|
|
A COMPARISON OF STREAM-AQUIFER INTERACTIONS AT THE LOCAL AND THE REGIONAL SCALE
SENGSCHMITT D.1, SCHMALFUSS R.2,
BLASCHKE A.P.3
Institut of
Hydraulics, Hydrology and Water Resources Management, Technical University of
Vienna, Karlsplatz 13, 1040 Vienna,
Austria
1 Technical University of Vienna, Karlsplatz
13/E223, 1040 Vienna, Austria
phone: +43-1 58801-22321
fax: +43-1 58801-22399
e-mail: seng@bimb.tuwien.ac.at
2 Verbundplan GmbH, Parkring 12, 1011 Vienna,
Austria
phone: +43-1 53605-54827
fax: +43-1 53605-54809
e-mail: SchmalfussR@verbundplan.at
3 Technical University of Vienna, Karlsplatz
13/E223, 1040 Vienna, Austria
phone: +43-1 58801-22325
fax: +43-1 58801-22399
e-mail: blaschke@bimb.tuwien.ac.at
Stream-aquifer interactions are
of fundamental importance for studies of natural groundwater recharge and to
design irrigation and drainage systems or wells near streams. Field
investigations between surface water and groundwater are normally a regional
scale problem. In order to quantify the seepage rates of wells near rivers the
stream-aquifer interaction at the local scale is becoming more important. For
fundamental research on physical processes that control the stream-aquifer
interaction the process scale (physical scale) has to be focused. To integrate
fundamental microscopic research (e.g.: laboratory investigations) into
practical river basin management the question if dominant processes in one
scale can be found in another scale is of fundamental importance. This
investigation presented here will compare seepage rates from an 11.93 km
long Danube River reach in the Vienna area (regional scale) starting with the
impoundment of the river by the hydro-power dam "Freudenau" in March 1996 with
linear upscaled seepage rates from a cross-section within this reach (local
scale) until August 1998. Our comparison shows a good correspondence
between the two scales during the impoundment phase 1. But we were able to
show that linear upscaling is only possible if the following restrictions are
taken into account: ·
similar geohydraulic boundary conditions over the whole study site must be
assumed, · a
calibration factor has to be found due to the possible spatial variability of
geological and hydrological parameters and · the influence of the riverbed clogging
processes have to be analysed with the help of a seepage model. The reason is
that linear upscaling is only possible if the clogging processes have no
substantial influence on the seepage rates. In our investigation a decrease in
the leakage-factor of the streambed of 5 times during phase 1 at the local
scale shows a reduction in the seepage rates of only 10%. We could also show
that the change in the groundwater temperature from 2°C during winter to 20°C
during summer caused a periodic change in the seepage rates of 50%.
Keywords: : stream-aquifer interaction; seepage
rate; clogging processes
Stream-aquifer interactions
effect the dynamics of the adjacent aquifer due to flood events in the stream
(Blaschke et al., 1998) and are therefore of fundamental importance for studies
of natural groundwater recharge (Mitchell-Brucker et al., 1996; Serrano et al.,
1998) and to design irrigation and drainage systems or wells near streams
(Hantush, 1965). Also the exchange of contaminants from the river into the
aquifer and vice versa is a problem intimately related to the hydraulics of the
stream-aquifer system. The parameters that affect hydraulic exchange include
aquifer geometry, water-groundwater pressure difference (gaining-, loosing stream)
and the hydraulic properties themselves. Field investigations between surface
water and groundwater are normally a regional scale problem. In order to
quantify the seepage rates of wells near rivers the stream-aquifer interaction
at the local scale is becoming more important. For fundamental research on
physical processes that control the stream-aquifer interaction the process
scale (physical scale) has to be focused. To integrate fundamental microscopic
research (e.g.: laboratory investigations) into practical river basin
management the question if dominant processes in one scale can be found in
another scale is of fundamental importance. Only comparison between scales
which are orders of magnitude apart can give an answer. This investigation
presented here will compare seepage rates from an 11.93 km long river
reach (regional scale) with seepage rates calculated at a cross-section within
this reach (local scale) and can therefore be seen as a study to illuminate the
topic of scale problems.
