Sie sind auf Seite 1von 15

Hydrogeology Journal (2018) 26:2753–2767

https://doi.org/10.1007/s10040-018-1831-1

REPORT

Evapotranspiration capture and stream depletion due to groundwater


pumping under variable boreal climate conditions: Sudogda River Basin,
Russia
Sergey Grinevskiy 1 & Elena Filimonova 1 & Victor Sporyshev 1 & Vsevolod Samartsev 1 & Sergey Pozdniakov 1

Received: 17 December 2017 / Accepted: 7 July 2018 / Published online: 26 July 2018
# Springer-Verlag GmbH Germany, part of Springer Nature 2018

Abstract
Groundwater pumping and changes in climate-induced recharge lead to lower groundwater levels and significant changes in the
water balance of a catchment. Water previously discharged as evapotranspiration can become a source of pumpage. Neglecting
this effect leads to overestimated streamflow depletion. A small river basin (Sudogda River Basin, Russia) with a boreal climate
and with long-term records of groundwater head and streamflow rate (showing that the measured stream depletion is less than the
pumping rate) was investigated. The role of evapotranspiration in the water balance was analyzed by a hydrogeological model
using MODFLOW-2005 with the STR package; the annual variation in recharge was obtained with the codes Surfbal and
HYDRUS. The Sudogda River Basin was classified according to landscape and unsaturated-zone texture classes, and for each
classified zone, the unsaturated-zone flow simulation was used to calculate the annual recharge dynamics for the observation
period. Calibration of the regional flow model was conducted using flow and head observations jointly for two steady-state flow
conditions—natural (before pumping started) and stressed (pumping). The simulations showed that pumped water originates
from three sources: intercepted baseflow (75% of the annual total pumping rate), the capture of groundwater evapotranspiration
discharge plus increased groundwater recharge (17%), and induced stream infiltration (8%). Additionally, multi-year precipita-
tion records were analyzed to detect any long-term recharge and pumping water-budget changes. The results showed that
increasing groundwater recharge by natural precipitation leads to (1) decreased intercepted baseflow and induced streamflow
infiltration and (2) increased intercepted evapotranspiration discharge, thereby reducing stream depletion.

Keywords Groundwater recharge/water budget . Evapotranspiration . Numerical modeling . Calibration . Russia

Introduction directly connected with an upper unconfined aquifer at the


river valley, where the whole aquitard between the aquifers
The majority of pumping fields are situated in river valleys. has been eroded.
The drawdown of shallow groundwater levels (GWLs) due to Traditionally, three sources of pumped water have been
pumping in small river basins leads to significant changes in considered: aquifer storage, induced stream infiltration, and
the water balance of the catchment and harmful effects on reduced groundwater discharge to the river (intercepted
stream and riparian landscapes within the river valley. This baseflow; Todd and Mays 2005). As pumping continues,
report investigates pumping of a confined aquifer which is the release of water due to aquifer storage decreases—over
the long term, changes in storage go to zero and the ground-
water flow becomes steady state. In this case, the pumpage
Electronic supplementary material The online version of this article is supplied only by intercepted baseflow and induced stream
(https://doi.org/10.1007/s10040-018-1831-1) contains supplementary
material, which is available to authorized users. infiltration, together defined as stream depletion (SD),
which is directly equal to the total pumping rate (TPR).
* Elena Filimonova The problem of SD due to pumping has been extensively
ea.filimonova@yandex.ru studied by many researchers. Early analytical solutions
were obtained for simplified stream–aquifer systems
1
Geology Faculty, Moscow State University, Leninskie Gory, GSP_1, (Theis 1941; Glover and Balmer 1954; Hantush 1965;
Moscow, Russia Shestakov 1965). Later analytical and semi-analytical
2754 Hydrogeol J (2018) 26:2753–2767

expressions have been developed to quantify SD in more stream–evapotranspiration interaction by model experiments
complicated systems: partially penetrated thin (line-width) and showed that the amount of ET capture varies from ~2 to
streams (Bochever 1966; Minkin 1972; Hunt 1999, 2003), 9.4% of the groundwater pumpage, and results from the case
heterogeneous aquifers (Butler Jr et al. 2001), leaky aqui- study in the Platte River (Nebraska, USA) indicated that ~11%
fers (Bakker and Anderson 2003; Zlotnik 2004; Butler et al. of the pumped water was supplied by ET capture. The role of
2007), bounded aquifers (Cherepanskiy 2005; Yeh et al. ET in the water balance of the TPR was also investigated by
2008), and a two-layer leaky aquifer system (Hunt 2009). Rassam et al. (2007) using the Theis (1941) solution for esti-
Jenkins (1968), using the principle of the superposition, mating stream depletion and the Shah et al. (2007) decay
developed the Theis (1941) solution for estimation of SD model for assessing the ET decline implemented in the GIS
caused by intermittent pumping. Wallace et al. (1990), for platform TIME. Dimensionless plots for ET capture for two
the Theis model, and later Darama (2001), for the Hantush different land-covers with a sandy loam soil are obtained
model, gave analytical solutions for cyclic pumping of wells. based on pumping rates, distance from pumping well to the
Regression dependencies for the estimation of SD created by river and aquifer transmissivity.
cyclic pumping were obtained by Filimonova and Shtengelov In humid boreal environments such as the northern
(2013). Wilson (1993), Chen (2003) and Chen and Yin (2004) European part of Russia, the effect of ET change due to
obtained equations for separately estimating induced stream pumping has not been well studied yet. The typical relations
infiltration and baseflow reduction. between mean annual values of groundwater recharge (GWR)
Recently, numerical models have been used to analyze or ET and shallow GWL depth for boreal climate conditions
stream–aquifer interaction in more realistic systems. Various and different landscapes are shown in Fig. 1 (Grinevskiy and
authors have analyzed the errors introduced by the assump- Pozdniakov 2013). According to these relations, there is a
tions in analytical models (Sophocleous et al. 1995; Conrad Bcritical^ GWL depth where mean annual flow through the
and Beljin 1996). Since that time, numerical models have water Table (W) is zero, and this Bcritical^ depth differs for
been used to estimate SD at the regional scale (Filimonova different landscapes. If the GWL is higher than the Bcritical^
and Baldenkov 2015; Kendy and Bredehoeft 2006; Langhoff depth, ET takes place, which increases up to a potential value
et al. 2006; Rodríguez et al. 2006; Wang et al. 2015; Woessner corresponding to a zero level depth. Otherwise, if the GWL is
2000; Zume and Tarhule 2008) or for model experiments lower than the Bcritical^ depth, GWR takes place, which in-
(Baalousha 2012; Bredehoeft and Kendy 2008; Chen and creases up to a maximum constant value for the deeper GWL;
Shu 2002; Fox and Durnford 2003; Lackey et al. 2015). hence, lowering of the GWL by pumping may cause not only
The previously mentioned methods are used for SD esti- ET capture but also increased GWR. Both the ET capture and
mation, but none take into consideration the effect of ground- Bextra^ GWR caused by lowering of a shallow GWL are the
water evapotranspiration (ET) on the surface–groundwater in- additional sources of pumped water and allow SD to be less
teraction. In a river valley with a shallow groundwater level, a than the TPR.
significant part of the natural groundwater flow formed due to This study analyzed the relatively typical case of a water-
recharge can be discharged through ET. Lowering of the shal- intake well field within a small river basin. The Sudogda River
low water table by pumping can lead to ET decreasing, which Basin has a long history of stream discharge observations
is a source of pumped water but has no influence on river flow. before and after the development of the water-pumping well
In such cases, SD will be less than the TPR; however,
neglecting the ET capture may result in overestimating SD.
Numerous papers have focused on the estimation of ET losses
(Nachabe et al. 2005; Ping et al. 2014; De Silva et al. 2008;
Shah et al. 2007; Szilagyi et al. 2007; Zhang et al. 2004 and
other), and a review of ET modeling is presented by Doble
and Crosbie (2017).
Nevertheless, very few investigations have analyzed the
role of ET capture in water pumpage. Darama (1996) obtained
an analytical solution for SD, including the effects of linear
variation in captured evaporation. He showed that the assump-
tion of linear variation in evaporation as a function of water-
table depth overestimated the reduction in evaporation and
underestimated the SD. Later, Darama (2004) investigated
the influence of nonlinear variation in evaporation with depth Fig. 1 Relations between mean annual flow through the water Table (W)
to the water table on a steady-state SD rate using a modeling and its depth (Zg) for different vegetation cover and soil types under
approach. Chen and Shu (2006) analyzed groundwater– boreal climate conditions (Grinevskiy and Pozdniakov 2013)
Hydrogeol J (2018) 26:2753–2767 2755

