Catena 152 (2017) 268–276

Contents lists available at ScienceDirect

Catena

journal homepage: www.elsevier.com/locate/catena

Catchment-scale surface water-groundwater connectivity on China's

Loess Plateau
Zhi Li a,⁎, Xueqing Lin a, Anna E. Coles b, Xi Chen a
a
State Key Laboratory of Soil Erosion and Dryland Farming on the Loess Plateau, College of Natural Resources and Environment, Northwest A&F University, Yangling 712100, China
b
Global Institute for Water Security & School of Environment and Sustainability, University of Saskatchewan, Saskatoon, Saskatchewan S7N 3H5, Canada

a r t i c l e i n f o a b s t r a c t

Article history: Thick loess deposits (up to 200 m deep) and a large unsaturated zone (up to 100 m below the soil surface) com-
Received 16 July 2016 plicate the interactions between surface water and groundwater in China's Loess Plateau, especially in catch-
Received in revised form 13 January 2017 ments characterized by tableland-gully topography in this region. In the Heihe watershed, electrical
Accepted 18 January 2017
conductivity (EC), chloride (Cl) concentrations, and stable water isotope data (δD and δ18O), obtained during
Available online xxxx
dry and wet seasons in the period of 2012–2014 are used to determine the relationship between surface water
Keywords:
and groundwater. EC and Cl concentrations are found to be unsuitable for analyzing the relationship between
Groundwater surface water and groundwater in this watershed, due to the impacts of chemical weathering. The stable water
Surface water isotope data indicate a gaining surface water system, i.e. groundwater and gully runoff both recharge streamflow.
Connectivity Further, groundwater appears to dominate the hydrologic cycle in the watershed. Therefore, water resources
Stable water isotopes management on China's Loess Plateau should pay more attentions to groundwater protection.
Loess Plateau © 2017 Elsevier B.V. All rights reserved.

1. Introduction CLP, the relationships between streamflow reduction and groundwater

Water shortages, including runoff reduction and groundwater de-
pletion, have been observed worldwide (Milliman et al., 2008; Rodell
et al., 2009; Wada et al., 2010; Xu et al., 2010). For effective manage-
ment of water resources, it is fundamental to determine whether sur-
face water-groundwater connectivity is present or not, and in what
manner this connectivity exists (e.g. as a connected - a gaining or losing
surface water system - or disconnected system - completely disconnect-
ed system or a transitional state) (Banks et al., 2011; Brunner et al.,
2011; Chiogna et al., 2014; Penna et al., 2014; Song et al., 2006;
Woessner, 2000). Although important, this information is difficult to
be obtained in some regions, especially for those with the complicated
hydrologic system.
China's Loess Plateau (CLP) is subject to severe soil erosion and
water shortages due to intensive rainstorms, steep landscapes, low veg-
etation cover, highly erodible loess soil, and a semiarid to arid climate
(Li et al., 2009). Over the past 50 years, the land surface condition has
been substantially changed by the soil conservation measures (Sun et
al., 2015), and the climate has become drier and warmer (Li et al.,
2010). As a result, both streamflow and groundwater storage have
been observed to decrease in most catchments (Gao et al., 2015; Gao
et al., 2016; Zhang et al., 2008; Zhao et al., 2014). However, on the
⁎ Corresponding author.
E-mail address: lizhibox@126.com (Z. Li).
depletion are still poorly understood.
The hydrological processes on CLP are complicated due to a very
thick loess layer (up to 200 m) and a deep unsaturated zone (up to
100 m). For deep unsaturated zones at steady state with low water con-
tent, moisture movement is dominated by the upward movement of
vapor, and groundwater recharge is negligible (Ross, 1984). Groundwa-
ter recharge on CLP in the form of piston flow is therefore limited due to
low precipitation and the existence of dried soil layers with low water
contents (Li, 1983). According to this assumption, the groundwater like-
ly contributes little to surface water. However, hydrometric-based
hydrograph separation shows that groundwater contributes more
than 50% to streamflow (Dou et al., 2009; Zhu et al., 2010), suggesting
strong surface water-groundwater connectivity. Importantly, however,
the type of surface water-groundwater connectivity remains unclear.
CLP consists of two main landscapes: hilly-gully regions and table-
land-gully regions, with different hydrologic systems. In the hilly-gully
region, the steep slopes make rainfall discharge quickly into gullies;
however, in the tableland-gully region, almost all rainfall infiltrates
into soils on the flat tablelands. The differences in the hydrologic sys-
tems are likely to cause different interactions between surface water
and groundwater (Liu et al., 2010; Song et al., 2009; Su et al., 2009). In
the hilly-gully region, observations have revealed significant surface
water-groundwater interactions, with fluctuating stable water isotope
concentrations of groundwater in the middle reaches due to varying
water sources including precipitation, direct surface runoff, isotopical-
ly-enriched surface water, and/or lateral recharge of adjacent

