Você está na página 1de 16

1488

ditions and to fulll different purposes, they are often In consideration of these problems with recharge
difcult to compare, especially with respect to spatial estimation,
and
the two parallel objectives of this study are to:
temporal scales (Risser et al. 2009; De Vries and Simmers
2002). The parallel use of at least two conceptually 1. Detect structural errors and inconsistencies between
different approaches has been recommended to allow two contrasting approaches for estimating recharge:
cross-checking of results and to establish their plausibility
percolation calculated with a soil-water balance model,
(see e.g., Risser et al. 2009; Healy and Cook 2002;
and water-table uctuation methods
Scanlon et al. 2002).
2. Improve overall capabilities for using water-table
Among the many methods of groundwater recharge uctuation methods to estimate recharge on a regional
determination which are available, one group is partic- scale by addressing the problem of hidden recharge
ularly attractive due to its straightforwardness and
simplicity. Water-table uctuation (WTF) methods eval- In line with these objectives, this paper presents:
Zhang
Jie & Janinvan
Heyden & David
& that any
uate changes
groundwater
levels Bendel
assuming
Roland
Barthel
increase in the water table is most likely a result of 1. A review and description of existing approaches for
recharge. It is somewhat surprising that relatively few calculating the potential groundwater rise, in particmethods for estimating groundwater recharge include an
ular, the master recession curve (MRC) method
Keywords
Groundwater
budget .
analysis of
groundwater
level uctuations,
as groundwater
Abstract
Despite
a long history
of related research,
developed
by Heppnerrecharge/water
and Nimmo (2005)
Water-table
uctuation
(WTF)
method
. Groundwater
level
levels are the
only
directlyrecharge
measurable
result
of recharge.
quantifying
and
verifying
is still
a major
2. The development
of new
WTF
methods
based on the
time
series
.
Germany
.
Groundwater
monitoring
One
of
the
reasons
for
this
might
be
that
the
relationchallenge. The combination and comparison of conmethod developed by Heppner and Nimmo (2005)
ship between
groundwater
changes
and recharge
ceptually
different
methodshead
has been
recommended
as
3. The application of existing and newly developed WTF
is strategy
not particularly
straightforward
simple (Note:
a
for evaluating
rechargeor
estimates.
In this
methods to real-world data
since actual
measurements
are used,
these methods
areA comparison of existing and newly developed WTF
article,
recharge
estimates from
water-table
uctuation
4.
sometimes
referred
to as direct
methods,
which to
is, Introduction
(WTF) methods
are combined
with
and compared
methods, with a special focus on the aforementioned
however,
misnomer,
as the
ofdiscretized
this article problems with hidden recharge, in order to determine
the
resultsa of
the spatially
andremainder
temporally
I. Voss (Voss 2005) writes in the editorial for
will clearly show). model PROMET (processes of radi- Clifford
soil-water-balance
the best method to be used in conjunction with the
Hydrogeology
Journals special issue The Future of
The mass
work and
presented
has two
objectives.
The
ation,
energyhere
transfer).
As main
PROMET
and
research question
Hydrology:
rst
of
these
is
related
to
a
problem
encountered
in
the
WTF methods rely on different measurable variables, 5.
a A description of one potential option for combining
large-scale integrated
research
project
GLOWA-Danube,
comparison
of these two
contrasting
techniques
allows and cross-validating WTF methods with the results of a
Hydrogeology
is soil-water
far from being
a typical
where theassessment
authors task
develop and
calibrate a physically
improved
ofwas
the to
plausibility
of recharge
based
balance
modelquantitative
(PROMET)
science.
.
Each
model
and
each
prediction
is a
regional-scale
groundwater
ow model
for methods
a large catchestimates.
An enhanced
approach
to WTF
is
hypothesis
and
there
are
rarely
true
tests
of
these
ment
with
an
area
of
77,000
km
2
(Barthel
et
al.
2005;
presented. The approach assumes that in the case of no
Wolf et al. there
2008)exists
usingapercolation
data provided
by the predictions. . In other words, hydrogeologists
recharge,
maximum possible
potential
physically
spatially
distributed
soil-water
balancebelieve that their approach is a meaningful and useful
decline
forbased,
any given
groundwater
level.
The primary
way to
proceed, butthe
cannot
prove it objectively.
Project
background:
GLOWA-Danube
project
model PROMET
(Mauser
and Bach
2009).
Initial
conclusion
is that
WTF methods
are
excellent
foranalysis
showed that the
the plausibility
relationshipofbetween
zone percoladetermining
spatiallyroot
distributed
The
This
GLOWA-Danube
statement is particularly
project wasrelevant
carried out
to groundwater
from January
tion
and groundwater level changes
was not
always as
regional-groundwater-recharge
estimation
approaches
2001 to October
which is2010
dened
by an
(e.g.,
interdisciplinary
Hiscock 2005)consortium
as the
expected
(see Barthel
2006). Percolation
sometimes
and
for detecting
inconsistencies
in available
models.recharge
of
amount
of surface water which reaches the permanent
occurred without
anyderived
changefrom
in groundwater
level, and
Recharge
estimates
WTF approaches
17 research
water
table. groups,
The accurate
private
quantication
companies, and
of groundwater
state
in other
cases
groundwater
levelsfor
rose
even when
alone
are,
however,
not suitable
regional-scale
agencies
recharge
is
essential
for
the
sustainable
management
of
simulated
percolation
did
not
occur.
Often,
quantity
and
recharge estimation due to (1) their strong dependency
and aimed at resources
groundwater
providing (De
an integrated
Vries and Simmers
approach2002)
to
and
dynamics
of simulated
percolation
observed
groundon
local data,
applicability
of whichand
is limited
to only
predicting
one of the major boundary conditions for many, if
waterspecic
levels were
not related
in the
expected
way
very
conditions,
and (2)
their
sensitivity
to forms
changes
not
all, groundwater-related
in the hydrological cycle
calculations
due to global
and models
change
(see
in
(dampened,
delayed,
inuences
other
than distorted).
recharge. These observations led
the Sanford 2002; Scanlon et al. 2002). Unfortunately, it
e.g.
to the idea of trying to validate and to detect structural
Upper
not possible
Danube to
Catchment
measure of
groundwater
southern Germany
rechargeand
directly.
errors of the percolation results from PROMET using is
north- and Redshaw (1979) and Schirmer (2008) point
groundwater level measurements in combination withRushton
ern Italy
out
groundwater
(UDC; approx.
recharge
77,000
determination
km2; Fig. 1).
asGLOWAbeing one of
WTF methods.
Danube
the
main
considers
challenges
both
in
hydrogeology.
the
inuence
of
Lysimeters
natural
changes
used in in
The
second
objective
is
related
to
a
more
general
Received: 10 June 2010 / Accepted: 24 July 2011
the eld
ecosystem
cannotsuch
takeas
measurements
climate changes,
directly
as well
at the
as water
those
interest in
using
WTF methods
Published
online:
2 September
2011 to enhance the possibilities
ofon and are further limited by their expense and the
for appropriately determining and validating rechargetable,
* Springer-Verlag 2011
social changes, e.g., changes in land use or water confact
a
sumption.
they provide
Further information
details are provided
at single by
points
Barthel
only.et al.
regional scale. WTF methods suffer from two differentthat
Z.
Jie : D. Bendel
: R. Barthel
())a known specic yield (Sy) Scientists
(2005,
and Barthel
practitioners
(2011);
have,
Ludwig
therefore,
et al. (2003);
long sought
Mauser
problems:
(1) they
require
(see 2010);
Institute of Hydraulic Engineering (IWS),
and
develop
Bach (2009);
methods
and
for
Nickel
determining
et al. (2005).
recharge, among
Eq. 1), which can be very difcult to determine (Healyto
Universitaet Stuttgart(
The UDC is a heterogeneous, mountainous catchment.
and
Cook 2002),
and (2)
it is difcult
to distinguish them
Pfaffenwaldring
61, 70569
Stuttgart,
Germany
Altitudes
(1)
water-balance-based
within the catchment
methods,
range
(2)from
approaches
287 to 4,049
basedm
between
recharge-induced groundwater changes and
e-mail:
roland.barthel@iws.uni-stuttgart.de
above sea level (a.s.l.). Climatic and hydrological variaon
those
Tel.:
+49-711-68566601
bles vary
Darcys
law,
as follows
(3) tracer
over
techniques,
a large spatial
(4) numerical
range: precipmodeling
resulting from other processes. In particular, it is difcult
J.
itation from
(5)650
empirical
to overmethods,
2,000 mm/year,
and (6) average
so-calledannual
direct
toHeyden
determine whether or not actual recharge is being methods,
Bavarian Environment Agency(
temperature
(see
comment
from
on their
4.8 to
directness
+9C, evaporation
in the following)
from 450 to

