Water consumption of agriculture and natural ecosystems along the Tarim River, China.

Full text

Water allocation and water consumption of irrigated agriculture and
natural vegetation in the Aksu-Tarim river basin, Xinjiang, China
Niels Thevs
a
,
*
, Haiyan Peng
a
, Ahmedjan Rozi
a
, Stefan Zerbe
b
, Nurbay Abdusalih
c
a
Institute of Botany and Landscape Ecology, University of Greifswald, Grimmer Strasse 88, 17487 Greifswald, Germany
b
Faculty of Science and Technology, Free University of Bozen-Bolzano, Piazza Universit
a 5, 39100 Bolzano, Italy
c
Institute of Resource and Environmental Sciences, Xinjiang University, Shengli Lu 14, 830046 Urumqi, China
article info
Article history:
Received 10 January 2013
Received in revised form
23 May 2014
Accepted 26 May 2014
Available online xxx
Keywords:
Central Asia
Cotton
Evapotranspiration
Remote sensing
Riparian vegetation
Water resource management
abstract
A significant part of the world's largest river basins are located in areas of arid and semi-arid climate,
such as the Amu Darya, Jordan, Murray-Darling, Yellow River, and Aksu-Tarim river basin. These river
basins are experiencing water scarcity resulting in conflicts between upstream and downstream, con-
flicts between water users, and degradation of the natural ecosystems. Therefore, in many river basins,
including the Aksu-Tarim river basin, water quota systems have been established, in order to allocate
water under scarcity. The Aksu-Tarim river basin (NW China) has developed into one of the most
important cotton production areas worldwide. In this paper, we aim at assessing the water consumption
through irrigated agriculture, mainly cotton, and natural vegetation in the Aksu-Tarim river basin against
the background of this water quota system. Firstly, we map the evapotranspiration (ET
a
) as water con-
sumption of irrigated agriculture and natural vegetation in the Aksu-Tarim river basin. Secondly, we
calculate water balances and relate them to the water quota system. We employed the remote sensing
method Simplified Surface Energy Balance Index (S-SEBI), in order to map ET
a
based on MODIS satellite
images for the growing seasons 2009, 2010, and 2011. Thereby, the MODIS products 8-day land surface
temperature (MOD11A2), 16-day albedo (MCD43A3), and 16-day NDVI (MOD13A1) were used. The ET
a
of
cotton ranges from 884 to 1198 mm. The ET
a
of the natural vegetation of a total coverage ranges from 715
in 2009 to 960 mm in 2011, clearly following the annual runoff of the Aksu and Tarim River. The water
balance of the Aksu-Tarim river basin is 3.25 to 3.73 km
3
,0.1e0.53 km
3
, and 3.55 to 4.12 km
3
in
2009, 2010, and 2011, respectively. The water quotas along the Aksu River and the upper reaches of the
Tarim are exceeded by water consumption, while the quotas along the middle and lower reaches are not
met. Considerable amounts of groundwater, including fossil groundwater, are exploited for irrigation
along the Aksu and Tarim River, which must be regarded as exploitation of a non-renewable resource.
©2014 Elsevier Ltd. All rights reserved.
1. Introduction
A significant part of the world's largest river basins are located
in areas of arid and semi-arid climate, such as the Amu Darya,
Jordan, Murray-Darling, Yellow River, and Tarim river basin
(Central Asia Atlas, 2012; GLOWA Jordan, 2008;Glantz, 2010;
ICWC, 1992; Murray Darling Basin Commission, 2006; Song et al.,
2000; Tang and Deng, 2010; Zhu et al., 2003). These river basins
are experiencing water scarcity resulting in conflicts between
upstream and downstream regions, conflicts between water users,
and degradation of the natural ecosystems. In these areas, agri-
culture, mostly irrigated agriculture, is the largest consumer of
water (Howell, 2001). Given the background of increasing pop-
ulations, increasing demand for agricultural products, and thus
increasing water demands, water distribution in river basins be-
tween upstream and downstream as well as between water users
plays and will play an important role for societies within such
river basins (FAO, 2012). In many river basins, water quota systems
have been established in order to allocate distinct amounts of
water to different users and different sections of a river as
reviewed by Molle (2009). In Central Asia, quota systems have
been introduced between countries for the Amu Darya and Syr
*Corresponding author. Tel.: þ49 3834 864131.
E-mail address: thevs@uni-greifswald.de (N. Thevs).
Contents lists available at ScienceDirect
Journal of Arid Environments
journal homepage: www.elsevier.com/locate/jaridenv
http://dx.doi.org/10.1016/j.jaridenv.2014.05.028
0140-1963/©2014 Elsevier Ltd. All rights reserved.
Journal of Arid Environments xxx (2014) 1e11
Please cite this article in press as: Thevs, N., et al., Water allocationand water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

Darya, which are the two tributaries of the Aral Sea (ICWC, 1992),
on provincial level for the Heihe River, China (Chen et al., 2006),
and for the Tarim River according to tributaries and river sections
(Tang and Deng, 2010).
The Tarim Basin, which includes the Tarim and its tributaries,
located in Xinjiang, Northwest China (Fig. 1), harbors 54%
(352,200 ha) of the world's riparian Populus euphratica Oliv. forests
(http://whc.unesco.org/en/tentativelists/5532/). Those forests form
a mosaic of riparian forests, wetlands, shrub vegetation, and small
stands of herbaceous vegetation (Thevs et al., 2008; Zhang et al.,
2005) and provide habitat for wildlife (http://whc.unesco.org/en/
tentativelists/5532/). The P. euphratica forests are the only forests
in the Tarim Basin. Those forests and the wetlands are the most
productive ecosystems of the drylands in the Tarim Basin (Thevs
et al., 2007,2012). Furthermore, the Tarim Basin has become the
world's most important cotton production region with a total
annual cotton lint production of 2.1 million t, i.e. 8.85% of the world
production in 2010 (Xinjiang Statistics Bureau, 2011;http://faostat.
fao.org/). In 2011, the share of the cotton lint production in Xinjiang
of the worldwide production climbed to 11% (USDA, 2013;http://
faostat.fao.org/). Half of the cotton in the Tarim Basin is produced
along the Aksu and Tarim River (Feike et al., 2014; Xinjiang
Statistics Bureau, 2011). Therefore, it is relevant to show how the
water quota system under intensive and increasing agriculture
works.
The Tarim Basin covers an area of 1.02 million km
2
, and is home
to a population of 9.02 million people (Tan and Zhou, 2007). The
area of irrigated land has increased all over the Tarim Basin, from
706,000 ha in 1949, over 1,330,000 ha in 1980 and 1,412,000 ha in
1990 (Xia,1998), in 2010 to 1,600,000 ha (Xinjiang Statistics Bureau,
2011). Due to the arid climate with an annual precipitation of
30e70 mm (Liu, 1997; Tang and Deng, 2010), all agriculture de-
pends on irrigation with river water being the most important
source for water. However, during the past five decades there has
been a slight increase of runoff from the headwaters into the Aksu
River, but due to land reclamation along the Aksu, the inflow from
the Aksu into the Tarim at Aral shows a decreasing trend (Tang and
Deng, 2010; Xu et al., 2005,2010).
Population growth and agricultural development, partly driven
by resettlement of people from other Chinese provinces to Xinjiang,
resulted in degradation of the natural ecosystems and desiccation of
the two terminal lakes of the Tarim river basin, Lop Norand Taitema,
by beginning of the 1970s (Feng et al., 2005; Hao et al., 2009; Song
et al., 2000; Zhang, 2006; Zhang et al., 2003). In order to balance
the water use between economic development (mainly irrigated
agriculture) and environmental flow along the Tarim, the Xinjiang
Government developed a water distribution program and water
quota system, which will be introduced in section 3(Peng et al.,
2014; Song et al., 2000; Tang and Deng, 2010; Zhang, 2006). The
aim of this paper is to assess the water consumption through
Fig. 1. Map of the Tarim Basin with its administrative units.
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e112
Please cite this article in press as: Thevs, N., et al., Water allocation and water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

