rainfall over the two black boxes (one black box) shown in Fig. 2a (Fig. 2c) is about 0.91 (1.21) mm day−1. The mean of those time series has been removed in Fig. 3a, and the trend and mean have been removed in Fig. 3b. Averaged station

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over...

Fig. 2 Heterogeneous correlation coefficient maps of the first SVD patterns between the rainfall and the GTS on the TP in July during 1980–2012. (a) and (c) [(b) and (d)] are the left (right) heterogeneous maps for the rainfall (GTS). (a) and (b) are derived from the weather station data. (c) and (d) are derived from the GPCP precipitation and

ERA-Interim GTS. A marker on a grid indicates that correlation coefficient is significant at the 10% level according to Student’s t test. The station (GPCP) rainfall averaged in the two black boxes (the one black box) in (a) (c) is used in Fig. 3

GTS data is used but not the expansion coefficients of the first SVD pattern for the GTS because the correlation between the averaged station GTS and expansion coefficients is 0.96 and 0.87 for the station data and reanalysis, respectively. Furthermore, using the averaged station GTS on the TP enables direct comparison rainfall anomalies with GTS anomalies (Fig. 3c). Figure 3c is a scatter plot between the averaged station GTS on the TP and the rainfall over NWC-WM, which are the same time series shown in Fig. 3b but without being standardized. The correlation coefficient between the GTS on the TP and the rainfall of station (GPCP) data in July during 1980–2012 is − 0.46 (− 0.49), which is significant at the 10% level. In addition, the years of 1981, 1986, 1988, 1991,

1994, 2000, 2001, 2006, and 2010 have the positive GTS anomalies that are greater than the standard deviation of 0.5. The years of 1983, 1984, 1985, 1992, 1993, 1995, 1998, 2003, 2004, 2007, and 2012 have the negative GTS anomalies that are smaller than the standard deviation of − 0.5 (Fig. 3b). Figure 3c shows that when the GTS anomalies on the TP are below the − 0.5 standard deviation, rainfall anomalies are generally positive over NWC-WM. For the station and GPCP rainfall, the exception years with negative/zero rainfall anomalies are 1983, 1985, and 2004. When the GTS anomalies on the TP are above 0.5 standard deviation, rainfall anomalies are generally negative over NWC-WM. For the station (GPCP) rainfall, the exception years with positive/zero rainfall anomalies are 1988 and 1994 (1994 and 2006). Generally,

Y. Zhou et al.

the years with negative (positive) GTS anomalies, the rainfall anomalies are significant over northern XJ and part of western Mongolia at about 46° N (northern GS at about 42° N) in both the station and GPCP rainfall. This indicates that the relationship between the anomalous GTS on the TP and rainfall over NWC-WM is not symmetrical in terms of the spatial distributions of rainfall anomalies. Besides of the anomalous rainfall over NWC-WM, rainfall anomalies are also significant over the eastern (southeastern) TP during the years with negative (positive) GTS anomalies for both the station and GPCP rainfall. 3.2 Atmospheric circulation This subsection presents the atmospheric circulation associated with the relationship between the GTS on the TP and rainfall over NWC-WM. Firstly, the mean geopotential height at 150 and 500 hPa and column moisture flux in July during 1980–2012 are shown in Fig. 5a. Column moisture flux ps

*

(g−1 ∫10 q V dp, Trenberth 1991) is the integration of the spe*

Fig. 3 a Time series of the anomalous station GTS averaged on the TP (black solid line), station rainfall averaged over northwestern China (STA, blue dash line), and GPCP rainfall averaged over NWC-WM (red dash line) in July during 1980–2012. The areas for averaging are shown in Fig. 2a and c. b The standardized time series of the GTS and rainfall with their trend being removed. The trend and mean of the three time series are removed, and then divided by their standard deviations, respectively. Thus, y-axis shows the standardized value, and the black dash lines are the reference lines of the 0.0 and ± 0.5 standard deviations of GTS. c Scatter plot between the GTS averaged on the TP (x-axis, K) and the station (blue circle) and GPCP (red dot) rainfall (y-axis, mm day−1) in July during 1980–2012. The vertical dash lines show the ± 0.5 and ± 1.0 standard deviations of GTS, and the horizontal blue (red) dash lines show the ± 0.5 standard deviations of the station (GPCP) rainfall

when the GTS anomalies on the TP are positive (negative), the rainfall anomalies are negative (positive) over NWC-WM, which is consistent with the results derived from the SVD analysis. Figure 4a and c (4b, d) shows the anomalous rainfall composites during the years with the GTS anomalies on the TP that are smaller (greater) than − 0.5 (0.5) standard deviation. Before conducting the composite, the trend and mean of the rainfall at each grid during 1980–2012 are removed. In Fig. 4, the top (bottom) panels are for the station (GPCP) rainfall. The left (right) panels are for the negative (positive) GTS anomalies on the TP. A marker on a grid indicates the anomalous rainfall that is significantly different from the mean rainfall in July during 1980–2012 at the 10% level according to Student’s t test. The patterns of composite anomalies are generally consistent with the SVD analysis for both station and GPCP rainfall. Moreover, it is interesting to find that during

cific humidity (q) multiplied by horizontal winds (V ) from the ground surface (ps, surface pressure) to the top of the atmosphere (10 hPa), and then divided by the acceleration of gravity (g = 9.8 m s−2). Typically, the South Asian high has a center around 30° N at 150 hPa (Wu et al. 2015, 2016; Zhang et al. 2016; Yan et al. 2016; Liu et al. 2017; shaded in Fig. 5a), and the northwestern Pacific subtropical high has a center around 25° N at 500 hPa. Moisture is transported from India and the Bay of Bengal to the TP and from the Bay of Bengal and South China Sea to eastern China by the Asia summer monsoon. At the middle latitudes, moisture is transported eastward from the Atlantic Ocean to NWC-WM. Over the TP, the moisture from northwestern China and the Bay of Bengal meets together, and then a convergence of column moisture flux is found. Furthermore, the vertical circulation (Fig. 5b, c) between 80° and 100° E shows that ascent is to the south of and over the TP. After removing the zonal mean, there are warm centers over the TP at about 300 hPa, positive geopotential height center over the TP at about 150 hPa, and negative geopotential height below 400 hPa surrounding the TP. To the north of the TP, there is a vertical circulation with an ascent over the north slope of the TP (Fig. 5b, c) and a descent between 40 and 50° N (Fig. 5c). These findings are consistent with those found by Liu et al. (2001) and Wu (2004). The descent between 40° and 50° N to the north of the TP is the reason for the dry climate over NWC-WM (He et al. 1987; Manabe and Broccoli 1990; Duan and Wu 2005). However, the mean state is not sufficient to explain the interannual variation of the rainfall over NWC-WM. Therefore, Fig. 6a and c (b, d) shows the composite maps of the anomalous circulation according to the years with the GTS anomalies on the TP that are smaller (greater) than − 0.5

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over...