Combination of soil-water balance models and water-table


uctuation
methods for evaluation and improvement of groundwater recharge

Lazarettstr. 67, 80636 Mnchen, Germany

Hydrogeology Journal (2011) 19:


19:14871502
14871502

DOI 10.1007/s10040-011-0772-8

1489

Fig. 1 The Upper Danube Catchment: a location and topography; b simplied geological map

(60%) water in the area. The hydrogeology of the UDC


catchment area is characterized by alluvium. Figure 2
is dened by four major zones: the Alps, the Molasse shows the location of 30 wells specially selected to
Basin (Tertiary), the Jurassic Karst, and the Crystallinedemonstrate study results.
Basement Complex (Fig. 1).
In the UDC, groundwater level times series data
collected at least weekly, if not more frequently, are Methods and models used
available for each of approximately 3,700 groundwater
observation wells. The longest time series available Soil-water balance model PROMET
contains data collected between 1918 and 2006. Many
PROMET (processes of radiation, mass and energy
of these wells are located in shallow aquifers in the transfer; Mauser and Bach 2009), is a process-based
alluvium found alongside the rivers of the catchment,soil-water balance model which simulates the uxes of
and are screened at depths between 1 and 20 m. While
energy and matter on the land surface and in which
wells constructed in alluvium represent 60% of all
mass and energy balances are strictly conserved. It is
wells in the project area, only 510% of the total
raster-based, has a 1-km2 resolution and runs on an

Fig. 2 Selected groundwater observation wells (30) in the Upper Danube Catchment for the demonstration of results
Hydrogeology Journal (2011) 19: 14871502