irrigated agriculture and natural vegetation in the Aksu-Tarim river
basin against the background of this water quota system.
Firstly, we map the evapotranspiration (ET
a
) as water con-
sumption and calculate the net water loss of irrigated agriculture
and natural vegetation in the Aksu-Tarim river basin. Secondly, we
relate these results to the water distribution program and water
quota system. We employed the remote sensing method Simplified
Surface Energy Balance Index (S-SEBI) after Roerink et al. (2000)
and Gowda et al. (2008), in order to map ET
a
based on MODIS
satellite data for the vegetation seasons 2009, 2010, and 2011.
2. Study area
The study area is the Aksu-Tarim river basin (Figs. 1 and 3). Until
the 1970s, the three tributaries Aksu, Hotan, and Yarkant flowed
together at Aral and formed the Tarim River. Today, only the Aksu
River discharges perennially into the Tarim River as shown in Fig. 1
(Song et al., 2000). Therefore, in this paper we consider the Aksu-
Tarim river basin as one river basin with the two sub-basins of
the Aksu and the Tarim.
The headwaters of the Aksu are located north of Aksu City in the
Central Tianshan (Fig. 1) where snow and glacier melt, as well as
summer rainfall, deliver water into the Aksu River's headwaters
(Chou, 1960; Giese et al., 2005; Song et al., 2000). The long-term
average of water discharge from the headwaters into the Aksu
River just upstream of Aksu City is 8.06 km
3
/a (Fig. 3). Within each
year, about 75% of the annual runoff is discharged during July,
August, and September, which results in annual summer floods
(Song et al., 2000; Tang and Deng, 2010).
In 2009, this was the driest year within the three consecutive
years 2007, 2008, and 2009, during which the Tarim ceased to flow
downstream of Yingbaza in June (pers. observation of the authors).
In 2009, the discharge of the Tarim reached an extreme lowthat the
Tarim ceased to flow downstream of Xayar during spring and early
summer. During summer, which is the flood season, the discharge
in the middle and downstream section was lower than the average
discharge during spring, which is the low discharge time. In 2010,
the Tarim experienced one of the highest summer floods since the
1950s. Therefore, river branches and large areas under natural
vegetation were flooded. River branches and depressions remained
submerged until May/June 2011 (Fig. 2).
The climate is extremely arid and continental, as shown in
Table 1. From the foothills of the Tianshan Mountains over the
Tarim River toward the Taklamakan Desert, the annual precipita-
tion decreases from about 130 mm to less than 30 mm. Within the
study area, the precipitation ranges from about 70 mm north of the
Tarim River to 30 mm south of the Tarim (Liu, 1997). This is re-
flected by the climate data given in Table 1. The stations Kuqa and
Korla are located close to the foothills of the Tianshan, (i.e. north of
the Tarim River), while Aral and Tikanlik are located at the Tarim
River.
The natural vegetation is a mosaic composed of riparian forests,
wetlands, and shrub vegetation with P. euphratica Oliv, Phragmites
australis Trin. (ex Steud)., and Tamarix ramossisima Ledeb., or hal-
ophytes as key-stone species (Thevs et al., 2008). P. euphratica and P.
australis are obligate phreatophytes. These plants continuously
exploit the groundwater (Gries et al., 2003). Tamarix ramosissima is
a facultative phreatophyte, i.e. it strives to continuously connect
with the groundwater, but also can survive on moist soil without
tapping the groundwater directly (Cleverly et al., 2002; Gries et al.,
2003).
3. The water quota system in the Aksu-Tarim river basin
The water quota system for the Tarim River and its tributaries
was developed by 2005 and shall be enforced at the latest by 2020
(Peng et al., 2014; Thevs, 2011). Under average conditions, the Tarim
shall receive the following amounts of water: 3.42 km
3
/a from the
Aksu River, 0.90 km
3
/a from the Hotan River, and 0.33 km
3
/a from
the Yerkant River, which total to 4.65 km
3
/a at Aral, where the three
tributaries of the Tarim flow together (Table 2 and Fig. 3). Addi-
tionally, 0.45 km
3
/a have to be released from the Kaidu-Konqi River
into the Tarim lower reaches at Qala (Table 3 and Fig. 3).
Under average conditions, which means an annual inflow of
8.06 km
3
into the Aksu River (Table 2), the net water loss along the
Fig. 2. River stretch of the Tarim from Yingbaza to Iminqak. NDVIs from Landsat TM (path 144 row 31) from 2009, 2010, and 2011 (dates are written in the images). a: Dry river bed,
b: natural riparian vegetation, c: cotton fields, d: river bed during summer flood 2010 with submerged river banks, e: submerged depressions during flood season 2010, f: de-
pressions remaining submerged until spring 2011.
Table 1
Mean Air Temperature and precipitation at the climate stations Aral, Kuqa, Korla,
and Tikanlik (http://www.weatherbase.com). Aral, Kuqa, and Korla are indicated in
Fig. 1. Tikanlik is located 120 km southeast of Yuli (Fig. 2).
Climate station Aral Kuqa Korla Tikanlik
Time of record [a] 32 39 32 33
Position 40.30

N,
081.03

E
41.43

N,
082.57

E
41.45

N,
086.07

E
40.37

N,
087.42

E
Elevation [m a.s.l.] 1013 1100 933 847
Annual average air temperature [

C] 10 11 11 10
Average January air temperature [

C] 7778
Average July air temperature [

C] 25 25 26 26
Annual average precipitation [mm] 42 62 52 34
Table 2
Annual inflows into the Aksu River and into the Tarim River starting in Aral from the
three tributaries Aksu, Yarkant, and Hotan [km
3
](Tang and Deng, 2010). 50%
probability in this table refers to average conditions, as found back in Table 3.
Probability 90% 75% 50% 25%
Inflow into Aksu River upstream of irrigated land 6.68 7.25 8.06 8.85
Inflow into the Tarim at Aral from
Aksu 2.52 2.64 3.42 4.2
Yarkant 0 0 0.33 0.56
Hotan 0.2 0.64 0.9 1.55
Total inflow into the Tarim at Aral 2.72 3.28 4.65 6.31
Total inflow into the river basin of the Aksu and Tarim 6.88 7.89 9.29 10.96
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e11 3
Please cite this article in press as: Thevs, N., et al., Water allocationand water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

Aksu River must not exceed 4.64 km
3
/a (TMB, 2005) so that
3.42 km
3
/a flows into the Tarim (Fig. 3). Under low water inflow
conditions, i.e. 90% probability in Table 2 (6.68 km
3
/a), only
2.52 km
3
/a must be released into the Tarim River. Similar water
quota regulations are also applied for the Yarkant and Hotan River
(TMB, 2005).
Water quotas for the Tarim River are more detailed regarding
the water use of different sectors. Based on the hydrologic data of
the period from 1957 to 2000, the maximum annual water inflow at
Aral Station was 6.96 km
3
/a while the minimum was 2.56 km
3
/a.
Therefore, ten scenarios varying from 2.5 to 7.0 km
3
/a are made for
water quotas along the Tarim according to the inflow at Aral. For the
Tarim River, no probabilities as for the tributaries (Table 2) are used.
In Table 3, only the water allocation for the minimum, average, and
maximum inflow scenario are shown. When an average annual
inflow at the Aral Station is given (4.65 km
3
/a), the annual water
quota for the upper reaches of the Tarim is 2.01 km
3
/a in total with
0.41 km
3
/a allocated for irrigated agriculture, and 1.60 km
3
/a for
environmental flow.
Under these average conditions, 2.64 km
3
/a water must be
released into the middle reaches (Table 2 and Fig. 3), The annual
water quota for the middle reaches, i.e. from Yingbaza until Qala, is
2.14 km
3
in total, with 0.35, 0.12, and 1.66 km
3
allocated for irri-
gated agriculture, oil industry, and environmental flow,
respectively.
The remaining 0.51 km
3
/a, together with 0.45 km
3
/a carried
through channels from the Konqi River, which amount to 0.96 km
3
/
a, must be released from the Qala Gauging Station to the lower
reaches of the Tarim. Along the river stretch from the Qala Gauging
Station to the Daxihaizi Reservoir, 0.46 km
3
/a and 0.15 km
3
/a are
allocated for irrigation and environmental flow, respectively.
Finally, 0.35 km
3
/a water, as environmental flow, shall be released
from the Daxihaizi Reservoir into the lower reaches toward the
Taitema Lake which is the current terminal lake of Tarim river
(TMB, 2005; Tang and Deng, 2010). The water quota for irrigation is
kept constant from average inflow at Aral to the maximum inflow at
Aral, while the water quota for environmental flow changes ac-
cording to the inflow at Aral (Table 2;Tang and Deng, 2010).
4. Methods
4.1. Evapotranspiration mapping
In this paper, we mapped the actual evapotranspiration (ET
a
)of
the Aksu-Tarim river basin for the growing seasons 2009, 2010, and
2011, in order to retrieve the amounts of water consumed by irri-
gated agriculture and natural vegetation for representative sites
and on a river basin scale. We used the following MODIS satellite
data products, in order to cover the whole Aksu-Tarim river basin:
8-day land surface temperature (MOD11A2), 16-day albedo
(MCD43A3), and 16-day NDVI (MOD13A1).
The growing season was defined to last from 1st of April to 31st
of October in each year. Cotton as the major crop in the study area is
planted at the beginning of April and harvested until the end of
October (Protze, 2011). P. euphratica, as a keystone species of the
natural vegetation, flowers in April, and the leaves shoot in May and
fall until the end of October (Xinjiang Linkeyuan Zhisha Yanjiusuo,
1989).
ET
a
can be determined through: i) climate station data, either by
calculating a reference ET applying crop coefficients (Allen et al.,
1998) or through the Bowen Ratio method (e.g. Malek and
Fig. 3. Water inflow from the tributaries Aksu, Yarkant, and Hotan into the Tarim mainstream and water quotas along the Tarim mainstream under average conditions [km
3
/a], after
Thevs (2011) with I: irrigation and industry, E: environmental flow, O: oil exploitation. AeB: upper reaches, BeC: middle reaches, CeD: upper section of lower reaches, D and below:
lower reaches.
Table 3
Annual water quotas [km
3
/a] fixed within the Scheme of Surface Water Distribution
for water withdrawal for economic activities and environmental flow along the
upper, middle, and lower reaches of the Tarim River starting in Aral. Aral is located at
the confluence of the three tributaries Aksu, Yarkant, and Hotan. The three columns
refer to minimum, average, and maximum water inflow at Aral and respective water
allocation (Tang and Deng, 2010).
River section and water user Inflow at Aral and
respective water allocation
Minimum Average Maximum
Upper reaches
Inflow at Aral 2.50 4.65 7.00
Agriculture 0.37 0.41 0.41
Environmental flow 0.72 1.60 2.62
Runoff released into middle reaches 1.42 2.64 3.97
Middle reaches
Runoff at Yingbaza 1.42 2.64 3.97
Oil exploitation 0.12 0.12 0.12
Agriculture 0.30 0.35 0.35
Environmental flow 0.73 1.66 2.74
Runoff released into lower reaches 0.27 0.51 0.76
Lower reaches
Water transferred from Kenqi river 0.45 0.45 0.45
Runoff at Qala Station 0.72 0.96 1.21
Agriculture 0.41 0.46 0.46
Environmental flow 0.09 0.15 0.31
Runoff released into the lower
reaches below Daxihaizi
0.23 0.35 0.44
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e114
Please cite this article in press as: Thevs, N., et al., Water allocation and water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