http://dx.doi.org/10.1016/j.catena.2017.01.026
0341-8162/© 2017 Elsevier B.V. All rights reserved.
 Z. Li et al. / Catena 152 (2017) 268–276
groundwater (Liu et al., 2010). However, it is unclear about the interac-
tions (if any) between groundwater and surface water in the tableland-
gully region.
The use of hydrometric techniques is traditional but important for
identifying the role of surface water or groundwater in the hydrologic
system. Based on separation methods such as recession curve displace-
ment technique (Rutledge and Daniel, 1994) and digital filter method
(Arnold et al., 1995), the streamflow can be separated into groundwater
and rainfall-induced runoff. However, the results from hydrometric
methods have been challenged by the geochemical hydrograph separa-
tion technique; for example, stream discharge responds promptly to
rainfall but with dominant flows comprised of water that can be years
or decades old (Kirchner, 2003; McDonnell and Beven, 2014). The trac-
er-based methods, especially those using the stable isotopes of water,
provide a useful technique for hydrograph separation. They can be
used to qualitatively interpret or quantify the role of each component
by mass balance methods (Brkić et al., 2016; Klaus and McDonnell,
2013; Liu et al., 2015; Srdoc˛ et al., 1982).
The objective of this study is to determine the surface water-ground-
water connectivity in the tableland-gully region of CLP. Specifically, sev-
eral aspects will be analyzed and discussed: (i) How do the
hydrochemical characteristics of the surface water and groundwater
vary over space and time? (ii) What is the relationship between surface
water and groundwater? (iii) What implications does the identified sur-
face water-groundwater connectivity have for water resource
management?
2. Materials and methods
2.1. Study area
The study site is the Heihe catchment in the southern region of CLP
(Fig. 1). The catchment has an area of 1506 km2. The annual mean pre-
cipitation is 571 mm with 55% falling in July through September, and the
mean annual temperature is 9.4 °C (1961–2012). The annual mean run-
off depth is 46.4 mm (1961–2012) accounting for 9% of the annual mean
precipitation. Low flows occur from December to June, while high flows
occur from July to September. The water table depth fluctuates between
30 and 100 m below the soil surface.
The tableland-gully region consists of two landforms, i.e. flat table-
lands and gullies with steep slopes. On tablelands, almost all rainfall in-
filtrates into soils and recharges groundwater, and then groundwater
discharges to gullies and the main channel. Rainfall falling in the gullies
Fig. 1. Sampling sites in the Heihe watershed.
269
is directly converted to runoff due to the steep slopes (approximately
36% of the whole catchment has a slope gradient greater than 15°). As
the main channel of the river cuts deeply below the bedrock of the ta-
bleland, it also obtains groundwater from confined aquifers. Therefore,
streamflow in the main channel can obtain water from gully runoff
and groundwater. However, it is not clear whether the streamflow in
the main channel can also recharge groundwater.
The location of the tablelands and gullies in the Heihe catchment are
unevenly spatially distributed. All the tablelands are located in the
lower reach, while the upper reach is dominated by gullies. With a
loess layer thicker than 150 m (Fig. 2), the tablelands have great poten-
tial for water storage; therefore, the spatial variation in terrain might
have impacts on the water quantity from different sources.
The hydrogeological conditions are similar across the whole catch-
ment (Fig. 2). The horizontal Neogene/Cretaceous mudstone and sand-
stone beds (N2 and KZ series) are covered by three layers of loess. The
top layer is composed of Malan Loess (Q3, upper Pleistocene) with an
approximate thickness of 15 m. The middle layer is composed of Lishi
Loess (Q2, middle Pleistocene) with a thickness of 71–74 m, which
forms an aquifer due to unconsolidated sediments of relatively high po-
rosity. The third layer originates from Wucheng Loess (Q1, lower Pleis-
tocene) with a thickness of 22–78 m, which forms an aquitard due to its
low permeability.
2.2. Hydrochemical water sampling and data analysis
Samples for precipitation, runoff in the tributary gullies (hereafter
referred to as gully runoff), groundwater along the main channel (re-
ferred to as groundwater), and streamflow in the main channel have
been collected in the Heihe watershed. Precipitation was sampled on a
daily basis from one site in the lower reach in 2005, 2010 and during
the period 2012–2014. Streamflow, gully runoff and groundwater
were sampled twice per year during the period 2013–2014 (see the
sampling sites in Fig. 1). The first sampling campaign in June is repre-
sentative of the dry season due to the greatest evapotranspiration and
the lowest flow rates, while sampling campaigns during August to Octo-
ber are representative of the wet season due to the highest precipitation
amounts and flow rates (Fig. 3). Groundwater samples were taken from
both wells and springs. To investigate the impacts of rainfall on
streamflow, we collected samples after four separate rainfall events
(Table 1) in June 2012, August 2012, September 2013, and June 2014.
For all samples, with the exception of rainfall samples, EC, pH, and
temperature were measured in-situ using a portable instrument,
270 Z. Li et al. / Catena 152 (2017) 268–276

Fig. 2. Hydrogeological profiles of the study area.

