Abstract

In this study, a Lagrangian particle dispersion model, Flexible Particle (FLEXPART), is employed to simulate the trajectories of global air parcels during 2000–09 with the purpose of revealing the moisture sources of the semiarid grasslands of China, especially on precipitation days. Based on land-cover and precipitation data, two areas of semiarid grasslands are identified: one in North China and one in the Tibetan Plateau. Using the FLEXPART simulation results, air parcels reaching these two target regions are traced back for 10 days to examine their temporal variations in position (longitude, latitude, and altitude) and specific humidity. The moisture sources of these semiarid grasslands are discussed for different precipitation categories. Moreover, the contributions of different moisture sources to the precipitation in the target regions are computed and compared. The results indicate that the moisture released in the target regions is substantially from the Eurasian continent, in both summer and winter. During May–September, the southern and eastern adjacent land areas seem to be the main moisture sources of rainfall in the grasslands of North China, while the Eurasian continent on the north and west tends to be the predominant contributor to the rainfall over the grasslands of the eastern Tibetan Plateau. During October–April, moistures released in both target regions principally originate from the Eurasian continent on the north and west. Overall, although the moisture uptake over oceanic sources is also considerable, most released moisture over the target regions is from the Eurasian continent throughout the year, while little of the contribution of oceanic sources is due to great loss of moisture en route.

1. Introduction

Precipitation in China is significantly related to the atmospheric moisture transport over East Asia. An increase in atmospheric precipitable water is associated with an increase of precipitation in most regions of China and vice versa (Zhai and Eskridge 1997). The climate in China is principally dominated by East Asian monsoons. In summer, three branches characterize the atmospheric moisture transport over China (Simmonds et al. 1999). The robust southwesterly water vapor transport by Indian monsoon, which carries moisture from the Arabian Sea and the Bay of Bengal, and the southerly transport by East Asian summer monsoon, which carries moisture from the South China Sea and tropical Pacific Ocean along the west margin of the western Pacific subtropical high, bring abundant water vapor into southern, eastern, and central China. The westerly water vapor transport by midlatitude westerlies is relatively weak and mainly exerts impact on northern China. In winter, because of the significant influence of the Siberian high, northwesterly cold–dry airflow prevails over most parts of China, carrying moisture from the Eurasian continent (Zhang et al. 1997; Gong et al. 2001; Wu and Wang 2002; Zhou 2011). On the other hand, the northwestern part of China is located in the innermost center of the Eurasian continent, where the summer monsoon as well as its associated moisture transport can hardly reach into, thus suffering a warm and arid climate (Shi et al. 2007).

The semiarid area of China is located in the transition zone between the southeastern monsoonal regions and the interior arid regions, with the annual precipitation generally ranging from 200 to 400 mm. The ground in this location is mainly covered by grasses or shrubs and is quite vulnerable to climate change (Chuluun and Ojima 2002; Christensen et al. 2004; Gong et al. 2004; Yu et al. 2013).

A decreasing trend in precipitation that began in the 1960s has been found in the semiarid regions of North China (Yatagai and Yasunari 1995; Wang 2001; Liu et al. 2005; Huang et al. 2011). A number of studies have been conducted to determine how the precipitation in the semiarid regions of China is related to Asian monsoons (Yatagai and Yasunari 1995), the El Niño–Southern Oscillation (ENSO) (Wang and Li 1990; Simmonds et al. 1996), Arctic sea ice (Li 1995; Liu et al. 2012; Li and Wang 2013), circumglobal teleconnection (Ding and Wang 2005; Huang et al. 2011), and so on. Speculatively, the declining of Arctic sea ice in recent decades and its atmospheric impacts (Screen et al. 2013a,b) may have an implication for the decreasing precipitation in the semiarid regions of China. However, there are few works concentrating on the critical issue of where the atmospheric moisture over the semiarid regions of China originates from. Clarifying this problem would further the understanding of the semiarid climate in China and improve related predictions. Trenberth (1999) estimated the roles of moisture advection and local evaporation in atmospheric moisture recycling and found that the global annual mean recycling, which refers to the contribution of local evaporation to local precipitation, which for 500-km scales is 9.6%. This finding demonstrates the importance of water vapor transport for global precipitation. Numaguti (1999) examined the origin and transport processes of atmospheric water using an atmospheric general circulation model. The results suggest that most of the winter precipitation over the Eurasian continent is supplied by evaporation from oceans and that most summer precipitation is supplied by evaporation from the continental surface. Nevertheless, the question of whether the precipitated water in the semiarid regions of China originates from distinct oceanic and continental areas in winter and summer needs to be more specifically examined.

There have been numerous studies focusing on atmospheric moisture transport and its relationship with the precipitation over China (e.g., Zhai and Eskridge 1997; Simmonds et al. 1999; Sun et al. 2011; Wang and Chen 2012; Wei et al. 2012; Sun and Wang 2013): most of which used the conventional Eulerian method that diagnoses the temporal variations of meteorological elements over a fixed region. A main disadvantage of the Eulerian method in determining the moisture sources of precipitation is that it cannot trace the air from possible moisture source regions to the target precipitation region or backtrack in the opposite direction, making it difficult to reconstruct a reasonable “source–receptor” relationship of atmospheric water. An alternative method is to use a Lagrangian numerical model to simulate the trajectories of specific air parcels and their variation in position and specific humidity with time. The Flexible Particle (FLEXPART) dispersion model is a sophisticated Lagrangian particle dispersion model (Stohl et al. 2005) that was originally designed to calculate the long-range and mesoscale dispersion of air pollutants from point sources (Stohl et al. 1998, 2002). This model has been extensively used in recent years to study the global water cycle (Stohl and James 2004, 2005; Stohl et al. 2008; Sodemann and Stohl 2009; Gimeno et al. 2010; Viste and Sorteberg 2013). Stohl and James (2004) estimated the global distribution of moisture flux results from the FLEXPART model and found that its accuracy matched that obtained with the Eulerian method, indicating a good performance in simulating the global atmospheric water cycle. Recently, Drumond et al. (2011) surveyed the moisture sources of different regions in China based on FLEXPART results. Gimeno et al. (2013) examined the impacts of intensification of oceanic moisture sources on continental precipitation. Chen et al. (2012, 2013) identified the moisture sources of the Tibetan Plateau (TP) and the Yangtze River basin and quantitatively compared the contributions of different source regions by estimating the moisture uptake of backtracked air parcels over each source region. However, most studies concerning China among the above-mentioned works focused on the climatology of the atmospheric water cycle during boreal summer, whereas our concern is where the moisture comes from on precipitation days of summer and winter, which may be not exactly the same as climatology. Moreover, even if the air parcels absorbed a vast amount of moisture when passing over a source region, there is the possibility that only a small amount of moisture would remain when these air parcels enter the target region. This phenomenon could occur for several reasons en route, such as terrain obstruction, precipitation caused by dynamically forced ascending motion and deep convection, and interaction among air parcels of different physical properties.

In view of the abovementioned considerations, this study attempts to clarify three questions: 1) From which regions does the atmospheric moisture over the semiarid grasslands of China originate? 2) Does the atmospheric moisture over the semiarid grasslands of China come from different sources in winter and summer? 3) What is the contribution of each identified source to the precipitated water over the semiarid grasslands? To address these questions, FLEXPART was utilized to backtrack air parcels entering the semiarid grasslands in China during 2000–09. Although the simulating period is of a relative short length, which may misrepresent certain features of climate to some extent because of the low-frequency variability of climate system, it is sufficient to reveal the basic characteristics of hydrological cycle over East Asia. The model was driven by the National Centers for Environmental Prediction–Climate Forecast System Reanalysis (NCEP–CFSR) 6-hourly forecast data (Saha et al. 2010). The parcel trajectories and moisture sources for different precipitation conditions were approached separately.

The outline of this paper is as follows: Section 2 describes the FLEXPART model and the methodology used in this study. The sources and routes of moisture transported to the semiarid grassland regions of China under different precipitation conditions are identified in section 3. The contributions of potential moisture source regions to the total moisture released in the target regions are quantitatively computed and compared in section 4. Section 5 provides conclusions and discussions.

2. Model and methodology

a. Model description

The FLEXPART Lagrangian particle dispersion model is a comprehensive tool for studying atmospheric transport (Stohl et al. 2005) and has now been updated to version 9.02 (http://www.flexpart.eu/). FLEXPART computes the trajectories of a large number of particles that represent air parcels, and it traces their variations in three-dimensional position, temperature, specific humidity, air density, and so on. FLEXPART can be set up in a “domain filling” mode, wherein the entire atmosphere is represented by particles of equal mass. The particles are distributed in the model domain proportionally to the air density, and they move freely in the atmosphere. Moreover, FLEXPART can simulate forward in time to trace particles from their sources or backward in time to backtrack particles from given receptors. The input data of FLEXPART can be gridded data from the European Centre for Medium-Range Weather Forecasts (ECMWF) or from the NCEP.

In the present study, FLEXPART was run globally forward in time for the period 2000–09 using the domain-filling mode, with a total of 1 million particles released. The input data were derived from NCEP–CFSR 6-hourly products, including land cover, temperature, relative humidity, and wind, in 42 levels with a resolution of 0.5° × 0.5° (Saha et al. 2010). The model output comprises the identity number, three-dimensional position (latitude, longitude, and altitude), temperature, specific humidity, air density, and mass of each particle, recorded at 6-h intervals. The unit of altitude is meters above ground level (AGL). Air parcels residing over the target domains were backtracked for 10 days, which is the average residence time of water vapor in the atmosphere (Trenberth 1998; Numaguti 1999; Trenberth 1999), to investigate their behavior in moisture uptake and release.