field. The preliminary analysis of the stream-discharge-time comprises two interrelated blocks: transformation of precipi-
series shows that the decrease in the minimal river flow is tation on the land surface calculated by the code Surfbal
significantly smaller than the total water intake and that this (Grinevskii and Pozdnyakov 2010; Grinevskiy and
flow is subject to large temporal variability. Thus, the purpose Pozdniakov 2017) and moisture transfer in the unsaturated
of this report is to investigate the influence of long-term cli- zone simulated by HYDRUS (Šimůnek et al. 2009).
mate variability on changes in the water balance in the
Sudogda River Basin under natural and disturbed conditions
associated with groundwater pumping, to study the sustain-
ability of river bank pumping in a well field in a small river Study area
basin under boreal climate conditions.
Investigations were performed with a hydrogeological Meteorology, hydrology and hydrogeology
model using MODFLOW-2005 (Harbaugh 2005) with the
STR package (Prudic 1989). The hydrogeological model de- The Sudogda field of pumping wells (SFPW) is located 30 km
scribes the groundwater flow due to water exchange between to the southwest of the city of Vladimir, western Russia
a shallow aquifer and the landscape and due to groundwater/ (Fig. 2). It has met the water demands of Vladimir since
surface-water (GW–SW) interaction. In addition to the 1995. The climate of the study area is humid boreal. The mean
groundwater flow model, the hydrogeological simulation in- annual weather data, obtained from the Vladimir meteorolog-
cludes nonlinear variation in the water flux through bound- ical observing station are as follows: temperature is 4.56 °С,
aries such as recharge, evapotranspiration, and gain or loss by total precipitation is 650 mm, and total potential evapotrans-
the stream. MODFLOW provides a package for estimating piration is ~550 mm. The study area is a low, hilly glacio-
groundwater ET. The required parameters were obtained by fluvial plain with a developed stream network. Areas favor-
supporting simulations of the diffusive recharge and ground- able to wetlands are produced by an exceedance of precipita-
water ET formation due to the transformation of precipitation tion over potential evapotranspiration, flooding of the flood-
at the surface and in the unsaturated zone. The recharge model plain in spring, and discharge of confined aquifers through

Fig. 2 Location of the study area and groundwater level drawdown in the CP aquifer. AB is the line of hydrogeological cross-section shown in the
following figure; the grey square is the area where Sudogda pumping well sites and observation wells are located
2756 Hydrogeol J (2018) 26:2753–2767

hydrogeological gaps (pathways of hydraulic connectivity The aquitard overlying the CP aquifer consists of Permian-
where the aquitard between two aquifers has been eroded.) Jurassic clays with a total thickness of 20–40 m. In the
The basin’s main stream is the Sudogda River, which flows Sudogda River Valley, the whole aquitard has been eroded,
to the north into the large river Klyazma. The total length of and the CP aquifer is connected directly to the upper
the Sudogda River is 116 km, and the total catchment area is Quaternary (Q) aquifer (Fig. 3). The SFPW is situated in this
1,900 km2. Forested areas occupy 68% of the area, and wet- area. The underlying aquitard is composed of Upper
lands occupy less than 1%. According to measurements from Carboniferous clays with a thickness of 5–22 m.
stream gauging stations (GS) shown in Fig. 2, the riverbed Widespread Quaternary glacio-fluvial, glacial, alluvial and
width is 10–15 m, and the maximum width is 40–45 m near marsh sediments cover the eroded surface of the old rocks.
the city of Sudogda. The riverbed depth ranges from 0.5 to The thickness of the Quaternary sandy, sandy-loamy, loamy
2 m and the streamflow velocity of the Sudogda River is 0.1– and peat sediments is irregular and ranges from 0 to 30 m. The
0.5 m/s with the east riverbank in many areas being signifi- Q aquifer is unconfined and characterized by low values of
cantly peaty and marshy. hydraulic conductivity.
The largest tributaries of the Sudogda River are the Yastreb The Q aquifer discharges via ET into the rivers, and by
River, the Poboyka River, the Pechenka River, the Yada River, leakage into the underlying CP aquifer. The relatively low
the Voyninka River, and the Serduga River. The streamflow horizontal hydraulic conductivity of the shallow aquifer pre-
rates of the tributaries of the Sudogda River are 0.2 m3/s or vents lateral GW flux, and practically the entire recharge in the
less, except for the Serduga River with a streamflow rate of Quaternary sediments infiltrates into the confined CP aquifer.
0.38 m3/s. The Serduga River contains large amounts of water In the carbonate sediments, due to the high values of trans-
due to groundwater discharge; a lot of springs are also located missivity, groundwater flows to the Sudogda River. In the
on the valley slopes and in the floodplain. areas of floodplains and directly in the riverbed, the CP aquifer
The main aquifer of the Sudogda River Basin is comprised discharges through the Q aquifer into the Sudogda River. The
of Upper Carboniferous - Lower Permian (CP) fissured rocks, values of the lowest 30 daily streamflow rates, obtained by
which are the major object of these investigations. The sedi- several measurements from gauging stations (GS; Fig. 4), are
ments contain limestones, dolomites with clay, malm or gyp- used to characterize the groundwater discharge and baseflow.
sum interlayers. The total thickness of the CP confined aqui- The Sudogda River Basin, bounded by GS 2 and GS 17
fer, which corresponds to the upper active fissured zone, is (Fig. 2), has a catchment area of 1,490 km2. The total low-
30–50 m (Fig. 3). The transmissivity values vary widely from flow rate of the Sudogda River with an exceedance probability
200 to 1,000–5,000 m2/d. of 50% is equal 3.26 m3/s, and the baseflow value is 2.19 L/s/