SevenGo Duo pro™ SG78 (Mettler Toledo). Stable water isotope compo-
sitions (δD and δ18O) of all water samples were determined by the LGR
IWA-45EP water-vapor isotope analyzer with a precision of 0.2‰ for δD
and 0.03‰ for δ18O for liquid water. Cl concentrations of samples taken
in 2014 were obtained using a DIONEX ICS-1100 Ion Chromatography
System.
The spatiotemporal EC, Cl, and stable water isotope (δD and δ18O)
data were analyzed to give insights into the water exchange among
the streamflow, gully runoff and groundwater. As Cl was only deter-
mined for 2014, and it had similar spatiotemporal variability as EC (de-
scribed in the Results), EC instead of Cl was analyzed in detail.
samples in 2014, the values were not presented; however, they showed
good agreement with EC.
There is some spatial variation in hydrochemistry (Fig. 4). EC and Cl
concentrations in streamflow are relatively uniform throughout the
catchment, albeit with a slight increase in concentration with distance
downstream. By comparison, gully runoff and groundwater exhibit
greater spatial variability through the catchment, and both are relatively
uniform in the upper reach and generally increasing in the lower reach.
Like EC and Cl, the δD of streamflow is also relatively uniform through-
out the catchment, with the exception of those samples collected fol-
lowing rainstorms (e.g. June 2012). The δD of gully runoff and
groundwater both become more depleted with distance downstream.

3. Results 3.2. Temporal variation in hydrochemical indices

3.1. Spatial variation in hydrochemical indices
The 2012–2014, catchment-averaged EC, δD, and δ18O data were
shown in Table 2. EC is the lowest in the streamflow, the highest in
the groundwater, and intermediate in gully runoff. For both δD and
δ18O data, the groundwater samples are more depleted (more negative)
than the gully runoff samples, which in turn are more depleted than the
streamflow samples. As Cl concentrations were only determined for the
The seasonal averages (separated by dry season and wet season) of
EC, δD, and δ18O data for streamflow, gully runoff and groundwater
were calculated to determine the overall seasonal differences (Table
2). The hydrochemical data for groundwater appear temporally-invari-
ant since they remain largely constant during the different seasons. By
comparison, the isotopic composition of streamflow and gully runoff
in dry seasons is more depleted (more negative) than in wet seasons.
For example, streamflow has a δD of −64.7‰ and −63.7‰ for the dry
and wet seasons respectively, while gully runoff has the corresponding
values of − 68.3‰ and − 66.7‰. This seasonal difference in isotopic
composition conflicts with what we would have expected; due to limit-
ed water inputs and strong evaporation, we would have expected that
the dry season samples would be more enriched (more positive) than
those in the wet season.
The seasonal difference was further investigated by comparing sam-
ples from different years and different seasons (Fig. 4). EC and δD of

Table 1
Sampling dates and the rainfall conditions before sampling.

Sampling day Antecedent rainfall

Day Amount, mm δ18O, ‰ δD, ‰

2012-06-30 2012-06-29 23.5 Not sampled Not sampled

2012-08-27 2012-08-26 14.1 −8.7 −57.5
2013-06-29 – – – –
2013-09-01 2012-08-28 16.9 −8.6 −58.4
2014-06-24 2013-06-19 14.9 −7.5 −54.0
Fig. 3. Mean monthly precipitation (P), potential evapotranspiration (ET0) and runoff (R)
2014-10-17 – – – –
in the Heihe watershed.
 Z. Li et al. / Catena 152 (2017) 268–276 271

Table 2 groundwater do not exhibit clear seasonal variation, with the δD of

Spatiotemporal variations in electrical conductivity (EC) and stable water isotopic compo- groundwater, especially, being very similar across all seasons. This sug-
sitions of streamflow, gully runoff and grounwater.
gests that gully runoff and groundwater are less sensitive to rainfall than
Streamflow Runoff Groundwater streamflow, which might indicate different water sources.
EC δD δ18O EC δD δ18O EC δD δ18O
3.3. Dual isotope comparison
Spatial variations, averaged over both dry and wet seasons
Whole 628 −64.2 −9.0 636 −67.5 −9.3 839 −69.7 −9.7
watershed Dual isotope (δD and δ18O) comparison was used to provide infor-
Upper reach 627 −63.7 −9.1 544 −65.0 −8.9 701 −68.1 −9.6 mation about water sources and identify the relationships between dif-
Lower reach 630 −64.7 −8.9 736 −70.3 −9.7 918 −70.7 −9.7
ferent water components (Fig. 5). The volume-weighted precipitation
Temporal variations, averaged over the whole watershed isotopic composition anchors most data points of streamflow, gully run-
Dry season 643 −64.7 −8.9 587 −68.3 −9.3 836 −69.7 −9.6 off, and groundwater, which implies that the water source of this water-
Wet season 580 −63.7 −9.1 675 −66.7 −9.2 841 −69.7 −9.7
shed is predominantly precipitation (Fig. 5a); however, the intersection
Temporal variations, averaged over different reaches point of the evaporation line (EL) and local meteoric water line (LMWL)
Upper Dry 640 −63.5 −8.9 471 −66.4 −9.1 795 −68.5 −9.7 also suggests other water sources, which will be subsequently
reach Wet 558 −63.6 −9.1 625 −62.7 −8.7 648 −68.1 −9.6
interpreted.
Lower Dry 642 −66.0 −8.8 714 −71.3 −9.8 872 −70.8 −9.7
reach Wet 605 −63.8 −9.0 722 −70.0 −9.7 1008 −71.3 −9.9
The isotopic composition of streamflow in dry seasons spans a wider
range than that in the wet season (Fig. 5b), implying that the isotopic
Units: EC, us/cm; δD and δ18O, ‰.
composition in wet seasons is more homogeneous. The differences are
perhaps caused by different flow velocities through the catchment.
Low flows in the dry season are subject to more and stronger evapora-
streamflow during dry seasons typically are more positive and more tion. As a result, its isotopic composition becomes more positive with
enriched, respectively, than those in wet seasons. However, it is appar- distance downstream, than it would do in the wet season when high
ent that they are influenced by antecedent rainfall events. For example, flows accelerate the movement of water downstream. For streamflow
a rainfall with the amount of 23.5 mm occurred before the June 2012 data, the slope of the EL is 4.81 and its intersection with the LMWL is
sampling, which coincided with EC and δD of streamflow being much more positive in dual isotope space than the volume-weighted precipi-
depleted in comparison to those in August 2012. Gully runoff and tation, which implies that streamflow originates from direct