Bingham, 1993) as used by Hou et al. (2010) to determine evapo-
transpiration of P. euphratica at the Heihe River in Inner Mongolia,
China, ii) lysimeters (e.g. Sammis, 1981), iii) eddy co-variance
measurement devices (e.g. Cleverly et al., 2002; Scott et al.,
2008), or iv) using the residual of the water balance equation
(e.g. Reddy et al., 2012). These methods measure ET
a
at a specific
point but cannot cover large, diverse, and remote areas. Therefore
algorithms have been developed, in order to map ET
a
on the basis of
satellite images (Landsat, MODIS, or ASTER), e.g. SEBAL
(Bastiaanssen, 1995; Bastiaanssen et al., 2002,2005), METRIC (Allen
et al., 2005), SEBS (Su, 2002), and S-SEBI (Roerink et al., 2000;
Sobrino et al., 2005, 2007), as reviewed by Gowda et al. (2007,
2008) and Senay et al. (2011).
On the basis of the Simplified Surface Energy Balance Index (S-
SEBI), developed by Roerink et al. (2000), the latent heat flux can be
calculated as:
LE ¼LðR
n
GÞ(1)
with LE, L,R
n
, and Gbeing latent heat flux [W/m
2
], evaporative
fraction, net radiation [W/m
2
], and soil heat flux [W/m
2
], respec-
tively. When calculating daily values instead of instantaneous
values, G¼0(Sobrino et al., 20 05, 2007) so that Eq. (1) is simplified:
LE
d
¼L
d
R
nd
(2)
with LE
d
,L
d
, and R
nd
referring to daily latent heat flux sum [MJ],
daily evaporative fraction, and daily net radiation sum [MJ].
In the S-SEBI (Roerink et al., 2000; Sobrino et al., 2007) it is
assumed that net radiation (R
nd
) is converted into evapotranspi-
ration, i.e. potential evapotranspiration (ET
pot
). Following this
assumption, ET
a
, can be calculated as follows:
ET
a
¼LET
pot
(3)
Thereby, we estimated R
nd
, and thus also ET
pot
, on the basis of the
date, geographical position, transmissivity of the atmosphere, land
surface temperature (MOD11A2), and albedo (MCD43A3), applying
the module i.evapo.potrad (http://grasswiki.osgeo.org/wiki/
AddOns/GRASS_6#GIPE) of the software package GRASS GIS 6.4
(http://grass.fbk.eu/). The transmissivity was estimated from the
data of a climate station set up in Iminqak (Fig. 1).
The evaporative fraction (L) is the part of the ET
pot
, which is
realized as ET
a
. Thus, over well-watered vegetation, e.g. wetland
vegetation, we can assume Lz1 and ET
a
zET
pot
. The land surface
temperature is low, because the net radiation is consumed by
evapotranspiration. In contrast, at places without any vegetation or
other moisture, Land ET
a
are zero. Here, the land surface tem-
perature is high, because the net radiation goes into the sensible
heat flux. Between these two extremes it is assumed that Land
land surface temperature have a linear relationship (Roerink et al.,
2000).
The evaporative fraction is calculated as follows:
L¼TH Tx
TH TC (4)
whereby TH, TC, and Tx refer to the land surface temperatures
(MOD11A2) at the hot pixel, cold pixel, and pixel, for which Lis
calculated, respectively. The satellite records TH, TC, and Tx at its
overpass time, which is between 10:50 and 11:10 local time for
MOD11A2 according to the day view time channel. So, Lis calcu-
lated for the overpass time. According to Bastiaanssen (2000) and
Farah et al. (2004),Lremains constant during daytime so that L
calculated for satellite overpass time can be used to calculate daily
ET
a
.
Cold pixels were chosen from wetlands with dense reed vege-
tation interrupted by small open waters, which do not fall dry
during the growing season. Hot pixels were selected from areas free
of vegetation but lying adjacent to the riparian vegetation along the
Tarim River. Hot pixels could be selected from fallow fields as
suggested by Waters et al. (2002), because there were none avail-
able, which fitted into one of the MODIS pixels. At the hot pixels,
L¼0 and ET
a
¼0. Thus, all the radiation is converted into sensible
heat flux so that the land surface temperature is high.
Following this approach, 8-day ET
a
maps were produced ac-
cording to the 8-day MOD11A2 satellite images. These ET
a
maps
were aggregated to monthly ET
a
maps and finally summed up to ET
a
maps for the growing seasons 2009, 2010, and 2011. The MOD11A2
product has a resolution of 1 km by 1 km so that all ET
a
maps were
produced in this resolution.
In Iminqak, in 2009 and 2011 a climate station was operated, in
order to measure ET
a
for a validation of the MODIS ET
a
. The climate
station was installed on the ground of the ranger station in ImInqak
due to security reasons. This station is located near the edge of the
riparian forest. The climate station was equipped with sensors for
incoming and outgoing radiation (pyranometers CMP3, Kipp &
Zonen) and ventilated air temperature/humidity sensors in two
different heights so that ET
a
was calculated with the Bowen Ratio
method (Malek and Bingham, 1993). One air temperature/humidity
sensor was mounted 2 m above soil surface, while the other sensors
were mounted 10 m above soil surface level. Electricity supply
came from two 12 V car accumulators, which were charged by the
nearby ranger station. Data were recorded every 0.1 s and stored by
a data logger at 15 min intervals.
4.2. Retrieve ET
a
data from irrigated agriculture and natural
vegetation
The ET
a
maps of each year's growing season were laid over a map
of irrigated land in order to retrieve the ET
a
of the whole irrigated
land area. The average annual precipitation (Liu, 1997) was sub-
tracted from ET
a
, in order to calculate the net water loss of the
irrigated land (Scott et al., 2008). The map of the irrigated land was
manually digitized from Landsat satellite images from 2009 (cf.
Thevs, 2011). ET
a
values for cotton, as the dominant crop, were
retrieved from MODIS pixels (1 km 1 km), which were
completely covered by irrigated fields. Thus, places listed in Table 4
between 3 and 39 pixels of the MODIS ET
a
maps were selected
manually. Riverbed pixels were selected manually from MODIS
pixels, as well. Pixels, which were completely covered by the
riverbed, qualified as riverbed pixels.
MODIS ET
a
pixels, from which we read the ET
a
of the natural
vegetation, were selected from two Quickbird satellite images.
These Quickbird images each cover a cross section through the
Tarim River and adjacent natural vegetation south of Xayar and at
Iminqak (Fig. 3). The two Quickbird images date from 2009/07/15
and 2009/06/09, correspondingly. As Quickbird has a ground res-
olution of 0.6 m 0.6 m, only single trees and shrubs are visible. For
each MODIS pixel, the total vegetation coverage in percent was
estimated visually.
Table 4
Sum of ET
a
during the growing seasons (2009e2011) measured with the climate
station Iminqak and detected through remote sensing.
Year Sum of ET
a
during growing season [mm]
Climate station Remote sensing Deviation [%]
2009 612 611 0.2
2010 794
2011 836 929 10.8
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e11 5
Please cite this article in press as: Thevs, N., et al., Water allocationand water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

