Linkages of surface air temperature variations over Central Asia with large-scale climate patterns

In this study, we investigate the dominant modes of surface air temperature variations of the cold season (from November through to the next March) and the warm season (from May to September) over Central Asia, and their associations with large-scale climate patterns for the period of 1979–2016. The first two modes of the cold season surface air temperature (CSAT) over Central Asia, obtained by empirical orthogonal function (EOF) analysis, feature the monopole structure and the north-south dipole pattern, respectively. For the warm season surface air temperature (WSAT), the leading two EOF modes are characterized by the homogenous structure and the northwest-southeast seesaw pattern, respectively. Further analysis indicates that the large-scale atmospheric circulation anomalies play key roles in the CSAT and WSAT variations over Central Asia. The CSAT variation over Central Asia is closely related to the Scandinavia pattern (SCAND) and the Arctic Oscillation (AO), while the WSAT variation is tightly tied to the East Atlantic/Western Russia pattern (EAWR) and the North Atlantic Oscillation (NAO). These large-scale climate patterns tend to cause the CSAT and WSAT anomalies over Central Asia via their effects on regional geopotential heights, warming advections, and other processes. Positive geopotential height anomalies and increased downward solar radiations generally favor positive SAT anomalies over Central Asia. Moreover, the warm advections are also conducive to the formation of positive SAT anomalies over Central Asia. Our findings are expected to facilitate the improvement of understanding and predicting the CSAT and WSAT variations over Central Asia.


Introduction
About 65 million people live in Central Asia, which is a key region connecting Asia and Europe. Due to fragile ecological environment, Central Asia is extremely sensitive to climate change (Lioubimtseva and Henebry 2009;Huang et al. 2017). Surface air temperature (SAT) is one of the most important climate variables, and its anomalies have profound impacts on natural ecology and human society over Central Asia and many other regions of the globe (IPCC 2013). Understanding the spatiotemporal changes in SAT over Central Asia is of great significance to the policymaking of climate change adaptation strategy.
Previous studies demonstrated that the SAT over Central Asia has been significantly increasing since the beginning of the 20th century, faster than global warming (Lioubimtseva et al. 2005;Zhang et al. 2011;Hu et al. 2014;Chen et al. 2017). Compared with springtime, summertime, and autumntime, the warming of wintertime is the strongest. The EOF analysis showed that the annual and seasonal average SAT variations over Central Asia are characterized by the same-sign pattern, followed by the dipole type in East-West or North-South (Wang et al. 2008;Chen et al. 2009;Zhang et al. 2018;Zhu et al. 2020).
Large-scale climate patterns such as the Arctic Oscillation (AO), the North Atlantic Oscillation (NAO), and the Scandinavia pattern (SCAND) have been shown to play a vital role in influencing the climate variations in Eurasia (Barnston and Livezey 1987;Hurrell and Van Loon 1997;Wallace 1998, 2000;Koide and Kodera 1999;Steinbrecht et al. 2001;Besselaar et al. 2010;Hoy et al. 2013;Guo and Li 2016;Wang et al. 2016;Hu et al. 2017;Zhuang et al. 2018;Chen et al. 2018Chen et al. , 2019. The importance of a specific largescale climate pattern to the climate variation depends on the region and the season. For example, it is widely recognized that the El Niño-Southern Oscillation (ENSO) has a significant impact on the East Asian summer monsoon system (Yang and Lau 1998;Wu et al. 2009;Ren et al. 2015;Zhang et al. 2016;Santoso et al. 2017). Wu et al. (2010) indicated that the summer SAT over Northeast China (NEC) tends to be anomalously lower (higher) than normal in El Niño (La Niña) developing years before the mid-1970s, whereas in the 1980s and 1990s, their relationship is weakened or even becomes opposite. The positive phase of the NAO or AO tends to cause warm SAT anomalies over large parts of north Eurasia, and cool SAT anomalies over southern Europe and the Middle East during winter (Gong et al. 2001;Wu and Wang 2002;Li and Wang, 2003;Liu et al. 2014;Wang et al. 2017;Chen et al. 2019). However, limited studies have been conducted to understand the associations of large-scale climate patterns with the SAT variations over Central Asia (Wang et al. 2008;Chen et al. 2009;Yao et al. 2014).