DOI 10.1007/s10040-011-0772-8

1490

1491

hourly
Limitations
time step.
of the
PROMET
WTF method
is comprised
are (1)of
that
eight
other Master recession curve approach (Delin et al. 2007)
causescomponents.
model
of groundwater
Thelevel
soil hydraulic
uctuations
and
unrelated
soil temperto In the master recession curve approach by Delin et al.
(2007), the potential decline curve is determined using a
recharge
ature
component
(e.g., pumping,
is the most
direct
important
evapotranspiration,
to this study.earth
It
nonlinear regression equation known as the master
tides,developed
was
etc.) are neglected,
to model soil-water
(2) that itcontent
cannot and
account
soil for
recession curve (MRC). More specically, the potential
recharge occurring
temperature,
as well
atas
a steady
the vertical
rate (e.g.,
and lateral
when ows
the of
recharge
water
in and
rateout
is equal
of theto
unsaturated
the discharge
soilrate,
as a as
function
is the decline
case curve is dened by the natural log of the
difference in elevation between the actual groundwater
forinltration,
of
long, low intensity
exltration,
precipitation
percolation,
events,
and capillary
the method
would
rise.
The
predict
component
no recharge),
uses an(3)
approach
that regionalizing
described results
by level and the asymptote that the groundwater recession
approaches (i.e., the MRC). Unlike the graphical
may signicantly
Eagleson
(1978). Soils
reduce
arethe
divided
accuracy
into of
four
thelayers,
calculated
approach, the MRC approach avoids subjectivity once
recharge
each
withbecause
a different
groundwater
layer thickness
level which
uctuations
depends on
the necessary parameters have been calculated. It also
measured
the
soil type. Phillips equation is explicitly solved
in a well
under
simplifying
are only representative
boundary conditions
for a small
and assumparea of a yields a better recharge approximation than the RISE
approach does because it accounts for groundwater
heterogeneous
tions.
In general,
aquifer,
excess(4)
water
thatpercolates
the key parameter,
into the
specic
underlying
soil layer, and except in the bottom layer, decline in the presence of recharge.
yield (Swater
excess
y), is acan
function
also move
of time
laterally
and is difcult
into neighboring
to measure,
Fig. 3 Potential groundwater decline curve (dashed red line) and
elements. Water percolating through the deepest soil measured groundwater level (solid blue line) resulting from
in an unconned aquifer (after Healy and Cook 2002)
layer enters the groundwater component and is con- recharge
Master recession curve approach (Heppner and Nimmo
ceptually considered as groundwater recharge. The 2005)
term
percolation
as used the
in this
articlegroundwater
refers to the The
Approaches
for determining
potential
(Fig. MRC
3, dashed
line),byhere
referred
as the (2005)
potential
approach
Heppner
andtoNimmo
water percolating out of the deepest soil layer into thegroundwater
decline
The determination
of this
assumes thatdecline
there iscurve.
a functional
relationship between
groundwater
component.
Contrary
to assumptions
During the last
few decades,
a number
of attemptsmade
have
potential
decline
the groundwater
main objective
of most
groundwater
levelisand
decline,
orrelated
more
by
themade
PROMET
authors,the
percolation
cannot
beof these
been
to improve
WTF method.
Most
methods
andthat
is also
the groundwater
target of the levels
improvements
specically,
higher
lead to higher
considered
identical
to groundwater
recharge
(seeto described
have focused
on estimating
groundwater
rise due
in this paper.
hydraulic gradients
(for example towards rivers) and thus
Barthel
2006
explanations).
recharge
(h),for
and
more specically on the question of
(R) is calculated
for aHere,
givengroundwater-level
time interval
toRecharge
higher groundwater
declines.
how to determine the potential groundwater decline (see
(t)
as the
potential
multipliedlevel
by
decline
is dened
as groundwater
the change inrise
the(h)
groundwater
Fig. 3). Attempts to improve the calculation of h have
the
specic
yield
(S
y
)
(Healy
and
Cook
2002):
after one time interval (e.g., one day) has lapsed.
been made by various authors including, for example,Different
The
water-table
uctuation
(WTF)
method
Moon et al. (2004), who modied the WTF method to regression
dhDh
(linear, power, bin mean, bin
median)
Sy functions
1
take into account cumulative precipitation in periods R
Syavailable data in order to describe the
are
tted
to
History
when and background of the WTF method
dtDt
Methods
for estimating
groundwater
recharge
from relationship between groundwater level and groundwater
the cumulative
water-level
rise reached
a maximum
decline.
The potential groundwater decline can then be
groundwater