Fig. 3 Cross-section of Sudogda


River Basin
Hydrogeol J (2018) 26:2753–2767 2757

Fig. 4 Stream discharge with an


exceedance probability of 50%,
along the Sudogda River (left Y
axis), and gain in Sudogda River
baseflow (streamflow changes
between gauging stations except
tributaries; right Y axis)

km2. The streamflow rate increases irregularly (Fig. 4) due to the SFPW began in 1989, before the beginning of groundwa-
different conditions of the CP aquifer discharge, connected via ter abstraction, and included observations of groundwater
hydrogeological gaps, within the paleovalley of the Sudogda levels in dual completion wells in the CP and upper Q aquifers
River. The sections between GS 5–7 and GS 10–13 are char-
acterized by the maximum gains in baseflow.
The main tributaries of the Sudogda River, except the
Serduga River, have much lower values of gain in baseflow
of 0.2–0.6 L/s/km2, which is ten times less than the Sudogda
River. This can be explained by partial penetration of the
tributaries; hence, only the shallow aquifer can discharge into
these local rivers. Another reason for this difference is the
location of the Sudogda River Valley in the zone of the eroded
Permian aquitard, which produces a good hydraulic connec-
tion between the CP aquifer and the Sudogda River. The
Serduga River is situated in similar conditions.
The analysis of the hydrological conditions shows that
groundwater discharge associated with the CP aquifer occurs
into the riverbed of the Sudogda River. Moreover, discharge
into the local rivers is negligible. The distribution of
streamflow rates of the Sudogda River is characterized by
the heterogeneity of the connection between the surface water
and the groundwater due to zones of eroded aquitard. The
maximum gain in baseflow is located in the sections between
gauging stations 8 and 5 and between 13 and 10, situated near
the SFPW. The total gain in Sudogda River baseflow is
2.62 m3/s, which exceeds the average TPR of 0.62 m3/s.

Analysis of SFPW monitoring data

The SFPW consists of 17 pumping wells in the CP aquifer,


combined at three sites with 7, 6, and 4 pumping wells in the
Sudogda River Valley (Fig. 5), and started operation in 1995.
The annual TPR is almost constant and averages 53,430 m3/d Fig. 5 Location map of Sudogda pumping well sites, observation wells
(0.62 m3/s). Monitoring of groundwater and surface water in and GWL drawdown in the Q aquifer
2758 Hydrogeol J (2018) 26:2753–2767

(Fig. 5) and measurements of dry season flow in the Sudogda are obtained for GSs 5 and 7, which remained in natural
River and its tributaries in 1989, 2002 and 2014 (Fig. 1). flow conditions. Also, it can be noticed that the differences
in Q*i values between GS 13 and GS 15 are not so clear,
Groundwater levels because they are outside of the zone of groundwater
pumping influence. Additionally, the absolute streamflow
Significant groundwater level drawdown in both aquifers rates here are much larger than in upstream areas, so the
due to exploitation was observed during only the first years changes are comparable to the normal error of the
of pumping, and then the GWL changed insignificantly in streamflow measurement.
accordance with variability in seasonal and annual weather Thus, one can assume that the main river flow depletion
conditions (Fig. 6). The maximum decrease in the GWL of occurs downstream of GS 7, and using Q*7 values for natural
the CP aquifer in pumping wells is ~8 m with a radius of up conditions (1989) and measured river flow in 2002 and 2014
to 10 km from the water intake (Fig. 2), and the maximum Qi, the restored natural river flow QRi for 2002 and 2014 in the
decrease in the GWL of the Q aquifer is ~3 m with a radius zone of groundwater pumping influence can be calculated as
of up to 5 km (Fig. 5). follows:
For the most part of the GWL-drawdown area, the GWL of
CP aquifer is still higher than the GWL of Q aquifer (wells 2 QRi ¼ Qi ⋅Q*7 ð1Þ
and 2 g, Fig. 6), meaning there has been only a partial decrease
of CP aquifer discharge and partial river baseflow intercep- Then, the river flow depletion ΔQ was calculated as the
tion. However, in the zone of maximum CP aquifer GWL difference between restored and measured streamflow rates
drawdown, the difference in aquifer groundwater levels exhib- (Table S1 of the ESM). These values are quite similar in
ited the opposite pattern (wells 1 and 1 g, Fig. 6), indicating 2002 and 2014 for all stations upstream of GS 15, and their
leakage from the alluvial aquifer that was compensated by averages tend to increase from GS 8 downstream to GS 13,
induced stream infiltration. Thus, the intercepted groundwater which is located near the border of the pumping influence
baseflow and induced river infiltration leads to depletion of zone. The large difference in ΔQ values calculated for GS
the Sudogda River flow. 15 is probably associated with the error of the river flow
measurement as already mentioned. Flow measurements at
Estimation of streamflow depletion the Sudogda River tributaries during the pumping period
did not show any univocal changes compared to natural
Groundwater pumping influence on streamflow occurs inside conditions. This is probably due to the low permeability of
the area of GWL depression, which is limited to the stream the tributary’s riverbed and the small natural flow rates dur-
section between GS 5 and 13 (Fig. 2), and can be estimated by ing the dry period.
comparing between the measured low summer streamflow The total river flow depletion, calculated as an average
rates before groundwater pumping started (1989) and during for GS 13, is 0.43 m3/s. It is important that this value is less
the period of groundwater pumping (2002, 2014). than the average TPR of 0.62 m3/s, i.e., ~70% of it, which
The Sudogda River flow rates measured in 2002 and 2014 indicates that there are other sources providing a constant
are quite similar but differ significantly from the measure- TPR in steady-state flow conditions in addition to induced
ments in 1989—Table S1 of the electronic supplementary stream infiltration and intercepted baseflow. It can be as-
material (ESM) even outside the zone of groundwater sumed that this Badditional^ source of pumped water repre-
pumping influence, which can be explained by the different sents ET capturing in the area, where the natural GWL
amounts of natural water availability in these years. Therefore, depths of the Q aquifer were higher than the Bcritical^
a direct comparison of the river flow data is not appropriate; values. Additionally, in those parts of the pumping influence
however, although the absolute values of low streamflow in area where the natural GWL was near the Bcritical^ depth,
different years vary, it can be assumed that the relative river the GWL lowering may result in an increase in GWR. Thus,
flow increments Q*i = Qn/Qi between downstream (n) and both ET capture and Bextra^ GWR caused by shallow GWL
upstream (i) GSs in the dry season remain ~equal because they lowering are regarded as additional sources of pumped water
depend on the intensity of groundwater discharge. and are the reason that the Sudogda River flow depletion is
Due to pumping influence, the relative streamflow in- less than the TPR. Since these Badditional^ sources are
crement Q*i would be different from that of natural condi- strongly dependent on climate conditions, their total values
tions. This conclusion is confirmed by the data in Table S1 change seasonally and annually and lead to similar variabil-
of the ESM, which show that Q*i values in 2002 and 2014 ity in the Sudogda River flow depletion. The annual vari-
are less than those in 1989 downstream of GS 7 (Table S1 ability in the groundwater pumping sources and Sudogda
of the ESM), where the river flow depletion by the ground- River flow depletion was investigated by retrospective
water pumping occurs. The largest differences in Q*i ratios modeling using different annual meteorological data.
Hydrogeol J (2018) 26:2753–2767 2759