One of the most important issues concerning Lagrangian methods is the accuracy of trajectories, which depends on the calculation schemes. The numerical accuracy of FLEXPART has been improved to the second order by Petterssen’s scheme (Petterssen 1940) since its version 5.0, as a correction of the former “zero acceleration” scheme, which is accurate to the first order. In view that truncation error is a crucial error source for the computation of trajectories (Stohl 1998), Petterssen’s scheme should have greatly improved the numerical accuracy of trajectories. The importance of higher-order schemes to accuracy has been widely realized. In particular, Barras and Simmonds (2009) used a fourth-order Runge–Kutta method to calculate the backward trajectories, acquiring results of superior accuracy. Moreover, it should be noted that the calculation errors of trajectories increases as tracing time extends. The 10-day backtrace period in this study can be regarded as over which the trajectories are of acceptable accuracy (Stohl 1998).

b. Methodology

1) Target domains

A precise classification of land covers requires reliable and high-resolution data on vegetative cover. Hansen et al. (2000) analyzed imagery from the Advanced Very High Resolution Radiometer (AVHRR) satellites between 1981 and 1994 and distinguished 14 land-cover classes. This product is available at three spatial scales at a 1°, 8-km, and 1-km pixel resolution, thus providing a powerful tool for global climate change research.

The 1° × 1° AVHRR land-cover dataset was employed accompanied with monthly precipitation data derived from the Global Precipitation Climatology Project (GPCP) (Adler et al. 2003) to identify the semiarid grasslands in China. As shown in Fig. 1, the regions of annual precipitation ranging from 200 to 400 mm, which could be approximately regarded as the semiarid area of China, form a narrow northeast–southwest zone, which is mainly covered by grassland and shrubs. This sinuous belt stretches across eastern Inner Mongolia, the Loess Plateau, and the eastern TP. The grasslands in North China (NC) are principally characterized by typical steppe and desert steppe, while the eastern TP features high–cold steppe and meadow steppe because of its unique topography (Fig. 2). Further details about the vegetation distribution over China can be found in Hou (1983) and Liu et al. (2003).

Fig. 1.

Land cover (colors as follows: 0 is water; 1 is broadleaf evergreen forest; 2 is coniferous evergreen forest and woodland; 3 is high-latitude deciduous forest and woodland; 4 is tundra; 5 is mixed coniferous forest and woodland; 6 is wooded grassland; 7 is grassland; 8 is bare ground; 9 is shrubs and bare ground; 10 is cultivated crops; and 11 is broadleaf deciduous forest and woodland) and annual mean precipitation (contours; unit: mm) during 1981–94.

Fig. 1.

Land cover (colors as follows: 0 is water; 1 is broadleaf evergreen forest; 2 is coniferous evergreen forest and woodland; 3 is high-latitude deciduous forest and woodland; 4 is tundra; 5 is mixed coniferous forest and woodland; 6 is wooded grassland; 7 is grassland; 8 is bare ground; 9 is shrubs and bare ground; 10 is cultivated crops; and 11 is broadleaf deciduous forest and woodland) and annual mean precipitation (contours; unit: mm) during 1981–94.

Fig. 2.

Topography (colors; unit: m) and gauge stations of the semiarid grasslands in NC (in red) and the TP (in blue). The white rectangles are the D1 and D2 target regions for particle backtracking.

Fig. 2.

Topography (colors; unit: m) and gauge stations of the semiarid grasslands in NC (in red) and the TP (in blue). The white rectangles are the D1 and D2 target regions for particle backtracking.

Considering that the grasslands in NC and the TP are subject to different topography and climate regimes, the semiarid grasslands in China are separated into two target regions, as shown in Fig. 2. Daily records of 82 and 75 gauge stations that cover the semiarid grasslands in NC and the TP, respectively, were derived from the China Meteorology Administration observation archives (the red and blue dots in Fig. 2), which contain daily records of 756 gauge stations in China from 1951 to 2009. These records were used to identify different precipitation categories. Some of these employed stations are located in regions with an annual precipitation greater than 400 mm or lower than 200 mm but covered by grassland, as this study mainly focuses on the grassland zone. However, most of the stations have an annual precipitation ranging from 200 to 400 mm. Therefore, these stations should be representative of the semiarid grasslands in NC and the TP. Furthermore, to facilitate the backtracking of air parcels over the two target regions, two rectangular domains were designated: domain 1 (D1; 105°–115°E, 35°–45°N) and domain 2 (D2; 90°–100°E, 30°–40°N) for the grasslands in NC and the TP, respectively (shown in Fig. 2).

2) Three precipitation categories

Generally, a large-scale synoptic system has a horizontal length of 1000 km or more, corresponding to a range of a dozen degrees of longitude and/or latitude or more, whereas the horizontal dimensions of a middle- or small-scale system are usually below several hundred kilometers, corresponding to a range within several degrees of longitude and/or latitude. To fully understand the moisture sources of precipitation of different scales in the semiarid grasslands, the atmospheric water cycle associated with three categories of precipitation are analyzed and compared. Because the stations in the two target grassland areas both span approximately 20° of longitude and more than 10° of latitude (Fig. 2), the investigated three precipitation categories are defined as follows: 1) no precipitation (N), where none of the gauge stations in either target region has precipitation; 2) mid- and small-scale precipitation (MS), where less than 50% (but >0%) of the gauge stations in either target region has a daily precipitation greater than 0.1 mm; and 3) large-scale precipitation (L), where more than 50% of the gauge stations in either target region have a daily precipitation greater than 0.1 mm. Although criteria 2 and 3 are not very rigorous, they are adequate for reasonably identifying typical MS and L cases, considering their agreement with large- and mid–small-scale systems in horizontal ranges, as mentioned above. The statistical results regarding annual mean days, the precipitation amount and its percentage relative to the total seasonal precipitation of these three cases during 1979–2009 are shown in Tables 1 and 2 for the warm (May–September) and cold (October–April) seasons, respectively.

Table 1.

Annual mean days, precipitation amount, and the percentage of contribution to the seasonal total precipitation of N, MS, and L events during the warm season (May–September) of 1979–2009.

Annual mean days, precipitation amount, and the percentage of contribution to the seasonal total precipitation of N, MS, and L events during the warm season (May–September) of 1979–2009.
Annual mean days, precipitation amount, and the percentage of contribution to the seasonal total precipitation of N, MS, and L events during the warm season (May–September) of 1979–2009.
Table 2.

Annual mean days, precipitation amount, and the percentage of contribution to the seasonal total precipitation of N, MS, and L events during the cold season (October–April) of 1979–2009.

Annual mean days, precipitation amount, and the percentage of contribution to the seasonal total precipitation of N, MS, and L events during the cold season (October–April) of 1979–2009.
Annual mean days, precipitation amount, and the percentage of contribution to the seasonal total precipitation of N, MS, and L events during the cold season (October–April) of 1979–2009.

Table 1 indicates that the MS days account for a predominant part of the warm season of the grasslands in NC. With respect to precipitation, however, the L and MS precipitation both contribute nearly half to the total, being slightly higher for L. For the grasslands in the eastern TP, although there are less L days than MS days, the contribution of the L precipitation to the total precipitation during the warm season accounts for 61.9%, much greater than that of the MS precipitation. Thus, the L precipitation seems to be the main contributor to the summer precipitation of the semiarid grasslands in both NC and the TP. In contrast, during the cold season, the MS cases significantly dominate in both the number of days and the precipitation amount (Table 2), accounting for 60.8% and 90.7% of the winter precipitation over the grasslands in NC and the eastern TP, respectively.

Hence, the moisture sources of the L precipitation in the warm season and those of the MS precipitation in the cold season will be the emphasis of following discussion.

3) Identification of source regions

The Eulerian atmospheric water budget equation for an atmospheric column of unit area can be written as

 
formula

where is specific humidity, is evaporation, is precipitation, is the 3D velocity, and is the gravitational acceleration. Given that

 
formula

inserting Eq. (2) into Eq. (1), we can get

 
formula

The terms on the left- and right-hand sides can be cancelled. Further, considering that in pressure coordinates,

 
formula

which is the continuity equation in pressure coordinates, Eq. (3) finally becomes

 
formula

Equation (5) can be regarded as a Lagrangian formula of the atmospheric water budget, wherein is the surface freshwater flux. From FLEXPART output, changes in with time could be integrated to diagnose .

One of the advantages of the Lagrangian formula [Eq. (5)] over the Eularian formula [Eq. (1)] lies in that it interprets the atmospheric water budget in a concise way, without considering the role of convergences. Moreover, the Lagrangian method can also track forward or backward in time by evaluating the changes in along the trajectories of selected particles (Stohl and James 2005).

It should be noted is that the air parcels traced back over a certain area do not necessarily represent the whole air column over this region; however, we still use to denote the net uptake or release of moisture by these air parcels. In addition, as mentioned in section 1, even if a region has a high positive value, indicative of a strong moisture source, the target-bound air parcels passing over this region may not carry a massive amount of moisture into the target region. A number of studies have realized this point and quantified the contribution of different sources by backtracking the variation in specific humidify of air parcels (Stohl and James 2004, 2005; Sodemann et al. 2008; Chen et al. 2012, 2013). This problem will be specifically discussed in section 4 with regards to quantifying the contribution of different moisture source regions.