level time series were rst applied in the
value,
calculated
for a given groundwater level using these
1920s
(Meinzer
1923;
Meinzer
and Stearnsand
1929).
Howand Crosbie
et al.
(2005),
who considered
removed
regression
functions (Heppner et al. 2007). This approach
Specic yield
ever,
applications
relatively
andinhave
been
the Lisse
effect, remain
which refers
to anscarce
increase
groundhas
been
used
by otherin
authors
(e.g. Crosbie etrecharge
al. 2005;
S
y
,
a
crucial
parameter
WTF approach-based
questioned
critically
bydue
several
authors
(e.g., Johansson
water level that
is not
to direct
recharge.
The
calculations
(see
Eq.
1),
is
dened
as
the
ratio
of
the
1987). After 1990, Sophocleous (1991) developed a
method called hybrid water-uctuation which redrew water
volume
that drains from
a saturated
rock due to gravity to
attention to WTF methods (Ketchum et al. 2000).
New developments
and
enhancements
However, the most widely used method, and the onlythe total volume of the rock (e.g., Fetter 2001). However,
Graphical approach
this
denition
is Introduction,
only valid under
one addressed in detail in this paper, is the water-table
As stated
in the
oneequilibrium
motivation conditions
for this
In
the graphical
and Delin
2007),
in
small-scale
experiments,
and
is notand
directly
uctuation
(WTF)approach
method. (Lorenz
An excellent
review
of the and
study
was
the
potential
for
improvement
validation
potential
applicable
or useful recharge
in eld studies
requiring
the considtheory and application of this particular method is
of
the
groundwater
results
used
in
the
decline
curves
are extrapolated
manually in order to eration of numerous factors. Sy is in fact a function of
presented
by Healy
and Cook (2002).
GLOWA-Danube project through the combination of
obtain
media
porosity,
to the
water table,
WTF
Methoddepth
and the
physically
baseddrainage
soil-water
the potential groundwater decline based on a visual the
duration,
and
antecedent
moisture
conditions,
among
balance
model
PROMET.
This
potential
was
to be
inspection of the dataset. This approach is more
other
realized via the consideration of an additional measPrinciples
subjectiveof the WTF method
variables
(see Shah
and Ross
2009 for example).
The
urable parameter
(namely
groundwater
level) in the
The WTF method is based on the assumption that
parameter
is difcult to
measure,
and there isresults,
no efcient
checks
for
plausibility
of
PROMET
percolation
groundwater level rises in unconned aquifers are caused
which are typically validated using river discharges
by recharge. The water table is assumed to decline mainly
only.
due
to
drainage
into
rivers
(Schwartz
and
Zhang
2003).
It In a preliminary stage of the research presented in
RISE approach
this paper, several
WTF approaches
applied to the
is
also
assumed
that
drainage1998)
into rivers
occurs
during
The
RISE
approach
(Rutledge
makes
the simple
Advantages
and disadvantages
of WTFwere
methods
data andisthe
results compared
periods
when
groundwater
recharge
also occurs
takes place.
AnyWTF method
assumption
that
no groundwater
decline
in theavailable
The
straightforward,
easyto
tooutput
apply, from
PROMET.
This
comparison
led
to
the
conclusions
rise
in the
tableThe
duepotential
to recharge
(h; Fig. 3)
is thus
absence
ofwater
recharge.
groundwater
rise
(h)
and makes use of directly measurable data. In that
WTF-method
results and
PROMET results
do not for
comprised
of
two
parts:
the
actual
groundwater
rise
is simply dened as the absolute rise in groundwater (1)
comparison
to baseow
separation-based
methods,
under
certain
hydrogeological
conditions
that
related
level. to a given recharge event, and the potential agree
example, one of the biggest advantages is that and
it can
someto
ofcalculate
the assumptions
in temporal
WTF approaches,
groundwater
decline
which wouldlevel
havedoes
taken
in(2)
In cases where
the groundwater
in place
fact decline,
be used
rechargemade
in high
those hourly
concerning
the determination
of potencase
The actual
groundwater
rise can especially
be
h is of
setno
torecharge.
zero. Because
a potential
groundwater-level
resolution
(e.g.,
or daily).
Furthermore, the
tial
declines,
seem
to
be
inappropriate
(see
Zhang
measured,
whereas
the
potential
groundwater
decline
decline would occur in the absence of recharge, this WTF method directly involves the result of the
2009).
As a result,
several
modications
to the
must
approach is expected to result in an underestimation of
groundwater
recharge
process,
unlike many
of the
previously
describedwith
WTFunsaturated
approaches processes.
were made based
be
determined
indirectly.
The WTF
decline
in groundwater
recharge.
Compared
to other
methods,
however,level
it is
methods
concerned
Hydrogeology Journal (2011) 19:14871502
19: 14871502