Fig. 6 Annual variations in the


total pumping rate and
groundwater levels in the CP
aquifer (wells 1 g, 2 g) and Q
aquifer (wells 1, 2)

Hydrogeological model The hydraulic conductivity of the upper Q aquifer, which is


mostly composed of loamy deposits in the main part of the
The hydrogeological model of the Sudogda River Basin simulated area, varies from 0.1 to 2 m/d. Maximum values of
should be able to reflect the changes in the hydrologic balance up to 20 m/d are observed in the Sudogda River Valley with
components occurring in the unsaturated zone during GWL sandy alluvium. The maximum transmissivity values of the
lowering and should also be designed to simulate the SD CP aquifer of up to 5,000 m2/d are also located in the Sudogda
caused by pumping. Therefore, the hydrogeological model River Valley, where the carbonate sediments of the aquifer are
comprises three interconnected blocks: the groundwater flow strongly fissured and karstic. The wide range of vertical
model, the GWR and ET model, and the GW–SW interaction leakance values is mostly associated with variability in the
model. In other words, the groundwater flow model, a major Permian-Jurassic aquitard thickness. For hydrogeological
unit of the hydrogeological model, is supplemented with in- gaps, where the aquitard thickness is zero, the vertical
ternal nonlinear conditions that characterize the GWR and ET leakance is set to 10 d−1. The first model set was developed
depending on the shallow GWL depth as well as taking into for steady-state mean annual natural and pumping conditions.
account the river flow and the GW–SW interaction changes
caused by the pumping or natural climate variability. Groundwater/surface-water interaction simulation

The MODFLOW STR package (Prudic 1989) for simulation


Groundwater-flow-model development
of GW–SW interaction under natural and pumping conditions
was used. The differences in GW–SW interaction intensity are
A two-layer finite-difference MODFLOW model (Harbaugh
characterized by low river flow increments between GSs (see
2005) was developed using Processing Modflow to simulate
section ‘Meteorology, hydrology and hydrogeology’), so the
groundwater flow in the studied Sudogda River basin. The
model area of 1,805 km2 includes the Sudogda River Basin, Table 1 Groundwater flow model parameters
which also includes a segment of the main Klyazma River
(Fig. 2). The western, eastern and southern model boundaries Model parameter Parameter values
are watersheds and no-flow boundaries. The northern
Minimum Maximum
constant-head boundary coincides with the Klyazma River.
A finite-difference grid of 224 rows and 200 columns with Layer 1, Q aquifer:
uniform cell dimensions of 250 m × 250 m was used for sim- Top elevation (m) 95 180
ulation. The groundwater flow model consists of two major Bottom elevation (m) 71 169
layers: the unconfined Q aquifer and the confined CP aquifer, Horizontal hydraulic conductivity (m/d) 0.1 20
which are divided by a Permian-Jurassic aquitard of clay and Vertical leakance (d−1) 10−5 10
loam. The model top is the land-surface topography and the Layer 2, CP aquifer:
model bottom is an impermeable underlying aquitard. Top elevations (m) 65 107
The model area was divided into sub-regions to character- Bottom elevations (m) 40 40
ize the aquifer properties based on well data and pumping Transmissivity (m2/d) 200 5000
tests. The list of model parameters is presented in Table 1.
2760 Hydrogeol J (2018) 26:2753–2767

hydraulic conductivity of the river streambed Kriv was calcu-


lated for the river sections as follows:
mM l
K riv ¼ ð2Þ
bΔН
where Ml = ΔQ/l is the low river flow increment ΔQ per unit
river length l; b is the average width of the river, calculated
from river flow measurements; m is the average thickness of
the riverbed sediments, set as 0.5 m; ΔH is the average dif-
ference between groundwater and surface-water levels and is
defined as 0.1 m for the Sudogda River and 0.3 m for its
tributaries with lower riverbed permeability.
The other parameters for the STR package were obtained
by river flow measurements on GSs and were interpolated for
river grid cells between them. Manning’s roughness coeffi-
cients were calculated using Manning’s equation (Prudic
1989) and measured river parameters.

Groundwater recharge model

Estimation of groundwater recharge for the Sudogda River


Basin was based on the simulation of the water balance at
the land surface and the unsaturated flow inside the soil cover
and unsaturated zone down to the GWL. For the same climate Fig. 7 Schematic diagram of the GWR model, which includes two
conditions, these processes are generally identical for the uni- blocks: code SurfBal (Pozdniakov et al. 2015) and code HYDRUS-1D
form typical natural conditions of GWR, which is controlled (Šimůnek et al. 2009)
by a combination of the surface, vegetation type, soil and
underlying unsaturated zone properties, as well as the GWL
depth. Hence, the results of the simulation of GWR, obtained station were performed for all seven combinations of vegeta-
for each type (combination) of such conditions, can be extrap- tion and soil types (landscape types) obtained by overlaying
olated for the corresponding part of the river basin (Crosbie et the maps shown in the ESM. Since natural GWL depths in the
al. 2010; Grinevskii and Pozdnyakov 2010). The GWR model Sudogda River basin vary from less than 1 to 10 m, unsatu-
includes two blocks—the surface water and energy balance, rated flow simulations for each soil profile were implemented
and the subsurface unsaturated flow (Fig. 7). using different constant hydraulic heads (from 1 to 10 m) at
The well-known package HYDRUS-1D (Šimůnek et al. the lower boundary.
2009) was used for groundwater recharge simulations based Data for the unsaturated flow simulation were obtained
on long-term meteorological data with a daily resolution. For from field investigations at pilot sites in the Sudogda River
a boreal climate with a long winter period, raw meteorological basin with different types of soil and vegetation cover, and soil
data such as precipitation, cannot be used directly as the atmo- samples for laboratory study were taken. Additionally, data on
spheric boundary conditions in HYDRUS. These data must be the nearest similar soil profiles from a database provided by
preprocessed to take snow accumulation, snow melting, surface the Russian Soil Institute (V.V. Dokuchaev Soil Science
runoff, evaporation from snow and vegetation cover, and soil Institute 2016) were used.
freezing-thawing processes into account. For this purpose, In the results of the simulations, the groundwater recharge
SurfBal (Grinevskii and Pozdnyakov 2010; Grinevskiy and was treated as equal to the downward water flux (W) through
Pozdniakov 2013; Pozdniakov et al. 2015), a code for simulat- the lower boundary of the modeled soil profile. The mean
ing surface-water and energy balances and generating upper annual GWR value was obtained for each landscape type area
boundary conditions for HYDRUS-1D, was used. with different GWL depth (Zg) by averaging the recharge
Firstly, the areal division of the Sudogda River basin was values over the simulation period 1965–2015 (Fig. 8).
conducted using the typical recharge conditions such as veg- Negative values of the mean annual water flux W for shallow
etation, soil and subsoil type (see ESM). Then, GWR simula- GWL depths mean that, under these conditions, the total up-
tions by SurfBal and HYDRUS-1D using the daily meteoro- ward flux in the unsaturated zone is greater than the total
logical data for 1965–2015 from the Vladimir meteorological downward flux, and groundwater discharge by ET occurs.
Hydrogeol J (2018) 26:2753–2767 2761

– The elevation of the ET surface was set as the earth


surface
– The ET extinction depth d is equal to the GWL depth at
which the recharge rate becomes constant (W ≈ Wm)

Thus, the simultaneous use of the Recharge and EVT pack-


ages can approximate the relations of W(Zg) for MODFLOW
simulation, as shown in Fig. 9, and the corresponding param-
eters of W(Zg) and the MODFLOW packages for different
landscapes are listed in the ESM. In this way, mean annual
GWR or ET values are not directly assigned to the upper
boundary in MODFLOW but are calculated based on the sim-
ulated GWL depths. It should be noted that such an approxi-
mation of the W(Zg) curves is rather good when the depth to
the water table is less than the Bcritical^ values (Zg < Zg0) but
leads to significant errors when Zg > Zg0. Figure 8 shows that
such approximation errors are greater for loamy soil profiles,
where W(Zg) curves are more gentle.