3. Moisture sources and sinks

The climate of East Asia is profoundly influenced by the East Asian summer and winter monsoons, which mainly occur in May–September (Ding and Chan 2005) and October–April (Zhang et al. 1997), respectively. Although the semiarid regions are not as strongly influenced by the monsoons as eastern China, their moisture-supplying regime may differ noticeably between the warm and cold seasons (Numaguti 1999) due to the distinct structures of the atmospheric flow field. Henceforth, the moisture sources of the semiarid grasslands will be separately discussed for warm and cold seasons in sections 3a and 3b.

a. Warm season

To use a relative small number of trajectories to illustrate the vast amount of tracks of target-bound air parcels in a concise and reasonable way, a cluster analysis was conducted to obtain a number of cluster mean trajectories that represent all the real trajectories, using the method proposed by Dorling et al. (1992). The number of clusters was arbitrarily determined as 500 in this study, which is found to be sufficient to cover the spread of all real trajectories and also could keep the illustration clean and understandable. Moreover, considering that quite a part of the trajectories is taken by those at very high levels (e.g., 20 000 m), whereas this study focuses on atmospheric moisture that substantially cycled at relatively lower levels, the trajectories entirely above 16 000 m AGL (the altitude of 100-hPa isobaric surface is around 16 000 m AGL) were excluded before the cluster analysis. Worthy of note is that a single cluster mean trajectory may represent a considerable amount of trajectories.

Figure 3 shows the cluster mean trajectories of air parcels entering D1 and D2 on N, MS, and L days during May–September. Most parcels reaching D1 are from the middle and lower atmospheric layers of Eurasia and can be further backtracked to the upper levels over the Atlantic (Figs. 3a–c). It should be noted that the altitude transition along a trajectory is partly due to the unit of altitude used here being meters AGL rather than meters above mean sea level (MSL). On nonprecipitation days (Fig. 3a), few air parcels come from the south or east of D1. On MS days (Fig. 3b), a considerable number of air parcels are from the relatively lower atmospheric layers, mostly below 4000 m AGL, over southern and eastern China and its adjacent seas as well as the Bay of Bengal and the Arabian Sea: even more so for L days (Fig. 3c). This result implies that air from the south and east may have played a key role in the summer precipitation in D1, but its role in moisture supply requires deeper insight in terms of the moisture content and source region.

Fig. 3.

Cluster mean trajectories (number: 500) of target-bound air parcels reaching (top) D1 and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during May–September of 2000–09. The altitudes (unit: m AGL) of air parcels are represented by colors.

Fig. 3.

Cluster mean trajectories (number: 500) of target-bound air parcels reaching (top) D1 and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during May–September of 2000–09. The altitudes (unit: m AGL) of air parcels are represented by colors.

In contrast, with a view to the high elevation of the TP, most air parcels reaching into D2 are from high levels (above 8000 m AGL) over the middle and lower latitudes (Figs. 3d–f). On MS and L days (Figs. 3e,f), especially L days, a low-level branch via the Arabian Sea–Indian Peninsula–Bay of Bengal, which could be regarded as part of the Somali cross-equatorial flows (Findlater 1969), is strengthened compared with that on N days (Fig. 3d). Thus, it can be inferred that an enhanced Somali low-level jet may be crucial for the summer precipitation in D2. The Somali jet is recognized as one of the most important components of the Asian summer monsoon, and it plays a key role in water vapor transport between the Southern and Northern Hemispheres (Wang and Xue 2003; Joseph and Sijikumar 2004; Ding and Chan 2005). However, Zhu (2012) found that the correlation between the summer rainfall in China and the Somali low-level jet is rather weak, especially at the interannual scale. It is also suggested by Zhu (2012) that, although the low-level Australian cross-equatorial flows appear to bring much less moisture into East Asia than the Somali jet, the summer rainfall in China is more closely connected with the Australian cross-equatorial flows than the Somali jet at the interannual scale. Therefore, it is necessary to determine whether the Somali jet is truly an important moisture conveyer for summer rainfall in D2.

The potential contribution of the target-bound air parcels to the moisture supply could be inferred from their moisture content. Figure 4 shows the mean moisture content of target-bound air parcels 1–10 days before reaching corresponding target regions. A comparison between the moisture content of D1-bound air parcels on nonprecipitation and precipitation days indicates that precipitation in D1 involves more atmospheric moisture from northwestern, southern, and eastern China as well as the adjacent seas of China, the Indochina Peninsula, the Bay of Bengal, the Indian Peninsula, and the Arabian Sea, as shown in Figs. 4a–c. Similarly, there appears to be a strong moisture flow dominated by the Somali jet entering D2 on precipitation days via the Arabian Sea–Indian Peninsula–southern foothills of the Himalayas (Figs. 4d–f). It is also noteworthy that the moisture content over northwestern China is perceptibly larger on MS and L days (Figs. 4e,f) compared to nonprecipitation days (Fig. 4d). According to these results, moisture from southern and eastern China, the adjacent seas of China, the Indochina Peninsula, the Bay of Bengal, the Indian Peninsula, the Arabian Sea, and northwestern China may be key contributors of summer rainfall in the semiarid grasslands of China. However, a few questions need to be clarified before drawing a conclusion, such as which of these regions are moisture sources and how much of the moisture shown in Fig. 4 is transported into target regions and actually precipitated. The former question will be discussed in the following part of this section, and attempts will be made to shed light on the latter question in section 4.

Fig. 4.

Mean moisture content (unit: 109 kg) of target-bound air parcels in 1–10 days before reaching (top) D1 and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during May–September of 2000–09.

Fig. 4.

Mean moisture content (unit: 109 kg) of target-bound air parcels in 1–10 days before reaching (top) D1 and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during May–September of 2000–09.

Figure 5 shows the mean of the target-bound air 1–10 days before reaching the target regions under different circumstances. Intuitively, the red areas in Fig. 5 signify where there is a net uptake of moisture, whereas the blue areas signify where there is a net release of moisture. In other words, the red area represents moisture sources and the blue area represents moisture sinks. Although a large number of D1-bound air parcels come from western Eurasia (Figs. 3a–c) and carry a mass of moisture (Figs. 4a–c), it seems that western Eurasia is mainly a moisture sink area in the warm season (Figs. 5a–c), especially on nonprecipitation days. In contrast, the neighboring regions to the south and east of D1 as well as northwestern China–eastern Central Asia seem to be important moisture sources on MS and L precipitation days of D1 during the warm season (Figs. 5b,c). A relatively small quantity of moisture was taken over the western Bay of Bengal and the Arabian Sea (Figs. 5b,c). Moreover, the Indian Peninsula and the southern foothills of the Himalayas, which encounter abundant rainfall because of the robust Indian summer monsoon, are actually moisture sinks.

Fig. 5.

Mean (unit: mm day−1) of target-bound air parcels in 1–10 days before reaching (top) D1and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during May–September of 2000–09. Colored polygons in (c),(f) represent the ranges of source regions to be examined in section 4, as follows: in (c), the Eurasian continent on the north and west of D1 (in green), the southern and eastern adjacent land areas of D1 (in gray), the adjacent seas of China (in royal blue), the Bay of Bengal (in purple), and the Arabian Sea (in brown) and, in (f), the Eurasian continent on the north and west of D2 (in green), central and eastern China (in gray), the Bay of Bengal (in purple), and the Arabian Sea (in brown).

Fig. 5.

Mean (unit: mm day−1) of target-bound air parcels in 1–10 days before reaching (top) D1and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during May–September of 2000–09. Colored polygons in (c),(f) represent the ranges of source regions to be examined in section 4, as follows: in (c), the Eurasian continent on the north and west of D1 (in green), the southern and eastern adjacent land areas of D1 (in gray), the adjacent seas of China (in royal blue), the Bay of Bengal (in purple), and the Arabian Sea (in brown) and, in (f), the Eurasian continent on the north and west of D2 (in green), central and eastern China (in gray), the Bay of Bengal (in purple), and the Arabian Sea (in brown).

One the other hand, the distribution of moisture sources and sinks of D2-bound air partly resemble that of D1-bound air, as shown in Figs. 5d–f. The D2 region and northwestern China–eastern Central Asia seem to be major moisture sources for air entering D2 on N and MS days (Figs. 5d,e). On L days, more moisture over the western Arabian Sea and the Bay of Bengal is taken by the D2-bound air, but the magnitude of is still below that over northwestern China (Fig. 5f). In any case, the Indian Peninsula, the Himalayan foothills, the eastern Arabian Sea, and the eastern Bay of Bengal act more like moisture sinks than sources.