In previous research, a year is commonly divided into four seasons: spring (March-April-May, MAM), summer (June-July-August, JJA), autumn (September-October-November, SON), and winter (December-January-February, DJJ). However, this conventional definition may not be suitable for the Central Asia climate studies because Central Asia, located in the interior of Eurasia, has a typical continental climate with the characteristics of fairly colder in winter and rather hotter in summer, as well as short durations of spring and autumn. In this study, from the point of monthly average SAT, we firstly define November to March in the next year as the cold season and May to September as the warm season over Central Asia, and then verify the rationality of the definition according to atmospheric circulation evolutions. Next, the dominant modes of the cold season SAT (CSAT) and the warm season SAT (WSAT) over Central Asia are further identified. Finally, we explore the relationships of large-scale climate patterns with the CSAT and the WSAT variations over Central Asia.

Data and methods
Monthly mean surface air temperature used in this study was derived from Climatic Research Unit gridded dataset version 4.01 (CRU TS4.01) with a horizontal resolution of 0.5°× 0.5° (Harris et al. 2014). Large-scale climate indices, including the AO, the East Atlantic/Western Russia (EAWR), the NAO, the Niño 3.4, the Pacific/ North American (PNA), the Polar/Eurasia (POL), and the SCAND, were obtained from the National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center (CPC) (https://www.esrl.noaa.gov/psd/ data/climateindices/list/). To explain the physical processes associated with the CSAT and the WSAT variations over Central Asia, we also employed monthly atmospheric circulation variables consisting of 500-hPa geopotential height (H500), 850-hPa wind vector (V850) , and air temperature at 850 hPa (T850) from ERA5 reanalysis dataset produced by European Centre for Medium-Range Weather Forecasts with a spatial resolution of 0.5°× 0.5° (Hersbach et al. 2019). The shortwave radiations, longwave radiations, and sensible and latent heat fluxes are obtained from the National Centers for Environment Prediction-US Department of Energy (NCEP-DOE) AMIP-II reanalysis (Kanamitsu et al. 2002). The study period is from January 1979 to December 2016.
We use the EOF analysis (von Storch and Zwiers 1999) to describe the spatial-temporal variation characteristics of the CSAT and the WSAT over Central Asia. The correlation and regression analyses are applied to examine the relationships of the CSAT and the WSAT variations over Central Asia with large-scale climate patterns, and the associated physical mechanisms. We apply Students' t test to assess the statistical significance levels of the correlation and regression analyses.

Definition of the cold season and the warm season over Central Asia
As shown in Fig. 1, the monthly mean SAT values averaged over Central Asia are higher than 15°C from May to September and are all below 0°C for November to March in the next year for the period of 1979-2016. This suggests that we can define the months from November through to the next March as the cold season and the months from May through to September as the warm season over Central Asia. To further justify the suitability of this definition, the time series of winter and summer SAT over Central Asia for the period of 1979-2016 are firstly shown (Fig. 2). The mean SAT values over Central Asia are all less than −4°C in winter and above 19°C in summer.
We further analyze and compare the 850-hPa wind vector fields in March, April, May, September, October, and November with those in winter (December-January-February, DJF) and summer (June-July-August, JJA). Strong southwesterly flows are seen over Central Asia in winter (Fig. 3a), while northwesterly and northeasterly winds are observed in the northern and southern parts, respectively, in summer (Fig. 3b). When considering the wind field in every month other than summer and winter, it reveals that the wind fields in March and November resemble the winter fields ( Fig. 3a, c, h), while in May and September, the wind fields are similar to the summer fields (Fig. 3b,d,g). We also calculate the spatial correlation coefficients between the wind directions of 850-hPa wind fields in winter with those in every month from January to December over the region extending from 35°to 58°N and from 45°to 90°E. Results indicate that the correlation coefficients in cold months (November to March) all exceed 0.60, significant at the 99% confidence level, and they are all small in warm months (May to September) (Fig. 4). Combining the results of temperature and wind fields, therefore, the cold season is defined as the months from November to the next March, and the warm season includes the months from May to September over Central Asia.