Model run and results of the simulation

The model was calibrated using streamflow and head obser-


vations for two steady-state flow conditions. The first set
represents natural conditions before the beginning of
Sudogda water intake, while the second set represents con-
Fig. 8 Relations between mean annual groundwater flux through the ditions stressed by groundwater pumping. Each data set
water table (W) and GWL depth for different soil profiles in a meadow includes 16 head and 13 flow observations, whereby the
and b forest landscapes in the Sudogda River Basin heads were measured at the same observation boreholes
for the two data sets and the head measurements were aver-
The W(Zg) relations can be approximated as follows aged over several years to represent average steady-state
(Shestakov et al. 1982; Grinevskiy and Pozdniakov 2013): groundwater flow. River flows were measured at the same
  GSs and profiles as well and the river flow increment was
 Zg used instead of the total flow value, an approach chosen
W ¼ W m − W m −W exp −
0
ð3Þ
z0 with respect to the main research goal of estimating the
impact of groundwater pumping on river baseflow. The
where Wm and W0 are downward (positive) and upward
two groundwater flow conditions were simulated by two
(negative) water fluxes through the water table when Zg → ∞
models, which were calibrated jointly (see ESM).
or Zg → 0, respectively (z0 is a parameter). The Bcritical^ GWL
Calibration and sensitivity analysis was performed with
depth Zg0, where W = 0, can be found from Eq. (4) as follows:
UCODE (Hill et al. 2005). Each iteration of UCODE ran
Wm both models and read both simulation results at once. Such
Z 0g ¼ −z0 ln ð4Þ
W m −W 0 joint calibration allows observation data to be used more
effectively. As a result, the conductivity parameters of Q
To set the received relations of W(Zg), MODFLOW pack- and CP aquifers as well as leakance coefficients of the sep-
ages Recharge and Evapotranspiration (EVT; Harbaugh 2005) arating aquitard were calibrated (see ESM). The GWR and
were used together. Constant maximum recharge values Wm ET parameters, as well as streamflow parameters, did not
were assigned to the Recharge package for corresponding change during model calibration.
areas. At the same time, the following parameters were used The verification of calculated mean annual GWR and ET
in the EVT package: values, as well as their relations with GWL depths, was con-
ducted by the MODFLOW simulation of the natural steady-
– The maximum evapotranspiration rate ETM was calculat- state water-balance conditions. The difference between mean
ed as the sum of absolute values of W 0 and W m : annual total GWR and ET in the catchment area between GSs
ETM = |Wm| + |W0| 2 and 17 (Fig. 2) simulated by MODFLOW is 281.1 × 103 m3/
2762 Hydrogeol J (2018) 26:2753–2767

Groundwater intake simulation under long-term


climate variability conditions

Previous simulation results show that the mean annual steady-


state budget of Sudogda water intake consists of three sources:
intercepted groundwater flow to the river, induced stream in-
filtration, and the sum of intercepted ET and additional
(increased) GWR. Their proportions can change under natural
climate variations, and this could be the reason for the vari-
ability in the Sudogda River depletion.
The groundwater flow simulation for 1965–2015 was car-
Fig. 9 An example of W(Zg) approximation for MODFLOW simulation ried out to investigate the climate-induced variability in the
with simultaneous use of the Recharge and EVT packages. Dashed line Sudogda basin water budget under natural conditions and
shows linear ET decreasing in the EVT package; Wm represents constant
maximum GWR in the Recharge package
groundwater pumping. The GWL observations (Fig. 6) show
small natural variability before and during groundwater
pumping, and the GWL annual values are almost constant.
d and is quite similar to the observed low river flow increment
This result is the reason for simplifying the long-term ground-
with a 50% exceedance probability between these GS, which
water simulation and neglecting annual changes in aquifer
is 281.7 × 103 m3/d.
storage, as neglecting the storage variation allows one to esti-
mate the climatic variability in the groundwater budget more
Steady-state simulation under mean annual climate visibly.
conditions Thus, long-term simulations were performed with the cal-
ibrated model with different annual values of maximum GWR
The two sets of simulations with the calibrated model allow (Wm) in the Recharge package and different annual ETM
for estimation of the mean annual water budget for groundwa- values in the EVT package for 51 steady-state stress periods,
ter pumping by comparing the modeled water budgets under which were obtained from water balance simulations using
natural and pumping conditions. The major source of pumped SurfBal and HYDRUS-1D for 1965–2015. The simulations
water is intercepted river baseflow (75% of TPR). Another were carried out in two sets: under natural conditions and with
significant source is decreased groundwater ET (17% of an average TPR of 53.4 × 103 m3/d from 1995 to 2015.
TPR). Surprisingly, induced stream infiltration is a minor
source of pumped water (8% of TPR). The calculated total Simulation results
intercepted ET includes both actual intercepted ET and in-
creased GWR, since using the model approximation of The simulation results for natural conditions show rather
W(Zg) based on the Recharge and EVT MODFLOW packages good correlations of modeled total annual GWR, ET and
does not allow one to separate them correctly. The increased river baseflow with total annual precipitation from April
GWR is approximated as a part of total intercepted ET in the to March (Bhydrological^ year), as shown in Fig. 10. This
drawdown area, where natural GWL depths were below the annual sum of precipitation was chosen because all winter
Bcritical^ values. The increased GWR is ~3.5 × 103 m3/d (7% precipitation accumulates as snow cover, which usually
of the TPR), and the actual intercepted ET is 5.4 × 103 m3/d melts in March. Total ET has the greatest climate-
(10% of the TPR). induced natural variability because of its very strong de-
The simulated total depletion of the Sudogda River is pendence on the GWL depth (Fig. 8). It should be noted
0.51 m3/s, and this value is less than the TPR (83% of it) that the rather good correlations of simulated GWR and
but is greater (by 18%) than that calculated by the river ET with annual precipitation confirm the adequacy of the
flow analysis (0.43 m3/s). Such difference can be related GWR model, because GWR and ET values are not
both to the errors of SD estimation based on river flow data assigned directly but are instead calculated during ground-
and the inaccuracy of W(Zg ) model approximation for water flow simulation.
groundwater flow simulation. Despite these limitations, The water-intake annual budget (% of TPR), which was
the simulation results confirmed the importance of calculated for every year as the difference between the
intercepted ET and increased GWR for the water-intake model water budgets for natural and pumping conditions,
water budget and for SD. The decrease of the Sudogda is quite stable for 1995–2015, and the majority of it is
River depth, calculated by the STR MODFLOW package, intercepted river baseflow (Fig. 11). However, in wet
varies downstream of the GS 7 from 0.05 to 0.55 m, years, with increased GWR, induced stream infiltration de-
representing 5–20% of the natural river depth. creases to insignificant values, and this change is
Hydrogeol J (2018) 26:2753–2767 2763