4.3. Calculating water balances for the Aksu-Tarim river basin
Water balances were calculated after Wu et al. (2013), in order
to estimate how much the water quotas were met or exceeded:
PET OþI¼Dgw þDs(5)
with P being precipitation, ET referring to total evapotranspiration,
Obeing basin outflow, Ibeing inflow in the basin, and
D
gw and
D
s
referring to change in groundwaterand soil moisture storage. Basin
inflows are given Table 2. Thereby, we used 90% probability (min-
imum inflow), 25% probability (maximum inflow), and 50% prob-
ability (average inflow), in order to calculate water balances for
2009, 2010, and 2011, respectively, because measured inflow and
runoff data are not available to the public after 2005. Basin outflow
for the whole Aksu-Tarim river basin is zero, while outflow for the
Aksu river basin is given in Table 2. Precipitation has been taken
from Liu's(1997)precipitation map and subtracted from the ET
a
maps for the growing seasons 2009, 2010, and 2011, respectively, in
order to yield net water loss (NWL) of the Aksu-Tarim river basin.
Eq. (5) was simplified as follows:
INWL ¼Dgw þDs(6)
A negative term
D
gw þ
D
swould indicate that the water quota was
exceeded and groundwater and soil moisture was depleted, in or-
der to meet the water consumption of irrigation and natural
vegetation.
5. Results
The sum of the daily ET
a
values over the vegetation season 2009
measured at the climate station Iminqak nearly equals the ET
a
detected from the MODIS satellite images (Table 4). In 2011, the
MODIS ET
a
is 10.8% higher than the ET
a
measured at the climate
station. The mean ET
a
of agricultural land used for cotton along the
Aksu River, Tarim upper reaches, Puhui, and Lopnor, summed up
over the growing seasons 2009, 2010, and 2011, respectively,
ranged from 884 to 1198 mm (Table 5). In contrast, the mean ET
a
of
agricultural land used for cotton in Yingbaza, Dongkutan, and
Tikanlik ranged between 501 and 666 mm within the period of
2009e2011. Correspondingly, the mean ET
a
of other agricultural
land ranged from 884 to 1226 mm along the Aksu River and Tarim
upper reaches, while it only was between 656 and 674 mm in
Tikanlik.
In most of the locations listed in Table 5, the ET
a
of the vegeta-
tion season dropped slightly from 2009 to 2010 and increased again
from 2010 to 2011. This trend is most pronounced for Dongkutan,
Lopnor, and Tikanlik. In contrast, the ET
a
of the natural vegetation
increased slightly from 2009 to 2010 and sharply from 2010 to 2011.
For example, the mean ET
a
of natural vegetation of a total coverage
of 70% and more was determined with 798, 889, and 1115 mm in
2009, 2010, and 2011, respectively. Fig. 4 also reflects these two
trends.
The mean ET
a
at the hot anchor pixels did not change over the
study period and was 1, 5, and 1 mm in 2009, 2010, and 2011,
correspondingly. The mean ET
a
of the cold pixels was 1,687, 1,660,
and 1790 mm in 2009, 2010, and 2011, correspondingly, thus
showing an increase only from 2010 to 2011.
The net water loss from water reservoirs, agricultural land, and
natural vegetation along the whole Aksu and Tarim River ranges
from 10.13 to 10.61 km
3
/a in 2009, to 12.84e13.41 km
3
/a in 2011 as
shown in Table 6. The groundwater and soil moisture storage
(
D
gw þ
D
s)is3.25 to 3.73 km
3
,0.1e0.53 km
3
, and 3.55 to
4.12 km
3
in 2009, 2010, and 2011, respectively for the whole Aksu-
Tarim river basin, indicating a decreasing groundwater and soil
moisture storage in 2009 and 2011 (Table 6). For the Tarim River
alone, the corresponding
D
gw þ
D
sare 0.22 to 0.71 km
3
,
2.55e2.97 km
3
, and 0.18 to 0.76 km
3
. This corresponds to an
increase of the groundwater and soil moisture storage in 2010 and
slight decreases in 2009 and 2011. At the Aksu River,
D
gw þ
D
sis
negative during all years studied (Table 6). In 2009 and 2011, the net
Table 5
ET
a
[mm] of the agricultural land and natural vegetation along the Aksu and Tarim River from 2009, 2010 and 2011.The names of the different agricultural fields are displayed
in Fig. 1.nenumber of MODIS pixels, Std. Dev. estandard deviation, Sign.
a
¼0.05 eletters indicate significant differences of mean of ET
a
at
a
¼0.05 (Tukey post hoc test).
Land cover 2009 2010 2011
NET
a
mean [mm] Std. Dev. Sign.
a
¼0.05 NET
a
mean [mm] Std. Dev. Sign.
a
¼0.05 NET
a
mean [mm] Std. Dev. Sign.
a
¼0.05
Cold Pixels 10 1687 373 a 10 1660 298 a 10 1790 248 a
Irrigated land
Karatal 19 1145 22 b 19 1133 32 bcd 19 1143 37 bc
Bexerik 22 1164 103 b 22 1078 41 bcde 22 1124 58 bc
Aral 33 1198 119 b 33 1043 84 bcde 33 1164 96 bc
Bingtuan 23 1068 107 bc 23 1004 83 bcdef 23 1130 59 bc
Xayar 18 1115 81 b 18 884 69 efg 19 1027 33 bc
Yingbaza 4 644 172 de 4 641 189 ghij 4 614 276 ef
Dongkutan 3 560 126 de 3 504 98 jk 3 756 127 de
Puhui 39 1108 187 b 39 910 154 def 39 963 139 bcd
Lopnor 20 1063 74 bc 20 947 59 cdef 20 959 66 cd
Tikanlik 19 580 199 de 16 501 212 jk 19 666 159 e
River bed pixels 41 753 142 d 41 882 129 efg 41 1167 132 bc
Pixels natural vegetation coverage
70% 3 798 73 cd 3 889 116 def 3 1115 66 bc
60e70% 16 760 113 d 16 774 127 fghi 16 978 132 bcd
50e60% 26 715 149 de 26 780 166 fghi 26 960 236 bcd
40e50% 39 556 204 de 39 604 234 hij 39 756 286 de
30e40% 42 514 296 def 42 553 314 ij 42 688 356 e
20e30% 39 426 264 ef 39 470 285 jk 39 603 334 ef
10e20% 32 245 238 fg 32 284 221 kl 32 365 297 f
<10% 77 8 153 g 77 51 136 lm 77 71 174 g
Hot pixels (n¼124) 163g 554m 154g
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e116
Please cite this article in press as: Thevs, N., et al., Water allocation and water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

water loss on the irrigated land alone slightly exceeded the inflow
available for the Aksu River (Table 6).
Along the Aksu River, the net water loss of the irrigated agri-
culture was more than twice the net water loss of the natural
vegetation in 2009 and 2010, and almost twice the net water loss of
the natural vegetation in 2011. In contrast, along the Tarim River's
upstream and midstream section, the net water loss of the natural
vegetation was higher than the net water loss of the irrigated
agriculture (Fig. 4).
The net water loss through agricultural land along the upstream
section of the Tarim River (Table 6) was in the range of the corre-
sponding quota of 0.41 km
3
/a under average to high runoff condi-
tions and 0.37 km
3
/a under low runoff conditions, respectively
(Table 3). In 2009, the net water loss of the natural vegetation
(Table 6,Fig. 4) exceeded the corresponding quota fixed for low
inflow conditions of 0.72 km
3
/a. In 2011, the net water loss of the
natural vegetation (2.34e2.53 km
3
during the growing season) was
close to the corresponding water quota established for maximum
water inflow conditions (2.62 km
3
/a).
Along the middle and lower reaches, the water quotas for irri-
gation (Table 3) were not reached by the net water loss of irrigated
land (Table 6,Fig. 4) during the three years studied. The net water
loss of the natural vegetation along the mid-stream section of the
Tarim lagged far behind the respective water quotas for environ-
mental flow during 2009, 2010, and 2011. In 2009 and 2010, the net
water loss of the natural vegetation during the growing season was
0.53e0.55 km
3
and 0.77e0.79 km
3
, respectively (Table 6), while the
water quotas for environmental flow were 0.73 km
3
/a (low inflow
at Aral reflecting 2009) and possibly 2.74 km
3
/a (maximum inflow
at Aral reflecting 2010). In 2011, the net water loss of the natural
vegetation was 1.15e1.19 km
3
during the growing season (Table 6),
while the corresponding water quota was 1.66 km
3
/a (Table 3).
Along the lower reaches of the Tarim River, the net water loss of the
natural vegetation was 0.11 km
3
, 0.09 km
3
, and 0.27 km
3
during the
growing seasons 2009, 2010, and 2011, respectively. The corre-
sponding water quotas were 0.09 km
3
/a (low inflow at Aral
reflecting 2009), possibly 0.31 km
3
/a (maximum inflow at Aral
reflecting 2010), and 0.15 km
3
/a (average inflow at Aral reflecting
2011). Thus, in 2009 and 2011, the net water loss of the natural
vegetation was in the range of, or exceeded the water quota fixed
for environmental flow at the Tarim down-stream section.
In Fig. 4, it is visible that the net water loss of the irrigated land
drops from 2009 to 2010 and increases from 2010 to 2011, while the
net water loss of the natural vegetation increases from 2009 over
2010 to 2011. This trend is also visible when referring to land cover
classes in Table 5. The evaporation from water reservoirs is constant
during the three years, except for the lower reaches. There, the
evaporation from reservoirs triples from 2009 to 2011 (Table 6,
Fig. 4).
6. Discussion
6.1. MODIS ET
a
and ET
a
measured by climate station
The deviation of the MODIS ET
a
from the climate station ET
a
, i.e.
0.1% in 2009 and 10.8% in 2011, is within the range of such de-
viations reviewed by Bastiaanssen et al. (2005). The higher devia-
tion in 2011 can be explained by the water, which submerged river
branches and depressions along the Tarim until spring 2011 (Fig. 2).
Periodical open waters are also located within the MODIS pixel, in
which the climate station is located. Thus, in 2011 these periodi-
cally submerged depressions and river branches increase the
evapotranspiration of the MODIS pixel at the climate station.
Furthermore, we also have to recall on the discharge pattern of
the three years 2009e2011 (pers. observation by the authors, cf.
Fig. 2). In 2009, the Tarim River fell dry during spring and early
summer downstream of Xayar and did not carry a summer flood.
The summer flood in 2010 was one of the highest floods since the
1950s. River branches and depressions remained filled with water
until spring 2011. Thus, the groundwater layer was refilled all over
the Tarim flood plain. The summer flood in 2011 was an average
flood event. The surface water runoff is reflected in groundwater
level fluctuations (Chen et al., 2006; Ye et al., 2009).
6.2. ET
a
of cotton
The ET
a
of cotton along the Aksu River, Tarim upper reaches, in
Puhui, and in Lopnor, is 884 to 1198 mm, which is in the range of
other studies in Central Asia (Chapagain et al., 2006; Conrad et al.,
2007; Steduto et al., 2012). The low ET
a
for agricultural land used for
cotton and other crops in Yingbaza, Dongkutan, and Tikanlik may
be explained as follows. Along the middle and lower reaches of the
Fig. 4. Net water loss from water reservoirs, irrigated land, and natural vegetation (ET
a
eP) along the Aksu River and Tarim River (upstream, middle, and downstream section)
during the growing seasons 2009, 2010, and 2011.
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e11 7
Please cite this article in press as: Thevs, N., et al., Water allocationand water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