Therefore, moisture from continental areas, including the local domains and northwestern China–eastern Central Asia, tends to be a crucial contributor to the summer rainfall in the semiarid grasslands of China, while the influence of the Somali jet is not as remarkable as expected. However, considering that the summer rainfall in the semiarid grasslands of China is mostly contributed by L precipitation, in which case there is also a respectable amount of moisture taken from the ocean area by target-bound air (Figs. 4c,f, 5c,f), the role of the ocean should not be ignored. A comparison of the contributions from different sources will be discussed in section 4.

b. Cold season

In wintertime, air in the midlatitudes is primarily subject to zonal motions because of the strong midlatitude westerlies and is also frequently subject to meridional motions associated with synoptic-scale waves and eddies. The East Asian winter monsoon dominates over East Asia, exerting robust northwesterlies over eastern Eurasia. As shown in Fig. 6a, on nonprecipitation days, most D1-bound air parcels are from its west and north, with few parcels from its south. On MS and L days, more D1-bound air parcels come from the south and east of D1 (Figs. 6b,c), mainly over land areas. It seems that air from the south also plays a key role in bringing precipitation to D2 (Figs. 6d–f). Specifically, a good number of trajectories enter D2 on MS and L days via the Bay of Bengal and Indian Peninsula (Figs. 6e,f), whereas on nonprecipitation days this number is much smaller (Fig. 6d).

Fig. 6.

Cluster mean trajectories (number: 500) of target-bound air parcels reaching D1 (upper panels) and D2 (lower panels) on N (a, d), MS (b, e), and L (c, f) days during October–April of 2000–09. The altitudes (unit: m AGL) of the air parcels are represented by colors.

Fig. 6.

Cluster mean trajectories (number: 500) of target-bound air parcels reaching D1 (upper panels) and D2 (lower panels) on N (a, d), MS (b, e), and L (c, f) days during October–April of 2000–09. The altitudes (unit: m AGL) of the air parcels are represented by colors.

Accordingly, the moisture to the south of the two target regions is more closely linked to precipitation days than to nonprecipitation days, as shown in Fig. 7. However, this finding only indicates a potential contribution of moisture from the south. Whether these regions (e.g., the neighboring southern areas for D1 and the Bay of Bengal–Indian Peninsula for D2) are moisture sources is still unclear. Except for those nearby areas, farther regions, such as the Mediterranean Sea, the Red Sea–Middle East, and northern Atlantic, may also be important sources, especially on nonprecipitation days (Figs. 7a,d) and less notably on precipitation days (Figs. 7b,c,e,f). These regions are mostly water areas. This is quite different from the situation during summer, when the Mediterranean Sea, the Red Sea, and the northern Atlantic play very minor roles (Figs. 4, 5).

Fig. 7.

Mean moisture content (unit: 109 kg) of target-bound air parcels in 1–10 days before reaching (top) D1 and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during October–April of 2000–09.

Fig. 7.

Mean moisture content (unit: 109 kg) of target-bound air parcels in 1–10 days before reaching (top) D1 and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during October–April of 2000–09.

The corresponding map indicates that, except for northwestern China, the moisture conveyed to D1 on nonprecipitation days appears to largely originate from the Mediterranean Sea and the water area surrounding the Arabian Peninsula, including the Red Sea, the Aden Gulf, the western Arabian Sea, and the Persian Gulf (Fig. 8a). Additionally, small positive values occupy the midlatitude Atlantic at approximately 30°N, indicating a weak moisture source. On precipitation days (Figs. 8b,c), the over these remote sources decreases, while neighbor areas to the south and east of D1 have an increased contribution of moisture. As far as D2 is concerned, the local domain and the water areas surrounding the Arabian Peninsula are the major moisture sources on nonprecipitation days (Fig. 8d). On precipitation days, northwestern China, the northern Arabian Sea, and the western Bay of Bengal are additional moisture sources (Figs. 8e,f), more notable on L days. In the case of D2, the northern Atlantic behaves more like a moisture sink region rather than a source region.

Fig. 8.

Mean (unit: mm day−1) of target-bound air parcels in 1–10 days before reaching (top) D1 and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during October–April of 2000–09. Colored polygons in (b),(e) represent the ranges of source regions to be examined in section 4, as follows: in (b), the Eurasian continent on the north and west of D1 (in green), the southern and eastern adjacent land areas of D1 (in gray), the water areas surrounding the Arabian Peninsula (in brown), the Mediterranean Sea (in purple), and the northern Atlantic (in yellow) and, in (e), the Eurasian continent on the north and west of D2 (in green), the water areas surrounding the Arabian Peninsula (in brown), the Mediterranean Sea (in purple), and the northern Atlantic (in yellow).

Fig. 8.

Mean (unit: mm day−1) of target-bound air parcels in 1–10 days before reaching (top) D1 and (bottom) D2 on (a),(d) N; (b),(e) MS; and (c),(f) L days during October–April of 2000–09. Colored polygons in (b),(e) represent the ranges of source regions to be examined in section 4, as follows: in (b), the Eurasian continent on the north and west of D1 (in green), the southern and eastern adjacent land areas of D1 (in gray), the water areas surrounding the Arabian Peninsula (in brown), the Mediterranean Sea (in purple), and the northern Atlantic (in yellow) and, in (e), the Eurasian continent on the north and west of D2 (in green), the water areas surrounding the Arabian Peninsula (in brown), the Mediterranean Sea (in purple), and the northern Atlantic (in yellow).

c. Comparison between warm and cold seasons

Based on the above discussions, a summary of the main potential moisture sources of precipitation over D1 and D2 in warm and cold seasons is depicted in Table 3. All these sources can potentially make an appreciable contribution to the precipitation in the target regions, but their actual roles in supplying moisture to the semiarid grasslands needs to be further clarified. Considering that the precipitation in the warm season is contributed more by L precipitation than by MS precipitation (Table 1) and that the opposite occurs in the cold season (Table 2), the moisture sources in Table 3 were identified based on the moisture content and maps of the L and MS circumstances for the warm and cold seasons, respectively.

Table 3.

Main continental and oceanic moisture sources of D1 and D2 in the warm and cold seasons during 2000–09. Bolded sources are predominant in moisture content and the magnitude of .

Main continental and oceanic moisture sources of D1 and D2 in the warm and cold seasons during 2000–09. Bolded sources are predominant in moisture content and the magnitude of .
Main continental and oceanic moisture sources of D1 and D2 in the warm and cold seasons during 2000–09. Bolded sources are predominant in moisture content and the magnitude of .

In both the warm and cold seasons, the southern and eastern adjacent land areas of D1, which are strongly affected by the summer monsoons and are much moister than the semiarid regions, tend to be the foremost moisture sources of precipitation in D1. Another continental moisture source throughout the year appears to be northwestern China (here mainly referring to the area north of 40°N and west of 90°E). Being far from oceans, northwestern China is controlled by a dry climate (Shi et al. 2007), with an annual precipitation generally below 200 mm. Nevertheless, the intensive solar radiation renders a strong evaporation over northwestern China relative to its precipitation, especially in summer (Fig. 5), which makes it an atmospheric moisture source rather than a sink. In winter, the evaporation in northwestern China notably decreases because of reduced solar radiation and lower surface air temperature (Fig. 8). The difference of moisture sources of D1 between warm and cold seasons mainly lies in the oceanic sources. During the warm season, the oceanic moisture transported to D1 largely originates from the southern marine areas near China, such as the South China Sea and the Bay of Bengal (Fig. 5c). During the cold season, most of the oceanic moisture seems to be from remote western water areas, such as the Mediterranean Sea and the Red Sea (Fig. 8b). This difference can mainly be attributed to the southwesterlies associated with the summer monsoon and the northwesterlies dominated by the winter monsoon prevailing over East Asia in the two seasons, respectively.

With regard to D2, there is an evident difference in both continental and oceanic sources between the two seasons. Northwestern China–eastern Central Asia tends to be the predominant continental source on precipitation days of D2 during the warm season (Figs. 5e,f) but is notably weakened during the cold season (Figs. 8e,f), especially on MS days (Fig. 8e). As for oceanic sources, there is a westward shift from the water area south of D2 in the warm season (Fig. 5f) to the water area surrounding the Arabian Peninsula in the cold season (Fig. 8e), similar to that of D1.

It is noteworthy that the continental moisture sources in the cold season are not as strong as in the warm season. Therefore, the oceanic moisture is likely to play a more important role in the cold season relative to in the warm season. On the other hand, considering that the precipitation in the warm season accounts for 83.5% and 85.1% of the annual precipitation in the NC and TP grasslands, respectively, the moisture sources in the warm season essentially have greater implications for the target regions. However, all these factors require deeper quantitative insight in terms of the actual contribution to the target regions from the sources mentioned above.