Fig. 4 Composites of the rainfall anomalies (mm day−1) during the years with the negative (a, c) and positive (b, d) GTS anomalies on the TP for station (a, b) and GPCP (c, d) rainfall. A marker on a grid indicates the composite is significant at the 10% level according to Student’s t test. It is

noted that the interval of the shading is 0.1 mm day−1 from − 0.5 to 0.5 mm day−1, and 0.5 mm day−1 from − 3 to − 0.5 mm day−1 and 0.5 to 3 mm day−1, because the rainfall magnitudes over NWC-WM are smaller than elsewhere

(0.5) standard deviation. The years with anomalous negative/ positive GTS have been provided in the preceding subsection and are derived from Fig. 3. Figure 6 shows the anomalous geopotential height at 150 (shaded) and 500 hPa (contour), column moisture flux (vector; top panels), and the conver  ps * gence of column moisture flux [−∇⋅ g−1 ∫10 q V dp ,

and over the southern TP (northwestern China around 42° N) in Fig. 6c. The anomalies of the moisture convergence over southern TP and western Mongolia are significant at the 10% level. During the years with positive GTS anomalies (Fig. 6b, d), the patterns of the anomalous geopotential height, moisture flux, and moisture flux convergence are generally opposite with respect to those during the years with negative GTS anomalies. In general, the increase (decrease) of the rainfall is consistent with the moisture convergence (divergence) anomalies over western Mongolia and the southern TP during the years with the negative (positive) GTS anomalies on the TP. However, over northwestern China at 42° N, the divergence (convergence) of column moisture flux anomalies is inconsistent with the increase (decrease) of the rainfall during the years with negative (positive) GTS anomalies. Anomalous meridional circulation, air temperature, geopotential height, vertical velocity, and specific humidity are also provided for the years with negative and positive

Trenberth 1991; bottom panels] for the years with the negative (left panels) and positive (right panels) GTS anomalies on the TP. During the years with negative GTS anomalies, there are anomalous negative geopotential height centers at both the 150 and 500 hPa over northern GS in Fig. 6a. A cyclonic anomalous moisture flux is found over northwestern China and transports moisture from the Lake Baikal through northwestern China to the TP (Fig. 6a). Moreover, moisture is anomalously transported from the Bay of Bengal to the TP (Fig. 6a). Anomalous column moisture flux convergence (divergence) is found over western Mongolia around 50° N

Y. Zhou et al.

Fig. 5 a Mean geopotential height at 150 hPa (shaded, m) and 500 hPa (contours with the interval of 20 m) in July during 1980–2012. The contour lines of the 500-hPa geopotential height that are smaller than 5820 m are not shown. Vectors present the mean atmospheric moisture flux (kg m−1 s−1) in July during 1980–2012. b Vertical section of the mean air temperature (shaded, K), geopotential height (contours with

the interval of 20 m), vertical velocity, and meridional circulation (vector) between 80° and 100° E in July during 1980–2012. For the air temperature and geopotential height, the zonal mean is removed but not for the meridional and vertical winds. Vertical wind (Pa s−1) is multiplied by − 25. c Vertical section of the vertical velocity (in 10−2 Pa s−1) between 80° and 100° E in July during 1980–2012

GTS anomalies on the TP in Fig. 7. Figure 7a (Fig. 7b) shows that during the years with negative (positive) GTS anomalies, negative (positive) geopotential height anomalies are found over 42° N with its center between 200 and 150 hPa, and an anomalously cold (warm) air temperature center is also over 42° N at about 300 hPa; anomalously upward (downward) motion is over the TP and the region at about 50° N but anomalously downward (upward) motion is over the northern TP and the north slope of the TP. Figure 7c and d shows that the anomalous vertical motions over the TP and the region at 50° N are significant at the 10% level. During the years with negative (positive) GTS anomalies in Fig. 7e (Fig. 7f), specific

humidity anomalies decrease (increase) at 500 hPa around 40° N but increase (decrease) at the lower levels between 40° and 50° N. The increase (decrease) of the specific humidity at lower levels and anomalous ascent (descent) between 40° and 50° N agree with the increase (decrease) of the rainfall over NWC-WM during the years with negative (positive) GTS anomalies. Over the southern TP, the anomalous vertical motion is also consistent with the rainfall anomalies. Generally, when GTS is colder (warmer) than normal over the TP in July, there is a cold (warm) center of air temperature anomalies at about 300 hPa and negative (positive) geopotential height anomalies with centers between 200 and

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over...

Fig. 6 Top panels (a, b) are the composites of the anomalous geopotential height at 150 hPa (shaded, m) and 500 hPa (contours with the interval of 5 m) and moisture flux (kg m−1 s−1) for the years with the negative (a) and positive (b) GTS anomalies on the TP. Vectors with color and shaded areas are significant at the 10% level according to Student’s t

test. Bottom panels (c, d) are the composites of column moisture flux convergence (in 10−5 kg s−1) for the years with the negative (c) and positive (d) GTS anomalies on the TP. A marker on a grid indicates the composite is significant at the 10% level according to Student’s t test

150 hPa over northwestern China at about 42° N. An increase (decrease) in anomalous specific humidity at lower levels and ascent (decent) are found between 40° and 50° N. These findings are consistent with the positive (negative) rainfall anomalies over NWC-WM between 40° and 50° N. However, the significant relationship in the SVD patterns between GTS and rainfall does not mean an affirmative causality, and they may be coincidentally covariant and affected by other factors. Moreover, it is also difficult to determine the effect of the GTS on the TP on the rainfall changes over NWC-WM in the composites of anomalous circulation. Thus, in the next section, numerical sensitivity experiments are carried out by using RegCM4.1 to further explore the role played by the GTS over the TP in the rainfall changes over NWC-WM.

4 Numerical experiment and results 4.1 Model experiments RegCM4.1 (http://gforge.ictp.it) is used to conduct sensitivity experiments in the present study. The control (CON) and sensitivity experiments during 1982–2011 are performed without and with the anomalous GTS in the key regions on the TP, respectively, because GTS is an important variable in the surface energy balance. The key regions are shown in Fig. 1b. In Fig. 1b, the TPC (TPN) denotes the set of experiments with the key region covering the central and eastern TP except for the western TP (part of the northern TP and the north slope of the TP). The use of the key regions of TPC and TPN is firstly based on the SVD analysis (Fig. 2). In Fig. 2b (2d), it is found

Y. Zhou et al.

Fig. 7 Top panels (a, b) are the composites of the geopotential height (contours with the interval of 10 m), air temperature (shaded, K), and meridional circulation (vector with the vertical velocity multiplied by − 100) for the years with the negative (a) and positive (b) GTS anomalies over the TP. Shaded areas of air temperature are significant at the 10% level. The middle panels (c, d) are the composites of vertical

velocity (in 10−2 Pa s−1) for the years with the negative (c) and positive (d) GTS anomalies. Those anomalies are averaged between 80° and 100° E. The markers indicate the significance of the 10% level for the composites according to Student’s t test. The bottom panels (e, f) are the same as middle panels but for specific humidity (in 10−4 kg kg−1)

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over...

that the area with large correlation coefficients is over central and eastern TP (north and the north slope of the TP). Secondly, about 15 groups of extra sensitivity experiments have been conducted with various key regions over the TP (figures not shown). Among all those experiments, TPC and TPN are more typical than others. TPC and TPN do not include the western TP because of the lack of weather stations over western TP. Moreover, the correlation coefficients over the western TP are opposite to those over the eastern TP in Fig. 2b and are not significant in Fig. 2d. In the present study, the land model of the biosphere–atmosphere transfer scheme (BATS, Dickinson et al. 1993) is used in RegCM4.1. The tendency equation of the ground surface temperature in BATS is force restore rate equation (Eq. 1, Blackadar 1976; Deardorff 1978). ∂T s HA T s −T 2 ¼ −c1 −c2 ∂t ρ s cs d 1 τ1