Tarim River, large parts of the agricultural land is fallow, because
during 2007, 2008, and 2009 the Tarim River ceased to flow during
early summer, when irrigation demand increases (Bothe, 2010;
Thevs, 2011). Farmers thus exploit groundwater in order to over-
come the season during which the Tarim runs dry. The ground-
water supply can only satisfy the irrigation demand for a part of the
agricultural land, therefore nothing was planted or the crop was
lost on part of the agricultural land. As MODIS has a spatial reso-
lution of 1 km by 1 km, the pixels covering the agricultural land
along the middle and lower reaches contain land planted with
cotton and fallow land, which results in mixed pixels (Settle and
Drake, 1993). Therefore, the ET
a
of mixed pixels detected with
MODIS is an average of the ET
a
of cotton and fallow land.
Along the Aksu River and Tarim upstream there was no pro-
nounced water shortage during 2007e2009, because these regions
are located upstream from other water users. Additionally, in Xayar
water also can be diverted from the
€
Og€
an River into the irrigated
land along the Tarim River. In Puhui and Lopnor, though located in
the vicinity of the Tarim downstream section, all agricultural land is
planted with cotton. The ET
a
is similar to the Aksu River and the
Tarim upstream section, because these two areas receive water
from the Kenqi River rather than the Tarim (Song et al., 2000).
Due to the low runoff during 2009, more land remained fallow
along the Aksu and Tarim River than during the years before. This
may explain that the ET
a
of cotton decreased from 2009 to 2010. In
2010/2011 more water was available in reservoirs and natural de-
pressions so that more land was allocated for cotton than before.
6.3. ET
a
of the natural vegetation
The ET
a
of the natural vegetation strongly depends on the total
coverage. In Table 7, values of water consumption of P. euphratica,T.
ramosissima, and Elaeagnus angustifolia L., the keystone species
dominating the natural riparian vegetation in Central Asia, are
displayed.
The water consumptions reported by Khamzina et al. (2009) are
considerably higher than the corresponding MODIS ET
a
for natural
vegetation with a total coverage of 70% or more (Tables 5 and 7).
This might be explained through the high groundwater level under
the sites Khamzina et al. (2009) investigated and the presumably
high tree density, where the trees had been planted. The water
consumptions of T. ramosissima (Cleverly et al., 2002) correspond
with the MODIS ET
a
of this study (Tables 5 and 7). The water con-
sumption measured by Thomas et al. (2006) in Table 7 is lower
compared to the MODIS ET
a
(Table 5), due to the P. euphratica stand
measured by Thomas et al. (2006) which had a greater coverage
than 20%.
The water consumption of the natural vegetation along the
Tarim River increased from 2009 to 2010 and again from 2010 to
2011 (Tables 5 and 6,Fig. 4). This also is explained with the different
discharge into the Tarim in 2009, 2010, and 2011. During the
summer flood in 2010, which was among the highest since the
1950s, large areas and many river branches were flooded. The
evaporation from these surface waters increased the ET
a
detected.
Furthermore, the groundwater recharge was strongly enhanced
through this flood. Therefore, the water supply for part of the
natural vegetation might have increased, which results in higher
transpiration. This may apply for T. ramosissima, which being a
facultative phreatophyte mainly uses groundwater as water source,
but also can survive on moist soils (Busch et al., 1992). Thus, the
transpiration might be decreased, when the groundwater drops
and only soil moisture is available. However, the transpiration rate
might increase, when the groundwater level increases again. Hao
et al. (2010) assumed that hydraulic lift from the groundwater to-
ward the soil surface by P. euphratica contributes to the water
supply of herbs in the undergrowth of riparian forests. So, during
2009, when the groundwater level was lower than in 2010 and
2011, this hydraulic lift probably was lower, too, because
P. euphratica needed a larger share of the water uptaken from the
groundwater for itself. During 2010 and 2011, the hydraulic lift
increased so that the water consumption of the undergrowth
increased, too.
After the high flood in 2010, many river branches and de-
pressions carried water until spring 2011 (cf. Fig. 3). Also,
Table 6
Net water loss [km
3
/a] of water reservoirs, irrigated land, and natural vegetation along the Aksu and Tarim River and water balances for the Aksu-Tarim river basin during the
growing seasons 2009, 2010, and 2011. For some parts of the Aksu and Tarim river basin it is unclear whether they hydrologically belong to this river basin or neighboring
basins, e.g. in Xayar water can be divertedfrom the Tarim River or from the
€
Og€
an River. Therefore, ranges are given. Inflow data is taken from Table 2.Inflowfor the Ak su River is
the difference between inflow upstream of the irrigated land and outflow to the Tarim River (Table 2).
River stretch Aksu Tarim Total Aksu
and Tarim
Upstream Midstream Downstream Total
Growing season 2009
Water Reservoirs 0.20 0.05 0.00 0.05 0.10 0.30
Irrigated land 4.83 0.34e0.69 0.14 0.13 0.61e0.96 5.44e5.79
Natural vegetation 2.15 1.58e1.70 0.53e0.55 0.11 2.23e2.36 4.38e4.52
Total NWL 7.18 1.97e2.44 0.68e0.69 0.30 2.94e3.43 10.13e10.61
Inflow (IeO for the Aksu) 4.16 2.72 10.13
D
gw þ
D
s3.02 0.22 to 0.71 3.25 to 3.73
Growing season 2010
Water Reservoirs 0.21 0.06 0.00 0.06 0.12 0.33
Irrigated land 4.58 0.32e0.59 0.17 0.09 0.58e0.85 5.16e5.43
Natural vegetation 2.30 1.78e1.91 0.77e0.79 0.09 2.65e2.80 4.95e5.10
Total 7.09 2.16e2.57 0.94e0.96 0.24 3.34e3.76 10.43e10.86
Inflow (IeO for the Aksu) 4.65 6.31 10.96
D
gw þ
D
s2.44 2.55e2.97 0.1e0.53
Growing season 2011
Water Reservoirs 0.21 0.06 0.00 0.15 0.21 0.42
Irrigated land 5.04 0.41e0.76 0.25 0.20 0.85e1.21 5.90e6.25
Natural vegetation 2.75 2.34e2.53 1.15e1.19 0.27 3.76e3.99 6.51e6.74
Total 8.00 2.81e3.35 1.40e1.44 0.62 4.83e5.41 12.84e13.41
Inflow (IeO for the Aksu) 4.64 4.65 9.29
D
gw þ
D
s3.36 0.18 to 0.76 3.55 to 4.12
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e118
Please cite this article in press as: Thevs, N., et al., Water allocation and water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

