As a plateau with the highest average altitude in the world, the Tibetan Plateau (TP) is known as the “Roof of the World” and the “Third Pole of the World” (Qiu, 2008). Its remarkable dynamic and thermal effects not only directly affect the atmospheric circulation and climate of the TP, but also have a significant impact on the atmospheric circulation system and climate pattern in the Northern Hemisphere (Zhou et al., 2009).

Ye et al. (1957) discovered that the TP is a heat source in summer based on analysis of the TP’s thermal effects. Subsequently, a large number of studies have been performed to investigate the thermal characteristics and the thermal effects of the TP (Yanai et al., 1992; Zhao and Chen, 2001; Zhang and Qian, 2002; Ma et al., 2004, 2005; Wu et al., 2004; Ma et al., 2011). Ye et al. (1979) found that sensible heating plays an important role over the TP and its thermodynamic properties have obvious seasonal changes. Yanai et al. (1992) found the importance of heating effect of near-surface atmosphere in the thermal effects of the TP. Zhong et al. (2006) found that the dynamic and thermal effects of the TP on the atmosphere are mainly achieved through interaction of the underlying surface with atmosphere and the exchange of mass and energy in a turbulent manner. Ma et al. (2006) pointed out that the thermal and dynamic effects and the mass–energy exchange process in the TP have a major impact on climate change in China, Asia, and the world. Through examination of surface heat flux over the TP, Yang et al. (2010) found that the surface sensible heat flux of the TP is decreasing at a rate of 2% per decade. Yang et al. (2014) also found that the climate change and the decline in wind speed over the TP lead to a decline in sensible heat flux. Wu et al. (2016) found that there are different feedback processes for the sensible heating and latent heating over the main body of the TP.

Many studies have shown that the surface heat source of the TP has an important impact on the atmospheric circulation in the Asian monsoon region (Duan et al., 2003; Zuo et al., 2011). Duan et al. (2004) found that the reversal of the meridional temperature gradient due to the thermal effect of the TP is an important factor causing monsoon onset, so it can be used to predict the monsoon outbreak over the Bay of Bengal. Bao et al. (2010) discovered that if the thermal effect of the TP as an atmospheric heat source is weakened, it will lead to a weakening of the Asian monsoon intensity and an increase in summer drought in China. Therefore, it is of great significance to study the distribution and variation of surface heat flux over the TP.

The monsoon is a phenomenon in which the prevailing wind direction changes with seasons, and the Asian monsoon region is the most significant monsoon region in the world. Previous studies have shown that the heating effect of the TP on the atmosphere is an important reason for the formation and maintenance of summer circulation and outbreak of the South Asian summer monsoon (ASM). Zhang and Qian (2002) discussed the mechanism of the heating effect of ground heat source on the onset of the ASM. Duan and Wu (2005) used the NCEP/NCAR reanalysis data to study the influence mechanism of thermal forcing of the TP on the subtropi-cal Asian summer climate. Their results showed that because the TP is a huge high-level heat source, a shallow cyclonic circulation occurs near the TP ground surface and above which forms a strong anticyclonic circulation. According to the barotropic vorticity equation, large-scale stable airflows are converged in the lower layer and diverged in the upper layer east of the TP. However, the western side of the TP is characterized by an inverted structure; that is, the airflows are diverged in the lower layer but converged in the higher layer. Therefore, the pumping effect generates updrafts and downdrafts in the air column on the east and west sides of the TP, respectively. In addition, Tamura and Koike (2010) emphasized the important role of convective heating in the seasonal evolution of the ASM.

Most of previous studies have focused on the impact of the TP heating field on the ASM, but studies about the long-term variation trend of surface heat fluxes over the TP and associated impacts on the onset mechanism of the ASM are still insufficient. Based on the ERA-Interim reanalysis data from 1979 to 2016, the present study intends to identify the long-term variation characteristics of surface heat fluxes over the TP before and after the ASM onset, so as to better analyze how the surface heating affects the monsoon evolution, especially with the data from the Third Tibetan Plateau Atmospheric Science Experiment (TIPEX-III) in future. This kind of research is conducive not only to predicting the ASM onset and its impact on China’s weather and climate, but also to preventing the disastrous weather and reducing the econo-mic losses in the TP downstream regions. These match with the objectives of the TIPEX-III.