Fig. 4. Spatiotemporal variations in electrical conductivity (a –c), chloride concentration (d –f) and δD isotopic compositions (g –i) of streamflow, gully runoff and groundwater over 2012–
2014. Water samples are collected twice per year, one in June and the other during flood season mainly from July to September.
272 Z. Li et al. / Catena 152 (2017) 268–276

Fig. 5. Dual isotope comparison of streamflow, gully runoff and groundwater (panel b, c and d use the same legend. The equations represent the LMWL in a; and EL in b, c and d).

Fig. 6. Comparison of electrical conductivity in streamflow, gully runoff and groundwater.

 Z. Li et al. / Catena 152 (2017) 268–276
precipitation and surface runoff from precipitation but is subject to
strong evaporation.
The isotopic composition of gully runoff (Fig. 5c) and groundwater
(Fig. 5d) both span a large range. The EL slopes are 6.09 and 5.55, respec-
tively, and the intersections of EL and LWML are more negative in dual
isotope space than the volume-weighted precipitation, suggesting that
they are not only sourced from direct precipitation and rapid runoff of
precipitation, but also from some other water with more negative isoto-
pic composition.
3.4. Interactions between surface water and groundwater
The EC and δD of streamflow, gully runoff, and groundwater for each
sampling campaign were plotted together (EC: Fig. 6; δD: Fig. 7) to in-
terpret the potential interactions between surface water and ground-
water. The EC data of gully runoff and groundwater typically are lower
than those of streamflow in the upper reach, while higher than those
of streamflow in the lower reach (Fig. 7). Cl concentrations show a sim-
ilar spatial pattern to EC. According to mass balance, water with higher
values of EC or Cl should be the sink of that water with smaller values of
EC or Cl. Accordingly, gully runoff and groundwater appear to recharge
streamflow in the upper reach, while the opposite is true in the lower
reach, where streamflow appears to recharge gully runoff and
groundwater.
However, these findings demonstrated by the EC and/or Cl concen-
trations are not supported by the isotope mass balance (Fig. 7). In
most cases, streamflow is more enriched (more positive values for δD)
than gully runoff, which in turn is more enriched than groundwater.
Therefore, it is unlikely that the groundwater, whose δD isotope compo-
sition is depleted, is recharged by the streamflow with more positive
isotopic composition.
Fig. 7. Comparison of isotopic compositions in streamflow, gully runoff and groundwater (dashed lines stand for the isotopic compositions of rainfall before sampling, details can be found
in Table 1; rainfall before the sampling in June 2012 was not sampled).
273
4. Discussion
4.1. Appropriate hydrochemical indices for hydrologic systems in deposi-
tional environments
For the Heihe catchment in the CLP region, the focus of this study,
different hydrochemical indices suggest differing flow paths and direc-
tions of recharge. While EC and Cl concentrations suggest flow from sur-
face water to groundwater, the isotopic data indicate flow from
groundwater and runoff to streamflow. These disparate flow path re-
sults highlight the importance of using the appropriate index/indices
to interpret hydrological processes.
Theoretically, the stable isotopes of water should more accurately in-
dicate hydrological processes and flow paths, since they are compo-
nents of the water molecule and inherent labels. By comparison, ion
concentration can be influenced by myriad factors, such as chemical
weathering, and atmospheric and anthropogenic inputs. With a long
formation period, the thick loess deposits have higher ion concentration
with depth due to a gradual decrease in chemical weathering (Chen et
al., 2001; Nie et al., 2015). Accordingly, the high EC or Cl concentration
in groundwater is perhaps not due to recharge from surface water but
from the temporal evolution of the loess deposits. Therefore, if the soil
information is taken into account, the flow path interpreted from EC
and Cl concentrations is also from groundwater and gully runoff to
streamflow, which is in agreement with that revealed by the isotopic
data.
The above results imply that the ion concentrations in water from
deep loess layers are not suitable for approaches that use mass balance
equations. This conclusion is supported by the results of Xiao et al.
(2016) and Chen et al. (2005). From river water sampling across the
CLP, Xiao et al. (2016) found that the total dissolved solids (TDS) were
274 Z. Li et al. / Catena 152 (2017) 268–276
much higher than the global average and other large rivers in China. The
dissolved solutes were mainly derived from rock weathering, but with
minor anthropogenic and atmospheric inputs. Further, for the period
1958–2000, an increased trend in TDS was detected for the CLP and
the change trend coincided with a significant decrease in water dis-
charge (Chen et al., 2005). The decrease in streamflow was attributed
to increasing regulations of reservoirs, substantial conversion from
farmland to forest and grassland, and intensive water withdrawal, as