ð1Þ

in which Ts and HA are ground surface temperature (GTS) and the sum of heat fluxes to atmosphere, respectively. ρs and cs are the density and specific heat of soil, respectively. d1 is a soil depth influenced by the diurnal temperature cycle. T2 is the mean soil temperature over the layer d2, which is a soil depth influenced by the annual temperature cycle. τ1 is diurnal period. c1 and c2 are constants and equal to 2π1/2 and 2π, respectively. In the set of the experiment of TPC (TPN), the forcing of 2 × 10−4, 4 × 10−4, and 8 × 10−4 K s−1 is added on the right side of Eq. (1) in the key region at each time step in the model run of July in the three subset experiments, which are denoted as TPC + 2, TPC + 4, and TPC + 8 (TPN + 2, TPN + 4, and TPN + 8), respectively. After adding the forcing, ground surface heating (HA in Eq. 1) in the model is adjusted by the GTS, and GTS is then adjusted by the surface processes during the simulation. Anomalous GTS is the difference of GTS between the sensitivity experiments and CON. In addition, another three subsets of experiments are conducted with the same negative forcing added on the right side of Eq. (1) in the key region of TPC (TPN) and denoted as TPC − 2, TPC − 4, and TPC − 8 (TPN − 2, TPN − 4, and TPN − 8). In the experiment of each subset, 30 runs are performed corresponding to the years of 1982–2011, and the model is integrated from June 1st to August 1st in each year. GTS forcing is only added in July in each run of the sensitivity experiments, and June is taken as spin-up time. The RegCM4.1 is configured with the Lambert projection with the reference parallels at 30° and 60° N (Fig. 1a). The model horizontal and vertical resolutions are 60 × 60 km and 18 sigma levels, respectively. The integral time step in the model is 60 s. The land model used is the BATS. Moreover, the planetary boundary layer scheme of Holtslag et al. (1990), the lateral boundary scheme of linear relaxation, and the

cumulus convection scheme of Grell et al. (1994) are adopted. The RegCM4.1 is driven by the 6-hourly ERA-Interim reanalysis and monthly NOAA OISST during June 1st to August 1st of 1982–2011. The model results at 18 sigma levels are saved at daily interval and interpolated to 17 pressure levels for analysis. The performance of the model to simulate rainfall, circulation, and surface heat flux is shown in Fig. 8. Figure 8 is obtained by using the data of CON and ERA-Interim reanalysis during July. All the pattern statistics are calculated over the area of 12.5° N–52.5° N and 70° E–110° E, which includes the areas of TP and NWC-WM. The model results are interpolated to 2.5° × 2.5° grids. Figure 8 shows that the pattern correlations for the rainfall, meridional wind, and surface sensible and latent heats are between 0.7 and 0.8. The correlations for geopotential height, zonal wind, upward long wave radiation, and GTS are greater than 0.9. Except for rainfall, the standard deviations of the variables of CON are 0.75– 1.5 times of those of the observation. The standard deviation of the rainfall of CON is about 1.6 times of that of the observation. This is because the resolution of the RegCM4.1 is finer than that of the observations. Thus, the model can well describe topography, which causes the large variation of the rainfall. Generally, the RegCM4.1 can well simulate the rainfall, circulation, and surface heat flux, and can be adopted in the present study. 4.2 Rainfall anomalies In the model, the anomalous rainfall in July is obtained by composing the differences between the sensitivity experiments and CON. For example, the rainfall of CON is subtracted from the rainfall of a subset sensitivity experiment to get the anomalous rainfall. The rainfall differences with a samples size of 30 (1982–2011) are obtained. It is noted that both sensitivity and control experiments share the same boundary conditions. After that, the average of the 30-year differences is computed. A significant test with the level of 10% is conducted by using Student’s t test with the null hypothesis that the average of differences is zero. This procedure is carried out for all the sensitivity experiments, including the TPC ± 2, TPC ± 4, TPC ± 8, TPN ± 2, TPN ± 4, and TPN ± 8. The anomalous rainfall for the experiment of TPC is presented in Fig. 9. The left (right) panels of Fig. 9 show the anomalous rainfall with negative (positive) GTS anomalies in the region TPC shown in Fig. 1b. The top, middle, and bottom panels in Fig. 9 are for TPC ± 2, TPC ± 4, and TPC ± 8, respectively. In each panel of Fig. 9, the box over the areas of northeastern XJ, northern GS, western IM, and Mongolia covers the same region as that shown in Fig. 2c, where a significant relationship between rainfall and GTS is found in the observation. In Fig. 9, when the GTS forcing on the TP in the model is negative (positive) with any magnitude,

Y. Zhou et al.

Fig. 8 Taylor diagram for pattern statistics of rainfall, geopotential height at 150 and 500 hPa (GPH150 and GPH500), zonal wind at 10 m and 150 hPa (U10 and U150), meridional wind at 10 m and 150 hPa (V10 and V150), surface sensible heat flux (SE), surface latent heat flux (LH), surface upward long wave radiation (ULR), and ground surface temperature (GTS). The pattern statistics are calculated between the model results of CON and ERA-Interim reanalysis. The radial distance from the origin to a red point is proportional to the standard deviation of observation. The centered root mean square difference between the model results (red points) and observation (REF) is proportional to their distance apart. The correlation between the model results and observation is presented by the azimuthal position of the red points

the rainfall anomaly is generally positive (negative) over NWC-WM, especially at 45° N (42° N). This is generally consistent with the composite maps of the observed rainfall anomalies shown in Fig. 4c and d. Furthermore, as the negative (positive) GTS forcing on the TP decreases (increases), anomalous rainfall increases (decreases) over NWC-WM in Fig. 9a, c, and e (Fig. 9b, d, f). These results are also consistent with the relationship between GTS and rainfall, which is found in the observations in Section 3. Besides of the rainfall changes in the box shown in Fig. 9, a positive (negative) rainfall anomaly is also found over southern XJ and western TP (central and eastern TP), when the GTS forcing is negative (positive) in the model. These results do not agree with those found in the observation in Section 3. Figure 10 is the same as Fig. 9 but for the experiment of TPN. In Fig. 9, the patterns of the rainfall anomalies over NWC-WM (black box) are almost the same as those of the experiment of TPC. As the negative (positive) GTS forcing being decreased (increased) over region TPN, rainfall increases (decreases) over NWC-WM in Fig. 10a, c, and e (Fig. 10b, d, f). This agrees with those found in Fig. 9 and the observations in Section 3. When GTS forcing is negative (positive) in region TPN, rainfall anomalies are negative and positive (positive and negative) over the northern and southern (southern and northern) TP, respectively. Moreover, the