groundwater levels and soil moisture contents were higher
compared to 2010 and 2009. Therefore, such surface waters already
contributed to ET
a
from the beginning of the vegetation season in
2011. Additionally, more water was available for the vegetation
from the beginning of the vegetation season in 2011, which pre-
sumably resulted in an increased ET
a
. As the summer flood in 2011
reached average levels, no water shortage occurred during the
vegetation season, therefore, the ET
a
of the natural vegetation
increased again from 2010 to 2011.
6.4. Water balance and water quotas
The net water loss along the upper, middle, and lower reaches of
the Tarim during the growing seasons 2009 to 2011 ranged be-
tween 1.97 and 3.35 km
3
, 0.68 and 1.44 km
3
, and 0.3 and 0.62 km
3
,
respectively. Xu et al. (2005) give values for water consumption,
which perhaps refer to the net water loss of 1.7 km
3
, 2.3 km
3
, and
0.61 km
3
for the upper, middle, and lower reaches of the Tarim. The
differences between the results of Xu et al. (2005) and the results of
this study can be explained as follows.
The area under irrigation along the Tarim has increased until
now with the largest increases along the upper reaches. Therefore,
the water consumption and thus net water loss along the upper
reaches increased from the time span of the data of Xu et al. (2005)
to the data from 2009 to 2011 of this study. Along the middle
reaches dykes were built until 2004 (Tang and Deng, 2010), which
cut off part of the inland delta along the middle reaches so that the
net water loss decreased from Xu et al. (2005) toward the results of
our study.
The total amount of water consumed in the Aksu-Tarim river
basin exceeded the inflow during 2009 and 2011 (Table 6).
Considering only the Aksu, consumption exceeded inflow during all
three years. This finding agrees with Xu et al. (2010), who found
that there was an increasing trend of the inflow into the Aksu River,
while the runoff from the Aksu into the Tarim at Aral (Fig. 1)
showed a decreasing trend. The gap between inflow and water
consumption partly can be explained through exploitation of
groundwater, as Siebert et al. (2006) indicate that 10%e20% of the
irrigated area in the Tarim Basin is equipped for irrigation with
groundwater. Groundwater exploitation for irrigation from fossil
groundwater aquifers also was documented around Yingbaza
(Thevs, 2011). Furthermore, the MODIS ET
a
of the sparse vegetation
may have been over-estimated (cf. Table 7).
The water consumption for agriculture and environmental
flow along the Tarim River downstream of Yingbaza, i.e. middle
and lower reaches of the Tarim River, falls behind the respective
quotas, while the quotas for the Tarim upstream of Yingbaza are
met or exceeded. This finding agrees with Peng et al. (2014),who
investigated the perspective of land users and other stakeholders
with regard to the implementation of the water quota system of
the Aksu-Tarim river basin. During 2010, there was a surplus in
groundwater and soil moisture storage of 2.55e2.97 km
3
along
the Tarim. This surplus is due to the extremely high summer
flood of that year as discussed in section 6.3. In 2010, there was
increase in groundwater storage, but also in soil moisture, at
least in areas close to river branches and depressions, which
carried water until spring 2011 (Fig. 2). Therefore, in this paper
we refer to
D
gw þ
D
srather than assuming
D
s¼0 as suggested
by Wu et al. (2013). In 2011, the surplus groundwater and soil
moisture from 2010 was partly consumed as indicated by the
negative
D
gw þ
D
sof 0.18 to 0.76 km
3
. However, as the
surplus of 2010 was much larger than
D
gw þ
D
sin 2011, the
flood in 2010 built up groundwater and soil moisture storage for
water users for longer than 2011.
Similar water balance studies at Basin level under arid climate
conditions were carried out by Guerschmann et al. (2008) for the
Murray-Darling Basin, where irrigated crops and floodplain vege-
tation showed a negative water balance, too, Barnett and Pierce
(2008) for the Colorado River, Wu et al. (2013) for the Hai River
Basin in Northern China, and by Hochmuth et al. (submitted for
publication) for the Heihe river basin in Northwestern China. In
the Hai Basin, the target water consumption as defined after sci-
entific assessments was exceeded by 6.73 km
3
/a during 2002e2007
(Wu et al., 2013). As, according to Wu et al. (2013), this excess water
consumption cannot be avoided by only introducing water saving
measures, the South to North Water Transfer Project is needed, in
order to carry water from the Yangze river basin into the Hai River
Basin. For the Aksu-Tarim river basin such water transfer is not an
option, because the neighboring large river basins (Indus, Syr
Darya, Amu Darya, or Ili) are located outside of China, are far away,
and water stressed themselves (Shen and Chen, 2010) so that a
water transfer would transfer water stress from one basin to the
other, too. For the Colorado River a reduction of inflow of 10%e30%
is expected due to climate change. Now already all water is
distributed among water consumers. Therefore, the question of
who has to reduce its water consumption by how much must be
solved (Barnett and Pierce, 2008). This is similar to the situation in
the Aksu-Tarim river basin. If the water consumption upstream is
reduced to a level, which does not require exploitation of fossil
groundwater and which is within the stipulated water quota, the
Table 7
Water consumption (ET
a
) of riparian vegetation types in Central Asia over the growing season.
Vegetation type ET
a
[mm] Source Corresponding natural
vegetation of this study
(cf. Table 3)
Populus euphratica, Tarim Basin (China), groundwater
4 m, tree density 2300e3400 trees/ha
192e392 Thomas et al. (2006) 10%e20% total coverage
Tamarix ramosissima, Tarim Basin (China) 92e180 Thomas et al. (2006) 10%e20% total coverage
Populus euphratica, Ejina (Heihe Basin, Inner Mongolia, China) 447 Hou et al. (2010) 20%e30% total coverage
Elaeagnus angustifolia, Khorezm, Uzbekistan, planted on
shallow (0.9e2 m) groundwater
1250 Khamzina et al. (2009) Exceeds MODIS ET
a
of natural
vegetation with total coverage
of 70% and more
Populus euphratica, Khorezm, Uzbekistan, planted on
shallow (0.9e2 m) groundwater
1030 Khamzina et al. (2009) Exceeds MODIS ET
a
of natural
vegetation with total coverage
of 70% and more
Tamarix ramosissima, Rio Grande (USA), LAI ¼2.5, not flooded 740 Cleverly et al. (2002) 50%e70% total coverage in 2009,
i.e. no summer flood
Tamarix ramosissima, Rio Grande (USA), LAI ¼3.5, flooded
during spring
1220 Cleverly et al. (2002) Total coverage 70% and more in
2011, i.e. vegetation partly
submerged.
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e11 9
Please cite this article in press as: Thevs, N., et al., Water allocationand water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