Based on the new definition, there are altogether 37 cold seasons and 38 warm seasons during the study period of 1979-2016. The cold season of 1979 refers to November 1979 to March 1980, and the 1979 warm season is from May to September in 1979. The dominant modes of the CSAT and the WSAT variations over Central Asia and their linkages with large-scale atmospheric circulation patterns will be explored and discussed in the following sections.

Dominant modes of the anomalous CSAT and the WSAT variations over Central Asia
We perform the EOF analysis to investigate the temporal and spatial characteristics of the anomalous CSAT and WSAT variations with respect to the 1979-2016 climatological values over Central Asia ( Fig. 5 and Fig. 6). The first leading EOF mode of the anomalous CSAT over Central Asia (EOF1), explaining 84.5% of the total variance, displays a same-sign structure with the maximum center located over western Kazakhstan (Fig. 5a). The corresponding EOF1 time series displays a pronounced interannual variability with a period cycle of 2-3 years and weak interdecadal cycle around 10 years ( Fig. 5b and Fig. 7a) and is highly related with the regional averaged anomalous CSAT over Central Asia (r = 0.99, the 99.9% confidence level), indicating that the EOF1 of the anomalous CSAT dominates the variation characteristics of the anomalous CSAT over Central Asia.
The second EOF mode (EOF2) of the anomalous CSAT over Central Asia describing 8.7% of the total variance is characterized by a north-south seesaw pattern (Fig. 5c). The corresponding EOF2 time series exhibits an obvious variation ( Fig. 5d) with a period cycle of 6-9 years from the power spectrum (Fig. 7b).
The EOF1 mode of the anomalous WSAT over Central Asia features a homogeneous structure, which explains 66.9% of the total variance (Fig. 6a). The corresponding EOF1 time series shows a substantial upward trend, and the power spectrum analysis displays a period of 2-3-year cycle variations (Figs. 6b and Fig. 7c). The EOF2 mode o f t h e an o m a l o u s W S A T o v e r C e n t r a l A s i a i s  Fig. 7d). In total, the EOF1 and EOF2 account for more than 81% of the total variance of the anomalous WSAT over Central Asia. 3.3 Large-scale atmospheric circulations related to the CSAT and WSAT anomalies over Central Asia Figure 8 presents regressions of H500, V850, and T850 anomalies onto the EOF1 and EOF2 time series of the anomalous CSAT over Central Asia for the period of 1979-2016. The significantly positive H500 anomalies related to the EOF1 mode of the anomalous CSAT over Central Asia appear over the regions from Central Siberia to Central Asia and Western China, while the negative H500 anomalies are observed in the Scandinavian Peninsula and Eastern Europe in the cold season (Fig. 8a). Correspondingly, an anomalous anticyclonic circulation exists over Central Siberia and surrounding areas, whereas an anomalous cyclonic circulation dominates the Scandinavian Peninsula and Eastern Europe (Fig. 8b).
For the EOF2 mode of the anomalous CSAT, the significantly positive H500 anomalies are seen over high-latitude areas, southern part of Central Asia, and many subtropical areas, while the significantly negative H500 anomalies mainly appear over Europe and Northwestern Asia (Fig. 8c). For the 850-hPa wind field, the westerly flow is weakened over Northern Central Asia and the southerly flow is slightly enhanced over Southern Central Asia (Fig. 8d). Figure 9 shows regressions of H500, V850, and T850 anomalies onto the EOF1 and EOF2 time series of the anomalous WSAT over Central Asia for the period of 1979-2016. Corresponding to the EOF1 mode of the anomalous WSAT over Central Asia, there are strong positive H500 anomalies over the region extending from Eastern Europe to Central Asia, accompanied by an anomalous anticyclonic circulation in the 850-hPa wind field in the warm season ( Fig. 9a, b). For the EOF2, significantly negative H500 anomalies are observed over the region from Eastern Europe to Northwestern Central Asia with an anomalous cyclone in the 850-hPa wind field (Fig. 9c, d). Regardless of the cold season or the warm season, the positive H500 anomalies can provide favorable conditions for the warm SAT anomalies over Central Asia and other regions may be via the enhancements of solar radiation reaching land surface and subsidence warming.