The study site is part of the
river reach of the Danube River in the Vienna area, Austria. Within this reach
the river consists of a main channel (Danube River), which was impounded in
March 1996 by the hydro-power dam "Freudenau", and a flood bypass canal
(New Danube) along the left Danube bank (see Fig. 1). Between these two
channels there is an artificial island called "Donauinsel". The discharge at
the New Danube is regulated by means of weirs: the inlet weir
(river-km 1938.08) at the upstream end of the New Danube (used to regulate
the flow into the New Danube), weir 1 (river-km 1926.15) and
weir 2 (river-km 1918.30; used to maintain the water level in the New
Danube).Outside the flood periods, the water level in the New Danube is
regulated by the weirs 1 and 2, resulting in two calm, lake like surfaces
(upper reservoir, lower reservoir). The Danube River flow is regulated, and the
discharge varied from 770-6600 m3 s-1 (average
1700 m3 s-1) during the study period
(March 1996 until August 1998). At the location "Schulschiff" (SSF,
river-km 1931.4) we installed instrumentation to investigate the seepage
rates through the "Donauinsel" at the local scale (see chapter 4). This
cross-section is situated within the upper reservoir of the New Danube and the
adjacent Danube River reach, where we investigate the seepage rates through the
"Donauinsel" at the regional scale (see chapter 3). In March 1996 the
water level at the hydro-power dam "Freudenau" was increased to about
159.0 m a.s.l. which caused a water level increase at the
cross-sections "inlet weir " to about 160.8 m a.s.l., at SSF to
about 159.4 m a.s.l. and at "weir 1" to about
159.1 m a.s.l. (phase 1). Twenty months later (in
November 1997) the water level was increased a second time by additional 2.5 m
at the dam causing an increase at the cross-sections "inlet weir " to
about 162.0 m a.s.l, at SSF to about 161.5 m a.s.l. and at
"weir 1" to about 161.4 m a.s.l. (phase 2). During
phase 1 and phase 2 the water level at the upper reservoir of the New
Danube was about 157.65 m a.s.l. (±0.4 m). Since the impoundment in
March 1996 there has been a permanent groundwater flow from the Danube
through the artificial island to the upper reservoir of the New Danube due to
the existing difference in the pressure head of about 2.3 m (phase 1)
and 3.9 m (phase 2), respectively. In a certain part of the right
Danube River bank there is an impervious wall which impedes the exchange
between surface water and groundwater. The aquifer underneath the riverbed
consists of quarternary sandy gravel with a hydraulic conductivity of about
5 x 10-3 m s-1 and a thickness of
some decimeters to a few meters. It is underlain by an impermeable layer of
silt and clay (see Fig. 1).
Figure 1: study
site
The seepage rates through the
"Donauinsel" to the 11.93 km long upper reservoir of the New Danube (QIN,R,
see Fig 1) can be calculated with a water balance from the upper reservoir
(Schmalfuss, 1998). The change of the storage of the reservoir (DVRES) is equal to the groundwater
inflow at the right bank (QIN,R) plus the precipitation (P) minus
evaporation (E) multiplied with the reservoir surface area (AR = 1.97 x 106 m2)
minus the outflow at the left bank (QOUT,L,) and at weir 1 (QOUT,W,
see eqn. 1).
(1)
Weir 1 is closed during
night (5:00 p.m. to 7:00 a.m.) and the outflow during this period QOUT,W
is therefore zero. The outflow at the left bank QOUT,L can be
estimated with a groundwater-model to 0.4 m3 s-1.
The precipitation (P) and evaporation term (E) is measured at a climate station
situated at the "Donauinsel" (see Fig. 1). Putting this information into
equation 1 the water balance simplifies during night to
(2)
DVRES
can be calculated measuring the water level at weir 1 after closing the
weir (WC) and before opening it the next day (WO) with
the following equation
(3)
The infiltration term QIN,R
expressed in m3 s-1 can now be examined by combining
equation 2 and 3 and dividing it through the closing time of weir 1 (Dt)
(4)
This daily calculated values were averaged over
a period of one week and are shown in Figure 4 (closed dots) from
March 1996 to the end of August 1998.