issue must be tackled, how the resulting reduction of water con-
sumption will be distributed. Hochmuth et al. (submitted for
publication) found a slightly positive water balance for the Heihe
River, including its downstream, so that it was concluded that the
water quotas there were fulfilled. This might be explained, as the
water allocation along the Heihe has received more attention than
the water allocation along the Tarim. In the case of the Heihe river
basin the Central Government of China urged the two provinces
Gansu and Inner Mongolia, which share that river basin, to enforce
the water quotas (Zhang, 2005), while in the Tarim Basin the
Province Government has to enforce the water quotas so that the
political pressure is lower.
7. Conclusion
We mapped the evapotranspiration (ET
a
) for irrigated agricul-
ture, natural vegetation, and water reservoirs in the Aksu-Tarim
river basin, Xinjiang, China for the growing seasons 2009, 2010,
and 2011. Water balances were calculated and compared with the
water quotas fixed for the Aksu and Tarim River. We used the S-SEBI
approach to map ET
a
. This approach yielded accurate results for
irrigated agriculture and dense natural vegetation, but presumably
over-estimated the ET
a
of sparse vegetation.
The water balance for the Aksu river basin was negative during
2009, 2010, and 2011. The water quota fixed for the Aksu River
does not differentiate between water consumption of irrigated
agriculture and natural vegetation. The water consumption of the
natural vegetation, in addition to the irrigated agriculture, resul-
ted in a negative water balance. Further research works should
investigate more closely the ET
a
of the natural vegetation in the
Aksu river basin, and the water quotas for the Aksu River should
be refined regarding the natural vegetation, i.e. environmental
flow.
In the Tarim river basin, the water quota allocated for irrigated
agriculture and environmental flow along the mid and down-
stream area of the Tarim River are not met, while the water con-
sumption along the Tarim upstream often exceeds the quotas
given. Thus, the limits specified for the Tarim upstream and the
Aksu River must be enforced in order to ensure sufficient water
supply to mid and downstream areas of the Tarim. Considerable
amounts of groundwater, including fossil groundwater, are
exploited for irrigation along the Aksu and Tarim River, which
must be regarded as exploitation of a non-renewable resource.
The groundwater exploitation must be limited to a renewable
amount without an impact on groundwater supply to the natural
vegetation, in order to maintain the natural vegetation as stipu-
lated in the water quota system in the Aksu-Tarim river basin.
Water users along the mid and downstream river stretches of the
Tarim depend on the enforcement of rules and to some extent, on
good will of the water users upstream, in order to secure the water
amount needed for irrigation and environmental flow. A situation
of non-secure water supply downstream poses risks to irrigated
agriculture. Therefore, we propose that, in addition to the
enforcement of rules, land use systems should be based on
indigenous plants. These may include Apocynum venetum as a
textile plant or Glycyrrhiza spec. and Alhagi sparsifolia as medicinal
plants. These should be proposed because they are species that
exploit the groundwater and remain productive in water scarce
years, as seen in 2009.
Acknowledgements
We thank the Bauer-Hollmann Foundation and the Rudolf and
Helene Glaser-Foundation for the funding of this study within the
Junior Research Group Adaptation Strategies to Climate Change and
Sustainable Land Use in Central Asia (Turkmenistan and Xinjiang,
China).
References
Allen, R.G., Pereira, L.S., Raes, D., Smith, M., 1998. Crop Evapotranspiration e
Guidelines for Computing Crop Water Requirements. FAO Irrigation and
Drainage Paper 56. FAO, Rome.
Allen, R.G., Tasumi, M., Morse, A., Trezza, R., 2005. A Landsat-based energy balance
and evapotranspiration model in Western US water rights regulation and
planning. Irrigat. Drain. Syst. 19, 251e268.
Barnett, T.P., Pierce, D.W., 2008. When will Lake Mead go dry? Water Resour. Res.
44, W3201.
Bastiaanssen, W.G.M., 1995. Regionalization of Surface Flux Densities and Moisture
Indicators in Composite Terrain. Ph.D. Thesis. Wageningen University,
Wageningen.
Bastiaanssen, W.G.M., 2000. SEBAL-based sensible and latent heat fluxes in the
irrigated Gediz Basin, Turkey. J. Hydrol. 229, 87e100.
Bastiaanssen, W.G.M., Ahmad, M., Yuan, C.M., 2002. Satellite surveillance of evap-
orative depletion across the Indus Basin. Water Resour. Res. 38, 1273e1282.
Bastiaanssen, W.G.M., Noordman, E.J.M., Pelgrum, H., Davids, G., Thoreson, B.P.,
Allen, R.G., 2005. SEBAL Model with remotely sensed data to improve water
resources management under actual field conditions. J. Irrigat. Drain. Eng.,
85e93. Jan./Feb.
Bothe, J., 2010. The Water Use of Cotton Cultivation in the Tarim River Basin in
Northwest China. Bachelor Thesis. University of Osnabrück, Osnabrück,
Germany.
Busch, D.E., Ingraham, N.L., Smith, S.D., 1992. Water uptake in woody Phreatophytes
of the southwestern United States: a stable isotope study. Ecol. Appl. 2,
450e459.
Central Asia Atlas, 2012. http://www.caatlas.org/ (assessed 28.12.12).
Chapagain, A.K., Hoekstra, A.Y., Savenje, H.H.G., Gautam, R., 2006. The water foot-
print of cotton consumption: an assessment of the impact of worldwide con-
sumption of cotton products on the water resources in the cotton producing
countries. Ecol. Econ. 60, 186e203.
Chen, Y.N., Ziliacus, H., Li, W.H., Zhang, H.F., Chen, Y.P., 2006. Ground-water level
affects plant species diversity along the lower reaches of the Tarim river,
Western China. J. Arid Environ. 66 (2), 231e246.
Chou, T., 1960. The problem of channel shifting of the middle reaches of the Tarim
River in Southern-Sinkiang. In: Murzayev, E.M., Chou, L. (Eds.), Natural Condi-
tions of Sinkiang. Chinese Academy of Sciences, Peking, pp. 87e133. Prirodnyye
usloviya Sin'tsziana NAUK. Peking (in English).
Cleverly, J.R., Dahm, C.N., Thibault, J.R., Gilroy, D.J., Allred Coonrod, J.E., 2002. Sea-
sonal estimates of actual evapotranspiration from Tamarix ramosissima stands
using three-dimensional eddy covariance. J. Arid Environ. 52, 181e197.
Conrad, C., Dech, S.W., Hafeez, M., Lamers, J.P.A., Martius, C., Strunz, Gh, 2007.
Mapping and assessing water use in a Central Asian irrigation system by uti-
lizing MODIS remote sensing products. Irrigat. Drain. Syst. http://dx.doi.org/
10.1007/s10795-007-9029-z.
FAO, 2012. Coping with Water Scarcity. FAO Water Reports 38. FAO, Rome.
Farah, H.O., Bastiaanssen, W.G.M., Feddes, R.A., 2004. Evaluation of the temporal
variability of the evaporative fraction in a tropical watershed. Int. J. Appl. Earth
Observ. Geoinform. 5, 129e140 .
Feike, T., Mamitimin, Y., Li, L., Abdusalih, N., Doluschitz, R., 2014. Development of
agricultural land and water use and its driving forces in the Aksu-Tarim Basin,
P.R. China. Environ. Earth Sci. http://dx.doi.org/10.1007/s12665-014-3108-x.
Feng, Q., Liu, W., Si, J.H., Su, Y.H., Zhang, Y.W., Cang, Z.Q., Xi, H.Y., 2005. Environ-
mental effects of water resource development and use in the Tarim river basin
of Northwestern China. Environ. Geol. 48, 202e210.
Giese, E., Mamatkanov, D.M., Wang, R., 2005. Wasserressourcen und deren Nutzung
im Flussbecken des Tarim (Autonome Region Xinjiang/VR China). Zentrum für
Internationale Entwicklungs- und Umweltforschung/ZEU), Universit€
at Giessen,
Giessen, Germany.
Glantz, M.H., 2010. Creeping environmental disasters: Central Asia's aral seas. In:
Kostianoy, A.G., Kosarev, A.N. (Eds.), The Aral Sea Environment, The Handbook
of Environmental Chemistry, vol. 7. Springer, Heidelberg, pp. 305e316.
Gowda, P.H., Chavez, J.L., Colaizzi, P.D., Evett, S.R., Howell, T.A., Tolk, J.A., 2007.
Remote sensing based energy balance algorithms for mapping ET: current
status and future challenges. Am. Soc. Agric. Biol. Eng. 50, 1639e164 4.
Gowda, P.H., Chavez, J.L., Colaizzi, P.D., Evett, S.R., Howell, T.A., Tolk, J.A., 2008. ET
mapping for agricultural water management: present status and challenges.
Irrigat. Sci. 26, 223e237.
Gries, D., Zeng, F., Foetzki, A., Arndt, S.K., Bruelheide, H., Thomas, F.M., Zhang, X.M.,
Runge, M., 2003. Growth and water relation of Tamarix ramosissima and Populus
euphratica on Taklamakan desert dunes in relation to depth to a permanent
water table. Plant Cell Environ. 26, 725e736.
GLOWA Jordan, 2008. GLOWA Jordan River: an integrated approach to sustainable
management of water resources under global change. http://www.glowa-
jordan-river.de/ (accessed 20.04.12).
Guerschmann, J.-P., Van Dijk, A.I.J.M., McVicar, T.R., Van Neil, T.G., Li, L., Liu, Y., Pe~
na-
Arancibia, J., 2008. Water balance estimates from satellite observations over the
Murray-Darling Basin. A report to the Australian Government from the CSIRO
Murray-Darling Basin Sustainable Yields project, CSIRO Australia. http://www.
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e1110
Please cite this article in press as: Thevs, N., et al., Water allocation and water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028