To investigate how the warm CSAT and WSAT anomalies over Central Asia are induced, we examined the effects of surface heat flux and advection. Figures 10 and 11 show the CSAT-related and WSAT-related anomalies of surface net radiation (NR), sensible heat flux (SHF), net shortwave The EOF1. c, d The EOF2 (H500, V850, and T850 denote 500-hPa geopotential height, 850-hPa wind vector, and 850-hPa air temperature, respectively. The black dots in a and c and the color shadows and black arrows in b and d indicate the 95% confidence levels based on a two-sided Student's t test. The contour intervals of H500, V850, and T850 are 3 gpm, 5 m·s −1 , and 0.2°C respectively. Central Asia is marked by the black lines) Fig. 9 The same as Fig. 10, but for the anomalous WSAT over Central Asia radiation (SWR), and net longwave radiation (LWR) from1979 to 2016. Note that SWR is equal to net downward shortwave radiation minus net upward shortwave radiation, and LWR is equivalent to the difference between net downward longwave radiation and net upward longwave radiation. For the NR, it is calculated by the sum of SWR and LWR. Positive NR represents downward direction and can warm the surface. Corresponding to the EOF1 of the anomalous CSAT, positive NR is observed throughout most Central Asia (Fig. 10a), which is dominated by downward SWR (Fig. 10b). This is likely attributed to the clear sky accompanied by the anomalous anticyclone (Fig. 8a). Although the LWR tends to cool the ground (Fig. 10c), its effect is weaker than the SWR (Fig. 10 b and c). These results indicate that the ground surface receives more heat from the downward solar radiation to warm the surface (Fig.10a). When the ground is heated and warmer than the surface air, it will lead to upward sensible heat flux to warm the air (Fig. 10b), thereby contributing constructively to the warmer CSAT (Fig. 8b).
For the EOF2 of the anomalous CSAT, negative CSAT anomalies over northeastern parts of Central Asia are attributed to significantly negative NR anomalies dominated by the cooling effect of upward LWR (Figs. 8d and 10f, h), while weaker positive NR anomalies exist in the middle parts of Central Asia, which contribute to the formation of positive CSAT anomalies in these regions (Figs. 8d, 10f). However, the spatial distribution of the NR anomalies (Fig. 10f) is different from that of the CSAT anomalies to some extent (Fig. 8d). Positive CSAT anomalies over western parts of Central Asia cannot be explained by NR increase.
Corresponding to the EOF1 of the anomalous WSAT, significantly positive NR anomalies are seen around western and northern parts of Central Asia where positive WSAT anomalies center is observed (Figs. 11a, 7b), which is mainly due to downward SWR increase (Fig. 11b) brought by positive H500 and anomalous anticyclone at V850 (Fig. 9a, b). The positive WSAT anomalies appear over Eastern Central Asia (Fig. 9b); nevertheless, it is controlled by negative NR anomalies due to upward SWR and LWR increases (Fig. 11a-c). These indicate that the formation of the EOF1 of anomalous WSAT over Central Asia can be partly explained by the NR anomalies.
Corresponding to the EOF2 of the anomalous WSAT, the spatial distribution of NR anomalies (Fig. 10f) is similar to that of the WSAT anomalies, both showing a northwestsoutheast dipole distribution (Figs. 6c, 9d, 11f). Negative WSAT anomalies correspond to negative NR anomalies over the northwestern part of Central Asia, while the opposite phenomenon is observed in Southeastern Central Asia. The above analysis indicates that surface heat flux changes over many regions of Central Asia can explain the formation of CSAT and WSAT anomalies associated with the first two EOF modes (Figs. 12 and 13).