Due to the high altitude and complex topography of the TP, the observation stations over the TP are sparse and unevenly distributed. The observation data of heat flux over the TP are relatively scarce, with great limitations in time and space. Therefore, gridded reanalysis data covering long time periods have been chosen to investigate the thermal effects of the TP.

A large number of studies have used several reanaly-sis datasets for comparative verification. Yanai and Tomita (1998) used NCEP/NCAR reanalysis data to study the heat source distribution in the Northern Hemisphere. Lan et al. (2005) compared the heat sources calculated by NCEP2 reanalysis data and ECMWF interim reanalysis (ERA-Interim) data. Wang and Wang (2007) compared and analyzed the differences between NCEP and ERA reanalysis data over the TP. Annamalai et al. (1999) studied the ASM by comparing the reanalysis data of ERA and NCEP and found that the ERA data are superior to the NCEP data in terms of diabatic heating in the Asian monsoon region. These results indicate that the ERA-Interim data have good applicability and high precision over the TP, and are more suitable for calculating the heat sources over the TP. Therefore, this study adopts the monthly average surface sensible and latent heat fluxes from the ERA-Interim data with a horizontal resolution of 0.75° × 0.75° during the period of 1979–2016.

Variations of surface heat fluxes over the TP before and after the onset of ASM during 1979–2016 are studied through averaging the surface heat fluxes of the TP from March to October so as to obtain the overall flux variation trend. The interannual variation trend of surface heat flux over the TP from 1979 to 2016 are analyzed by using the Mann–Kendall (MK) trend test. The MK trend test is a non-parametric statistical method, also known as a non-distribution test. For this method, the sample does not need to follow a certain distribution, and it is not interfered by a few abnormal values, which is more suitable for type variables and ordinal variables; and calculations with this method are relatively simple.

As an important phenomenon in the earth’s climate system, abrupt climate changes can be divided into four categories: changes in mean, changes in variability, seesaw jump, and transit jump. The actual situation is often compounded by two or more types of abrupt changes. Because the MK method is relatively accurate for the test of abrupt changes in mean, it is selected as the abrupt change analysis method for surface heat fluxes over the TP (Fu and Wang, 1992).

It is well known that the TP is a strong heat source in summer and a weak cold source in winter. The shift from the cold source to the heat source often occurs between March and May, before the ASM bursts (Duan et al., 2004; Yanai and Wu, 2006).

According to the monthly mean distribution of surface sensible heat flux from March to October over the TP during 1979–2016 (Fig. 1), the main body of the TP is dominated by sensible heat in March. This is because the solar radiation in spring is enhanced and the albedo is reduced, resulting from the melting of snow on the surface of TP. Moreover, the increase in surface wind speed over the TP also affects the change in sensible heat flux. Because the western TP is relatively arid and the vegetation there is sparse, high values of sensible heat flux are found in the western TP while low values are in the eastern TP. In addition, high values of sensible heat flux is also found in the southeastern TP. In April, surface sensible heat flux increases rapidly over the TP and the distribution of “west high and east low” is more obvious. The high value area of sensible heat flux in the western TP continues to expand and a new high value area appears in the northern TP. On the contrary, the high value area in the southeastern TP shrinks and is confined only in the southeast part of the original area. In May, the high value area of sensible heat flux in the southeastern TP continues to shrink, and the sensible heat flux decreases in the central TP. The high sensible heat flux in the western TP in May is comparable in magnitude to that in April, but the high value area continues to reduce in size. The sensible heat flux in the northern TP in May is equivalent to that in April, but the high value area is slightly expanding.

Figure 1 Monthly mean distributions of surface sensible heat flux (W m–2) from (a–h) March to October over the Tibetan Plateau during 1979–2016. The black line indicates the terrain height of 3000 m (Same in Fig. 2).

Figure options

With the warm and humid air and abundant water vapor brought by the ASM, the high value area of sensible heat flux in the southeastern TP attenuates in June. The high value area in the western TP continues to shrink but still shows an obvious distribution of “west high and east low”, while the high value area in the north remains unchanged. In July, affected by the monsoon flows prevailing in South Asia, the sensible heat flux in the TP continues to decline and the high value area in the western TP continues to shrink, while that in the northern TP is less affected by the monsoon, so the area and intensity there remain unchanged. In August, the TP is featured with low sensible heat flux except for the northern TP where a high value zone (of about 100 W m^–2) is maintained.