DOI 10.1007/s10040-011-0772-8

1492

on (1) a theoretical evaluation of the weaknesses in Continuous decline procedure (CDP)


existing approaches and (2) results from the application
The CDP denes potential declines as groundwater-level
of existing methods to data from selected observationdeclines which are continuous for at least 2 weeks. It
wells in the project area.
assumes that in dry periods, groundwater levels are not
The groundwater-level/groundwater-decline relation-affected by recharge (Fig. 4).
ship approach, developed in this paper, extends the
MRC approach by Heppner and Nimmo (2005), which
establishes a relationship between groundwater levelsNo precipitation/percolation procedure
and
The no precipitation/percolation procedure (NPP) considall positive groundwater declines, thereby making it ers groundwater-level declines which occur during periods
possible to determine groundwater declines from a given
without corresponding precipitation (or percolation) to be
groundwater level. They state that one limitation of their
potential declines. The underlying assumption is that if
approach is the poor relation between water-table the
elevation and decline rate. This poor relationship is most
source of recharge (precipitation/percolation) is zero, then
likely a result of the assumption that all positive groundthe corresponding declines occur in the absence of
water declines are potential declines and not only those
that occur in the absence of recharge, therefore, risking
underestimation of groundwater recharge. Johansson
Maximum decline procedure
(1987), for example, only used data from November to
April in order to get the true recession caused only The
by maximum decline procedure (MDP) only takes the
maximum
declines in each groundwater level range into
groundwater ow. He assumes that the most rapid
account.
Assuming
that groundwater discharge varies
recessions in that period occur on occasions of no
recharge. The groundwater-level/groundwater-declinewith
groundwater level, the largest groundwater declines
relationship approach proposes a functional relationship
not between groundwater levels and all groundwater (maximum declines) in each groundwater level range will
declines, but rather between groundwater levels and most likely occur during periods of no recharge. In this
potential groundwater declines (i.e., the groundwater study, the uppermost two declines have been selected
declines occurring in the absence of recharge). The from
difculty lies in the isolation of potential declines. To groups of ten (Fig. 6).
For all three procedures, the selected potential
address this challenge, three alternative procedures have
decline data were divided into groundwater level
ranges called bins. A bin is a range of groundwater
level elevations and was rst introduced by Heppner
1. Continuous decline procedure (CDP)
and Nimmo (2005). Within each bin groundwater level
2. No precipitation/percolation procedure (NPP)
changes can be treated differently to account for
3. Maximum decline procedure (MDP):