A further inspection suggests that the advection of mean temperature by anomalous winds is constructive (Fig. 12c), while the advection of anomalous temperature by mean wind is destructive (Fig. 12b) for the EOF1 of anomalous CSAT (Fig. 8b). To better understand this process, we also show the climatological mean winds overlaid on the EOF1of anomalous CSAT-related temperature anomalies (Fig. 14a) and the EOF1 of anomalous CSAT-related wind anomalies overlaid on the climatological mean temperature (Fig. 14b). Results show that the anomalous wind can advect the warm air northeastward toward Central Asia, leading to warming tendency over Central Asia (Figs. 12c, 14b, 8b).
For the EOF2 of anomalous CSAT, the CSAT-related warm center is right located to the southwestern of Central Asia; therefore, the warm air is advected northeastward toward Central Asia by the climatological mean wind (Fig.  14c); what's more, the anomalous wind can also advect the warm air from low latitudes to Southern part of Central Asia (Fig. 12e). These contribute to warming tendency CSAT anomalies over Southern Central Asia (Fig. 8d). The warm advections accompanied by the climatological winds (Figs. 12d, e and 14c) are also conducive to the formation of the CSAT anomalies over the Northern part of Central Asia. However, due to the cancellation of the cold advections associated with the anomalous winds (Figs. 12f, 14d), the positive CSAT variations over the northern part of Central Asia weaken, even the negative CSAT anomalies appearing in Northeast Central Asia (Fig. 8d).
For the EOF1 of the anomalous WSAT over Central Asia, the warm WSAT anomalies over Central Asia tend to be attributed to the advection of warm air related to the climatological winds, which stem from the anomalous warm center located in Northwestern Central Asia (Figs. 9b and 15a). In contrast, the anomalous anticyclone (Fig. 9b) favors cooling tendency over Central Asia (Fig. 9b) because the climatological winds blow up-gradient of the anomalous temperature field (Fig. 15b). Despite this seemingly destructive effect, this anomalous anticyclone is crucial for the formation of the anomalous warm center over Northeastern Central Asia Net radiation is calculated as net shortwave radiation plus net longwave radiation. Net shortwave radiation is calculated as net downward shortwave radiation minus net upward shortwave radiation, and net longwave radiation is calculated as net downward longwave radiation minus net upward longwave radiation. Central Asia is marked by the black lines) (Figs. 9b and 15a) because the southerly wind anomalies of the anticyclone blow down-gradient of the climatological mean temperature field (Fig. 15b).
Corresponding to the EOF2 of WSAT over Central Asia, the WSAT-related cold center is right located to the northwest of Central Asia (Fig. 15c); the cold air advection accompanied Fig. 11 The same as Fig. 10, but for the anomalous WSAT over Central Asia> by the climatological northwesterly winds therefore leads to negative WSAT anomalies over the northwestern portion of Central Asia (Figs. 9d and 15c). However, the warm WSAT anomalies over eastern and southern parts of Central Asia (Fig. 9d) are partly attributed to the warm air advection brought by the southwesterly wind in the south of anomalous cyclone (Fig. 15d).

Linkages of large-scale climate patterns with the CSAT and WSAT variations over Central Asia
Previous studies suggest that the large-scale atmospheric circulation changes may play an important role in the SAT anomalies over Central Asia (Victoria et al. 2001;Wang and Wei 2012;Zhai et al. 2016;Zhou et al. 2016;Luo et al. 2019). In the analysis that follows, we further investigate the largescale climate pattern anomalies associated with the first two EOF modes of CSAT and WSAT variations over Central Asia. Results show that the EOF1 time series of the anomalous CSAT over Central Asia is closely related to the SCAND in the cold season with a correlation coefficient of −0.63 for 1979-2016, significant at the 99% confidence level. Meanwhile, the EOF2 time series of the anomalous CSAT over Central Asia is highly correlated with the AO and the NAO in the cold season. Compared with the AO, correlation with the NAO is much weaker (Fig. 16a). These results suggest that the SCAND and the AO have substantial roles to play in influencing the CSAT variation over Central Asia. Correlations between the other large-scale climate indices shown in Fig. 16a and the first two EOF time series of the anomalous CSAT over Central Asia are not significant.