rainfall anomalies over the southern TP in TPN are generally consistent with the findings in the observation in Section 3. For the experiment of TPC (TPN), the scatter plot of the rainfall anomalies averaged in the black box shown in Fig. 10 against the GTS anomalies over region TPC (TPN) is shown in Fig. 11a (Fig. 11b). In both of the experiments, as the GTS forcing being increased, GTS anomalies increase in the key region and rainfall decreases over NWC-WM. The correlation coefficients between the GTS anomalies in the key region and rainfall anomalies over NWC-WM are − 0.48 and − 0.34 for the experiments of TPC and TPN, respectively. The sample size for each set of experiments is 180, and the correlations between the GTS and rainfall anomalies are significant at the 10% level. The results in Fig. 10 are consistent with those in the observation in Section 3 and in Figs. 9 and 10. In Fig. 11, the averages of the GTS anomalies over region TPC (TPN) are about − 9.89, − 6.07, − 2.35, 0.86, 1.62, and 3.02 K (− 7.11, − 4.08, − 1.84, 0.93, 1.88, and 3.57 K) in the experiments of TPC − 8, TPC − 4, TPC − 2, TPC + 2, TPC + 4, and TPC + 8 (TPN − 8, TPN − 4, TPN − 2, TPN + 2, TPN + 4, and TPN + 8), respectively. For the experiment of TPC (TPN), the magnitudes of the GTS anomalies in TPC − 2, TPC + 2, TPC + 4, and TPC + 8 (TPN − 2, TPN + 2, and TPN + 4) are generally comparable to the magnitudes of the observations. In next subsection, the anomalous atmospheric circulation is compared between TPC − 2 and TPC + 8 (TPN − 2 and TPN + 4) but not between the experiments with the same GTS forcing. This is because TPC − 2 and TPC + 8 (TPN − 2 and TPN + 4) generally have the magnitudes of the GTS anomalies over key regions at the same level. Lastly, with the same magnitude of the GTS forcing on the TP, the experiments with negative GTS forcing tend to cause the stronger magnitude of GTS anomalies than those with positive GTS forcing (Fig. 11). This will be discussed at the end of this section, after analyzing the anomalous circulation, which is responsible for this phenomenon. 4.3 Anomalous atmospheric circulation The anomalous atmospheric circulation in the model is derived by using the same method that is used to obtain the anomalous model rainfall. The anomalous geopotential height at 150 and 500 hPa, moisture flux, and column moisture flux convergence are shown in Fig. 12 for the experiment of TPC. Left panels are for TPC − 2, and right panels are for TPC + 8. When the GTS forcing over region TPC is negative (positive), a negative (positive) anomalous center of 150-hPa geopotential height is over the northern TP, and anticyclonic (cyclonic) moisture flux is over the TP in Fig. 12a (Fig. 12b), which induces the significant divergence (convergence) of the moisture flux over the TP in Fig. 12c (Fig. 12d). Moisture is transported from the TP to the

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over...

Fig. 9 Rainfall anomalies (mm day−1) of the experiments with the negative (a, c, e) and positive (b, d, f) GTS forcing over region TPC shown in Fig. 1b. The top (a, b), middle (c, d), and bottom (e, f) panels correspond to the experiments with the magnitude of the GTS forcing of 2 × 10−4, 4 × 10−4, and 8 × 10−4 K s−1, i.e., TPC ± 2, TPC ± 4, and TPC ± 8, respectively. A marker on a grid indicates the rainfall anomaly is

significant at the 10% level according to Student’s t test. The interval of the shading is 0.1 mm day−1 from − 0.5 to 0.5 mm day−1, and 1 mm day−1 from − 4 to − 1 mm day−1 and 1 to 4 mm day−1, because the rainfall magnitudes over NWC-WM are smaller than elsewhere. The black box in each panel covers the same region as that shown in Fig. 2c

surrounding areas of the TP (from the surrounding areas of the TP to the TP). At 500 hPa in Fig. 12c (Fig. 12d), there are positive and negative (negative and positive) geopotential height anomalies over the TP and XJ, respectively. However, the anomalous patterns shown in Fig. 12 are not consistent with the observation shown in Fig. 6. Figure 13 is the same as Fig. 12 but for the experiment of TPN. The left panels are for TPN − 2, and the right panels are

for TPN + 4. In Fig. 13, the anomalous centers of geopotential height at 150 hPa are further north than those shown in Fig. 12, which is quite similar to the observations (Fig. 6). Moisture is transported from the TP to northern GS (from northern GS to the TP) in TPN − 2 (TPN + 4) in Fig. 13a (Fig. 13b), and the moisture convergence is significantly negative and positive (positive and negative) over the northern TP and the north slope of the TP in Fig. 13c (Fig. 13d),

Y. Zhou et al.

Fig. 10 Same as Fig. 9 but for the experiments of TPN

respectively. Except that there is no anomalous geopotential height center at 500 hPa over the TP, the patterns of the anomalous geopotential height at 500 hPa in Fig. 13c are similar to that shown in Fig. 12c, with the anomalous centers located over XJ in the experiment of TPN − 2. However, the anomalous centers are over northern GS in the observations (Fig. 6a). In Fig. 13d, the anomalous geopotential height at 500 hPa is over the northern slope of the TP, which is close to the location of the anomalous centers in the observations (Fig. 6b). The anomalous meridional geopotential height, air temperature, vertical circulation, vertical velocity, and specific humidity between 80 and 100° E are further examined and

shown in Figs. 14 and 15 for the experiments of TPC and TPN, respectively. In Fig. 14a (Fig. 14b) for TPC − 2 (TPC + 8), there is a negative (positive) anomalous center of air temperature at about 300 hPa. An negative (positive) anomalous center of geopotential height is at 150 hPa, and anomalous geopotential height is positive (negative) below 400 hPa over the TP. Descent (ascent) is over the TP, and ascent (descent) is over the region between 40 and 50° N. The vertical velocity at about 45° N (42° N) is significantly negative (positive), indicating an upward (downward) motion over this region in Fig. 14c (Fig. 14d). Specific humidity significantly decreases and increases (increases and decreases)

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over... Fig. 11 Scatter plots of the anomalous rainfall (y-axis, mm day−1) over the black box shown in Fig. 10 against the GTS anomalies (x-axis, K) over the regions of TPC (a) and TPN (b). Green, blue, and red dots represent the experiments with the magnitude of the GTS forcing of 2 × 10−4, 4 × 10−4, and 8 × 10−4 K s−1, i.e., TPC ± 2, TPC ± 4, and TPC ± 8, respectively. The black line in each panel is the linear fit of the GTS anomalies to rainfall anomalies for the 180 runs in the experiments of TPC/TPN. The correlation coefficients between the rainfall and GTS anomalies are − 0.48 and − 0.34 for the experiments of TPC and TPN, respectively

below 400 hPa over the TP and to the north of the TP in Fig. 13e (Fig. 13f) for TPC − 2 (TPC + 8). The anomalous positive (negative) specific humidity and upward (downward) motion between 40° and 50° N are responsible for the positive (negative) rainfall anomalies over NWC-WM, especially over the regions at about 45° N (42° N). Figure 15 is the same as Fig. 14 but for the experiments of TPN − 2 and TPN + 4. In Fig. 15a (Fig. 14b) for TPN − 2 (TPN + 4), negative (positive) centers of anomalous air temperature and geopotential height are found around 40° N at 300 and 150 hPa, respectively. Near the surface of the north slope of the TP, there are negative (positive) air temperature anomalies, and geopotential height anomalies are positive (negative) but very shallow. Downward and upward (upward and downward) motion occur over the north slope of the TP and between 40° and 50° N, respectively. In addition, Fig. 15c (Fig. 15d) shows that the descent (ascent) over the north slope of the TP is significant, and ascent (descent) is significant over the southern TP and between 40° and 50° N, especially over the regions near 45° N (42° N). In Fig. 15e (Fig. 15f), specific humidity significantly decreases (increases) at the lower levels over the north slope of the TP and increases (decreases) between 44° and 54° N (40° and 50° N). Therefore, the increase (decrease) of the specific humidity at lower levels and upward (downward) motion between 40° and 50° N are the reasons for the positive (negative) rainfall anomalies over NWC-WM.

