3.1. Regionalization of China
Based on the climate, vegetation, and topography, Luo [27
] divided China into 7 regions (Figure 1
a). Following this work, Huang [28
] re-divided China into 3 regions by merging Regions 1, 2, 3, 4, and part of 5 as the monsoon-affected region, leaving the rest as the Tibetan Plateau (Region 6 and part of 5) and the arid region (Region 7). Generally, these two regionalization schemes are used for the discussion of climatic and geographic features in China. Araguas-Araguas et al. [7
] developed a new regionalization scheme based on the moisture sources dominating the precipitation in China (Figure 1
b). There are five major air masses that dominate the pluviometric regime of China [7
]: (1) the polar air mass originating in the Arctic; (2) the westerlies with recycled continental air mass over central Asia; (3) the tropical-maritime air mass originating in the northern Pacific; (4) the equatorial-marine air mass originating in the western equatorial Pacific; and (5) the equatorial-marine air mass originating in the Indian Ocean. Terzer et al. [29
] developed a new model, named the regionalized cluster-based water isotope prediction approach (RCWIP), to predict point and large scale spatial-temporal patterns of the stable precipitation isotopes. In their work, they defined 36 climatic clusters, with 4 of them being related to China, whose representative stations are Chiang Rai, Erenhot, Lhasa, and Shanghai, respectively. Their criterion for the clusters were the differences of the climatic variables of temperature, precipitation amount, and vapor pressure between different stations.
The regionalization by Luo [27
] incorporates geographical features in China, but it disregards the factor of moisture sources. As a result, one area that has a similar moisture source throughout is divided into two regions. For example, Regions 3 and 4 have the same moisture source from the South Pacific. However, in some cases, regions that have different moisture sources are treated as a single region. For instance, Region 7 is affected by both the westerlies and the arctic air masses. The regionalization by Huang [28
] merges all the regions affected by monsoons but overlooks the different geographic features. Although the regionalization by Araguas-Araguas et al. [7
] is based on moisture sources, some regions overlap (Figure 1
). Furthermore, the extent of some moisture sources is still controversial. Tian et al. [12
] found that the northern limit of summer monsoons from the Indian Ocean is in the middle of the Tibetan Plateau, at approximately 34°–35° N, which is farther north than the extent adopted by Araguas-Araguas et al. [7
] and Johnson and Ingram [10
]. Li et al. [30
], Zhou et al. [31
], and Xu et al. [32
] found that the East Asian summer monsoons (EASM) can reach the Qaidam Basin (farther west than 100° E), whereas previous studies indicated that the direct influence of EASMs reach as far west as 100° E in China [7
]. Pang [34
] summarized the moisture sources over Northwest China on the basis of a meta-analysis of water isotopes in the region, demonstrating that the Arctic polar air mass could reach as far as the Junggar Basin and that summer monsoons from both the Indian and Pacific Oceans has little impact.
When comparing and summarizing from the regionalization schemes of Luo [27
], Huang [28
], and Araguas-Araguas et al. [7
], we find that each region could have its own isotopic patterns affected by moisture sources and other factors. This has also been verified by the recent isotopic studies in China that showed the dominating factor affecting isotopes should be moisture sources and then the local geographical meteorological factors [11
]. To understand the temporal and spatial variation of precipitation isotopes, we devise a new regionalization scheme in this work. The primary factor we consider for this scheme is the moisture source. Thus, we divide the northern continental portion of China into Northwest China (Region I: the westerlies domain) and North China (Region II: the arctic domain). Because the South and North Pacific-dominated regions partially overlap owing to their similar moisture sources, we treat Northeast China as a single region (Region III: the northeast domain) due to its climatological and geographical differences from the rest of the region (Region IV: the Pacific domain). As in the previous regionalization, the Tibetan Plateau is treated as a single region (Region V) because the climatology and tectonic features of the Tibetan Plateau are quite different from those of other regions in China. The Tibetan Plateau plays an important role in affecting regional climatology [7
]. This region impacts both the westerlies and the summer monsoons. The high mountains block the westerlies and split the jet stream, which moves to the south and north of the plateau [7
]. The elevated heating by the Tibetan Plateau of the atmosphere plays a fundamental role in the formation and maintenance of the summer circulation, at least over Asia. The onset of the Bay of Bengal monsoon (BOBM) and the EASM can be linked to the thermal and mechanical forces of the Tibetan Plateau [37
Compared with the regionalization by Araguas-Araguas et al. [7
], our regionalization keeps the primary criterion as the moisture sources. However, the improvements can be found in our regionalization by taking into account the geographical and climatic factors. The overlapping regions do not exist as a result of combining the regionalization schemes of Luo [27
], Huang [28
], and Terzer et al. [29
]. The extent of Regions IV and V is ascertained by considering the recent findings of the Indian Ocean and Pacific monsoons mentioned above. The region of the Tibetan Plateau is highlighted in light of its effect on the regional climate.