well as many other anthropogenic processes (Zhang et al., 2008; Zhao
et al., 2014). Therefore, although human activities play a minor role in
directly increasing dissolved solutes, they can indirectly influence the
ion concentration by altering the streamflow.
Severe soil erosion on the CLP is thought to have greatly influenced
nutrient loss (such as organic matter, nitrogen and phosphorus) and
enriched eroded sediment (Zheng et al., 2005); thereby increasing the
ion concentration in runoff. If this is the case for this study, the potential
perturbation could have caused surface water to have a higher ion con-
centration than groundwater; however, this is not true since EC of
groundwater is much higher than that of surface water in the Heihe wa-
tershed. Indeed, despite dry or wet seasons, the water samples have lit-
tle sediment due to the substantial soil conservation practices
implemented since the 1950s (Li et al., 2009). According to an evalua-
tion for a small watershed (the Nanxiaohe catchment) on the CLP, ap-
proximately half of the total transported water and 94.8% of the total
transported soil and nutrients have been locally retained in the selected
catchment by the soil conservation practices (Xu et al., 2012). Therefore,
the potential impacts of soil loss and sediment on ion concentration can
be neglected.
Considering the impacts of depositional environments and human
activities, the use of ion concentration for interpreting the hydrological
cycle in thick depositional environments should be treated with cau-
tion. Otherwise, the conclusion about water flow paths and recharge/
discharge relationships from traditional hydrological indices might be
inaccurate. Instead, stable isotope data are good tools for hydrologic
cycle analysis in depositional environments since they are not influ-
enced by chemical weathering or human activities.
4.2. Surface water-groundwater connectivity
The isotopic composition of surface water, i.e. streamflow and gully
runoff, is spatially variable (Fig. 4). The streamflow in the lower reach
and the upstream gully runoff is more variable in terms of its isotopic
composition. Temporally, the isotopic composition of streamflow in
dry seasons is more positive than that in wet seasons; however, no ob-
vious seasonal pattern is detected for gully runoff. The isotopic compo-
sition of groundwater is likely to be stable over space and time.
The spatiotemporal variations in the isotopic composition of gully
runoff and groundwater are very similar, especially for those in the
lower reach, but different from those of streamflow. In most cases, the
isotopic composition of gully runoff and groundwater are more negative
than that of streamflow, (although the opposite situation occurs when
large rainstorms prior to sampling are detected) (Fig. 7). However,
even with large rainstorms, the isotopic composition of gully runoff
and groundwater does not change vastly when compared with those
samples without antecedent rainfall (Fig. 4). Therefore, the surface
water-groundwater connectivity system is likely to be a gaining surface
water system, i.e. groundwater and runoff to streamflow.
This conclusion of gaining surface water system is consistent with
the flow regime in Fig. 3. In the first half of the year (January to June),
streamflow did not significantly change despite increasing monthly pre-
cipitation. This was due to low antecedent soil water content and high
evapotranspiration, which implies that streamflow originates from
groundwater in the dry season. In the second half of the year (July to De-
cember), although the streamflow was sensitive to rainfall inputs (Fig.
3), the isotopic compositions in streamflow are more negative than
rainfall and more positive than groundwater (Tables 1 & 2), which
suggests that groundwater still greatly contributes to streamflow. Espe-
cially, though the precipitation in November to December is similar as
that in the first half of the year, the larger streamflow is caused by the
delayed groundwater discharge.
4.3. Role and sources of groundwater
The surface water-groundwater connectivity system is closely relat-
ed to the topography. The stream channel cuts deeper than the bedrock
of the tablelands and is at the smallest surface elevation of the region.
Therefore, it is not surprising that the gully runoff will flow into the
stream channel. However, interestingly, the stable isotopic composition
of groundwater is always more depleted than that of streamflow, which
suggests that groundwater only flows to streamflow and not vice versa.
Some of the groundwater samples are from springs with higher eleva-
tion than streamflow, which are unlikely to obtain recharge from
streamflow. Many groundwater samples are from wells in aquifers
near the stream channel, which should be completely connected to
streamflow; however, they have stable isotopic compositions more de-
pleted than those of streamflow even without antecedent rainfall. These
results imply that even the groundwater in aquifers near the stream
channel cannot be recharged by streamflow. Overall, therefore, the
groundwater in the tableland-gully region of the CLP is highly likely to
dominate the hydrologic system since surface water hinges on it.