4. SEEPAGE RATES AT
THE LOCAL SCALE
The seepage rates at the
cross-section SSF (local scale; situated approximately in the middle of the
study river reach, see Fig. 1) can be examined with the help of an
analytical steady state, one dimensional profile model (Sengschmitt et al.,
1998). This model splits the true two dimensional groundwater flow in to a one
dimensional vertical flow through the clogged riverbed and a one dimensional
horizontal flow in the aquifer below the streambed and the "Donauinsel" (see
Fig. 1). The boundary condition at the right Danube bank is a zero flux
condition due to the impervious wall (see Fig. 1). The left Danube bank is
also assumed to be impervious, which can be supposed due to the substantial clogging
of the block lining (Ingerle, 1991; Steiner et al., 1998). At any time t the
seepage rate Qt at location SSF can be calculated with the help of
equation 5 (see Sengschmitt et al., 1998)
(5)
where 
,
,
and 
,
provided that the following quantities are
known:
KF,10 = 6 x 10-3 m s-1
...... hydraulic conductivity of the
aquifer below the streambed at 10°C
KD,10 = 3 x 10-3 m s-1
...... hydraulic conductivity of the
"Donauinsel" at 10°C
n10 = 1.31 x 10-6 m2 s-1... kinematic viscosity at 10°C
nt (m2 s-1) ........................ kinematic viscosity at temperature t (see
Fig. 3)
hGW = 5 m
...................... thickness of
the aquifer below the streambed
L1 = 265 m
..................... width of the
Danube River at location "Schulschiff"
L2 = 220 m
..................... width of the
"Donauinsel" at location "Schulschiff"
H (m) .............................. upper reservoir hydrograph (related to
the elevation of the impermeable layer below the Danube River, see Fig. 2)
DH (m) ............................ difference between the water level in the
Danube River and the upper reservoir of the New Danube (see Fig. 2)
l10 (s-1) ........................... leakage factor of the clogged Danube
Riverbed at 10°C (see Fig. 2)
Figure 2: time series of the leakage
factor (losed dots), the pressure differences between the water level in the
Danube River and the upper reservoir of the New Danube (open dots, DH) and the upper reservoir hydrograph related
to the elevation of the impermeable layer below the Danube River (H)
Figure 3: time series of the seepage rates
at the location SSF (open dots) and the kinematic viscosity of the groundwater
(closed dots)
The kinematic viscosity nt, the differences DH
and H and the leakage factor of the clogged streambed l10 are strongly time dependent (see Fig. 2 and 3). The leakage factor
of the streambed was examined from April 1996 with the help of two
multi-level-piezometers situated at the riverbed (see Fig. 1) and the
analytical profile model presented above. The kinematic viscosity was derived
from the groundwater temperature which is equal to the temperature within the
Danube and the differences DH
and H were calculated for the time for which the leakage factor of the
streambed is known. Figure 3 shows the resulting seepage rates in m2 s-1
during steady state conditions in the Danube River (flood periods were not
considered!).
Upscaling the results at
location SSF we multiplied the seepage rates with the length of the upper
reservoir of 11930 m. Site SSF does not represent the mean hydrological
and geological conditions of the study site leading to some differences between
the seepage rates from the linear upscaled local scale and the regional scale.
A calibration factor has therefore be calculated for phase 1 and 2. For
phase 1 this factor was found to be 0.81 using data from the first five
weeks of phase 1 (Mai to June 1996). Figure 4 shows a further
good correspondence between the seepage rates at the two scales for
phase 1. The short time variation in the seepage rates (#1, #2, #4, #5,
#6, #7) can be found in both scales often having the same magnitude. Only the
peak in September 1996 (#3) calculated at the regional scale can not be
found at the local scale. A seasonal variability with high seepage rates during
summer and low seepage rates during winter can also be found in both scales
showing that an linear upscaling from the location SSF to the whole length of
the upper reservoir for phase 1 is possible. As a second step we used our
analytical model to analyse the influence of the two parameters - groundwater
temperature and leakage factor of the clogged streambed - on the seepage rates
during phase 1. The results show a periodic change of the seepage rates of
50% due to the change of the groundwater temperature from 2°C during winter to
about 20°C during summer. The effect of a 5 times lower leakage factor of
the streambed at site SSF during phase 1 (from about 5 x 10-5
in Mai 1996 to about 1 x 10-5 s-1 at
the end of phase 1, see Fig. 2) results in a reduction of the seepage
rates of only 10%. A calibration factor for phase 2 showing a good
correspondence could not be found. Taken the same factor of 0.81 as for
phase 1 the seepage rates between the two scales differ about 1 m3 s-1
at the beginning of phase 2 but move together after five months (at the
end of April, see Fig. 4). The reason for this discrepancy must be an
underestimation of the real seepage rates at the location SSF during the first
five months. The assumption of an impervious left bank which we used in our
analytical model was not fulfilled after additional impounding in
November 1997 (begin of phase 2). The new penetrated bank areas had a
good connection to the adjacent aquifer which was not closed during winter with
low suspended sediment concentrations of about 10 mg l-1 (see
Fig. 4). We suggest that the flood event at the end of April with
suspended sediment concentrations of about 200 mg l-1 clogged
the new penetrated areas very quickly leading to an further good correspondence
between the seepage rates at the two scales.