With the gradual withdrawal of the ASM, the sensible heat flux in the western TP increases gradually in September, and the TP again shows a distribution of “west high and east low” as a whole. After the ASM withdrawal, sensible heat flux in the TP decreases as a whole, and the high value area of sensible heat flux in the northern TP remains unchanged but the intensity decreases. In October, the sensible heat flux in the TP shows a distinct distribution of “west high and east low.” However, due to the decline of the overall temperature in the plateau, the overall intensity of sensible heat flux in the TP is around 100 W m^–2.

In summary, the surface sensible heat flux in the TP shows a distribution of “west high and east low” before the onset of the ASM, and it continues to increase and reaches its maximum in May. With the burst and development of the ASM, the surface sensible heat flux weakens rapidly, and there is only a high value area in the northern part of the TP. In September–October after the dissipation of the ASM, the sensible heat flux returns to the distribution of “west high and east low”, but its intensity is greatly reduced compared with that in spring, only about 100 W m^–2. This is also consistent with the distribution and variation of surface sensible heat flux over the TP calculated by Zhao and Chen (2000) using station data. Thus, it is inferred that the ERA-Interim reanalysis data are reasonable over the TP. In general, the distribution of sensible heat flux over the TP before and after the onset of the ASM conforms to the conclusion of Yao et al. (2015).

Surface heat fluxes consist of both sensible heat flux and latent heat flux. In order to fully reveal the variation characteristics of surface heat fluxes around the ASM over the TP during 1979–2016, the monthly mean distributions of surface latent heat flux from March to October over the TP during 1979–2016 are analyzed in the following (Fig. 2).

Figure 2 Monthly mean distributions of surface latent heat flux (W m–2) from (a–h) March to October over the Tibetan Plateau during 1979–2016.

Figure options

As demonstrated in Figs. 1, 2, the heat flux over the TP is mainly sensible heat in spring. In March and April, latent heat flux over the TP surface is generally weak (lower than 100 W m^–2), with relatively high values in the east and low values in the west. Meanwhile, low values of latent heat flux appear over the areas of high values of sensible heat flux in the southeastern, western, and northern TP.

Due to the abundant water vapor brought by the outbreak of the southwesterly monsoon in South Asia, latent heat flux in the eastern TP in May increases rapidly and its overall distribution of “east high and west low” is more obvious. Similar to March and April, the high value areas of sensible heat flux in the western and northern TP still correspond to the low value areas of latent heat flux. With the development of ASM, the high value area of latent heat flux over the TP expands from east to west and the magnitude continues to increase in June. The overall distribution is still east high and west low, and the low value area of latent heat flux still corresponds to the high value area of sensible heat flux. In July, the abundant water vapor and rich precipitation brought by the ASM further enhance the surface latent heat flux over the plateau, although it remains decreasing gradually from east to west. Furthermore, there is a low value area of latent heat flux in the northern TP corresponding to the high value area of sensible heat flux. In August, with the gradual dissipation of monsoon, surface latent heat flux over the TP gradually reduces. Although latent heat flux still shows the distribution of “east high and west low,” the high value area gradually recedes eastward. There is still a low value area of latent heat flux in the north corresponding to the high value area of sensible heat flux.

As the southwesterly monsoon dissipates, latent heat flux in the TP attenuates sharply in September and October, and the overall magnitude drops to around 100 W m^–2, which is equivalent to that of sensible heat flux. The area of surface latent heat flux in the plateau also tends to shrink southwestwards. The latent heat flux in the western TP is weakened. However, the surface latent heat flux in the plateau region still shows the distribution of “east high and west low.”

In summary, the surface latent heat flux over the TP is weak before the onset of the ASM, and presents a distribution of “east high and west low.” The southeastern, western, and northern parts of TP are all occupied by low values of latent heat flux but high values of sensible heat flux. With the outbreak and development of the ASM, surface latent heat flux in the TP increases rapidly and shows a distribution of “east high and west low.” There is a low value zone of latent heat flux in the northern TP, where a high value zone of sensible heat flux appears. As the ASM weakens, surface latent heat flux in the TP suddenly drops to around 100 W m^–2, basically equivalent to the quantity of sensible heat flux. The surface latent heat flux over the TP still shows a distribution of “east high and west low,” and the low value area of latent heat flux corresponds to the high value area of sensible flux. To be concise, based on the analysis of surface heat fluxes over the TP before and after the ASM onset, the TP surface is mainly dominated by sensible heat flux in spring; the main body of the TP is dominated by latent heat flux during the monsoon period; and the surface sensible and latent heat fluxes in the plateau are basically equivalent in magnitude after the withdrawal of the ASM.

