Study Area

The Ngari region study area lies in the western Tibetan Plateau (Fig. 1). It borders on the Nagqu (Nakchu) region, west of the Tanggula Mountains, and stretches to the western portion of the Himalayas in the west and southwest. It joins the middle section of the Kangdese (Gangdise) Mountains, and ends on the southern side of the Kunlun Mountains in the north. This region has an annual average temperature of 0°C to 2°C, and an average annual rainfall of 73 mm. About 80% of the annual precipitation falls in the summer season from June to September. Rainfall also varies spatially across the region, decreasing gradually from east to west (Ye and Gao, 1979); northwestern Ngari region has an average annual rainfall less than 50 mm.

Figure 1.

Map of the precipitation sampling sites in the western Tibetan Plateau and adjacent regions.

The meteorological stations of Gaize (Gertse, Gertze) and Shiquanhe (Senge Khabab) are located, respectively, in the east and the west of Ngari region. Shiquanhe lies in a narrow river basin. The climate is dry with an annual precipitation of about 75 mm. Gaize lies on the southeastern Changtang (Changthang, Qiangtang) Plateau, with some lakes and marshland in the vicinity. The precipitation at Gaize exceeds that at Shiquanhe, averaging nearly 190 mm per annum.

Sampling and Data Collection

The Gaize meteorological station launched a program to monitor stable isotopes in precipitation in 1998. Sample collection in 1999 was incomplete. The Shiquanhe station collected precipitation samples from 1999 to 2002 (Table 1). Both stations sampled each precipitation event. Rainfall samples were collected and immediately sealed in plastic bottles. Snow and other solid precipitation were collected on clean porcelain plates, put into clean plastic bags, and sealed. After the samples melted at room temperature, they were processed in the same manner as rainfall. All the samples were stored frozen in a cold laboratory until analyzed. The Frontier Observational Research System for Global Change, Japan, made the deuterium measurements, with an analytical precision of 1.0‰. The State Key Laboratory of Cryosphere and Environment, Chinese Academy of Sciences, Lanzhou, China, measured oxygen isotope compositions of all the precipitation samples using a MAT-252 mass spectrometer with a precision of 0.2‰. Both laboratories measured δ¹⁸O in some samples from both sampling stations for comparison and the results agreed quite well. In addition, we use isotopic data from Hetian (Hotan) and New Delhi. These stations were established as part of the International Atomic Energy Agency/World Meteorological Agency (IAEA/WMO) Global Network of Isotopes in Precipitation (GNIP), whose measurements include δ¹⁸O, δD, temperature, and precipitation. The records at Hetian (station number 5182800) are from 1988 to 1992, and at New Delhi (station number 4218200) are from 1961 to 2001 (available at  http://isohis.iaea.org). Precipitation was collected in monthly amalgamated samples, and analyzed at the IAEA laboratory in Vienna, Austria, for both δ¹⁸O and δD, with an analytical precision of 0.1‰ and 1.0‰, respectively (Araguás-Araguás et al., 1998).

Table 1

Characteristics of the sampling sites and of the water sampling.

The precipitation sampling program recorded the duration of each precipitation event, air temperature on the ground (recorded by a bimetallic thermograph), precipitation amount, and other relevant meteorological data (before and after each precipitation event).

Temperature Dependence of δ¹⁸o in Precipitation

Figure 2 shows the relationships between δ¹⁸O in precipitation and air temperature (the “air temperature” is the averaged value of the temperature at the start and end of an event) in recent precipitation events at the Shiquanhe and Gaize meteorological stations. The correlation coefficient between δ¹⁸O in precipitation and temperature at Shiquanhe is 0.43, within a 0.01 confidence limit. The correlation coefficient at Gaize is much poorer, only 0.22, also within a 0.01 confidence limit.

Figure 2.

Relationship between δ¹⁸O and temperature in recent precipitation events at Shiquanhe and Gaize.

We find better correlations between monthly precipitation-weighted means of δ¹⁸O and monthly temperatures (Fig. 3) than those for individual precipitation events (Fig. 2). Results from a linear regression least squares fit using weighted monthly means of the data can be expressed as follows:

(1)

(2)

Figure 3.

The variations of precipitation-weighted monthly δ¹⁸O and monthly mean temperature at four stations (Hetian, Shiquanhe, Gaize, and New Delhi).