Additionally, the upward (downward) motion over the southern TP is consistent with the positive (negative) rainfall anomalies over the southern TP. The patterns of anomalous circulations in the experiment of TPN (Figs. 13 and 15) are more similar to the observation (Figs. 6 and 7) than those of the experiment of TPC. Moreover, the rainfall anomalies are generally the same over NWC-WM in both of the experiments of TPC and TPN. Therefore, the rainfall changes over NWC-WM are more sensitive to the GTS anomalies over the northern TP and the north slope of the TP (region TPN) than those over the southern TP. Anomalous surface sensible heat (SE), surface latent heat (LH), the latent heat release related to rainfall (LHR), and surface upward longwave radiation (ULR) are computed over region TPC (TPN) for TPC − 2 and TPC + 8 (TPN − 2 and TPN + 4; Fig. 16). In the experiment of TPC − 2 (TPC + 8), the anomalous SE, LH, LHR, and ULR are − 6.90, − 11.40, − 20.72, and 0.62 (10.97, 8.44, 15.82, and 4.06) W m−2, respectively. All the heating anomalies are significant at the 10% level, except for the ULR of TPC − 2. This indicates that SE, LH, and LHR significantly cause the anomalous circulation over the TP, but the LHR is the dominant heating in the experiment of TPC. Over region TPN, the anomalous SE, LH, LHR, and ULR for TPN − 2 (TPN + 4) are − 3.91, − 4.16, − 6.78, and − 0.51 (5.56, 0.95, 1.13, and 2.14) W m−2, respectively. These heating anomalies are all significant at the 10%

Y. Zhou et al.

Fig. 12 Anomalous geopotential height at 150 hPa (shaded, m) and moisture flux (vector, kg m−1 s−1) for TPC − 2 (a) and TPC + 8 (b). Anomalous geopotential height at 500 hPa (contours with the interval of 1 m) and moisture flux convergence (shaded, in 10−5 kg s−1) for

TPC − 2 (c) and TPC + 8 (d). Vectors shown and shaded areas are significant at the 10% level according to Student’s t test. The geopotential height at 150 hPa and moisture flux convergence share the same color bar but have different units

level, except for the ULR in TPN − 2 and the LH and LHR in TPN + 4. In the experiment of TPN − 2 (TPN + 4), which has negative (positive) GTS forcing, LHR (SE) is the dominant heating that causes the anomalous circulation. Furthermore, through analyzing the rest of the experiments (TPC + 2, TPC ± 4, TPC − 8, TPN + 2, TPN − 4, and TPN ± 8), the same results are obtained (figures not shown). As shown in Fig. 11, the experiments with negative GTS forcing tend to induce stronger magnitudes of GTS anomalies in the key regions than those with positive GTS forcing in both of the experiments of TPC and TPN, while the same magnitude of the GTS forcing is used. This can be explained by the anomalous circulation shown in Figs. 14 and 15. For the experiments with negative GTS forcing, the anomalous negative GTS anomalies over the TP cause downward motion over the TP. Thus, less condensation heating associated with precipitation is over the TP (Fig. 9), which decreases the GTS on the TP. After that, negative GTS anomalies are enhanced over the TP and result in the reinforcement of the downward motion over the TP. This positive feedback between the GTS and downward motion over the TP causes the large magnitude of the GTS anomalies on the TP. On the contrary, for the circumstance with positive GTS forcing on the TP, the heat

on the TP surface is brought upward by the ascent over the TP, and there is no such positive feedback between the GTS and upward motion as that in the circumstance with negative GTS forcing. Therefore, the magnitudes of GTS anomalies over the TP are stronger in the experiments with negative GTS forcing than those with positive GTS forcing. Overall, the rainfall changes over NWC-WM are more sensitive to the GTS anomalies over the region of TPN than those over the southern TP. The negative GTS anomalies over region TPN mainly decrease the latent heat related to rainfall locally, and cause descent and ascent over region TPN and between 40° and 50° N, respectively. Moreover, specific humidity increases at lower levels between 44° and 55° N. Therefore, anomalously upward motion and the increase of specific humidity induce the positive rainfall anomalies over NWC-WM, especially over the region at 45° N. When positive GTS forcing is added in region TPN in the model, sensible heat is the dominant heating that causes the upward motion over region TPN but downward motion between 40° and 50° N. Moisture is pumped up from northwestern China to the TP, and specific humidity decreases at lower levels between 40° and 50° N. Therefore, anomalously downward motion and the decrease of the humidity cause the negative rainfall anomalies

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over...

Fig. 13 Same as Fig. 11 but for TPN − 2 (left panels) and TPN + 4 (right panels), and the contour intervals in the bottom panels are 0.5 m

over NWC-WM, especially over the region at 42° N. Those findings derived from the sensitivity experiments are generally consistent with those derived from observation in Section 3.

5 Discussion A comparison among the results of the experiments of TPC and TPN and those of the observations shows that the anomalous circulation of the experiment of TPN is generally consistent with that of the observations but not with the experiment of TPC. Over the southern TP, the vertical motion in the experiment of TPN is generally similar to that of the observations but is weaker. This indicates that in the observations, the rainfall changes over the southern TP are partly influenced by the GTS anomalies over the region TPN, and there should be other factors that have strong effects on the rainfall changes over the southern TP. Meanwhile, in reality, those factors may suppress the effects of the GTS anomalies over southern TP because it can be found that the vertical motion in the experiment TPC is opposite to that in the observations over the southern TP. Except for the western TP, the experiment of TPC can generally represent the effects of the GTS anomalies over the central and eastern TP, and that of TPN can represent the effects of GTS anomalies over the northern TP and the north slope of the TP. Therefore, the experiment of TPN can

be considered as the experiment of TPC with the effects of the GTS anomalies over the southern TP being suppressed. Moreover, the experiment of TPN also reflects the effects of the GTS anomalies over the north slope of the TP. Comparing the circulation of the experiment of TPN with that of the observations shows that the anomalous center of 500-hPa geopotential height is over northern GS in the observation but over XJ (the northern slope of the TP) in the experiment of TPN − 2 (TPN + 4). In addition, the anomalous vertical circulation between 40° and 50° N is mainly at lower levels over GS at about 42° N in the observation but at middle levels over 40° N in the experiment of TPN. We suppose that this is because the factors that suppress the effects of the GTS anomalies over southern TP can also influence the effects of the GTS anomalies over region TPN. This influence over region TPN is weaker than that over the southern TP but can cause the air temperature anomalies over the TP tilting further north and suppresses the vertical circulation to the lower levels between 40° and 50° N. However, those factors suppressing the effects of the GTS anomalies over the TP are still not clear and cannot be considered in the present sensitivity experiments and require further study. From the anomalous circulation of the observations in Fig. 6, it is supposed that the northwestern Pacific subtropical high (NWPSH) could be one of the factors. During the years with negative GTS anomalies, the NWPSH generally stays at

Y. Zhou et al.

Fig. 14 Anomalous meridional circulation, vertical velocity, and specific humidity between 80° and 100° E for TPC − 2 (a, c, e) and TPC + 8 (b, d, and f). Top panels (a, b) show the anomalous air temperature (shaded, K), geopotential height (contours with the interval of 2 m), and meridional circulation (vector with the vertical velocity multiplied by − 103). Shaded

areas of air temperature are significant at the 5% level. Middle (c, d) and bottom (e, f) panels are for vertical velocity (in 10−3 Pa s−1) and specific humidity (in 10−4 kg kg−1), respectively. The panels at each row share the same color bar. Marker indicates the anomalies are significant at the 10% level according to Student’s t test

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over...