Fig. 4 Selection of groundwater-level/groundwater-decline data recorded daily for well 529 in January 1992, corresponding to a
period of
Hydrogeology Journal (2011) 19:14871502

DOI 10.1007/s10040-011-0772-8

1493

Fig. 5 Selection of groundwater-level/groundwater-decline data recorded daily for well 529 in July 1990, corresponding to a period of
potential decline as dened by the no precipitation/percolation procedure (red bars)

averaged into bin-average points which are then usedgroundwater-level time-series data available for the
to determine groundwater-level/groundwater-decline UDC. The groundwater level/groundwater-decline relarelationships (Fig. 6).
tionship approach was applied using each of the three
procedures developed to determine potential declines,
namely CDP, NPP, and MDP. The results are compared
with those resulting from application of the MRC
Application
approach by Heppner and Nimmo (2005) and from the
RISE approach.
In order to meet the objectives outlined in the
For the purposes of this study, 30 representative
introduction, different WTF approaches, including
those newly developed in this study, were applied to observation wells were selected for analysis (Fig. 2) from

Fig. 6 Selection of groundwater level/groundwater-decline data recorded daily for well 529 in 1990, corresponding to a period of
potential decline as dened by the maximum decline procedure, showing all calculated declines, maximum declines, bin-average
points,
Hydrogeology Journal (2011) 19: 14871502