clw.csiro.au/publications/waterforahealthycountry/mdbsy/technical/V-
WaterBalanceEstimates.pdf.
Hochmuth, H., Thevs, N., He, P., Water allocation and water consumption of irri-
gation agriculture and natural vegetation in the Heihe River watershed, NW
China, Environ. Earth Sci. (Submitted for publication).
Hao, X.M., Chen, Y.N., Li, W.H., 2009. Impact of anthropogenic activities on the
hydrologic characters of the mainstream of the Tarim River in Xinjiang during
the last past 50 years. Environ. Geol. 57, 435e445.
Hao, X.M., Chen, Y.N., Li, W.H., Guo, B., Zhao, R.F., 2010. Hydraulic lift in Populus
euphratica Oliv. from the desert riparian vegetation of the Tarim River Basin.
J. Arid Environ. 74 (8), 905e911.
Hou, L.G., Xiao, H.L., Si, J.H., Xiao, S.C., Zhou, M.X., Yang, Y.G.,2010. Evapotranspiration
and crop coefficient of Populus euphratica Oliv forest during the growing season
in the extreme arid region northwest China. Agric. Water Manag. 97, 351e356.
Howell, T.A., 2001. Enhancing water use efficiency in irrigated agriculture. Agron. J.
93, 281e289.
ICWC (Interstate Commission for Water Coordination of Central Asia), 1992.
Agreement of five Central Asian States. http://www.icwc-aral.uz/statute1.htm
(accessed 02.12.12).
Khamzina, A., Sommer, R., Lamers, J.P.A., Vlek, P.L.G., 2009. Transpiration and early
growth of tree plantations established on degraded cropland over shallow sa-
line groundwater table in northwest Uzbekistan. Agric. For. Meteorol. 149,
1865e1874 .
Liu, M.G., 1997. Atlas of Physical Geography of China. China Map Press, Beijing.
Malek, E., Bingham, G.E., 1993. Comparison of the Bowen ratio-energy balance and
the water balance methods for the measurement of evapotranspiration.
J. Hydrol. 146, 209e220.
Molle, F., 2009. Water scarcity, prices and quotas: a review of evidence on irrigation
volumetric pricing. Irrigat. Drain. Syst. 23, 43e58.
Murray Darling Basin Commission, 2006. Murray Darling Basin Agreement. http://
www.comlaw.gov.au/Details/C2011C00160/Html/Text#_Toc289261269
(accessed 03.12.12).
Peng, H.Y., Thevs, N., Ott, K., 2014. Water distribution in the perspectives of
stakeholders and water users in the Tarim River Catchment, Xinjiang, China.
J. Water Resour. Protect. http://dx.doi.org/10.4236/jwarp.2014.66053.
Protze, M., 2011. Baumwollanbau im Tarim-Becken (China) e
€
Okonomie und
Wasserrechte. Diploma Thesis. University of Greifswald, Greifswald, Germany.
Reddy, J.M., Muhammedjanov, S., Jumaboev, K., Eshmuratov, D., 2012. Analysis of
cotton water productivity in Fergana Valley of Central Asia. Agric. Sci. 3,
822e834.
Roerink, G.J., Su, B., Menenti, M., 2000. S-SEBI A simple remote sensing algorithm to
estimate the surface energy balance. Phys. Clim. Earth B 25, 147e157.
Sammis, T.W., 1981. Yield of alfalfa and cotton as influenced by irrigation. Agron. J.
73, 323e329.
Scott, R.L., Cable, W.L., Huxman, T.E., Nagler, P.L., Hernandez, M., Goodrich, D.C.,
2008. Multiyear riparian evapotranspiration and groundwater use for a semi-
arid watershed. J. Arid Environ. 72, 1232e1246.
Senay, G.B., Leake, S., Nagler, P.L., Artan, G., Dickinson, J., Cordova, J.T., Glenn, E.P.,
2011. Estimating basin scale evapotranspiration (ET) by water balance and
remote sensing methods. Hydrol. Process. 25, 4037e4049.
Settle, J.J., Drake, N.A., 1993. Linear mixing and the estimation of ground cover
proportions. Int. J. Remote Sens. 14, 1159e1177.
Shen, Y.J., Chen, Y.N., 2010. Global perspective on hydrology, water balance, and
water resources management in arid basins. Hydrol. Process. 24, 129e135.
Siebert, S., Hoogeveen, J., Frenken, K., 2006. Irrigation in Africa, Europe and Latin
America. Update of the Digital Global Map of Irrigation Areas to version 4,
Frankfurt Hydrology Paper 5. Institute of Physical Geography, University of
Frankfurt, Frankfurt am Main, Germany.
Sobrino, J.A., G
omez, M., Jim
enez-Mu~
noz, J.C., Olioso, A., Chehbouni, G., 2005. A
simple algorithm to estimate evapotranspiration from DAIS data: application to
the DAISEX campaigns. J. Hydrol. 315, 117e125.
Sobrino, J.A., G
omez, M., Jim
enez-Mu~
noz, J.C., Olioso, A., 2007. Application of a
simple algorithm to estimate daily evapotranspiration from NOAAeAVHRR
images for the Iberian Peninsula. Rem. Sens. Environ. 110, 139e148.
Song, Y.D., Fan, Z.L., Lei, Z.D., Zhang, F.W., 2000. Research on Water Resources and
Ecology of TarimRiver, China. Xinjiang Peoples Press, Urumqi, China (in Chinese).
Steduto, P., Hsiao, T.C., Fereres, E., Raes, D., 2012. Crop Yield Response to Water. FAO
Irrigation and Drainage Paper 66. FAO, Rome.
Su, Z., 2002. The Surface Energy Balance System (SEBS) for estimation of turbulent
heat fluxes. Hydrol. Earth Syst. Sci. 6, 85e99.
Tan, X.W., Zhou, H.Y., 2007. Integrated management and hydroelectric development
in the Tarim river basin. Collection of Papers for the First Xinhua Forum of
Hydroelectric Development and Green Future. Beijing, February 2007 (in
Chinese).
Tang, D.S., Deng, M.J., 2010. On the Management of Water Rights in the Tarim River
Basin. China Water Power Press, Beijing (in Chinese).
Thevs, N., 2011. Water scarcity and allocation in the Tarim Basin: decision structures
and adaptations on the local level. J. Curr. Chin. Aff. 3, 113e137.
Thevs, N., Zerbe, S., Gahlert, F., Mijit, M., Succow, M., 2007. Productivity of reed
(Phragmites australis Trin. ex. Staud.) in continental-arid NW China in relation
to soil, groundwater, and land-use. J. Appl. Bot. Food Qual. 81, 62e68.
Thevs, N., Zerbe, S., Peper, J., Succow, M., 2008. Vegetation and vegetation dynamics
in the Tarim River floodplain of continental-arid Xinjiang, NW China. Phyto-
coenologia 38, 65e84.
Thevs, N., Buras, A., Zerbe, S., Kühnel, E., Abdusalih, N., Ovezberdyyeva, A., 2012.
Structure and wood biomass of near-natural floodplain forests along the Central
Asian rivers Tarim and Amu Darya. Forestry 81, 193e202.
Thomas, F.M., Foetzki, A., Arndt, S.K., Br uelheide, H., Gries, D., Zeng, F.J.,
Zhang, X.M., Runge, M. , 2006. Water use by pe rennial plants in the transition
zone between river oasis and de sert in NW China. Basic Appl. E col. 7,
253e267.
TMB (Tarim Management Bureau), 2005. Scheme of surface water distribution in
the tarim river basin (four source streams and one mainstream). http://www.
tahe.gov.cn/e/action/ShowInfo.php?classid¼106&id¼7375 (accessed 12.08.12).
USDA, 2012. China ePeoples Republic of cotton and products annual. GAIN Report
Number: CH12031. http://www.thefarmsite.com/reports/contents/
chinacotmay12.pdf (accessed 30.12.13).
Waters, R., Allen, R.G., Tasumi, M., Trezza, R., Bastiaanssen, W., 2002. SEBAL Surface
Energy Balance Algorithms for Land eIdaho Implementation eAdvanced
Training and Users Manual. University of Idaho, USA.
Wu, B.F., Jiang, L.P., Yan, N., Perry, C., Zeng, H.W., 2013. Basin-wide evapotranspi-
ration management: concept and practical application in Hai Basin, China.
Agricultural Water Management. http://dx.doi.org/10.1016/j.agwat.2013.09.021.
Xia, D.K., 1998. Changing and water resource of the Tarim River in Xinjiang. J. Arid
Land Resour. Environ. 12, 7e14 (in Chinese).
Xinjiang Linkeyuan Zaolin Zhisha Yanjiusuo, 1989. Huyanglin Gengxin Fuzhuang
Jishu Yanjiu. Xinjiang Linkeyuan Zaolin Zhisha Yanjiusuo, Urumqi (in Chinese).
Xinjiang Statistics Bureau, 2011. Xinjiang Statistical Yearbook of 2010. China Sta-
tistics Press, Beijing.
Xu, H.L., Ye, M., Song, Y.D., 2005. The dynamic variation of water resources and its
tendency in the Tarim River Basin. J. Geogr. Sci. 15, 467e474.
Xu, Z.X., Liu, Z.F., Fu, G.B., Chen, Y.N., 2010. Trends of major hydroclimatic vari-
ables in the Tarim River basin during the past 50 years. J. Arid Environ. 74,
256e267.
Ye, Z.X., Chen, Y.N., Li, W.H., Yan, Y., Wan, J.H., 2009. Groundwater fluctuations
induced by ecological water conveyance in the lower Tarim River, Xinjiang,
China. J. Arid Environ. 73 (8), 726e732.
Zhang, J., 2005. Barriers to water markets in Chinas Heihe River basin. Res. Rep.
Econ. Environ. Progr. Southeast Asia. No. 2005-RR8.
Zhang, J.B., 2006. Water management issues and legal framework development of
the Tarim river basin. In: Wallance, J., Wouter, P. (Eds.), Hydrology and Water
Law eBridging the Gap: a Case Study of HELP Basins. IWA Publishing, London,
pp. 108e142.
Zhang, H., Wu, J.W., Zheng, Q.H., Yu, Y.J., 2003. A preliminary study of oasis evo-
lution in the Tarim Basin, Xinjiang, China. J. Arid Environ. 55, 545e553.
Zhang, Y.M., Chen, Y.N., Pan, B.R., 2005. Distribution and floristics of desert plant
communities in the lower reaches of Tarim River, southern Xinjiang, People's
Republic of China. J. Arid Environ. 63 (4), 772e784.
Zhu, Z.P., Giordano, M., Cai, X.M., Molden, D., Hong, S.C., Zhang, H.Y., Lian, Y., Li, H.,
Zhang, X.C., Zhang, X.H., Xue, Y.P., 2003. Yellow River comprehensive assess-
ment: basin features and issues. IWMI and YRCC. http://www.iwmi.cgiar.org/
Publications/Working_Papers/working/WOR57.pdf (accessed 02.12.12).
N. Thevs et al. / Journal of Arid Environments xxx (2014) 1e11 11
Please cite this article in press as: Thevs, N., et al., Water allocationand water consumption of irrigated agriculture and natural vegetation in the
Aksu-Tarim river basin, Xinjiang, China, Journal of Arid Environments (2014), http://dx.doi.org/10.1016/j.jaridenv.2014.05.028