Time series and regression trendlines for averaged surface heat fluxes from March to October over the TP during 1979–2016 are shown in Fig. 3. According to the trendlines in Fig. 3, the surface sensible heat flux in the plateau exhibits a relatively weak downward trend as a whole. In contrast, the surface latent heat flux in the TP increases slightly. These are consistent with the fact revealed by Zhu et al. (2001) that the climate of the TP was significantly warmer and was warmer than that in the surrounding areas at the same latitude.

Figure 3 Temporal evolutions of averaged (March–October) surface sensible (thick solid line) and latent (thin solid line) heat fluxes (W m–2) over the Tibetan Plateau during 1979–2016. The dotted lines denote the trend of sensible heat flux (black) and latent heat flux (grey), respectively.

Figure options

In order to better understand the interannual variation trend of surface heat fluxes over the TP, the MK test is carried out. Figure 4 shows the MK test results for averaged surface heat fluxes (W m^–2) over the TP from March to October during 1979–2016. The black solid line is the forward sequence (UF) line and the gray dotted line is the backward sequence (UB) line. The upper and lower horizontal gray solid lines represent the significance level of 0.05, corresponding to ordinates of the MK critical values. The trend of the sequence can be further investigated by analyzing the statistical sequences UF and UB lines, and the abrupt change time as well as the region can also be clarified. If the UF value is greater than 0, it indicates an upward trend; while if UF is less than 0, it is a downward trend. If they exceed the 0.05 significance level line, a significant increase or decrease occurs. As shown in Fig. 4, the surface sensible heat flux in the TP shows an upward trend from 1979 to 1997, but a downward trend after 1997. From 1982 to 1987, the surface sensible heat flux exceeds the 0.05 significance level line, indicating that the surface sensible heat flux has increased significantly during this period. Contrary to the trend of sensible heat flux, surface latent heat flux over the TP decreases first and then increases. The UF line for surface latent heat flux exceeds the 0.05 significance level line during 1997–2006, indicating that the surface latent heat flux over the TP has increased greatly in these years. Overall, the surface sensible heat flux and latent heat flux over the TP show the opposite trends.

Figure 4 The Mann–Kendall (MK) test statistics for averaged (March–October) surface (a) sensible and (b) latent heat fluxes over the Tibetan Plateau during 1979–2016. The UF (UB) line denotes forward (backward) sequential statistic. The upper and lower horizontal gray solid lines represent the significance level of 0.05.

Figure options

In order to better analyze the variation characteristics of surface heat fluxes with time over the TP, the plateau is divided into two parts, east and west, with 90°E as the boundary. Time series and associated regression trendlines of averaged surface heat fluxes from March to October over the (a) western and (b) eastern TP during 1979–2016 are shown in Fig. 5.

Figure 5 As in Fig. 3, but for (a) western and (b) eastern TP.

Figure options

The surface sensible heat flux in the western TP has shown a slight upward trend during 1979–2016; in contrast, the surface latent heat flux in the western TP has shown a downward trend (Fig. 5a). The change of heat fluxes in the eastern TP is opposite to that in the western TP. The surface sensible heat flux in the eastern TP is decreasing, but the latent heat flux is rising (Fig. 5b). Compared with that in the western TP, a more obvious trend of heat fluxes occurs over the eastern TP, suggesting that the eastern TP has become more humid, similar to the observed climate variation trend over the TP in recent years. Affected by a series of factors such as more arid underlying surface and harsher local climate, the western TP seems to become warmer and drier.

The MK test is carried out for surface heat fluxes in the eastern and western TP (Fig. 6; legends are the same as in Fig. 4). It can be seen that in the western TP, surface sensible heat flux rises first, then declines, and finally rises again. In contrast, surface latent heat flux in the western TP shows a trend of decreasing first, then rising, and finally decreasing. Thus, the trends of the two are opposite. Similarly, surface sensible heat flux in the eastern TP shows a trend of rising first and then decreasing, with the UF line exceeding the 0.05 significance level line after 2005, indicating that the downward trend of surface sensible heat flux over the eastern TP after 2005 is more obvious. The surface latent heat flux in the eastern TP shows a trend of first decreasing and then increasing, with the UF line exceeding the 0.05 significance level line in 1998, indicating that the rising trend of latent heat flux is more obvious since 1998.