Fig. 10 Correlation of simulated


annual river baseflow,
groundwater recharge, and
evapotranspiration with total
precipitation of the
Bhydrological^ year (from April
to March) for 1965–2015. R2 is
the coefficient of determination

compensated by increasing ET capture. This pattern is also Summary and discussion


shown in Fig. 12, where increasing GWR, caused by in-
creasing total precipitation, leads to nonlinear changes in 1. Analysis of monitoring data from the active Sudogda
the water intake budget, including increases in the groundwater intake field shows that groundwater
intercepted baseflow and intercepted ET but decreases in pumping takes place under almost steady-state annual hy-
the induced stream infiltration. As already mentioned, the drodynamic conditions. The comparison of the Sudogda
total values of intercepted ET include both actual River dry season flow before and during pumping showed
intercepted ET and increased GWR. the SD under the pumping influence and was estimated to
Climate-induced changes in the groundwater intake budget be 0.42 m/s. The main uncertainties in the SD assessment
lead to corresponding variability in the Sudogda River deple- may be related to errors in the river flow measurements
tion, as shown in Fig. 13. The maximum river depletion is and to differences in natural water conditions among the
45.5 × 103 m3/d (85% of TPR) for years with minimal compared years. The differences between the measured
GWR and decreases to 41.8 × 103 m3/d (78% of TPR) for river flow data at the same stream GS for years with sim-
years with maximum GWR. ilar water conditions (2002 and 2014) are for most stations

Fig. 11 Changes in the Sudogda water-intake water budget (% of TPR) compared with annual GWR variations from 1995 to 2015. ISI is induced stream
infiltration; ΔBF and ΔET are intercepted river baseflow and captured ET, respectively
2764 Hydrogeol J (2018) 26:2753–2767

Fig. 12 Dependence of Sudogda


water intake budget sources from
total GWR for 1995–2015. ISI is
induced stream infiltration; ΔBF
and ΔET are intercepted river
baseflow and captured ET,
respectively

less than 0.02 m/s (~5% of estimated SD), as shown in shallow GWL drawdown area, and these processes are
Table S1 of the ESM. The relative values of river flow regarded as the reason why the Sudogda River SD is less
increment between GSs, which were used to estimate the than the TPR. The significant role of intercepted ET in the
natural river flow for the pumping period, helped to re- water intake budget and in the river depletion process is
duce the uncertainty due to natural water conditions already known for arid and semi-arid climate conditions
among compared years Thus, considering the possible (Darama 2004; Rassam et al. 2007; Chen and Shu 2006)
uncertainties, the estimated Sudogda River depletion of but has not been studied under boreal climate conditions.
0.42 m/s is less than the average TPR of 0.62 m/s, 3. The dependence of the mean annual flux through the wa-
representing ~70% of the TPR. ter table on the GWL depth W(Zg) was simulated using
2. Analysis of the Sudogda River Basin hydrogeology sup- SurfBal and HYDRUS 1D for different typical landscapes
ports the conclusion that the estimated difference between of the Sudogda River basin. The calculated mean annual
SD and TPR under steady-state conditions is connected W(Zg) relations show that there is a Bcritical^ GWL depth
with additional sources in the water intake budget that do at which the mean annual flow through the water Table
not affect the river flow. Since pumping leads to shallow (W) is zero, and this Bcritical^ depth differs for different
GWL drawdown, these additional water sources include landscapes (Fig. 8). This finding indicates that in the area
capturing of ET and increasing the GWR within the of Sudogda water intake influence, where the natural

Fig. 13 Dependence of the


Sudogda River depletion on
climate-induced variability in to-
tal GWR
Hydrogeol J (2018) 26:2753–2767 2765

shallow GWL depths of the Q aquifer were above the 7% of the TPR, respectively). Thus, for boreal climate
Bcritical^ depth, the GWL lowering by pumping has led conditions, both intercepted ET and increased GWR are
to ET capture. In contrast, in areas where the natural water significant for pumping water budgets and the influence
table was near the Bcritical^ depth, the GWL lowering has of pumping on river flow. The simulated total depletion of
resulted in increased GWR. Thus, water balance simula- the Sudogda River is 0.51 m3/s, and this value is less than
tions confirm significant variability in mean annual GWR the TPR (83% of it), but significantly more than that cal-
and ET due to GWL changes, and calculated W(Zg) pa- culated based on the results of the river flow analysis. The
rameters were used to account for the GWR and ET differences between the simulated and calculated SD
changes due to shallow GWL drawdown in the ground- values are related to several reasons. First, as already men-
water flow simulations. The approximation of the calcu- tioned, possible uncertainty in the calculated SD is related
lated W(Zg) relations in the MODFLOW simulations was to errors in the river flow measurements and differences in
performed by the joint use of the Recharge and EVT the natural water conditions of the compared river flow
packages, but this approximation has significant errors data. Second, the errors in the simulated total SD are re-
(Fig. 9). The maximum errors are confined to the GWL lated to uncertainties in the intercepted ET and increased
depth interval below the Bcritical^ depth to the depth of GWR model estimation due to W(Zg) approximation.
constant GWR; therefore, simulation of the increased Another reason may be related to the coarse model grid
GWR caused by GWL drawdown has significant uncer- within the shallow GWL depression area. The shape of
tainty. Analysis of the W(Zg) curves for different land- the W(Zg) curves (Fig. 8) shows that ET is most sensitive
scapes (Fig. 8) shows that the maximum simulation errors to small changes in the GWL near surface; therefore, the
of increased GWR occur in loamy and peat soils with averaging of the earth surface and GWL data in the
gentle W(Zg) curves. The use of the ETS1 package with Sudogda basin model grid cells (with dimensions of
nonlinear segment variation in ET between the ET surface 250 × 250 m) may lead to significant errors in the ET
and the extinction depth (Banta 2000) would reduce these capture estimates. Despite these possible uncertainties in
simulation errors. the intercepted ET and SD estimates, the results of
4. Both flow and head observation data were used for model groundwater flow simulation confirmed that intercepted
calibration to make better use of the limited amount of ET and increased GWR due to lowering the shallow GWL
observation points. The main challenge of model calibra- by pumping are significant components in the Sudogda
tion emerged due to positioning of monitoring wells, water intake budget and lead to the total SD being less
which are located close to the water intake and thus pro- than the average TPR.
vide little information on the heterogeneity in the model 6. The long-term groundwater flow and GWR simulations
parameters. Therefore, the river flow observations were allow one to estimate the natural climate-induced variabil-
assumed to be more important and were given greater ity in the Sudogda water intake budget. The simulation
weight than the head observations. The joint use of the results show that increases in the GWR by natural precip-
two observation data types allowed good estimation of itation lead to nonlinear decreases in the induced stream
numerous model parameters and verification of the model infiltration and to increases in the intercepted ET and
heterogeneity. The simulated natural difference between intercepted baseflow. Climate variability leads to corre-
total GWR and ET in the Sudogda River basin equals spondent changes in Sudogda River depletion: maximum
281.1 × 103 m3/d and corresponds to the observed river SD corresponds to minimal GWR and is 45.5 × 103 m3/d
baseflow, which is 281.7 × 103 m3/d. These total values (85% of TPR), decreasing to 41.8 × 103 m3/d (78% of
confirmed the accuracy of the parameters used in the TPR) for years with maximum GWR (Fig. 13).
model. 7. Despite the possible uncertainties in the intercepted ET
5. The mean annual budget of the Sudogda water intake, and SD estimations, the main results of the investigation
estimated as the difference between the simulation results show that Badditional^ sources of riverbank groundwater
for natural and pumping conditions, consists of three in the intake water budget such as intercepted ET and
sources. The main part is intercepted baseflow, which increased GWR, are essential even in boreal climates
accounts for 75% of the average TPR. The second part and are the reason for climate-induced variability in SD,
is induced ET and increased GWR, representing 17% of which is less than the TPR. Future research will focus on
the TPR, and the smallest part is induced stream infiltra- the study of seasonal variability in ET capture and SD
tion, representing 8% of the TPR. The model approxima- using a fine grid model with the ETS1 package in
tion of the W(Zg) curves by the EVT and Recharge pack- MODFLOW.
ages did not allow separation of the induced ET and in-
creased GWR correctly, but approximate calculation Acknowledgements These investigations were supported by the Russian
Science Foundation No. 16-17-10187.
shows that their relative values are comparable (10 and
2766 Hydrogeol J (2018) 26:2753–2767