Collectively, China is divided into 5 regions (Figure 2
). All of the results and discussion regarding precipitation isotopes in China will be based on this regionalization.
3.2. Spatial Distribution of Stable Isotopes and D Excess
A clear pattern of the annual average isotopic distribution can be seen in Figure 3
a,b. The minimum δ18
O with a value of −15.9‰ is at Lhasa station, and the maximum value of −5.7‰ is at Haikou station. The minimum δ2
H is −117.2‰ at Lhasa station, and the maximum is −34.7‰ at Guilin station. Isotopes become depleted from south to north overall, which is consistent with the latitude effect reported globally [1
]. Though a similar depleted trend can be found from east to west, we could not attribute this trend to the continental or altitude effect because precipitation is formed by different moisture sources from East to West China.
The d excess ranges from 4.7‰ at Chengdu station to 14.8‰ at Guilin station (Figure 3
c). Most of the values are close to the global average of 10‰ [1
]. Comparing the spatial distribution of d excess with that of δ18
O (Figure 2
and Figure 3
), we find that in Regions I, III, and V, the contours of δ18
O are intensive, whereas the contours of d excess are scarce; in contrast, in Region IV, the δ18
O values change slightly, whereas the d excess varies as a complex feature. This pattern occurs because in Regions I, III, and V, δ18
O declines gradually along the track of the air masses from the westerlies, the North Pacific, and the Indian Ocean, respectively, whereas in Region IV (the monsoon region), precipitation mainly occurs in summer caused by the monsoons, maintaining consistent δ18
O values [38
]. This pattern fits well with the dominant role of moisture sources in controlling precipitation isotopes.
3.3. Seasonal Variations of Precipitation Isotopes and D Excess
Stations with a full-year record of stable isotopes are selected to show the seasonal variations and their relationship with moisture sources in each region.
(1) Region I
The Wulumuqi and Zhangye stations are used to show the seasonal variations in precipitation isotopes (Figure 4
). The δ18
O values range from −21.0‰ to −3.7‰, increase gradually from January to July, and then decrease from August to December. The d excess in this region has the largest amplitude (32.0‰) among all the regions. In summer, the d excess is lower than the global average of 10, whereas in winter, it is greater than 10. As noted above, precipitation in this region is mainly formed by moisture from the westerlies. Measuring the isotopic composition of ice core samples taken from a high-altitude glacier in the Tianshan Mountains and the isotopic composition of snow and firn taken from a high mountain of the Siberian Altay, Kreutz et al. [40
] and Aizen et al. [41
] suggested that seasonal changes in moisture sources and recycling in the Caspian Sea region during transport from the Atlantic to the sampling site in the Tianshan Mountains are responsible for the variability in d excess. Pang et al. [13
] and Kong et al. [15
] further noted that moisture recycling during summer and autumn amplified the seasonal change in d excess in this region.
(2) Region II
The Altay and Baotou stations are selected to show the seasonal variations in precipitation isotopes (Figure 4
). The isotopic change in Region II is very similar to that in Region I. Nevertheless, the d excess is very different—the d excess has a clear seasonal pattern in Region I, which does not exist in Region II. Comparing the d excess at Wulumuqi and Altay stations, we find they are similar in summer (8.4‰ at Wulumuqi and 7.8‰ at Altay) and different in winter (20.3‰ at Wulumuqi and 9.8‰ at Altay). However, the d excess at Baotou station is closer to that of Wulumuqi in winter, which are both larger than 14.0‰. This finding may show that in winter the only moisture source for Altay is the Arctic Ocean, whereas Baotou has an additional moisture source from the westerlies.