To illustrate the spatial variation (north to south) of the δ¹⁸O – T relationship near the western Tibetan Plateau, Figure 3 also shows data from adjacent regions. The Hetian meteorological station lies to the north, and the New Delhi station lies to the south (Fig. 1, Table 1). The correlation between δ¹⁸O in precipitation and air temperature at Hetian is strongest (r = 0.86, P < 0.01). At the other extreme, the oxygen isotopic compositions of precipitation at New Delhi show essentially no correlation with air temperature. Together the data show a decrease from north to south across the western Tibetan Plateau and adjacent regions in the significance of the linear regression least squares fit between δ¹⁸O and T.

Monsoon precipitation patterns can largely explain the spatial variations of the δ¹⁸O – T relationship. With the attenuation of monsoon precipitation from south to north, the correlation of δ¹⁸O – T gradually becomes stronger. The southwest monsoon conditions are very strong at New Delhi, and δ¹⁸O is low during summer monsoon precipitation, while summer temperatures remain high (Fig. 4d). In contrast, Hetian lies to the north of the Plateau, beyond the influence of the monsoon rains. Without the southwest monsoon control, δ¹⁸O in precipitation shows a very strong temperature dependence. The δ¹⁸O values and temperature are both very high in summer and low in winter (Fig. 4a). At Gaize and Shiquanhe, however, the δ¹⁸O values increase from January to June with the gradually increasing air temperature, but then appear to decrease at the end of July or early August until the end of the rainy season (Fig. 4c, 4b). This deviation in the δ¹⁸O – T correlation at the two stations coincides with the southwest monsoon reaching the western Tibetan Plateau for a short period in summer each year.

Figure 4.

Comparison of temperatures and δ¹⁸O trends in precipitation over the annual cycle at four meteorological stations: a–Hetian; b–Shiquanhe; c–Gaize; d–New Delhi. Plots for Hetian and New Delhi show monthly mean values, while for Shiquanhe and Gaize show event values.

Araguás-Araguás et al. (1998) discussed the spatial variations of oxygen isotope responses in southeast Asia (Fig. 5). A curve sketched on Figure 5 separates different regimes of δ¹⁸O response. North of the curve, δ¹⁸O in precipitation shows strong temperature dependence. South of the boundary, however, δ¹⁸O in precipitation is controlled more by the effect of strong monsoon activity. The mechanism of this control is described as a precipitation “amount effect.” Hetian lies in the “temperature effect” region and New Delhi lies in the “amount effect” region, which agrees with the positive correlation between δ¹⁸O in precipitation and temperature at Hetian and lack of correlation between δ¹⁸O in precipitation and temperature at New Delhi. On the other hand, the Ngari region, mid-way between Hetian and New Delhi, lies in the transition belt between “temperature effect” regions in the north and precipitation “amount effect” regions in the south (Fig. 5). The duration of the southwest monsoon activity in the Ngari region is not as long as that at New Delhi; as a result, the δ¹⁸O – T correlation at the Gaize and Shiquanhe stations is better than that at New Delhi. That is, temperature remains the dominant factor that controls the variations of δ¹⁸O in the Nagri region. But, the southwest monsoon precipitation results in the low δ¹⁸O values in late summer, weakening the relationship of δ¹⁸O in precipitation and air temperature (Yu et al., 2004). The δ¹⁸O – T correlation gets stronger with increasing latitude from south to north, corresponding directly with the monsoon impact.

Figure 5.

Areas of Southeast Asia with different δD and δ¹⁸O responses to changes of surface air temperature and amount of precipitation, based on the long-term monthly means of isotopic and meteorological data (adapted from Araguás-Araguás et al., 1998). The sketched curve separates different regimes of δ¹⁸O response.

The weaker δ¹⁸O – T correlation at Gaize than at Shiquanhe (Fig. 2) may correspond to southwest monsoon activity at Gaize being stronger and more frequent than at Shiquanhe. Gaize is closer to the Bay of Bengal, resulting in higher summer precipitation, and the southwest monsoon reaches Gaize earlier. In addition, the vegetation at Gaize is more abundant than that at Shiquanhe and there are small lakes and marshland in the region. Hence, the local evaporation intensity at Gaize is greater than at Shiquanhe. Accordingly, the precipitation resulting from local convection contributes a larger proportion of the rainfall so as to weaken the δ¹⁸O – T relationship. All these factors result in making the correlation between δ¹⁸O in precipitation and air temperature at Gaize weaker than at Shiquanhe.