References shallow groundwater. Hydrogeol J 25:3–23. https://doi.org/10.1007/


s10040-016-1470-3
Filimonova EA, Shtengelov RS (2013) The dependence of stream deple-
Baalousha HM (2012) Drawdown and stream depletion induced by a
tion by seasonal pumping on various hydraulic characteristics and
nearby pumping well. J Hydrolog 466–467:47–59. https://doi.org/
engineering factors. Hydrogeol J 21(8):1821–1832. https://doi.org/
10.1016/j.jhydrol.2012.08.010
10.1007/s10040-013-1053-5
Banta ER (2000) MODFLOW-2000, The U.S. Geological Survey Filimonova E, Baldenkov M (2015) A combined-water-system approach
Modular Ground-Water Model: documentation of packages for sim- for tackling water scarcity: application to the Permilovo
ulating evapotranspiration with a segmented function (ETS1) and Groundwater Basin, Russia. Hydrogeol J 24(2):489–502. https://
drains with return flow (DRT1). US Geol Surv Open-File Rep 00-
doi.org/10.1007/s10040-015-1325-3
466 Fox G A, Durnford D S (2003) Unsaturated hyporheic zone flow in
Bochever FM (1966) Evaluation of well-field yield in alluvial aquifers: stream/aquifer conjunctive systems. Adv Water Resour 26(9):989–
the impact of a partially penetrating stream (in Russian). VODGEO 1000. https://doi.org/10.1016/S0309-1708(03)00087-3
(Hydrogeol) 13:84–115 Glover RE, Balmer GG (1954) River depletion resulting from pumping a
Bredehoeft J, Kendy E (2008) Strategies for offsetting seasonal impacts well near a river. Trans AGU 35(3):468–470
of pumping on a nearby stream. Ground Water 46(1):23–29. https:// Grinevskii SO, Pozdnyakov SP (2010) Principles of regional estimation
doi.org/10.1111/j.1745-6584.2007.00367.x of infiltration groundwater recharge based on geohydrological
Butler JJ Jr, Zlotnik VA, Tsou M-S (2001) Drawdown and stream deple- models. Water Resour 37(5):638–652. https://doi.org/10.1134/
tion produced by pumping in the vicinity of a partially penetrating S0097807810050040
stream. Ground Water 39(5):651–659. https://doi.org/10.1007/ Grinevskiy S O, Pozdniakov S P (2013) The use of HYDRU S-1D for
s12665-009-0117-2 groundwater recharge estimation in boreal environments In:
Butler JJ Jr, Zhan Х, Zlotnik VA (2007) Pumping-induced drawdown and Šimůnek, J., M. Th. van Genuchten, R. Kodešová (Eds.).
stream depletion in a leaky aquifer system. Ground Water 45(5): HYDRUS software applications to subsurface flow and contaminant
178–186. https://doi.org/10.1111/j.1745-6584.2006.00272.x transport problems, Dept. of Soil Science and Geology, Czech
Chen XH (2003) Analysis of pumping-induced infiltration in gaining University of Life Sciences, Prague, Czech Republic, pp 107–118
streams. J Hydrol 275(1–2):1–11. https://doi.org/10.1016/S0022- Grinevskiy SO, Pozdniakov SP (2017) A retrospective analysis of the
1694(02)00320-7 impact of climate change on groundwater resources. Moscow
Chen XH, Shu LC (2002) Stream–aquifer interactions: evaluation of Univ Geol Bull 72(3):200–208. https://doi.org/10.3103/
depletion volume and residual effects from ground water pumping. S0145875217030036
Ground Water 40(3):284–290. https://doi.org/10.1111/j.1745-6584. Hantush MS (1965) Wells near streams with semipervious beds. J
2002.tb02656.x Geophys Res 70(12):2829–2838
Chen XH, Shu LC (2006) GW evapotranspiration captured by seasonally Harbaugh AW (2005) MODFLOW-2005, the US Geological Survey
pumped wells in river valleys. J Hydrol 318:334–347. https://doi. modular ground-water model: the ground-water flow process. US
org/10.1016/j.jhydrol.2005.06.025 Geol Surv Techniques Methods 6-A16
Chen X, Yin Y (2004) Semianalytical solutions for stream depletion in Hill E et al (2005) UCODE 2005 and six other computer codes for uni-
partially penetrating streams. Ground Water 42(1):92–96. https:// versal sensitivity analysis, calibration, and uncertainty evaluation.
doi.org/10.1111/j.1745-6584.2004.tb02454.x US Geol Surv Techniques Methods 6-A11
Cherepanskiy MM (2005) Teoreticheskie osnovy gidrogeologicheskih Hunt B (1999) Unsteady stream depletion from ground water pumping.
prognozov vliyaniya otbora podzemnyh vod na rechnoi stok Ground Water 37(1):98–102
[Theoretical base of hydrogeological forecasts of groundwater Hunt B (2003) Unsteady stream depletion when pumping from
pumping influence on streamflow]. NIA-Priroda, Moscow semiconfined aquifer. J Hydrol Eng 8:12–19. https://doi.org/10.
1061/(ASCE)1084-0699(2003)8:1(12
Conrad LP, Beljin MS (1996) Evaluation of an induced infiltration model
Hunt B (2009) Stream depletion in a two-layer leaky aquifer system. J
as applied to glacial aquifer systems. Water Resour Bull 32(6):1209–
Hydrol Eng 14(9):895–903. https://doi.org/10.1061/(ASCE)HE.
1220
1943-5584.0000063
Crosbie RS, McCallum JL, Walker GR, Chiew FHS (2010) Modeling
Jenkins CT (1968) Techniques for computing rate and volume of stream
climate-change impacts on groundwater recharge in the Murray-
depletion by wells. Ground Water 6(2):37–46
Darling Basin, Australia. Hydrogeol J 18(7):1639–1656. https://
Kendy E, Bredehoeft JD (2006) Transient effects of groundwater
doi.org/10.1007/s10040-010-0625-x
pumping and surface-water-irrigation returns on streamflow. Water
Darama Y (1996) Hydraulic interactions between groundwater and sur- R e s o u r R e s 4 2 ( 8 ) : W 0 8 5 1 5 . h t t p s : / / d o i . o rg / 1 0 . 1 0 2 9 /
face water resources caused by a pumping well. Tech J Civil Eng 2005WR004792
(IMO) 88:1167–1185 Lackey G, Neupauer RM, Pitlick J (2015) Effects of streambed conduc-
Darama Y (2001) An analytical solution for stream depletion by cyclic tance on stream depletion. Water 7:271–287. https://doi.org/10.
pumping of wells near streams with semipervious beds. Water 3390/w7010271
Resour Res 36(1):79–86. https://doi.org/10.1111/j.1745-6584. Langhoff JH, Rasmussen KR, Christensen S (2006) Quantification and
2001.tb00353.x regionalization of groundwater–surface water interaction along an
Darama Y (2004) Stream depletion by a pumping well including the alluvial stream. J Hydrol 320:342–358. https://doi.org/10.1016/j.
effects of nonlinear variation of captured evaporation from a phreatic jhydrol.2005.07.040
aquifer. Hydrol Sci J 49(3):443–460. https://doi.org/10.1623/hysj. Minkin EL (1972) Vzaimosvyaz podzemnyh i poverhnostnyh vod i ee
49.3.443.54346 znachenie pri reshenii gidrogeologicheskih i vodohozyaistvennyh
De Silva M S, Nachabe M H, Šimůnek J, Carnahan R (2008) Simulating zadach [Groundwater and surface water interaction and its impor-
root water uptake from a heterogeneous vegetative cover. J Irrig tance in solving some hydrogeological and water management prob-
Drain Eng 134(2):167–174. https://doi.org/10.1061/(ASCE)0733- lems]. Stroiizdat, Moscow
9437(2008)134:2(167) Nachabe M, Shah N, Ross M, Vomacka J (2005) Evapotranspiration of
Doble RC, Crosbie RS (2017) Review: current and emerging methods for two vegetation covers in a shallow water table environment. Soil Sci
catchment-scale modelling of recharge and evapotranspiration from Soc Am J 69(2):492–499. https://doi.org/10.2136/sssaj2005.0492
Hydrogeol J (2018) 26:2753–2767 2767

