The close relations of glacier MB with precipitation and surface air temperature are confirmed in many previous works3, 4, so we firstly investigate above relations over the Pamir Plateau in various season. As seen by table 1, it is obvious the glacier MB correlates well only with summer precipitation and T2m, which agree with previous conclusion19, 21. Then in following study we only reveal how the PM influences the summer precipitation and T2mover the Pamir Plateau.
Observation suggest that the glacier MB has a sensitive response to the height of 0℃ level, its rising and declining well are linked to the warming and cooling of air temperature29, 30, which plays an important role in causing the shrinking and expansion of glacier. Figure 2a displays the interannual variation of the glacier MB and the height of 0℃ level. It is clear that the glacier MB shows a strong correlation and height of 0℃ level with a correlation coefficient of -0.52over the99% significance level, which confirms that the summer height of 0℃ level has a quite close relation with the glacier MB both at interannual and decadal time scales. Meanwhile, the summer PMI correlates well with the glacier MB, their correlation coefficient is 0.61, which is over the 99% significance level (Fig. 2b).Then whether the summer PMI is linked to the glacier MB via the height of 0℃ level? Fig. 2c answers above question, it is clear that the summer PMI also shows strong correlation of -0.75 with glacier MB, which is over the 99% significance level.
How does summer PM influence the height of 0℃ level? Fig. 3a shows the regressions of summer temperature vertical structure (averaged for 50–80°E) against the PMI. During the positive PMI years the air temperatures become cooler in middle to upper troposphere from 600 to 300hPa between 35–45°N.Whether the cooling of middle to upper tropospheric temperatures causes the decline of height of 0℃ level? Previous observation has also indicated the decline of height of 0℃ level surrounding the Pamir Plateau31. Figure 3b displays the regressions of averaged temperature from 600 hPa to 300 hPa against the PMI. It is clear the PM can significantly influence the temperature in middle latitude along 40°N from 50–80° E. When the summer PM is strong, middle to upper tropospheric cooling will occur over the domain of 35–45°N, 50–80° E. In order to reveal the relationships between the middle to upper tropospheric temperature with the MB, height of 0℃ level and other meteorological factors, the middle-upper troposphere temperature index (MUTTI) is defined by the normalized air temperature in the middle-upper troposphere (600–300hPa) regionally averaged over the domain of 35–45°N, 50–80°Eduring the period of1980–2012.
As seen by Fig. 4a, the MUTTI shows a strong correlation and the glacier MB during 1980–2012 with a correlation coefficient of -0.56, which is over the 99% significance level. Meanwhile, the MUTTI correlates well with the height of 0℃ level (Fig. 4b), and the correlation coefficient is 075, which confirm our hypothesis in Fig. 3a. Furthermore, Fig. 4c& Fig. 4d display the interannual variations of the MUTTI, T2m and precipitation during the period of 1980–2012. It is obvious that the MUTTI is both well related to the T2mand precipitation, and their correlation coefficients are 0.66 and − 0.57, which are both over 99%significance level. Previous studies indicated that summer warming amplitude of T2m over the Tibetan Plateau could reach 0.69℃ per decade during the period of 1987-200732, compared to the same period, we take Tashkurghan station (37.78°N, 75.23°E) located in the Pamir Plateau as an example, the warming amplitude of T2m is only 0.36 ℃ per decade(data not shown), which is obviously slower than one over the Tibetan Plateau. The above results confirm that the cooling of middle to upper (600 − 300 hPa) tropospheric temperatures results in the decline of height of 0℃ level, which slows down the warming amplitude of T2mover the Pamir Plateau.
Then whether the cooling of middle to upper (600 − 300 hPa) tropospheric temperatures also can cause more precipitation over the Pamir Plateau? As seen by Fig. 5a, during the negative MUTTI years, there is an anomalous cyclone over the central Asia, and it plays an important role in occurring precipitation over the Pamir Plateau33, which contributes a good dynamic condition of large scale circulation for precipitation. Meanwhile, there is an anomalous anticyclone over the Arabian Sea and an anomalous cyclone over central Asia (Fig. 5b), respectively. The former (anticyclone) strengthens northward transport of water vapor from the Arabian Sea along 60°E, where the terrain height is below 1500 m34, and the water vapor from the Arabian Sea can be transported to middle latitudes (30–35°N). Then the latter (cyclone) continues to transport the water vapor clustered in middle latitudes to central Asia35. In summer, this is an important water vapor transport path associated with summer precipitation over the Pamir Plateau, which is different from the transport by westerly. So the cooling of middle to upper (600 − 300 hPa) tropospheric temperatures also can contribute to a good water vapor condition.
Based on above analysis, the cooling of middle to upper (600 − 300 hPa) tropospheric temperatures over central Asia plays an important role in influencing the air temperature and precipitation over the Pamir Plateau, which links the PM and glacier change. So a very important question is how the PM causes the cooling of mid-upper (600 − 300 hPa) tropospheric temperatures over central Asia? According to Fig. 6a, the PMI experiences an opposite of interannual variation and the MUTTI during 1980–2012 with a correlation coefficient of -0.63, which is over 99%significance level. When the summer PMI in positive years (Fig. 6b), the anomalous cyclone dominates the surface layer of Tibetan Plateau and results in more precipitation. More precipitation corresponded to stronger heating of latent flux, which plays an important heating role over the Tibetan Plateau in summer36, 37. An anomalous anticyclone in the upper troposphere over the northwest flank of the PM region would response to the strong heating of latent flux38, which is well matched by Fig. 6b. The anomalous north winds of west part of the anticyclone strengthen the cold advection transport from high latitudes to central Asia and cause the cooling of middle to upper tropospheric temperatures.