(3) Region III
The Qiqihar station is selected to show the seasonal variations in precipitation isotopes (Figure 4
). The δ18
O values in this region are lower than in other regions. The δ18
O values (−9.6‰) are higher in summer and lower (−25.1‰) in winter. In contrast, the d excess values are lower (3.9) in summer and higher (9.3) in winter. The temporal variations of δ18
O and d excess values in this region are similar to that in Region I. However, Figure 2
shows that moisture sources in Region III are quite different from those in Region I. In Region III, the moisture source in summer is the Pacific Ocean, which has lower d excess and high δ18
O values; however, the source in winter is the Arctic Ocean, which has higher d excess and lower δ18
O values [35
]. This difference in moisture sources leads to the temporal variation of precipitation isotopes and d excess.
(4) Region IV
To track the onset of the monsoon season, Region IV is further divided into the sub-regions of South China (SC), Central China (CC), and North China (NC) (Figure 4
The Haikou, Hong Kong, Guilin, and Liuzhou stations are selected to show the seasonal variations in precipitation isotopes in Region SC. Figure 4
shows that the δ18
O values begin to decrease in April, which is the onset of the monsoons [36
]. The observed decreasing δ18
O values could be explained by the precipitation amount effect. The d excess values change very little throughout the year, exhibiting features of a marine moisture source.
The Nanjing, Fuzhou, Guiyang, and Wuhan stations are selected to show the seasonal variations in precipitation isotopes in Region CC. The δ18O values in CC begin to decrease in May, which is a month later than in Region SC. The d excess in winter (15.0‰) is higher than in summer (8.0‰), which is different from Region SC. This finding implies that precipitation isotopes in winter have been affected by moisture from the north.
The Shijiazhuang, Tianjin, Xian, and Zhengzhou stations are selected to show the seasonal variations in precipitation isotopes in Region NC. Both the δ18
O and d excess values in this region vary more significantly than in Regions SC and CC. Obviously, the values are affected by moisture from the north and northwest. However, in summer, the moisture is mainly derived from the Pacific Ocean [38
]. The δ18
O values begin to decrease in May. Compared with the δ18
O values in Regions SC and CC, the decreasing δ18
O values reveal the onset of the monsoon season from April to June in different parts of Region IV.
(5) Region V
The Lhasa, Nyalam, and Yushu stations are selected to show the seasonal variations in precipitation isotopes (Figure 4
). The seasonal variation in δ18
O values in this region is similar to that in Region SC. However, the δ18
O values are lower than in Region SC because of the higher altitude in Region V. The d excess values in this region are high in winter and low in summer. It should be noted that the d excess values at Yushu station are different from those at Lhasa and Nyalam stations. In summer, the d excess at Yushu station (14‰) is much larger than that at Lhasa and Nyalam stations (10‰ and 5‰) because the moisture is mainly from the Indian Ocean at Lhasa and Nyalam stations in summer, whereas Yushu station is less influenced by Indian Ocean moisture [12
In summary, it is found that in each region, the moisture source plays a primary role in controlling the precipitation isotopes. What’s more, with the long-term observation at the GNIP stations (i.e. Hongkong, Wulumuqi), the seasonality remains robust. Compared with the existing regionalization scheme, such a new regionalization scheme helps to identify the factors affecting isotopic variability. However, when using the regionalization scheme, it should be noted that the similarity of isotopic seasonality at each station is incorporated as a factor for the regionalization scheme, but is not the only factor. For instance, the δ18O values are higher in summer and lower in winter at both Wulumuqi and Qiqihar stations, but we divide them into different regions due to different moisture sources and geographical factors. Even in one region, the same moisture source might play different roles at different stations, especially for the stations located at the boundary of two regions, such as Baotou in Region II, Shijiazhuang in Region IV, and Yushu in Region V. For example, both stations of Altay and Baotou are located in Region II of the Artic Region, but in winter the Artic moisture source affects the precipitation isotopes at Altay station more significantly than that at the Baotou station, while in summer the effects at both stations are similar. Therefore, the factors influencing the precipitation isotopes are complex. The regionalization scheme helps to illustrate the impact of moisture sources on the precipitation isotopes, but one has to keep in mind that other factors, including local metrological and geographical conditions, could also modify them.