Local Mwl in Western Tibetan Plateau and Adjacent Regions

Due to parallel fractionation during water cycling, δD and δ¹⁸O relate linearly in precipitation on a global scale. Craig (1961) first summarized and related these data, and defined the meteoric water line (MWL). The global MWL is δD = 8.0 δ¹⁸O + 10 for continental precipitation unaffected by evaporation.

The local meteoric water lines were calculated from individual precipitation events at the four stations in this study. The following regressions are based on samples for δD collected at Shiquanhe in 1999–2002 and at Gaize in 2000–2002. There exists a good linear relation between δD and δ¹⁸O in precipitation at all four stations, as follows:

(3)

(4)

(5)

(6)

The local MWLs show substantial spatial variation. At New Delhi, the regression slope (7.16) is lower than 8 and the intercept (4.14) is far below 10. In contrast, at the other three stations, the slopes and intercepts are close to those for the global MWL. In brief, in the south of the study region, the local MWL is shallower and offset below the global MWL, and in the middle and the north of the study region the local MWLs are close to the global line. In addition, the correlation between δD and δ¹⁸O in precipitation strengthens gradually (the correlation coefficient (r) increased gradually) from south to north across the western Tibetan Plateau and adjacent regions, resulting from the gradually decreasing impact of monsoon precipitation.

Seasonal and Spatial Variations of Deuterium Excess (d) in Precipitation

Kinetic effects influence the effective fractionation factors, and there exists an excess value in the relationship between δD and δ¹⁸O. This deuterium excess (d), defined by Dansgaard (1964) as d = δD − 8.0 δ¹⁸O, is used as an index for nonequilibrium conditions. The d value in precipitation reflects the evaporation conditions at the source of the moisture (Merlivat and Jouzel, 1979; Jouzel and Merlivat, 1984; Armengaud et al., 1998). Precipitation derived from sea surface evaporation at low latitude is characterized by low d values (<10‰). Under dry climate condition, the kinetic fractionation will be intensified during evaporation, and the d value in the subsequent precipitation will be high (>10‰). In contrast, under humid climate condition, the kinetic fractionation in evaporation will be weakened, resulting in low d (<10‰) in the subsequent precipitation (Merlivat and Jouzel, 1979; Clark and Fritz, 1997).

The d values in precipitation from the Shiquanhe and Gaize stations were calculated based on simultaneous measurements of δD and δ¹⁸O. In addition, we use records of d in precipitation at New Delhi and Hetian provided by GNIP. These results show obvious spatial and temporal variations in the western Tibetan Plateau and its surroundings (Fig. 6). The d values from individual precipitation events (crosses) at the four stations vary substantially, unlike those from the precipitation weighted monthly means (open circles). This results from the varying moisture sources and humidity during precipitation events.

Figure 6.

Seasonal variations of deuterium excess (d), in precipitation at four stations in the western Tibetan Plateau and adjacent regions. The crosses and open circles stand for individual data and weighted monthly means, respectively.

Due to the very sparse precipitation in winter, we would mainly compare the d values in summer at the four stations. The weighted average monthly d values at New Delhi are low, except for January and November, especially from May to October. The low d values in summer coincide with monsoon precipitation, representing moisture from surface evaporation of the Indian Ocean in high humidity. Southwest monsoon precipitation, representing moisture from the Indian Ocean, also causes low d values at Gaize in July.

Because southwest monsoon storms in summer at Shiquanhe are not as intensive as at Gaize, the weighted average monthly d values in summer at Shiquanhe are higher than those at Gaize, especially in June and July. Summer precipitation at Hetian displays no corresponding period of low d values, and the d values are the highest of the four stations. This reflects moisture coming from a drier source, beyond the influence of the southwest monsoon. This coincides with the local moisture recycling in dry climate conditions and westerlies precipitation.

From south to north, in summer, the southwest monsoon activity gradually weakens, resulting in decreasing humidity, and at Hetian the moisture origins are controlled by the westerlies and local moisture recycling. On the other hand, the westerlies gradually weaken from north to south through this region, such that they have little impact at New Delhi. At Shiquanhe and Gaize, the westerlies and monsoon interweave, with the southwest monsoon dominating in summer. That is, the western Tibetan Plateau is a transition zone between the regions dominated by the monsoon and the regions dominated by the westerlies. Accordingly, from south to north across our study regions, the summertime averaged d values (weighted annual means) gradually increase with the humidity decreasing (Fig. 7).

Figure 7.

Spatial variations of mean d in summer precipitation (June–August) over the years of observation at four stations across the western Tibetan Plateau and adjacent regions.