Ping W, Grinevsky SO, Pozdniakov SP et al (2014) Application of the Szilagyi J, Gribovszki Z, Kalicz P (2007) Estimation of catchment-scale
water table fluctuation method for estimating evapotranspiration at evapotranspiration from baseflow recession data: numerical model
two phreatophyte-dominated sites under hyper-arid environments. J and practical application results. J Hydrol 336:206–217. https://doi.
Hydrol B 8(519):2289–2230. https://doi.org/10.1016/j.jhydrol. org/10.1016/j.jhydrol.2007.01.004
2014.09.087 Theis CV (1941) The effect of a well on the flow of a nearby stream.
Pozdniakov SP, Vasilevskiy PY, Grinevskiy SO (2015) Estimation of Trans Am Geophys Union 22(3):734–738
groundwater recharge by flow in vadose zone simulation at the Todd DK, Mays LW (2005) Groundwater hydrology, 3rd edn. Wiley,
watershed with different landscapes and soil profiles. Engineering Chichester, UK
geology and hydrogeology, no. 29. Bulgarian Academy of Sciences, V.V. Dokuchaev Soil Science Institute (2016) Soil database of Russian
Sofia, Bulgaria, pp 47–58 Soil Institute. http://eng.esoil.ru. Accessed November 2016
Prudic DE (1989) Documentation of a computer program to simulate Wallace RB, Darama Y, Annable MD (1990) Stream depletion by cyclic
stream–aquifer relations using a modular, finite-difference, ground- pumping of wells. Water Resour Res 26(6):1263–1270
water flow model. US Geol Surv Open-File Rep 88-729 Wang P, Pozdniakov SP, Shestakov VM (2015) Optimum experimental
Rassam DW, Knight JH, Pickett T (2007) Modelling ET capture during design of a monitoring network for parameter identification at riv-
groundwater pumping. In: Proceedings of MODSIM07 Int. Cong. erbank well fields. J Hydrol 523:531–541. https://doi.org/10.1016/j.
on modelling and simulation. Modelling & Simulation Society of jhydrol.2015.02.004
Australia & New Zealand, Canberra, Australia, pp 1450–1456 Wilson JL (1993) Induced infiltration in aquifers with ambient flow.
Rodríguez LB, Cello PA, Vionnet CA (2006) Modeling stream-aquifer Water Resour Res 29(10):3503–3512
interactions in a shallow aquifer, Choele Choel Island, Patagonia, Woessner WW (2000) Stream and fluvial plain ground water interactions:
Argentina. Hydrogeol J 14(4):591–602. https://doi.org/10.1007/ Rescaling hydrogeologic thought. Ground Water 38(3):423–429.
s10040-005-0472-3 https://doi.org/10.1007/s10040-006-0110-8
Shah N, Nachabe M, Ross M (2007) Extinction depth and evapotranspi- Yeh H-D, Chang Y-C, Zlotnik VA (2008) Stream depletion rate and
ration from ground water under selected land covers. Ground Water volume from groundwater pumping in wedge-shape aquifers. J
45(3):329–338. https://doi.org/10.1111/j.1745-6584.2007.00302.x Hydrol 349(3):501–511. https://doi.org/10.1016/j.jhydrol.2007.11.
Shestakov VM (1965) Teoreticheskie osnovy ocenki podpora, 025
vodoponijeniya i drenaja [Theoretical base of estimation of an Zhang L, Hickel K, Dawes WR, Chiew FHS, Western AW, Briggs PR
afflux, drawdowns and drainage]. MSU, Moscow (2004) A rational function approach for estimating mean annual
Shestakov VM, Pashkovskii IS, Soyfer AM (1982) Hydrogeological in- evapotranspiration. Water Resour Res 40:W02502. https://doi.org/
vestigations on irrigated areas. Nedra, Moscow 10.1029/2003WR002710
Šimůnek J, Šejna M, Saito H, Sakai M, van Genuchten MT (2009) The Zlotnik VA (2004) A concept of maximum stream depletion rate for leaky
HYDRUS-1D software package for simulating the one-dimensional aquifers in alluvial valleys. Water Resour Res 40:W06507. https://
movement of water, heat, and multiple solutes in variably-saturated doi.org/10.1029/2003WR002932
media. Ver. 4.08, University of California, Riverside, CA Zume JT, Tarhule A (2008) Simulating the impacts of groundwater
Sophocleous M, Koussis A, Martin JL, Perkins SP (1995) Evaluation of pumping on stream–aquifer dynamics in semiarid northwestern
simplified stream–aquifer depletion models for water rights admin- Oklahoma, USA. Hydrobiol J 16:797–810. https://doi.org/10.
istration. Ground Water 33(4):579–588 1007/s10040-007-0268-8

Das könnte Ihnen auch gefallen