Figure 6 The Mann–Kendall (MK) test statistics for averaged (March–October) surface heat fluxes over the Tibetan Plateau (TP) during 1979–2016. (a) Sensible heat flux over western TP; (b) latent heat over western TP; (c) sensible heat flux over eastern TP; and (d) latent heat over eastern TP. The UF (UB) line denotes forward (backward) sequential statistic. The upper and lower horizontal gray solid lines represent the significance level of 0.05.

Figure options

In conclusion, in the western part of the TP, surface sensible heat flux has shown a slight upward trend during 1979–2016, while surface latent heat flux has shown a downward trend. Compared with the western TP, the eastern part of the TP has significantly decreasing surface sensible heat flux, but obviously increasing surface latent heat flux with a clear upward trend.

The monthly mean surface heat fluxes from the ERA-Interim reanalysis data over the TP and its surrounding areas from March to October during 1979–2016 are examined to analyze characteristics of the changes of surface heat fluxes before and after the onset of the South Asian summer monsoon. The following conclusions are obtained.

(1) In the period from March to May before the onset of the ASM, the TP main body is occupied by sensible heat flux, which has a distribution “west high and east low” and tends to decline with time. Compared with sensible heat flux, the latent heat flux is weaker in magnitude and variation range but is overall exhibiting a trend of enhancing. From June to August, with the outbreak of the ASM, the high value area of sensible heat flux in the western TP continues to shrink and the overall magnitude over the plateau continues to decline. In contrast, the latent heat flux gradually increases with an overall distribution of “east high and west low.” From September to October, after the fading of the ASM, surface sensible heat flux in the plateau region still shows the characteristics of “west high and east low,” but the overall magnitude decreases. The latent heat flux gradually recedes eastward with reduced strength, but the strength is basically equivalent to that of the sensible heat flux. This is also consistent with the distribution features of sensible and latent heat fluxes over the TP analyzed by Zhao et al. (2018) using the data from the TIPEX-III.

(2) From 1979 to 2016, surface sensible heat flux over the TP shows a slight downward trend, while surface latent heat flux shows a slight upward trend. Based on the MK test, it is found that surface sensible heat flux in the TP has risen first and then decreased during 1979–2016. In contrast, surface latent heat flux has a tendency to decline first and then rise, which conforms to the observed climate change over TP in recent years.

(3) Analysis of surface heat fluxes in the eastern and western TP reveals that the surface sensible heat flux in the western TP increases slightly, while the latent heat flux shows a downward trend and has a larger variation. Compared with the western TP, the eastern TP has more obvious changes in surface heat fluxes. The sensible heat flux in the eastern TP shows a downward trend, while the latent heat flux presents a relatively obvious upward trend, which might be related to the warming and moistening of the TP in recent years.

(4) The MK test on the surface heat fluxes in the eastern and western parts of the TP shows that sensible heat flux in the western TP has a trend of rising first, then decreasing, and finally rising again. In contrast, latent heat flux is firstly decreased, then increased, and eventually decreased again. However, the UF lines of sensible and latent heat fluxes do not exceed the 0.05 significance level lines overall, which indicates that the overall changes in surface heat fluxes in the western TP are not so significant. Compared with the western TP, the eastern TP has increasing (decreasing) first and then decreasing (increasing) sensible (latent) heat flux, and the UF lines of sensible and latent heat fluxes in the eastern TP have exceeded the 0.05 significance level lines in recent years, indicating that the eastern TP has become more humid (with increased water vapor content), and surface heat fluxes in the eastern TP has obviously changed in recent years. This suggests that the eastern TP might be more closely associated with the climate change.

This study is concentrated on the long-term variation characteristics of the surface heat fluxes over the TP before and after the onset of the South Asian summer monsoon. In the future, the interaction between surface heat fluxes and atmospheric heat sources over the TP will be investigated, through comparisons of the spatiotemporal distributions and variation trends of the surface heating and the heat sources above the TP in association with the development of the ASM.