The different spatiotemporal patterns of isotopic composition sug-
gest that the water sources of streamflow are different from those of
gully runoff and groundwater, which can be further confirmed by the
dual isotope comparison. The water sources (inferred by the intersec-
tion of the EL and LWML) of gully runoff and groundwater have more
depleted isotopic compositions than that of the volume-weighted pre-
cipitation (Fig. 5), which implies that they are not only from precipita-
tion but also from water with a more depleted isotopic composition.
According to our earlier study (Cheng et al., 2014), the isotopic compo-
sition of unconfined groundwater falls on the LMWL and is very similar
to that of precipitation during August and September. However, the pre-
cipitation in the past several centuries had a more depleted stable isoto-
pic composition (Tan et al., 2014), and the deep confined groundwater
is mostly from ancient water with more depleted isotopic signature
than local modern precipitation (Li et al., 2015). This suggests that the
gully runoff and groundwater are from the confined water with more
depleted isotopic composition.
4.4. Implications for water resource management
The gaining surface water system identified here is quite different
from that in the hilly-gully region on CLP. The isotopic composition of
surface water and groundwater in a watershed located in a hilly-gully
region indicated groundwater recharge by surface water during the
rainy season (Liu et al., 2010). The surface water-groundwater interac-
tion is a two-way connected system in the hilly-gully region, while a
one-way connected system in the tableland-gully region. This high-
lights the importance of protecting groundwater in the tableland-gully
region on CLP.
Considering the regional differences in groundwater storage, the
groundwater in tablelands should be of high priority for water resource
management. Due to the thick vadose zone with porous loess, the
groundwater storage in tablelands is very large. However, the high
water demands on tablelands, which are increasing due to a large,
growing population and land use change, is causing greater groundwa-
ter extraction and a gradual water table decrease (Wang et al., 2010).
The groundwater can be recharged by vertical infiltration and preferen-
tial flow (Gates et al., 2011; Huang et al., 2013; Lin and Wei, 2001, 2006;
Tan et al., 2016); however, the low precipitation and recharge rates can-
not balance the high pumping rate. Thus, excessive groundwater
pumping should be under strict regulation.
 Z. Li et al. / Catena 152 (2017) 268–276
Table 3
The land use change in the Heihe watershed during 1995–2010.
Types Area (hm2)
1995 2000 2005
Farmland 90,230 90,544 88,883
Forest 7163 7981 8439
Grassland 57,243 55,800 56,788
Water 25 25 31
Built-up land 1272 1583 1738
Unused land 0 0 54
Meanwhile, groundwater recharge has been affected by substantial
land use change (Table 3). According to four land use maps during
1995–2010, the three land use types, i.e. farmland, grassland and forest,
accounted for approximately 99% of the whole watershed. However,
compared with 1995, farmland in 2010 decreased by 3721 hm2, while
forest and grassland increased by 1317 and 1856 hm2, respectively.
The conversion from farmland to grassland and forest will certainly in-
crease actual evapotranspiration and further alter the other hydrologi-
cal processes. For example, many farmlands on tablelands have been
converted to fruiting trees with higher water requirement, which great-
ly prevents soil water movement and decreases groundwater recharge
(Huang et al., 2013; Wang et al., 2015). Therefore, to protect the water
resources and especially groundwater, adjustment of land structure
should be cautious. Otherwise, the revegetation will exceed the sustain-
able water resource limits, and further cause conflicting demands for
water between the ecosystem and humans (Feng et al., 2016).
5. Conclusion
Hydrochemical indices in streamflow, gully runoff and groundwater
were analyzed to interpret the surface water-groundwater connectivity
in a watershed with a depositional environment on China's Loess Pla-
teau. Stable water isotope data appear to be a better tool for analysis
of the hydrologic system in this region since ion concentrations (EC
and Cl) reflect loess characteristics and not just water movements and
flow paths. The use of EC and Cl to analyze hydrological connectivity,
flow paths and water transfers in loess environments should thus be
treated with caution. The isotopic composition of streamflow is variable
over reaches and seasons; however, that of groundwater and gully run-
off is spatiotemporally invariant, especially in the lower reach. In most
cases, the isotopic composition is the most depleted for groundwater,
the most enriched for streamflow, and intermediate for gully runoff.
The surface water-groundwater connectivity is thus a gaining surface
water system, i.e. from groundwater and gully runoff to streamflow.
These findings highlight the importance of groundwater protection in
this region. Due to the lack of effective tracers, the contributions of