Figure 4: time series of the seepage rates
at the regional scale (closed dots) and at the linear upscaled local scale
(open dots); time series of the suspended sediment concentration within the
Danube River (source: Wasserstraßendirektion Wien)
We compared seepage rates from a
11.93 km long Danube River reach in the Vienna area (regional scale) with
seepage rates at the cross section SSF within this reach (local scale). With
the help of a calibration factor of 0.81 (calibrated using data from the first
five weeks of phase 1) which takes the spatial variability of the
hydrological and geological parameters into account, we were able to find a
good correspondence between the two scales during phase 1. Both the short
time variation due to fluctuations in the water level within the Danube River
and the seasonal variability due to periodic changes in the groundwater
temperature were found in both scales, also having the same magnitude. With the
help of our analytical model used at the local scale we were able to show that
the change in the groundwater temperature from 2°C during winter to 20°C during
summer caused a periodic change of 50% in the seepage rates. In contrast to
that the seepage rates decrease only 10% as a consequence of 5 times lower
leakage factors of the clogged streambed at the end of phase 1 at site
SSF. Additional impounding of the Danube River in November 1997
(phase 2) leads to some differences between the time series of the seepage
rates at the two scales of about 1 m3 s-1.
After the first flood event at the end of April 1998 this differences
disappeared leading to a further good correspondence between the seepage rates
at both scales. This investigation shows that upscaling seepage rates from the
local to the regional scale is only possible if the following restrictions are
taken into account: ·
similar geohydraulic boundary conditions over the whole study site must be
assumed, · a
calibration factor has to be found due to the possible spatial variability of
geological and hydrological parameters, · the influence of the riverbed clogging
processes have to be analysed with the help of a seepage model. The reason is
that linear upscaling is only possible if the clogging processes have no
substantial influence on the seepage rates (like in our investigation for phase 1).
Blaschke, A.P., Gutknecht, D. (1998) Kurz- und langzeitige Änderungen in
Grundwassersystemen. Wiener Mitteilungen, Wasser - Abwasser - Gewässer, Nr.
148, 373-392
Hantush, M.S. (1965) Wells near Streams with Semipervious Beds. J. of Geophysical Research, Vol. 70, No. 12, 2829-2838.
Ingerle, K., (1991) Über die Flußbettdurchlässigkeit und die Sauerstoffzehrung des Uferfiltrates im Staubereich von Donaukraftwerken. Wasserwirtschaft, Nr. 81, 415-422
Mitchell-Bruker, S., Haitjema, H.M. (1996) Modeling steady state conjuctive groundwater and surface water flow with analytic elements. Water Resour. Res., Vol. 32, No. 9, 2725-2732.
Schmalfuss, R. (1998) Der Einfluß der Grundwassertemperatur auf den
Wasserhaushalt am Beispiel der Neuen Donau. Wiener Mitteilungen, Wasser -
Abwasser - Gewässer, Nr. 148, 373-392
Sengschmitt, D., Steiner, K.-H., Blaschke, A.P., Schmalfuss, R. (1998)
Einfluß der Kolmation auf den Grundwasserhaushalt am Beispiel des Stauraumes
Freudenau. Wiener Mitteilungen, Wasser - Abwasser - Gewässer, Nr. 148, 321-350
Serrano, S.E.,
Workman, S.R. (1998) Modeling transient stream/aquifer interaction with the
non-linear Boussinesq equation and its analytical solution. J. of Hydrology,
Vol. 206, 245-255.
Steiner, K.-H., Blaschke, A.P., Gutknecht, D. (1998) Auswirkungen
künstlicher Dekolmationsmaßnahmen: Fallstudie Altenwörth. Wiener Mitteilungen,
Wasser - Abwasser - Gewässer, Nr. 148, 289-319