3.4. Regional Meteoric Water Line (RMWL)
The regional δ2
O relationship, which is well known as the RMWL, provides a reference for interpreting the provenance of surface water and groundwater. The East Asia meteoric water line (EAMWL) of δ2
H = 7.92δ18
O + 9.2 was obtained by Araguas-Araguas et al. [7
]. This line is indistinguishable from the GMWL of δ2
H = 8δ18
O + 10. For regions where isotopic data are unavailable, the EAMWL is very useful as a reference for groundwater studies; however, it cannot represent the features of local precipitation for a given region because precipitation isotopes at each region could be modified by local climatic factors, including the origin of the vapor mass, sub-cloud evaporation during rainfall, and the seasonality of precipitation [43
]. Thus, in this case, the EAMWL cannot define the groundwater input. Taking these local factors into consideration, we present the meteoric water lines based on the monthly isotopic data for each region (Figure 2
) with a dominant single moisture source and similar climatological characteristics (Table 2
The slopes of RMWLs for Regions I–IV, which have an average of 7.19 with a deviation of 0.52, are slightly less than the slope (7.92) of EAMWL and the slope (8) of GMWL, while the slope of RMWL for Region V (the Tibetan Plateau) is 8.41, which is larger than all the other slopes (Table 2
). It is known that the sub-cloud evaporation decreases the slope of the meteoric water line, whereas moisture recycling increases it [13
]. Although the seasonality of precipitation could also modify the slope of RMWL, there are no significant differences of precipitation patterns between Region V and the other regions. Thus, we deduce that the Tibetan Plateau region (Region V) is affected by moisture recycling, whereas other regions are affected by sub-cloud evaporation. This conclusion maintains consistency with previous studies in the Tibetan Plateau, Northwest, North, South, and Eastern China [15
3.5. Spatial Extent of the Isotope–Climate Relationship
Given the parallel behavior of δ18
O and δ2
H, we use the δ18
O–temperature and δ18
O–precipitation amount relationships to reveal the effect of climate on precipitation isotopes. Figure 5
shows the regional variation in correlation coefficients between δ18
O, temperature, and precipitation amount. According to the available data, the correlation is significant when the correlation coefficient is larger than 0.3 (pale yellow region in Figure 5
a and dark blue region in Figure 5
b); otherwise, it is not significant.
a demonstrates that the temperature effect is significant in Regions I, II, and III, whereas in the other regions, it is not significant. Rozanski et al. [21
] claimed that the temperature effect was mainly observed in regions of middle and high latitudes, which is consistent with our findings. The gradient of the temperature effect in Region I is identified as 0.13–0.68‰/°C (Figure 6
a). The westerlies are the dominant moisture source in Region I [13
]. In the context of an individual moisture source in the region, a shift in moisture source to the North Atlantic Ocean and the alternation of trajectory in summer and winter are considered as the factors leading to enriched isotopes in summer and depleted isotopes in winter [12
]. The temperature effect in Region III is reported as increasing in significance from south to north [39
]. The moisture in Region III is controlled by the westerlies and polar air masses in winter and by the westerlies and the Pacific Ocean in summer (Figure 1
and Figure 2
). In winter, when temperatures are low, isotopes incorporated in both westerly and polar air masses are depleted; in contrast, in summer, when temperatures are high, precipitation isotopes from the westerlies and Pacific Ocean are enriched. Therefore, the temperature effect in Region III is significant. From south to north in Region III, the difference in moisture sources becomes increasingly prominent, which leads to a more significant temperature effect. Thus, we can draw the conclusion that the temperature effect in Region I is caused by the seasonal shift of the westerlies, whereas in Region III, the temperature effect is mainly attributed to the seasonal differences in moisture sources.
The correlation coefficients between stable isotopes and precipitation amount demonstrate the precipitation amount effect (Figure 5
b). In total, the correlation between isotopes and precipitation amount is not as significant as that between isotopes and temperature. The most significant parts are located in Southwest and Northeast China, but Figure 6
b illustrates that in Southwest China, the slope is negative (anti-correlation), whereas in Northeast China, it is positive. Precipitation events in Northeast China occur mainly in summer, when temperatures are high, whereas in winter, both precipitation and temperature are low. Given the significant temperature effect in Northeast China, it is reasonable to consider that precipitation amount has little impact on precipitation isotopes in this region. A significant precipitation amount effect was observed only in Southwest China and in some regions along the southeast coast. This finding implies that precipitation amount has a stronger effect in monsoon-controlled South China. Johnson and Ingram [10
] noted that in the monsoon-affected regions, multi-regression analysis may be particularly useful for capturing the effect of precipitation amount on isotopes. Considering the temporal and spatial variation of monsoon intensity, moisture source analysis is indispensable in addressing monsoon isotopic data.