each component to streamflow are not quantified in this study, which
therefore could be a focus of future work.
Acknowledgements
This study was jointly funded by the National Natural Science Foun-
dation of China (51179161), Special-Funds of Scientific Research Pro-
grams of State Key Laboratory of Soil Erosion and Dryland Farming on
the Loess Plateau (A314021403-C2), and Chinese Universities Scientific
Fund (2452015105). We greatly appreciate the anonymous referees for
their constructive suggestions that helped improving the manuscript.
References
Arnold, J.G., Allen, P.M., Muttiah, R., Bernhardt, G., 1995. Automated base flow separation
and recession analysis techniques. Ground Water 33, 1010–1018.
275
Banks, E.W., Simmons, C.T., Love, A.J., Shand, P., 2011. Assessing spatial and temporal con-
nectivity between surface water and groundwater in a regional catchment: implica-
tions for regional scale water quantity and quality. J. Hydrol. 404, 30–49.
Brkić, Ž., Briški, M., Marković, T., 2016. Use of hydrochemistry and isotopes for improving
the knowledge of groundwater flow in a semiconfined aquifer system of the Eastern
2010 Slavonia (Croatia). Catena 142, 153–165.
86,509 Brunner, P., Cook, P.G., Simmons, C.T., 2011. Disconnected surface water and groundwa-
ter: from theory to practice. Ground Water 49, 460–467.
8480
Chen, J., An, Z., Liu, L., Ji, J., Yang, J., Chen, Y., 2001. Variations in chemical compositions of
59,099
the eolian dust in Chinese Loess Plateau over the past 2.5 Ma and chemical
31
weathering in the Asian inland. Sci. China Ser. D Earth Sci. 44, 403–413.
1760 Chen, J., Wang, F., Meybeck, M., He, D., Xia, X., Zhang, L., 2005. Spatial and temporal anal-
54 ysis of water chemistry records (1958–2000) in the Huanghe (Yellow River) basin.
Glob. Biogeochem. Cycles 19, n/a.
Cheng, L., Liu, W., Li, Z., Chen, J., 2014. Study of soil water movement and groundwater re-
charge for the loess tableland using environmental tracers. Trans. ASABE 57, 23–30.
Chiogna, G., Santoni, E., Camin, F., Tonon, A., Majone, B., Trenti, A., Bellin, A., 2014. Stable
isotope characterization of the Vermigliana catchment. J. Hydrol. 509, 295–305.
Dou, L., Huang, M.B., Hong, Y., 2009. Statistical assessment of the impact of conservation
measures on streamflow responses in a watershed of the Loess Plateau, China.
Water Resour. Manag. 23, 1935–1949.
Feng, X., Fu, B., Piao, S., Wang, S., Ciais, P., Zeng, Z., Lu, Y., Zeng, Y., Li, Y., Jiang, X., Wu, B.,
2016. Revegetation in China's Loess Plateau is approaching sustainable water re-
source limits. Nat. Clim. Chang. (advance online publication).
Gao, Z., Zhang, L., Cheng, L., Zhang, X., Cowan, T., Cai, W., Brutsaert, W., 2015. Groundwa-
ter storage trends in the Loess Plateau of China estimated from streamflow records.
J. Hydrol. 530, 281–290.
Gao, Z., Zhang, L., Zhang, X., Cheng, L., Potter, N., Cowan, T., Cai, W., 2016. Long-term
streamflow trends in the middle reaches of the Yellow River Basin: detecting drivers
of change. Hydrol. Process. 30, 1315–1329.
Gates, J.B., Scanlon, B.R., Mu, X.M., Zhang, L., 2011. Impacts of soil conservation on ground-
water recharge in the semi-arid Loess Plateau, China. Hydrogeol. J. 19, 865–875.
Huang, T., Pang, Z., Edmunds, W.M., 2013. Soil profile evolution following land-use
change: implications for groundwater quantity and quality. Hydrol. Process. 27,
1238–1252.
Kirchner, J.W., 2003. A double paradox in catchment hydrology and geochemistry. Hydrol.
Process. 17, 871–874.
Klaus, J., McDonnell, J.J., 2013. Hydrograph separation using stable isotopes: review and
evaluation. J. Hydrol. 505, 47–64.
Li, J., Pang, Z., Froehlich, K., Huang, T., Kong, Y., Song, W., Yun, H., 2015. Paleo-environment
from isotopes and hydrochemistry of groundwater in East Junggar Basin, Northwest
China. J. Hydrol. 529 (Part 2), 650–661.
Li, Y.S., 1983. The properties of water cycle in soil and their effect on water cycle for land
in the loess region. Acta Ecol. Sin. 3, 91–101.
Li, Z., Liu, W., Zhang, X., Zheng, F., 2009. Impacts of land use change and climate variability
on hydrology in an agricultural catchment on the Loess Plateau of China. J. Hydrol.
377, 35–42.
Li, Z., Zheng, F., Liu, W., Flanagan, D.C., 2010. Spatial distribution and temporal trends of
extreme temperature and precipitation events on the Loess Plateau of China during
1961–2007. Quat. Int. 226, 92–100.
Lin, R., Wei, K., 2001. Environmental isotope profiles of the soil water in loess unsaturated
zone in semi-arid areas of China. Isotope Based Assessment of Groundwater Renewal
in Water Scarce Regions. International Atomic Energy Agency (IAEA), Vienna, Austria,
pp. 101–108.
Lin, R., Wei, K., 2006. Tritium profiles of pore water in the Chinese loess unsaturated zone:
implications for estimation of groundwater recharge. J. Hydrol. 328, 192–199.
Liu, X., Song, X.F., Zhang, Y.H., Xia, J., Zhang, X.C., Yu, J.J., Long, D., Li, F.D., Zhang, B., 2010.
Spatio-temporal variations of δ2H and δ18O in precipitation and shallow groundwa-
ter in the Hilly Loess Region of the Loess Plateau, China. Environ. Earth Sci. 63,
1105–1118.
Liu, Y., Liu, F., Xu, Z., Zhang, J., Wang, L., An, S., 2015. Variations of soil water isotopes and