Fig. 15 Same as Fig. 14 but for TPN − 2 (a, c, e) and TPN + 4 (b, d, f)

about 25° N, which could increase southwesterly winds to the south of the TP and moisture flux over the southern TP. Moreover, the upward motion over the TP is increased, especially over the southern TP. Additionally, the moisture transported to northeastern China decreases, and then the

latent heat release decreases locally. This could further decrease the geopotential height at the upper levels over northern China (Fig. 6a). This is because previous studies (Wu and Liu 2003; Liu et al. 2004; Ren et al. 2015) have found that the latent heat release associated with rainfall over eastern China

Y. Zhou et al.

Fig. 16 Surface sensible heat (SE), latent heat (LH), upward long wave radiation (ULR), and heat release associated with the rainfall (LHR) over the key regions for sensitivity experiments of TPC − 2, TPC + 8, TPN − 2, and TPN + 4

is important for the changes of the South Asian high. On the contrary, during the years with positive GTS anomalies, the NWPSH is further north than normal, and the southwesterly winds and moisture flux decrease over southern TP but increase over northeastern China. Thus, rainfall and latent heat release increase over northeastern China, and then cause the increase of the geopotential height over northern China (Fig. 6b). Therefore, the increase (decrease) of the southwesterly wind and moisture flux over southern TP associated with the NWPSH may suppress the effects of the GTS anomalies over the TP on rainfall changes, especially over the southern TP. However, the increase (decrease) of the latent heat release and geopotential height over northeastern China may enhance the effects of the GTS anomalies over the northern and the north slope of the TP. For the effects of the NWPSH, those are just guesses, which need further works to confirm. Besides the NWPSH, previous studies have found that the summer rainfall over the TP is influenced by the signals from the Western Hemisphere though Rossby waves (Liu and Yin 2001; Bothe et al. 2010, 2011; Liu et al. 2015). Through analyzing the composite of the anomalous stream function at 200 hPa in July in the observations, wave trains can be identified over the Northern Hemisphere (figure not shown). However, because regional climate model is used in the present study, it is difficult to tell whether those wave trains affect the circulation over the TP or are caused by the GTS anomalies over the TP.

Generally, the differences between the model results and observations of rainfall over the TP are because that the largescale circulation, which influences the rainfall and circulation over the TP, is not considered in the model experiments. Firstly, the rainfall changes over the TP are only affected by the GTS anomalies in the model. The decrease (increase) in GTS over the TP may be caused by the rainfall increases (decreases) over the TP. The rainfall changes over the TP are caused by large-scale circulation, whose influences on TP rainfall are much stronger than that of the GTS over the TP. Thus, the effects of the GTS on the rainfall changes are counteracted by the effects of large-scale circulation over the TP. Furthermore, the experiments are designed to explore the relationship between the GTS over the TP and rainfall over NWC-WM, but not the relationship between the GTS over the TP and rainfall over the TP. Thus, the large circulation that causes the rainfall changes over the TP is not considered in the model experiments. Secondly, the vertical circulation in the model does not tilt to the north in the model. The anomalous accent (descent) motion mainly happens over the TP in the model, which causes rainfall to increase (decrease) over the TP. In the observations, the vertical circulation caused by GTS anomalies tilts to the north, and the ascent (descent) motion mainly occurs at the upper levels to the north of the TP. This causes the increase (decrease) of the moisture to the north of the TP at the upper levels, but not the rainfall changes over the TP. Thirdly, the vertical circulation that caused by the GTS anomalies in the observation is shallower than that in the model. Therefore, the shallow vertical circulation generally affects the rainfall change over the NWC-WM in the observations. However, the vertical circulation in the model reaches the upper levels of the troposphere, which can affect the rainfall over the TP. Therefore, further studies are needed on the mechanism of the large-scale circulation.

6 Summary A significantly negative relationship is found between the GTS over the TP and the rainfall over NWC-WM through the SVD analysis by using both the station data and reanalysis in July during 1980–2012. In addition to the SVD analysis, this negative relationship is confirmed by both the correlation between GTS and rainfall and the composite of anomalous rainfall. The negative relationship shows that when the GTS anomalies over the TP are negative (positive), the rainfall increases (decreases) over NWC-WM. From the composite maps of rainfall anomalies, it is further found that rainfall significantly increases (decreases) over northern XJ and western Mongolia at about 46° N (over northern GS at about 42° N) during the years with the negative (positive) GTS anomalies over the TP. This indicates that the responses of the rainfall over NWC-WM to the GTS over the TP are not symmetrical

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over...

between the years of positive and negative GTS anomalies in terms of spatial distribution of rainfall anomalies. The atmospheric circulation related to the negative relationship between GTS and rainfall is further examined. In the observations, normally, moisture is transported eastward from the Atlantic Ocean to NWC-WM. After removing the zonal mean, a warm core is at about 300 hPa over the TP, positive geopotential height is over the TP at the upper levels (above 400 hPa) with its center at about 150 hPa, and negative geopotential height is at lower levels (below 400 hPa) over the TP and surrounding areas of the TP. Ascent is found over the TP, and descent is found between 40° and 50° N to the north of the TP, which causes the typically dry climate over NWCWM. During the years with negative (positive) GTS anomalies over the TP, according to the anomalous horizontal circulation, anomalous centers of negative (positive) geopotential height at both the 150 and 500 hPa are over northern GS at about 42° N. According to the anomalous meridional circulation between 80° and 100° E, during the years of negative (positive) GTS anomalies over the TP, anomalous negative (positive) air temperature and geopotential height are over northern GS with their centers at about 300 hPa and between 200 and 150 hPa, respectively. Downward and upward (upward and downward) motions are over the north slope of the TP and to the north of the TP at about 50° N, respectively, and upward (downward) motion is over southern TP. Specific humidity decreases (increases) at about 500 hPa over the north slope of the TP but increases (decreases) at lower levels over NWC-WM between 40° and 50° N. Generally, the positive (negative) rainfall anomalies correspond to the positive (negative) specific humidity anomalies at lower levels and upward (downward) motion over NWC-WM. Over the southern TP, the negative (positive) rainfall anomalies are also related to the downward (upward) motion locally. To determine the effects of the GTS anomalies over the TP on the rainfall anomalies over NWC-WM, two sets of sensitivity experiments with positive and negative GTS forcing added over the regions of TPC and TPN are conducted by using the RegCM4.1, as well as one set of control experiment. In each set of sensitivity experiments, six subsets of experiments are carried out, including three subsets with negative (positive) GTS forcing in key regions. In each subset experiment, there are 30 runs during 1982–2011. Through analyzing the anomalous rainfall in the sensitivity experiments, it is found that the negative (positive) GTS forcing over the TP can induce the positive (negative) rainfall anomalies over NWC-WM, especially at 45° N (42° N), which generally agrees with the results found in the observations. Compared the results of the experiments of TPC with TPN, it is further found that the rainfall changes over NWC-WM in July are more sensitive to the GTS anomalies over the region of TPN than those over the southern TP.