The EOF1 time series of the anomalous WSAT over Central Asia is tightly associated with the EAWR and the NAO in the warm season. Among them, the correlation coefficient with the EAWR index is the strongest, reaching up to −0.58, followed by the NAO index with the correlation coefficient of −0.53 (Fig. 16b). These correlations are all significant at the 99% confidence level based on Student's t test. The EOF2 time series of the WSAT over Central Asia is significantly correlated with the EAWR index (r = 0.36, the 95% confidence level). These indicate that the atmospheric circulation anomalies related to the EAWR and the NAO play an important role in modulating the WSAT anomalies over Central Asia.
We further examine the relationships of the SCAND and the AO to the CSAT variations over Central Asia for the period of 1979-2016 (Fig. 17). The positive phase of the SCAND tends to correspond to significantly negative CSAT anomalies over Central Asia in the cold season (Fig. 17a), whereas the AO shows a north-south dipole pattern, roughly similar to the EOF2 mode of the CSAT over Central Asia (Figs. 17b and 4c) with significantly positive correlation center of CSAT anomalies appearing over northeastern areas of Central Asia (Fig. 17b).
In the warm season, the anomalous WSAT over Central Asia is negatively correlated with the EAWR and the NAO for 1979-2016 (Fig. 10). Significant correlations appear over almost the whole Central Asia for both these two climate patterns (Fig. 18 a and b); the maximum correlation coefficient center mainly shows over western and central parts of Central Asia for the EAWR and the NAO (Fig. 18a 18b).
To further confirm the roles of the SCAND and the AO patterns in the variation of anomalous CSAT over Central Asia, we display H500 and V850 wind anomalies by regression onto the SCAND and the AO indices, as shown in Fig.  19. The regression results related to the SCAND show that the circulation structure is reminiscent of the positive SCAND pattern with positive H500 anomalies over Scandinavia and Western Russia (Barnston and Livezey 1987). During the positive phase of the SCAND pattern, Central Asia is Fig. 13 The same as Fig. 12, but for the anomalous WSAT over Central Asia dominated by significant negative H500 anomaly in the cold season with the center existing in the Lake Baikal (Fig. 19c), accompanying cold advection from Siberia to Central Asia led to the negative CSAT anomalies over Central Asia, especially in northeastern parts of Central Asia (Figs. 19d,9b,and 15b). For the large-scale atmospheric circulation associated with the AO pattern, its structure bears a similarity to the positive phase of the AO (Thompson and Wallace 2000). The spatial distribution displays that the Arctic regions are covered by negative H500 anomalies, while positive H500 anomalies mainly exist over those regions in middle latitudes with one center in Western Europe and the others in Lake Baikal. The positive CSAT anomalies over Northeastern Central Asia are affected by the significant southwesterly winds, while the negative CSAT anomalies over southwestern parts of Central Asia are attributed to anomalous significant northwesterly winds associated with the anticyclonic circulation in Western Europe (Figs. 19a,b and 9b,c). In the warm season, the significantly negative H500 anomalies associated with the positive EAWR pattern occupy the most areas of Central Asia with the notable negative center located in northwestern parts of Central Asia (Fig. 20c). Therefore, the cooling of the WSAT in Northwestern Central Asia is more significant due to the significant northwesterly wind on the north of the cyclone conveying cold and dry air flow from high latitude (Figs. 20d and 18a). However, the remarkably negative H500 anomaly center related to the NAO pattern is located in Central Asia (Fig. 20a), which has an impact on the WSAT over the whole region of Central Asia ( Fig. 20b and 18b).

Conclusions
Central Asia is characterized by a remarkable continental climate and has short durations of spring and autumn. Based on the analyses of monthly mean SAT and atmospheric circulation, we define the months from November through to the next March as the cold season and May to September as the warm season, and further investigate dominant modes of the CSAT and WSAT variations over Central Asia connected with largescale climate patterns for the period of 1979-2016.