4. Quantification of contribution from moisture sources

As mentioned previously, even if an area is a strong moisture source, there is a possibility that only limited moisture originating from the area could arrive in the target regions for various reasons. One important reason is terrain obstruction, especially considering the high elevation of the TP. As shown in section 3, the moisture evaporated from the western and northern Arabian Sea and the western Bay of Bengal, which is primarily conveyed by the Somali jet, is a potentially strong oceanic contributor to the summer precipitation in D2 (Fig. 5f). However, before arriving at D2, a large part of this mass of moisture must climb up the steep southern slope of the TP (Fig. 1), which implies a great loss of moisture due to adiabatic cooling. To examine the influence of the topography of the TP on the moisture transport by the Somali jet, the en route variations in specific humidity of D1- and D2-bound air parcels on L days of the warm season are shown in Figs. 9a,b, with a focus on the cluster mean trajectories via the Arabian Sea and the Bay of Bengal. Trajectories above 9000 m AGL (the altitude of 300-hPa isobaric surface is around 9000 m MSL) were excluded before cluster analysis in order to more clearly illustrate the terrain obstruction effect of the TP. The D1-bound air over the Bay of Bengal is divided into two branches: one of which is by way of Indochina Peninsula–southwest China–D1, while the other climbs over the TP and enters D2 from the west (Fig. 9a). It is easy to see that the specific humidity of the former branch changes little along the way, maintaining a level greater than 10 g kg−1. In contrast, there is a significant decrease in specific humidity of the latter branch after it climbs up the TP, from above 10 g kg−1 to below 6 g kg−1. Thus, it can be inferred that even air parcels from the same source may make a notably discrepant contribution to the precipitation in the target regions simply because of the different routes. A comparison in the residence times of target-bound air parcels indicates that there are more D2-bound air parcels via the Arabian Sea and the Bay of Bengal than D1-bound air parcels (Figs. 9c,d), over which the specific humidity of the air parcels mostly ranges from 10 to 20 g kg−1, as shown in Fig. 9b. Nonetheless, the specific humidity of nearly all these air parcels decreases below 6 g kg−1 as they climb the TP, and a considerable part is even below 2 g kg−1. Thus, although the Somali jet acts as a robust moisture conveyor, its contribution to the precipitation over the grasslands in the TP should be far less than expected. On the other hand, even though a number of air parcels taking moisture from the Arabian Sea and the Bay of Bengal could steer by the TP and reach into D1 without a great loss of moisture, the Arabian Sea and the Bay of Bengal may not be important moisture sources to D1, because the number of air parcels from the Arabian Sea and the Bay of Bengal to D1 is only a small number (Fig. 9c), in comparison to D2 (Fig. 9d).

Fig. 9.

Cluster mean trajectories (number: 500) via the Arabian Sea and/or the Bay of Bengal into (a) D1 and (b) D2 on L days during May–September. Regions over the Arabian Sea and Bay of Bengal via which the trajectories were selected to do cluster analysis are indicated by white rectangles correspondingly. The change in color in a trajectory denotes the variation of specific humidity (unit: g kg−1). Areas in yellow contour indicate topography greater than 4000 m: here principally representing the TP. (c),(d) The residence times of target-bound particles corresponding to (a),(b), respectively. The same relative units were used in (c),(d).

Fig. 9.

Cluster mean trajectories (number: 500) via the Arabian Sea and/or the Bay of Bengal into (a) D1 and (b) D2 on L days during May–September. Regions over the Arabian Sea and Bay of Bengal via which the trajectories were selected to do cluster analysis are indicated by white rectangles correspondingly. The change in color in a trajectory denotes the variation of specific humidity (unit: g kg−1). Areas in yellow contour indicate topography greater than 4000 m: here principally representing the TP. (c),(d) The residence times of target-bound particles corresponding to (a),(b), respectively. The same relative units were used in (c),(d).

Therefore, a good understanding of the relative importance of different moisture sources to the target regions requires not only a general view of the maps but also a consideration of the continuous change of the moisture originating from the examined sources along their way to the target regions. Sodemann et al. (2008) introduced a valid “moisture source attribution” method to calculate the contribution of each evaporation location along a trajectory to the precipitation at the target location, which takes into account the evaporation and precipitation along the trajectory and is also applied by Martius et al. (2013). This method is practical for attribution of precipitation at a certain point location within the target region to multiple point sources within an examined source region along trajectories. However, this study intends to calculate the total contribution of an examined source region to the total precipitation in the target region, which are both areal rather than points. In this regard, evaporation within and without the examined source region along trajectories should be treated differently. Moreover, precipitation without and within the target regions should also be treated differently. These two points are not considered in the original moisture source attribution method, because it focuses on point locations. Therefore, to make it more applicable to the current study, a few alterations were made to the original method—here we refer to the altered method as “areal source–receptor attribution method”—to estimate the contribution of the moisture source regions identified in section 3 to the precipitation in the semiarid grasslands of China in a reasonable way.

The areal source–receptor attribution method proposed here mainly includes the following seven steps:

  1. Demarcate the examined source region with bounder lines (Figs. 5c,f,8b,e).

  2. Find all the target-bound trajectories that have moisture released in the target regions and calculate the total moisture release .

  3. Among the trajectories selected in step 2, for the ith examined source region, find all the trajectories passing over this region that have uptake of moisture from there.

  4. Among the trajectories selected in step 3, find the first (forward in time) moisture uptake over the source region of each trajectory.

  5. Afterward, along its way to the target region, corresponding computation is to be done according to the following cases:

    • 5.1) If it is still within the range of the examined source region, then at a location of

      • evaporation: evaporation would add the moisture contribution of the source region, thereby the moisture contribution should be updated by , where is the evaporation, on the right side is the old , and on the left side is the new (the same below).

      • precipitation: precipitation would reduce the moisture contribution of the source region, and thereby the moisture contribution should be updated by , where is the precipitation and is the moisture content of the air parcel. Here it is assumed that the reduced percentage of equals the ratio of to , the same below.

    • 5.2) If it is neither within the range of the examined source region nor within the range of the target region, then at a location of

      • evaporation: evaporation outside the source region would add the moisture content and reduce the relative importance of the moisture contribution of the source region. However, the absolute amount of would not change. The change of along a trajectory is directly recorded in the simulation outputs.

      • precipitation: the same as step 5.1(ii).

    • 5.3) If it is within the range of the target region, then at a location of

      • evaporation: the same as step 5.2(i).

      • precipitation: the absolute contribution of to the precipitation in the target region could be counted by , the ratio of which to equals that of to . Moreover, considering that air parcels might resides over the target region for several time intervals, should also be updated by in preparation for the next calculation.

  6. After all the above calculation, sum all the absolute contributions from the ith source region and mark this value as .

  7. The contribution percentage of the ith source region to the total moisture release (here approximately regarded as precipitation) in the target region could be given by

 
formula

Figure 10 gives a schematic diagram of this method.

Fig. 10.

Sketch of the areal source–receptor attribution method proposed in this study to estimate contribution of the examined areal moisture source regions to the precipitation in the areal target regions. Notes along the trajectory correspond to the steps of this method described in section 4. Precipitation is marked in ascending process [steps 5.1(ii), 5.2(ii), and 5.3(ii)], and moisture uptake is marked in the descending process [steps 4, 5.1(i), 5.2(i), and 5.3(i)].

Fig. 10.

Sketch of the areal source–receptor attribution method proposed in this study to estimate contribution of the examined areal moisture source regions to the precipitation in the areal target regions. Notes along the trajectory correspond to the steps of this method described in section 4. Precipitation is marked in ascending process [steps 5.1(ii), 5.2(ii), and 5.3(ii)], and moisture uptake is marked in the descending process [steps 4, 5.1(i), 5.2(i), and 5.3(i)].

Using the above method, the contribution of different moisture source regions to precipitation in the target regions was quantitatively estimated. In addition, a rough estimation of the contribution of local evaporation in the target regions was also achieved, simply treating the target region itself as both a source and a receptor region at the same time. Considering that in the warm and cold seasons, the semiarid grasslands are dominated by L and MS precipitation, respectively, the computation for the warm season is based on the L cases, while that for the cold season is based on the MS cases. In section 3, significant moisture sources in the warm and cold seasons have been identified for D1 and D2. To quantify the moisture uptake from these regions and their contributions to the moisture released in the target regions separately, all examined regions were demarcated by polygons on maps shown in Figs. 5c, 5f, 8b, and 8e. It should be noted what shown in Figs. 5 and 8 is just the mean state of of a vast amount of air parcels. Sink regions or zero regions identified in Figs. 5 and 8 does not necessarily mean that they make no contribution of moisture at all. In consideration of this point, both significant and nonsignificant source regions were examined. In particular, a large area of the Eurasian continent on the west and north of the target regions, which covers significant source regions in the northwest of China and moisture sink regions in the western Eurasia, was examined for a full view on the contribution of Eurasian continent. Moreover, the contribution of northern Atlantic to D1 and D2 in the cold season was computed respectively to make a comparison.

A complete source–receptor relationship consists of the uptake from the source and the release in the target regions. Thus, before concluding the contribution of each source region to the precipitation in the target regions, a comparison of the moisture uptake from source regions is conducive to a comprehensive understanding of the hydrological cycle related to the semiarid grasslands of China. Figure 11 shows the ratio of moisture uptakes from each examined source region to the total moisture release in the target regions, which have taken into consideration the evaporation and precipitation within the source region following step 5.1 of the method proposed above. During the warm season, the moisture uptake by D1-bound air from the continental sources is of an amount as large as twice that released in D1, while the uptake from oceanic source regions is much smaller (Fig. 11a). Specifically, the uptake from the Eurasian continent on the north and west of D1 and the uptake from the southern and eastern adjacent land areas of D1 have an essentially equivalent large value, relative to that from the Arabian Sea and the Bay of Bengal, which is even smaller than the local uptake. In contrast, the D2-bound air has a considerable moisture uptake from oceanic source regions, which, however, is still below the uptake from the Eurasian continent (Fig. 11b). During the cold season, more moisture is taken from the Eurasian continent by D1-bound air than oceanic source regions (Fig. 11c), whereas the moisture uptake from the water area surrounding the Arabian Peninsula by D2-bound air significantly surpasses the uptake from the Eurasian continent and other source regions (Fig. 11d). Moisture uptake from the local regions of D1 and D2 are of relatively small values in both warm and cold seasons. The moisture uptake from northern Atlantic by target-bound air parcels in the cold season is comparable to that from other oceanic source regions (Figs. 11c,d) but not outstanding.