In the results of the experiment of TPN, the negative GTS forcing over the TP mainly causes the decrease of the latent heat release associated with rainfall over region TPN, and then downward and upward motions are found over the region of TPN and between 40° and 50° N, respectively. After that, anomalous negative air temperature and geopotential height are found at the upper levels to the north of the TP. Specific humidity is increased at lower levels between 44° and 54° N. Thus, rainfall is increased over NWC-WM. On the contrary, the positive GTS anomalies over the TP mainly induce the increase of the sensible heat, which is the dominated heating over region TPN, and cause the upward motion over region TPN. After that, air temperature and geopotential height increase at upper levels to the north of the TP. Descent appears between 40° and 50° N, and moisture is pumped up from northwestern China to the TP. Specific humidity increases over northern TP but decreases at the lower levels between 40° and 50° N. Therefore, the rainfall is decreased over NWCWM. Finally, this study mainly explores the simultaneous relationship between the GTS over the TP and rainfall over northwestern China and Mongolia in July. The GTS anomalies over the TP may be affected by some leading signals, for example, the snow cover/depth on the TP in winter and spring, the El Niño/Southern Oscillation, and the Quasi-Biennial Oscillation. In general, the significantly negative relationship between the GTS on the TP and rainfall over north western China and western Mongolia is identified, and the effects of the GTS on the TP on the rainfall changes are further confirmed. So far, in this study, it is found that the GTS over the TP can be taken as a bridge between large-scale atmospheric circulation anomaly and rainfall changes over NWC-WM. Acknowledgements This study is supported by the National Natural Science Foundation of China under grant nos. 91437109 and 91537102, and the Desert Meteorology Research Foundation of China under grant no. Sqj2013002. Y.Z. thanks the support from the National Natural Science Foundation of China under grant no. 41505056. We thank the anonymous reviewers for their insightful and constructive suggestions.

References Barlow M, Wheeler M, Lyon B, Cullen H (2005) Modulation of daily precipitation over Southwest Asia by the Madden–Julian oscillation. Mon Weather Rev 133:3579–3594 Blackadar AK (1976) Modeling the nocturnal boundary layer. In: Proceedings of the third symposium on atmospheric turbulence, diffusion and air quality. American Meteorological Society, Boston, pp 46–49 Bothe O, Fraedrich K, Zhu X (2010) The large-scale circulations and summer drought and wetness on the Tibetan Plateau. Int J Climatol 30:844–855 Bothe O, Fraedrich K, Zhu X (2011) Large-scale circulations and Tibetan Plateau summer drought and wetness in a high-resolution climate model. Int J Climatol 32:832–846

Y. Zhou et al. Bretherton CS, Smith C, Wallace JM (1992) An intercomparison of methods for finding coupled patterns in climate data. J Clim 5: 541–560 Chen J, Wu X, Yin Y, Xiao H (2015) Characteristics of heat sources and clouds over eastern China and the Tibetan Plateau in boreal summer. J Clim 28:7279–7295 Choi KS, Kim BJ, Zhang R, Nam JC, Park KJ, Kim JY, Kim DW (2016) Possible influence of south Asian high on summer rainfall variability in Korea. Clim Dyn 46:833–846 Cressman GP (1959) An operational objective analysis system. Mon Weather Rev 87:367–374 Deardorff JW (1978) Efficient prediction of ground surface temperature and moisture with inclusion of a layer of vegetation. J Geophys Res 83:1889–1903 Dickinson RE, Henderson-Sellers A, Kennedy PJ (1993) Biosphere– Atmsophere Transfer Scheme (BATS) version 1e as coupled to the NCAR community model. NCAR Tech. Note NCAR/TN3871STR, 72 pp. https://doi.org/10.5065/D67W6959 Ding Y, Wang Z (2008) A study of rainy seasons in China. Meteorol Atmos Phys 100:121–138 Duan AM, Wu GX (2005) Role of the Tibetan Plateau thermal forcing in the summer climate patterns over subtropical Asia. Clim Dyn 24: 793–807 Duan AM, Wu GX (2009) Weakening trend in the atmospheric heat source over the Tibetan Plateau during recent decades. Part II: connection with climate warming. J Clim 22:4197–4212 Duan AM, Wang M, Lei Y (2013) Trends in summer rainfall over China associated with the Tibetan Plateau sensible heat source during 1980–2008. J Clim 26:261–275 Grell GA, Dudhia J, Stauffer DR (1994) A description of the fifthgeneration Penn State/NCAR Mesoscale Model (MM5). Tech. Rep. TN-398+STR, NCAR, Boulder, Colorado, pp 121. https:// doi.org/10.5065/D60Z716B He H, Mcginnis JW, Song Z, Yanai M (1987) Onset of the Asian summer monsoon in 1979 and the effect of the Tibetan Plateau. Mon Weather Rev 115:1966–1995 Holtslag AAM, de Bruijn EIF, Pan HL (1990) A high resolution air mass transformation model for short-range weather forecasting. Mon Wea Rev 118:1561–1575 Hu J, Duan A (2015) Relative contributions of the Tibetan Plateau thermal forcing and the Indian Ocean surface temperature basin mode to the interannual variability of the east Asian summer monsoon. Clim Dyn 45:2697–2711 Jiang D, Fu J, Zhuang D, Xu X (2013) Dynamic monitoring of drought using HJ-1 and MODIS time series data in northern China. Nat Hazards 68:337–350 Li X, Luo Y, Guan Z (2014) The persistent heavy rainfall over southern China in June 2010: evolution of synoptic systems and the effects of the Tibetan Plateau heating. J Meteorol Res 28:540–560 Liu X, Yin ZY (2001) Spatial and temporal variation of summer precipitation over the eastern Tibetan Plateau and the North Atlantic oscillation. J Clim 14:2896–2909 Liu Y, Wu GX, Liu H, Liu P (2001) Condensation heating of the Asian summer monsoon and the subtropical anticyclone in the eastern hemisphere. Clim Dyn 17:327–338 Liu Y, Wu GX, Ren R (2004) Relationship between the subtropical anticyclone and diabatic heating. J Clim 17:682–698 Liu Y, Bao Q, Duan A, Qian Z, Wu G (2007) Recent progress in the impact of the Tibetan Plateau on climate in China. Adv Atmos Sci 24:1060–1075 Liu Y, Wu GX, Hong J, Dong B, Duan A, Bao Q, Zhou L (2012) Revisiting Asian monsoon formation and change associated with Tibetan Plateau forcing: II. Change. Clim Dyn 39(5):1183–1195 Liu B, Wu G, Mao J, He J (2013) Genesis of the south Asian high and its impact on the Asian summer monsoon onset. J Clim 26:2976–2991