The EOF analysis shows that the first leading mode of the CSAT variation over Central Asia, describing 84.5% of the total variance, is featured by a monopole pattern with the maximum center appearing over western Kazakhstan. The EOF2 mode of the CSAT over Central Asia is characterized by a south-north dipole pattern. The EOF1 and EOF2 modes of the WSAT over Central Asia mainly show homogeneous and northwest-southeast variations, describing 66.9% and 14.4% explanatory variances, respectively. The EOF1 and EOF2 together account for about 93% and above 81% of the total variances for the CSAT and the WSAT variations, respectively.
The atmospheric circulation patterns play a crucial role in driving the CSAT and the WSAT variations over Central Asia. For the EOF1 mode of the CSAT over Central Asia, warm anomalies correspond to significant positive H500 anomalies and anomalous southwesterly winds at V850 over Central Asia. These subsequently may heat the CSAT over Central Asia by enhancing the downward solar radiation, strengthening the subsidence warming and warm advection. For the EOF2 mode of the CSAT over Central Asia, positive CSAT anomalies over Southwestern Central Asia are related with the significant southwesterly winds carrying warm air from low latitudes, while negative CSAT anomaly is attributed to the significant northeasterly winds which bring cold air to Northeastern Central Asia from high latitudes.
The large-scale atmospheric circulations associated with the EOF1 of the WSAT over Central Asia show that the notably positive H500 anomalies are dominant over Central Asia, generally contributing to positive WSAT anomalies via subsidence warming. Corresponding to the EOF2 of the WSAT over Central Asia, the changes in the wind field play an important impact on the WSAT anomalies. The anomalous southwesterly winds at V850 are in favor of the positive WSAT anomalies over Southeastern Central Asia. By contrast, the negative WSAT anomalies over Northwestern Central Asia are caused by the anomalous northwesterly winds.
Combining the EOF and correlation analyses, the SCAND and the AO are identified to be closely related to the CSAT variations over Central Asia, while the EAWR and the NAO are supposed to influence the WSAT variations over Central Asia to varying degrees. For example, the negative phase of the SCAND tends to result in strongly warm anomalies over the whole Central Asia while the EAWR may mainly play a substantial role over the western part in the warm season. Further analysis indicates that these large-scale climate patterns tend to cause the CSAT and WSAT anomalies over Central Asia via their effects on regional geopotential heights, wind-induced warm advections, and other processes.
This study analyzes dominant modes of the CSAT and WSAT variations over Central Asia and their close linkages to large-scale climate patterns and may help to improve the SAT predictions over this region. Meanwhile, other driving factors such as land surface conditions should be further explored (Zhang et al. 2011;Wei and Dirmeyer 2012), and understanding thermodynamic and dynamic processes influencing the SAT variations in the cold and warm seasons need to be deepened in the future. Fig. 17 The spatial distributions of the correlation coefficients between the anomalous CSAT over Central Asia and SCAND index (a) and AO index (b) in the cold season. (The black dots indicate the regions where the correlations exceed the 95% confidence level) Fig. 18 The spatial distributions of the correlation coefficients between the anomalous WSAT over Central Asia and EAWR index (a) and NAO index (b) in the warm season. (The black dots indicate the region where the correlations exceed the 95% confidence level) Fig. 19 Regression maps of H500, V850, and T850 anomalies associated with the large-scale climate indices in the cold season for the period of 1979-2016. a, b AO index. c, d SCAND index. H500, V850, and T850 denote 500-hPa geopotential height, 850-hPa wind vector, and 850-hPa air temperature. (Black dots in a and c and the color shadows and black arrows in b and d indicate the 95% confidence levels based on a twosided Student's t test. The contour intervals of H500, V850, and T850 are 3 gpm, 5 m·s −1 , and 0.2°C, respectively. Central Asia is marked by the black lines) Fig. 20 Regression map of H500, V850, and T850 anomalies associated with the large-scale climate indices in the warm season for the period of 1979-2016. a, b NAO index. c, d EAWR index. H500, V850, and T850 denote 500-hPa geopotential height, 850-hPa wind vector, and 850-hPa air temperature. (Black dots in a and c and the color shadows and black arrows in b and d indicate the 95% confidence levels based on a twosided Student's t test. The contour intervals of H500, V850, and T850 are 3 gpm, 5 m·s −1 and 0.2°C, respectively. Central Asia is marked by the black lines)