effective contribution times of precipitation and throughfall to alpine soil water, in
Wolong Nature Reserve, China. Catena 126, 201–208.
McDonnell, J.J., Beven, K., 2014. Debates—the future of hydrological sciences: a (common)
path forward? A call to action aimed at understanding velocities, celerities and resi-
dence time distributions of the headwater hydrograph. Water Resour. Res. 50,
5342–5350.
Milliman, J.D., Farnsworth, K.L., Jones, P.D., Xu, K.H., Smith, L.C., 2008. Climatic and anthro-
pogenic factors affecting river discharge to the global ocean, 1951–2000. Glob. Planet.
Chang. 62, 187–194.
Nie, J., Stevens, T., Rittner, M., Stockli, D., Garzanti, E., Limonta, M., Bird, A., Ando, S.,
Vermeesch, P., Saylor, J., Lu, H., Breecker, D., Hu, X., Liu, S., Resentini, A., Vezzoli, G.,
Peng, W., Carter, A., Ji, S., Pan, B., 2015. Loess Plateau storage of Northeastern Tibetan
Plateau-derived Yellow River sediment. Nat. Commun. 6, 8511.
Penna, D., Engel, M., Mao, L., Dell'Agnese, A., Bertoldi, G., Comiti, F., 2014. Tracer-based
analysis of spatial and temporal variations of water sources in a glacierized catch-
ment. Hydrol. Earth Syst. Sci. 18, 5271–5288.
Rodell, M., Velicogna, I., Famiglietti, J.S., 2009. Satellite-based estimates of groundwater
depletion in India. Nature 460, 999–1002.
Ross, B., 1984. A conceptual model of deep unsaturated zones with negligible recharge.
Water Resour. Res. 20, 1627–1629.
Rutledge, A.T., Daniel, C.C., 1994. Testing an automated method to estimate ground-water
recharge from streamflow records. Ground Water 32, 180–189.
Song, X., Liu, X., Xia, J., Yu, J., Tang, C., 2006. A study of interaction between surface water
and groundwater using environmental isotope in Huaisha River basin. Sci. China Ser.
D Earth Sci. 49, 1299–1310.
276 Z. Li et al. / Catena 152 (2017) 268–276
Song, X.F., Liu, X., Xia, J., Zhang, X.C., Yu, J.J., Zhang, Y.H., 2009. Interactions between sur-
face water and groundwater in Chabagou catchment using hydrogen and oxygen iso-
topes. J. Basic Sci. Eng. 17, 8–20.
Srdoc˛, D., Obelić, B., Sliepc˛ević, A., Stichler, W., Moser, H., Horvatin, N., Geyh, M.A., 1982.
Isotope analyses of groundwaters of the north African plain. Catena 9, 253–263.
Su, X.S., Wan, Y.Y., Dong, W.H., Hou, G.C., 2009. Hydraulic relationship between Malianhe
River and groundwater: Hydrogeochemical and isotopic evidences. J. Jilin Univ.
(Earth Sci. Ed.) 39, 1087–1094.
Sun, W., Song, X., Mu, X., Gao, P., Wang, F., Zhao, G., 2015. Spatiotemporal vegetation
cover variations associated with climate change and ecological restoration in the
Loess Plateau. Agric. For. Meteorol. 209–210, 87–99.
Tan, H., Wen, X., Rao, W., Bradd, J., Huang, J., 2016. Temporal variation of stable isotopes in
a precipitation-groundwater system: implications for determining the mechanism of
groundwater recharge in high mountain-hills of the Loess Plateau, China. Hydrol. Pro-
cess. 30, 1491–1505.
Tan, L., An, Z., Huh, C.-A., Cai, Y., Shen, C.-C., Shiau, L.-J., Yan, L., Cheng, H., Edwards, R.L.,
2014. Cyclic precipitation variation on the western Loess Plateau of China during
the past four centuries. Sci. Rep. 4, 6381.
Wada, Y., van Beek, L.P.H., van Kempen, C.M., Reckman, J.W.T.M., Vasak, S., Bierkens,
M.F.P., 2010. Global depletion of groundwater resources. Geophys. Res. Lett. 37,
L20402.
Wang, R., Liu, W., Zhao, X., 2010. Dynamic characteristics of groundwater level on
Changwu tableland. Agric. Res. Arid Areas. 28, 48–52.
Wang, Y., Shao, M., Liu, Z., Zhang, C., 2015. Characteristics of dried soil layers under apple
orchards of different ages and their applications in soil water managements on the
Loess Plateau of China. Pedosphere 25, 546–554.
Woessner, W.W., 2000. Stream and fluvial plain ground water interactions: rescaling
hydrogeologic thought. Ground Water 38, 423–429.
Xiao, J., Zhang, F., Jin, Z., 2016. Spatial characteristics and controlling factors of chemical
weathering of loess in the dry season in the middle Loess Plateau, China. Hydrol. Pro-
cess. n/a.
Xu, K., Milliman, J.D., Xu, H., 2010. Temporal trend of precipitation and runoff in major
Chinese Rivers since 1951. Glob. Planet. Chang. 73, 219–232.
Xu, X., Li, M., Liu, B., Kuang, S., Xu, S., 2012. Quantifying the effects of conservation prac-
tices on soil, water, and nutrients in the loess mesa ravine region of the loess plateau,
China. Environ. Manag. 49, 1092–1101.
Zhang, X.P., Zhang, L., Zhao, J., Rustomji, P., Hairsine, P., 2008. Responses of streamflow to
changes in climate and land use/cover in the Loess Plateau, China. Water Resour. Res.
44 (W00A07.1–W00A07.12).
Zhao, G., Tian, P., Mu, X., Jiao, J., Wang, F., Gao, P., 2014. Quantifying the impact of climate
variability and human activities on streamflow in the middle reaches of the Yellow
River basin, China. J. Hydrol. 519, 387–398.
Zheng, F., He, X., Gao, X., Zhang, C.-E., Tang, K., 2005. Effects of erosion patterns on nutri-
ent loss following deforestation on the Loess Plateau of China. Agric. Ecosyst. Environ.
108, 85–97.
Zhu, R., Zheng, H., Liu, C., 2010. Estimation of groundwater residence time and recession
rate in watershed of the Loess Plateau. J. Geogr. Sci. 20, 273–282.