Liu G, Wu R, Zhang Y, Nan S (2014) The summer snow cover anomaly over the Tibetan Plateau and its association with simultaneous precipitation over the Mei-Yu-Baiu region. Adv Atmos Sci 31:755–764 Liu H, Duan K, Li M, Shi P, Yang J, Zhang X, Sun J (2015) Impact of the North Atlantic oscillation on the dipole oscillation of summer precipitation over the central and eastern Tibetan Plateau. Int J Climatol 35:4539–4546 Liu YM, Wang ZQ, Zhou HF, Wu GX (2017) Two types of summertime heating over Asian large-scale orography and excitation of potentialvorticity forcing II. Sensible heating over Tibetan–Iranian plateau. Sci China Earth Sci 60(4):733–744 Manabe SJ, Broccoli AJ (1990) Mountains and arid climate of middle latitude. Science 247:192–195 Pan Y, Gong D, Deng L, Li J, Gao J (2004) Smart distance searchingbased and DEM-informed interpolation of surface air temperature in China. Acta Geograph Sin 59(3):366–374 Pan W, Mao J, Wu G (2013) Characteristics and mechanism of the 10–20day oscillation of spring rainfall over southern China. J Clim 26: 5072–5087 Park HS, Chiang JCH, Bordoni S (2012) The mechanical impact of the Tibetan Plateau on the seasonal evolution of the south Asian monsoon. J Clim 25:2394–2407 Qian YF, Zheng YQ, Zhang Y, Miao MQ (2003) Responses of China’s summer monsoon climate to snow anomaly over the Tibetan Plateau. Int J Climatol 23:593–613 Rajagopalan B, Molnar P (2013) Signatures of Tibetan Plateau heating on Indian summer monsoon rainfall variability. J Geophys Res Atmos 118:1170–1178 Ren X, Yang D, Yang XQ (2015) Characteristics and mechanisms of the subseasonal eastward extension of the south Asian high. J Clim 28: 6799–6822 Sato T, Kimura F (2007) How does the Tibetan Plateau affect the transition of Indian monsoon rainfall? Mon Weather Rev 135:2006–2015 Schiemann R, Lüthi D, Vidale PL, Schär C (2008) The precipitation climate of Central Asia—intercomparison of observational and numerical data sources in a remote semiarid region. Int J Climatol 28: 295–314. https://doi.org/10.1002/joc.1532 Schiemann R, Lüthi D, Schär C (2009) Seasonality and interannual variability of the westerly jet in the Tibetan Plateau region. J Clim 22: 2940–2957 Si D, Ding Y (2013) Decadal change in the correlation pattern between the Tibetan Plateau winter snow and the East Asian summer precipitation during 1979–2011. J Climate 26:7622–7634 Song JH, Kang HS, Byun YH, Hong SY (2010) Effects of the Tibetan Plateau on the Asian summer monsoon: a numerical case study using a regional climate model. Int J Climatol 30:743–759 Song C, Ke L, Richards KS, Cui Y (2016) Homogenization of surface temperature data in high mountain Asia through comparison of reanalysis data and station observations. Int J Climatol 36:1088–1101 Trenberth KE (1991) Climate diagnostics from global analyses: conservation of mass in ECMWF analyses. J Clim 4:707–722 Wang H, Li D (2011) Correlation of surface sensible heat flux in the arid region of northwestern China with the northern boundary of the east Asian summer monsoon and Chinese summer precipitation. J Geophys Res 16:D19114 Wang A, Zeng X (2012) Evaluation of multireanalysis products with in situ observations over the Tibetan Plateau. J Geophys Res 117: D05102 Wang Z, Duan A, Wu G (2014) Time-lagged impact of spring sensible heat over the Tibetan Plateau on the summer rainfall anomaly in East China: case studies using the WRF model. Clim Dyn 42:2885–2898 Wei W, Zhang R, Wen M, Rong X, Li T (2014) Impact of Indian summer monsoon on the south Asian high and its influence on summer rainfall over China. Clim Dyn 43:1257–1269

Effects of the ground surface temperature anomalies over the Tibetan Plateau on the rainfall over... Wei W, Zhang R, Wen M, Kim BJ, Nam JC (2015) Interannual variation of the south Asian high and its relation with Indian and east Asian summer monsoon rainfall. J Clim 28:2623–2634 Willmott CJ, Matsuura K (1995) Smart interpolation of annually averaged air temperature in the United States. J Appl Meteorol 34:2577– 2586 Wu GX (2004) Recent progress in the study of the Qinghai-Xizhang Palateau climate dynamics in China. Quaternary Science 24(1):1–9 Wu GX, Liu Y (2003) Summertime quadruplet heating pattern in the subtropics and the associated atmospheric circulation. Geophys Res Lett 30(5):12011–12014 Wu TW, Qian ZA (2003) The relation between the Tibetan winter snow and the Asian summer monsoon and rainfall: an observational investigation. J Clim 16:2038–2050 Wu GX, Liu Y, Wang T, Wan R, Liu X, Li W, Wang Z, Zhang Q, Duan A, Liang X (2007) The influence of mechanical and thermal forcing by the Tibetan Plateau on Asian climate. J Hydrol 8:770–789 Wu GX, He B, Liu Y, Bao Q, Ren R (2015) Location and variation of the summertime upper troposphere temperature maximum over South Asia. Clim Dyn 45:2757–2774 Wu GX, Zhou HF, Wang ZQ, Liu Y (2016) Two types of summertime heating over the Asian large-scale orography and excitation of potential vorticity forcing I. Over Tibetan Plateau. Sci China Earth Sci 59(10):1996–2008 Xu X, Lu C, Ding Y, Shi X, Guo Y, Zhu W (2013) What is the relationship between China summer precipitation and the change of apparent heat source over the Tibetan plateau. Atmos Sci Lett 14:227–234

Yan Y, Liu Y, Lu J (2016) Cloud vertical structure, precipitation, and cloud radiative effects over Tibetan Plateau and its neighboring regions. J Geophys Res Atmos 121:5864–5877 Yanai Y, Tomita T (1998) Seasonal and interannual variability of atmospheric heat sources and moisture sinks as determined from NCEPNCAR reanalysis. J Clim 11:463–482 Yanai M, Wu GX (2006) Effects of the Tibetan Plateau. In: Wang B (ed) The Asian monsoon. Springer, Berlin, pp 513–549 Yang B, Qin C, Brauning A, Burchardt I, Liu J (2011) Rainfall history for the Hexi corridor in the arid northwest China during the past 620 years derived from tree rings. Int J Climatol 31:1166–1176 Zhang R, Koike T, Xu X (2012) A China–Japan cooperative JICA atmospheric observing network over the Tibetan plateau (JICA/Tibet project): an overview. J Meteorol Soc Japan 90C:1–16 Zhang P, Liu Y, He B (2016) Impact of east Asian summer monsoon heating on the interannual variation of the south Asian high. J Clim 29(1):159–173 Zhou Y, Huang A, Zhao Y, Yang Q, Jiang J, La M (2015) Influence of the sea surface temperature anomaly over the Indian Ocean in march on the summer rainfall in Xinjiang. Theor Appl Climatol 119:781–789 Zhu Y, Liu H, Ding Y, Zhang F, Li W (2015) Interdecadal variation of spring snow depth over the Tibetan Plateau and its influence on summer rainfall over east China in the recent 30 years. Int J Climatol 35:3654–3660