Fig. 11.

Ratios of moisture uptake over moisture source regions of (a),(c) D1 and (b),(d) D2 to the total moisture release in the target regions during (top) the warm season and (bottom) the cold season (unit: %). Line-filled and dot-filled bars denote uptake from continental and oceanic source regions, respectively. Uptakes from all continental and/or oceanic source regions are added and denoted by solid black and/or white bars, respectively. The names of source regions are abbreviated as follows: “N.W. Eurasia”: the Eurasian continent on the north and west of D1 in (a),(c) and D2 in (b),(d); “S.E. Land”: the southern and eastern adjacent land areas of D1; “Adj. Seas”: adjacent seas of China; “Ara. Sea”: the Arabian Sea; “B.B.”: the Bay of Bengal; “C.E. China”: central and eastern China; “W.A.A.P”: water area surrounding the Arabian Peninsula; “Med. Sea”: the Mediterranean Sea; and “N. Atlantic”: northern Atlantic.

Fig. 11.

Ratios of moisture uptake over moisture source regions of (a),(c) D1 and (b),(d) D2 to the total moisture release in the target regions during (top) the warm season and (bottom) the cold season (unit: %). Line-filled and dot-filled bars denote uptake from continental and oceanic source regions, respectively. Uptakes from all continental and/or oceanic source regions are added and denoted by solid black and/or white bars, respectively. The names of source regions are abbreviated as follows: “N.W. Eurasia”: the Eurasian continent on the north and west of D1 in (a),(c) and D2 in (b),(d); “S.E. Land”: the southern and eastern adjacent land areas of D1; “Adj. Seas”: adjacent seas of China; “Ara. Sea”: the Arabian Sea; “B.B.”: the Bay of Bengal; “C.E. China”: central and eastern China; “W.A.A.P”: water area surrounding the Arabian Peninsula; “Med. Sea”: the Mediterranean Sea; and “N. Atlantic”: northern Atlantic.

Because of evaporation and precipitation en route, moistures originating from different source regions can be greatly changed in the absolute amount and their proportion to the total moisture content when they reach the target regions, and thus their contribution to precipitation in target regions may be remarkably different from what expected from the uptake comparisons shown in Fig. 11. Figure 12 shows the contribution of each examined moisture source region to the total moisture release in the target regions, using the areal source–receptor attribution method. It reveals that moistures from the Eurasian continent make predominant contributions to the precipitation in the semiarid grasslands of China whether in the warm season (Figs. 12a,b) or in the cold season (Figs. 12c,d). During the warm season, moisture from the southern and eastern adjacent land area of D1 accounts for above 30% of the precipitation in D1 (Fig. 12a). The moisture from the vast Eurasia north and west of D1 makes an approximately equivalent contribution. In contrast, the contribution of moisture from the Arabian Sea and the Bay of Bengal almost can be ignored. Similarly, the contribution of moisture from the Eurasian continent north and west of D2 to the total moisture release in D2 far exceeds that of moisture from those oceanic source regions (Fig. 12b). During the cold season, moistures from the Eurasian continent north and west of the target regions make dominant contributions to the precipitation in both D1 and D2, with both accounting for nearly 50% of total moisture release in the target regions (Figs. 12c,d). The southern and eastern adjacent land area of D1 also seems to an important contributor to the precipitation in D1 (Fig. 12c). All examined oceanic moisture sources are of little importance to the precipitation in the target regions, including northern Atlantic. Specifically, although a large amount of moisture is taken by D2-bound air over the water area surrounding the Arabian Peninsula (Figs. 8e, 11d), the contribution of moisture originating from there to the precipitation in D2 seems to be below 5% (Fig. 12d). In both the warm and cold seasons, contribution of the local evaporation is at a level around 10% for D1 and around 5% for D2, respectively. Overall, the Eurasian continent seems to be the foremost moisture supplier to the precipitation in the semiarid grasslands of China in both warm and cold seasons, while oceanic moisture makes little contribution because of significant loss en route.

Fig. 12.

Contributions (i.e., obtained from the areal source–receptor attribution method) of moisture source regions of (a),(c) D1 and (b),(d) D2 to the total moisture release in the target regions during (top) the warm season and (bottom) the cold season (unit: %). Line-filled and dot-filled bars denote contributions from continental and oceanic source regions, respectively. Contributions from all continental and/or oceanic source regions are added and denoted by solid black and/or white bars, respectively. Abbreviations of source regions are as in Fig. 11.

Fig. 12.

Contributions (i.e., obtained from the areal source–receptor attribution method) of moisture source regions of (a),(c) D1 and (b),(d) D2 to the total moisture release in the target regions during (top) the warm season and (bottom) the cold season (unit: %). Line-filled and dot-filled bars denote contributions from continental and oceanic source regions, respectively. Contributions from all continental and/or oceanic source regions are added and denoted by solid black and/or white bars, respectively. Abbreviations of source regions are as in Fig. 11.

There are a few points that need to be noted on the computation results shown in Figs. 11 and 12. The range of the Eurasian continent on the north and west of the target regions is defined by simple polygons (Figs. 5c,f, 8b,e) to facilitate the computation process, which involves some water areas such as the Caspian Sea and the Black Sea. In consideration of their small area and less significance in fields, moisture uptake from these areas should be only a small part of the total, and thus the effect on the computation results would be slight. Likewise, the water area surrounding the Arabian Peninsula was defined by a rectangle that involves the land area of the Arabian Peninsula. In light of that the of target-bound air parcels over the Arabian Peninsula approximates zero (Figs. 8b,e), this effect could also be neglected.

In addition, it is also worth noting that the examined continental and oceanic moisture sources of D1 together could account for as much as 80% of the precipitation in D1 (Figs. 12a,c), whereas those of D2 could only account for approximately 60% of the precipitation in D2 (Figs. 12b,d), implying that there might be other important source regions of D2 that have not been realized. Thus, further computations were done to find out which important source region (or regions) has not been identified. After a series of tests, it is found that the moisture from the Indian Peninsula, which appears to be a significant moisture sink region in the warm season and a weak moisture sink region in the cold season, tends to make a considerable contribution to the precipitation in D2, unexpectedly (Figs. 5f, 8e). This source can account for 10% and 14% of the precipitation in D2 for the warm season and the cold season, respectively. This may be attributed to that, although the mean state of Indian Peninsula mainly appears to a be moisture sink, especially in summer, there is still a portion of moisture was taken from there to D2, which is not exhibited in the mean state of shown in Figs. 5 and 8.

5. Conclusions and discussion

Significant moisture sources of the semiarid grasslands of China identified in this study include southern and eastern China, northwestern China–eastern Central Asia, the adjacent seas of China, the Bay of Bengal, the Arabian Sea, water areas surrounding the Arabian Peninsula (Red Sea, Aden Gulf, western Arabian Sea, and Persian Gulf), and the Mediterranean Sea. Local evaporation is also a nonnegligible factor. Of these sources, the continental sources tend to be the major contributors of precipitation over the semiarid grasslands, while the oceanic sources make very little contribution. In fact, the uptake of moisture over oceanic sources could be of considerable amount. This, at first sight, may be surprising but, because of the terrain obstruction and the long path to the continental interior, a large proportion of oceanic moisture is lost en route and only a small portion can reach inland, which is one of the most important causes of the dry climate in the midlatitude Northern Hemisphere (Broccoli and Manabe 1992).

Spreading over very different geography, the grasslands in NC and the TP are subject to different moisture transport pathways. On L days of the warm season, moisture released over the grasslands of NC mostly originates from its neighboring southern and eastern wet regions, while the Eurasian continent on the north and west, which includes the significant moisture source region northwestern China–eastern Central Asia, seems to be the most important moisture source of the grasslands in the TP. More interestingly, on MS days of the cold season, the uptake by D2-bound air parcels over the water area surrounding the Arabian Peninsula is predominant among all the sources, but its contribution to the precipitation in D2 is quite minor. All these findings imply that the atmospheric water cycle related to the semiarid grasslands of China involves a very complex mechanism.

Conclusions of this study are based on the 10-day backtrace results. Despite that a 10-day period is the global mean recycling time of atmospheric water, this time can be significantly different from region to region. For this reason, two more computations were conducted to examine what differences there would be in 15- and 5-day backtrace cases. It is found that the significance of moisture sources concerning meridional water vapor transport is susceptible to the length of backtrace time. For instance, the magnitude of along Somali jet is perceivably enhanced and reduced in the 15- and 5-day backtrace cases, respectively. Thus, a better understanding of the hydrological cycle related to the semiarid grasslands of China may need a deeper insight into the recycling time of moisture over different regions. Another limit of this study is that the estimation of contribution from the local evaporation may be rough, seeing that at a precipitation location within the target region there is also evaporation, but it was not taken into account using the areal source–receptor attribution method. Previous studies have demonstrated that local evaporation generally plays a minor role in supplying moisture to local precipitation relative to moisture advection (Trenberth 1999; Sun and Wang 2013). However, the global hydrological cycle is projected to dramatically change under global warming (Loaiciga et al. 1996; Soden and Held 2006; Sun and Jiang 2012), whereby the relationships among precipitation at different scales and intensity, local evaporation, and moisture transport may vary to certain degree (Trenberth et al. 2003). Thus, an explicit evaluation of the role of local evaporation will be conducive to a comprehensive understanding of the hydrological cycle over these semiarid grasslands and its possible alteration in the future, which calls for further work. Despite all the attempts of this study, there is still much space for improvement on the method and computation in regards to the identification of moisture sources. Sodemann et al. (2008) introduced a sophisticated methodology that involves several specific steps to identify moisture sources, which may be conducive to a more elaborate work on this issue. Gimeno et al. (2012) compared the methods used to establish atmospheric moisture source–sink relationships, with a particular view on the Lagrangian method, and provided a discussion on the advantages and disadvantages associated with each method. Thus, a best achievement may be expected with an integration of the results from several practical methods.

Recent decades have seen significant interdecadal changes in precipitation pattern over China in the late 1970s (Gong and Ho 2002), the early 1990s (Wu et al. 2010), and the late 1990s (Zhu et al. 2011). The interdecadal changes of precipitation pattern are closely related to the interdecadal changes of water vapor transport over China (Sun et al. 2011). Therefore, further simulation running through a longer period that involves the interdecadal changes of hydrological cycle will be conducive to the knowledge of the interdecadal changes of moisture sources.

Acknowledgments

This study was supported by the National Natural Science Foundation of China (Grants 41210007 and 41130103).

REFERENCES

REFERENCES
Adler
,
R. F.
, and
Coauthors
,
2003
:
The Version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present)
.
J. Hydrometeor.
,
4
,
1147
1167
.
Barras
,
V.
, and
I.
Simmonds
,
2009
:
Observation and modeling of stable water isotopes as diagnostics of rainfall dynamics over southeastern Australia
.
J. Geophys. Res.
,
114
,
D23308
,
doi:10.1029/2009JD012132
.
Broccoli
,
A. J.
, and
S.
Manabe
,
1992
:
The effects of orography on midlatitude Northern Hemisphere dry climates
.
J. Climate
,
5
,
1181
1201
.
Chen
,
B.
,
X. D.
Xu
,
S.
Yang
, and
W.
Zhang
,
2012
:
On the origin and destination of atmospheric moisture and air mass over the Tibetan Plateau
.
Theor. Appl. Climatol.
,
110
,
423
435
,
doi:10.1007/s00704-012-0641-y
.
Chen
,
B.
,
X. D.
Xu
, and
T. L.
Zhao
,
2013
:
Main moisture sources affecting lower Yangtze River Basin in boreal summers during 2004–2009
.
Int. J. Climatol.
,
33
,
1035
1046
,
doi:10.1002/joc.3495
.
Christensen
,
L.
,
M. B.
Coughenour
,
J. E.
Ellis
, and
Z. Z.
Chen
,
2004
:
Vulnerability of the Asian typical steppe to grazing and climate change
.
Climatic Change
,
63
,
351
368
,
doi:10.1023/B:CLIM.0000018513.60904.fe
.
Chuluun
,
T.
, and
D.
Ojima
,
2002
:
Land use change and carbon cycle in arid and semi-arid lands of East and Central Asia
.
Sci. China
,
45C
,
48
54
.
Ding
,
Q. H.
, and
B.
Wang
,
2005
:
Circumglobal teleconnection in the Northern Hemisphere summer
.
J. Climate
,
18
,
3483
3505
.
Ding
,
Y. H.
, and
J. C. L.
Chan
,
2005
:
The East Asian summer monsoon: An overview
.
Meteor. Atmos. Phys.
,
89
,
117
142
,
doi:10.1007/s00703-005-0125-z
.
Dorling
,
S. R.
,
T. D.
Davies
, and
C. E.
Pierce
,
1992
:
Cluster analysis: A technique for estimating the synoptic meteorological controls on air and precipitation chemistry—Method and applications
.
Atmos. Environ.
,
26A
,
2575
2581
.
Drumond
,
A.
,
R.
Nieto
, and
L.
Gimeno
,
2011
:
Sources of moisture for China and their variations during drier and wetter conditions in 2000–2004: A Lagrangian approach
.
Climate Res.
,
50
,
215
225
,
doi:10.3354/cr01043
.
Findlater
,
J.
,
1969
:
A major low-level air current near the Indian Ocean during the northern summer
.
Quart. J. Roy. Meteor. Soc.
,
95
,
362
380
,
doi:10.1002/qj.49709540409
.
Gimeno
,
L.
,
A.
Drumond
,
R.
Nieto
,
R. M.
Trigo
, and
A.
Stohl
,
2010
:
On the origin of continental precipitation
.
Geophys. Res. Lett.
,
37
,
L13804
,
doi:10.1029/2010GL043712
.
Gimeno
,
L.
, and
Coauthors
,
2012
:
Oceanic and terrestrial sources of continental precipitation
.
Rev. Geophys.
,
50
,
RG4003
,
doi:10.1029/2012RG000389
.
Gimeno
,
L.
,
R.
Nieto
,
A.
Drumond
,
R.
Castillo
, and
R.
Trigo
,
2013
:
Influence of the intensification of the major oceanic moisture sources on continental precipitation
.
Geophys. Res. Lett.
,
40
,
1443
1450
,
doi:10.1002/grl.50338
.
Gong
,
D.-Y.
, and
C.-H.
Ho
,
2002
:
Shift in the summer rainfall over the Yangtze River valley in the late 1970s
.
Geophys. Res. Lett.
,
29
,
1436
,
doi:10.1029/2001GL014523
.
Gong
,
D.-Y.
,
S.-W.
Wang
, and
J.-H.
Zhu
,
2001
:
East Asian winter monsoon and Arctic Oscillation
.
Geophys. Res. Lett.
,
28
,
2073
2076
,
doi:10.1029/2000GL012311
.
Gong
,
D.-Y.
,
P.-J.
Shi
, and
J.-A.
Wang
,
2004
:
Daily precipitation changes in the semi-arid region over northern China
.
J. Arid Environ.
,
59
,
771
784
,
doi:10.1016/j.jaridenv.2004.02.006
.
Hansen
,
M. C.
,
R. S.
Defries
,
J. R. G.
Townshend
, and
R.
Sohlberg
,
2000
:
Global land cover classification at 1 km spatial resolution using a classification tree approach
.
Int. J. Remote Sens.
,
21
,
1331
1364
,
doi:10.1080/014311600210209
.
Hou
,
H. Y.
,
1983
:
Vegetation of China with reference to its geographical distribution
.
Ann. Mo. Bot. Gard.
,
70
,
509
548
,
doi:10.2307/2992085
.
Huang
,
G.
,
Y.
Liu
, and
R. H.
Huang
,
2011
:
The interannual variability of summer rainfall in the arid and semiarid regions of Northern China and its association with the Northern Hemisphere circumglobal teleconnection
.
Adv. Atmos. Sci.
,
28
,
257
268
,
doi:10.1007/s00376-010-9225-x
.
Joseph
,
P. V.
, and
S.
Sijikumar
,
2004
:
Intraseasonal variability of the low-level jet stream of the Asian summer monsoon
.
J. Climate
,
17
,
1449
1458
.
Li
,
F.
, and
H. J.
Wang
,
2013
:
Autumn sea ice cover, winter Northern Hemisphere annular mode, and winter precipitation in Eurasia
.
J. Climate
,
26
,
3968
3981
.
Li
,
Y. H.
,
1995
:
The characteristics of precipitation in the dry and semi-dry region of China and their connection with Arctic sea ice (in Chinese)
.
Meteorology
,
21
,
11
15
.
Liu
,
B.
,
M.
Xu
,
M.
Henderson
, and
Y.
Qi
,
2005
:
Observed trends of precipitation amount, frequency, and intensity in China, 1960–2000
.
J. Geophys. Res.
,
110
,
D08103
,
doi:10.1029/2004jd004864
.
Liu
,
J. P.
,
J. A.
Curry
,
H.
Wang
,
M.
Song
, and
R. M.
Horton
,
2012
:
Impact of declining Arctic sea ice on winter snowfall
.
Proc. Natl. Acad. Sci. USA
,
109
,
4074
4079
,
doi:10.1073/pnas.1114910109
.
Liu
,
J. Y.
,
D. F.
Zhuang
,
D.
Luo
, and
X.
Xiao
,
2003
:
Land-cover classification of China: Integrated analysis of AVHRR imagery and geophysical data
.
Int. J. Remote Sens.
,
24
,
2485
2500
,
doi:10.1080/01431160110115582
.
Loaiciga
,
H. A.
,
J. B.
Valdes
,
R.
Vogel
,
J.
Garvey
, and
H.
Schwarz
,
1996
:
Global warming and the hydrologic cycle
.
J. Hydrol.
,
174
,
83
127
.
Martius
,
O.
, and
Coauthors
,
2013
:
The role of upper-level dynamics and surface processes for Pakistan flood of July 2010
.
Quart. J. Roy. Meteor. Soc.
,
139
,
1780
1797
,
doi:10.1002/qj.2082
.
Numaguti
,
A.
,
1999
:
Origin and recycling processes of precipitating water over the Eurasian continent: Experiments using an atmospheric general circulation model
.
J. Geophys. Res.
,
104
,
1957
1972
,
doi:10.1029/1998jd200026
.
Petterssen
,
S.
,
1940
: Weather Analysis and Forecasting. McGraw-Hill, 505 pp.
Saha
,
S.
, and
Coauthors
,
2010
:
The NCEP Climate Forecast System Reanalysis
.
Bull. Amer. Meteor. Soc.
,
91
,
1015
1057
.
Screen
,
J. A.
,
C.
Deser
,
I.
Simmonds
, and
R.
Tomas
,
2013a
:
Atmospheric impacts of Arctic sea-ice loss, 1979–2009: Separating forced change from atmospheric internal variability
.
Climate Dyn.
,
doi:10.1007/s00382-013-1830-9, in press
.
Screen
,
J. A.
,
I.
Simmonds
,
C.
Deser
, and
R.
Tomas
,
2013b
:
The atmospheric response to three decades of observed Arctic sea ice loss
.
J. Climate
,
26
,
1230
1248
.
Shi
,
Y. F.
,
Y. P.
Shen
,
E.
Kang
,
D. L.
Li
,
Y. J.
Ding
,
G. W.
Zhang
, and
R. J.
Hu
,
2007
:
Recent and future climate change in northwest China
.
Climatic Change
,
80
,
379
393
,
doi:10.1007/s10584-006-9121-7
.
Simmonds
,
I.
,
D.
Bi
, and
B.
Yan
,
1996
:
Relationships between summer rainfall over China and ocean temperatures in the tropical western Pacific
.
J. Meteor. Soc. Japan
,
74
,
273
279
.
Simmonds
,
I.
,
D.
Bi
, and
P.
Hope
,
1999
:
Atmospheric water vapor flux and its association with rainfall over China in summer
.
J. Climate
,
12
,
1353
1367
.
Sodemann
,
H.
, and
A.
Stohl
,
2009
:
Asymmetries in the moisture origin of Antarctic precipitation
.
Geophys. Res. Lett.
,
36
,
L22803
,
doi:10.1029/2009gl040242
.
Sodemann
,
H.
,
C.
Schwierz
, and
H.
Wernli
,
2008
:
Interannual variability of Greenland winter precipitation sources: Lagrangian moisture diagnostic and North Atlantic Oscillation
.
J. Geophys. Res.
,
113
,
D03107
,
doi:10.1029/2007JD008503
.
Soden
,
B. J.
, and
I. M.
Held
,
2006
:
Robust responses of the hydrological cycle to global warming
.
J. Climate
,
19
,
5686
5699
.
Stohl
,
A.
,
1998
:
Computation, accuracy and applications of trajectories-a review and bibliography
.
Atmos. Environ.
,
32
,
947
966
.
Stohl
,
A.
, and
P.
James
,
2004
:
A Lagrangian analysis of the atmospheric branch of the global water cycle. Part I: Method description, validation, and demonstration for the August 2002 flooding in central Europe
.
J. Hydrometeor.
,
5
,
656
678
.
Stohl
,
A.
, and
P.
James
,
2005
:
A Lagrangian analysis of the atmospheric branch of the global water cycle. Part II: Moisture transports between Earth’s ocean basins and river catchments
.
J. Hydrometeor.
,
6
,
961
984
.
Stohl
,
A.
,
M.
Hittenberger
, and
G.
Wotawa
,
1998
:
Validation of the Lagrangian particle dispersion model FLEXPART against large-scale tracer experiment data
.
Atmos. Environ.
,
32
,
4245
4264
,
doi:10.1016/S1352-2310(98)00184-8
.
Stohl
,
A.
,
S.
Eckhardt
,
C.
Forster
,
P.
James
, and
N.
Spichtinger
,
2002
:
On the pathways and timescales of intercontinental air pollution transport
.
J. Geophys. Res.
,
107
,
4684
,
doi:10.1029/2001jd001396
.
Stohl
,
A.
,
C.
Forster
,
A.
Frank
,
P.
Seibert
, and
G.
Wotawa
,
2005
:
Technical note: The Lagrangian particle dispersion model FLEXPART version 6.2
.
Atmos. Chem. Phys.
,
5
,
2461
2474
.
Stohl
,
A.
,
C.
Forster
, and
H.
Sodemann
,
2008
:
Remote sources of water vapor forming precipitation on the Norwegian west coast at 60°N—A tale of hurricanes and an atmospheric river
.
J. Geophys. Res.
,
113
,
D05102
,
doi:10.1029/2007jd009006
.
Sun
,
B.
, and
D. B.
Jiang
,
2012
:
Changes of atmospheric water balance over China under the IPCC SRES A1B scenario based on RegCM3 simulations
.
Atmos. Oceanic Sci. Lett.
,
5
,
461
467
.
Sun
,
B.
, and
H. J.
Wang
,
2013
:
Water vapor transport paths and accumulation during widespread snowfall events in northeastern China
.
J. Climate
,
26
,
4550
4566
.
Sun
,
B.
,
Y.
Zhu
, and
H.
Wang
,
2011
:
The recent interdecadal and interannual variation of water vapor transport over eastern China
.
Adv. Atmos. Sci.
,
28
,
1039
1048
,
doi:10.1007/s00376-010-0093-1
.
Trenberth
,
K. E.
,
1998
:
Atmospheric moisture residence times and cycling: Implications for rainfall rates and climate change
.
Climatic Change
,
39
,
667
694
,
doi:10.1023/A:1005319109110
.
Trenberth
,
K. E.
,
1999
:
Atmospheric moisture recycling: Role of advection and local evaporation
.
J. Climate
,
12
,
1368
1381
.
Trenberth
,
K. E.
,
A.
Dai
,
R. M.
Rasmussen
, and
D. B.
Parsons
,
2003
:
The changing character of precipitation
.
Bull. Amer. Meteor. Soc.
,
84
,
1205
1217
.
Viste
,
E.
, and
A.
Sorteberg
,
2013
:
Moisture transport into the Ethiopian highlands
.
Int. J. Climatol.
,
33
,
249
263
,
doi:10.1002/joc.3409
.
Wang
,
H.
,
2001
:
The weakening of the Asian Monsoon circulation after the end of 1970’s
.
Adv. Atmos. Sci.
,
18
,
376
386
.
Wang
,
H.
, and
F.
Xue
,
2003
:
Interannual variability of Somali jet and its influences on the inter-hemispheric water vapor transport and on the East Asian summer rainfall
.
Chin. J. Geophys.
,
46
,
11
20
.
Wang
,
H.
, and
H.
Chen
,
2012
:
Climate control for southeastern China moisture and precipitation: Indian or East Asian monsoon?
J. Geophys. Res.
,
117
,
D12109
,
doi:10.1029/2012JD017734
.
Wang
,
W.-C.
, and
K.
Li
,
1990
:
Precipitation fluctuation over semiarid region in northern China and the relationship with El Niño/Southern Oscillation
.
J. Climate
,
3
,
769
783
.
Wei
,
J.
,
P. A.
Dirmeyer
,
M. G.
Bosilovich
, and
R.
Wu
,
2012
:
Water vapor sources for Yangtze River Valley rainfall: Climatology, variability, and implications for rainfall forecasting
.
J. Geophys. Res.
,
117
,
D05126
,
doi:10.1029/2011JD016902
.
Wu
,
B.
, and
J.
Wang
,
2002
:
Winter Arctic Oscillation, Siberian high and East Asian winter monsoon
.
Geophys. Res. Lett.
,
29
,
1897
,
doi:10.1029/2002GL0125373
.
Wu
,
R.
,
Z. P.
Wen
,
S.
Yang
, and
Y. Q.
Li
,
2010
:
An interdecadal change in southern China summer rainfall around 1992–93
.
J. Climate
,
23
,
2389
2403
.
Yatagai
,
A.
, and
T.
Yasunari
,
1995
:
Interannual variations of summer precipitation in the arid/semiarid regions in China and Mongolia—Their regionality and relation to the Asian summer monsoon
.
J. Meteor. Soc. Japan
,
73
,
909
923
.
Yu
,
E.
,
H.
Wang
,
J.
Sun
, and
Y.
Gao
,
2013
:
Climatic response to changes in vegetation in the northwest Hetao Plain as simulated by the WRF model
.
Int. J. Climatol.
,
33
,
1470
1481
,
doi:10.1002/joc.3527
.
Zhai
,
P.
, and
R. E.
Eskridge
,
1997
:
Atmospheric water vapor over China
.
J. Climate
,
10
,
2643
2652
.
Zhang
,
Y.
,
K. R.
Sperber
, and
J. S.
Boyle
,
1997
:
Climatology and interannual variation of the East Asian winter monsoon: Results from the 1979–95 NCEP/NCAR reanalysis
.
Mon. Wea. Rev.
,
125
,
2605
2619
.
Zhou
,
L.-T.
,
2011
:
Impact of East Asian winter monsoon on rainfall over southeastern China and its dynamical process
.
Int. J. Climatol.
,
31
,
677
686
,
doi:10.1002/joc.2101
.
Zhu
,
Y.
,
2012
:
Variations of the summer Somali and Australia cross-equatorial flows and the implications for the Asian summer monsoon
.
Adv. Atmos. Sci.
,
29
,
509
518
,
doi:10.1007/s00376-011-1120-6
.
Zhu
,
Y.
,
H.
Wang
,
W.
Zhou
, and
J.
Ma
,
2011
:
Recent changes in the summer precipitation pattern in East China and the background circulation
.
Climate Dyn.
,
36
,
1463
1473
,
doi:10.1007/s00382-010-0852-9
.