DOI 10.1007/s10040-011-0772-8

1494

1495

a total
Step
3.of
Determination
3,700 observation
of thewells
potential
available.
groundwater
Among the
rise:
precipitation
behavior
dataand
from
variations
climate stations
in groundwater
or percolation
level time
criteria
Using
used
the
to functional
make this relationship
selection were:
determined in data
series at different spatial locations.
the previous step, the potential groundwater
Thedecline
availability
of atgroundwater
least a 20-year-long
time
series
(i.e. the
level in the
case
of Step
1. Preprocessing:
choosing the
30 representaWhile
steps 14 rely After
on groundwater
head
time series,
withno
measurements
taken
at
least
weekly
recharge, see Fig. 3) was determined at each stepstive
observation
small ifgaps
and obvious
5 and
6 are onlywells,
applicable
spatially
distributed
Thetime
complete
absence
of gaps study
in data
or theThe
presence
step over
the 11-year
period.
sum
outliers
each
of the groundwater
level time
recharge
datainare
available
from other sources
(for
of only
small
data gaps,
and no obvious
of the
potential
groundwater
decline outliers
and the
series
were
removed
to the
produce
equidistant
example
from
PROMET
as in
present
case). time
No inuence
from nearby
pumping
or irrigation
actual groundwater
rise
is the potential
groundseries.
All time
series wereofmade
the same
length
In this
study,
the estimation
recharge
by WTF
Location
mainly
in
the
alluvium
water rise h over the period t.
(11 years),
theout
1stby
ofaJanuary
to the
approaches
was from
carried
Python 1990
computer
Step 4. Calculation of recharge: Using the calculated program
31stwhich
of December
2000.
Valuestime
wereseries
aggregated
calculated
recharge
using ve
rise h from
the
daily,
weekly,
and (RISE
monthly.
The potential
resulting groundwater
set of wells islevel
geographically
widely
different
WTF
methods
approach, MRC approach
previous
step, recharge
was calculated
using sites
Step
2. Determination
of(2005),
the functional
relationship
distributed
throughout
the project
area and includes
by Heppner
and Nimmo
CDP, NPP,
and MDP).
1. Inofthis
study, Sy values
derived from
between
declineagainst
and groundwater
with aEq.
range
hydrological
characteristics
and inuencThis output
wasgroundwater
saved and plotted
percolation,
pumping
tests Table
or other
related
data were
not precipitation,
level: Atand
rstgroundwater-level
all groundwater declines
werethe proing climatic
factors.
1 lists
the main
characteristics
data which
any of the observation wells. Values
for every
time stepThe
andcross-correlation
sorted by
of theavailable
selected for
wells.
gram calculated
extracted from
the database.
of the parameter
used
other locations
were
groundwater
level.
The largest
two
out of tenand a
Percolation
results from
theinsoil-water
balance
model
between
recharge and
percolation
was
calculated
therefore
collected
from in
regional
literature for each
declines
were selected
as maximum
PROMET
were made
available
daily resolution
scatter
plot showing
water table
vs. declinedeclines.
rate and the
notentirety
referenced
here
due
groundwater
level
range
model(reports,
cell (1 maps,
1 km)thesis,
over the
of the
UDC
and
corresponding
for After dividing
ttedthe
linear
function was
generated.
Speto number
and
limited
public
availability),
into values
10 binswere
the maximum
were
the period
1960 to
2006
through
the
courtesy ofand
Prof. cic
Dr. yield
calculateddeclines
for every
well. (For the
found(University
to range between
0.01
to 0.25 forof the sourceassigned
to their
appropriate
bin. The groundW. Mauser
of Munich,
Coordinator
code, please
contact
the corresponding
author.)
alluvium. Aproject).
Sy value of 0.08 for every time step
water declines in each bin were averaged to a
GLOWA-Danube
t was estimated
to be most
appropriate
for the
Calculations
were performed
for the
30 selected
wells. bin-average point and a functional relationship
majority
of the locations
considered
in theinstudy,
between groundwater decline and groundwater
A detailed
description
of each step
is provided
the
based
inthe
geological
level was then derived from a linear function
following
foron
thesimilarities
example of
MDP. Forconditions
the other and
values small
foundchanges
in the literature.
tted to the bin-averages.
procedures,
apply to step 2 (Note: For NPP,
Results
Step 5. Simulation of Sy using PROMET percolation
estimates for cross-validation: This additional
The following sections primarily describe results of the
or alternative step can be carried out to link
MDP which was newly developed in this study. Results
WTF
approach
results
and PROMET
output.
Table 1
Basic
characteristics
of sampled
groundwater
observation wells
the four other procedures
For each
observation
well, the percolation time derived from the application of
Geology
a
Land use
b
Well ID
Elevation Distance to nearestare
Depth
water Aquifer
alsotocompared
briey. It should
be
noted that the
series corresponding to
the
cell
in
which
the
[m.a.s.l.]
river [km]
table
[m]
thickness
[m]
quantity
of
results
and
subsequently
the
potential for
observation well is located were extracted from
analysis
are
considerable.
Here,
only
a
small
selection of
88
Q/Alluvium
676.96
0.72
1.88
2.05
Arable
land
PROMET. If it is assumed that PROMET
95
J/Karstic Limestone 529.68
0.42
14.29 is presented.
110.00
Discontinuous urban fabric
results
reliable values
for percolation
and
182 calculates
Tr/Sandstone
465.58
4.16
1.41
34.13
Complex cultivation patterns
percolation is equal
to recharge
246 thatQ/Alluvium
405.54
0.86 if aggre4.39
7.02
Pastures
320 gated
C/Sandstone
over long periods
387.74
of time (in
7.90this paper
1.71
19.50
Coniferous forest
529 11 years),
Q/Alluvium
0.10
1.51
4.62
Complex cultivation patterns
Eq. 1 can be366.32
used to calculate
Sy
563using J/Karstic
Limestone values.
440.15 This2.09
81.07
260.30
Maximum
decline
procedureConiferous
(MDP) forest
known recharge
provides
576
Q/Alluvium
325.37
2.71
2.10 2 provides4.78
Arable
Table
an
overview
of
theland
main results for the
the option
for validation
and in-depth
analysis
663
P/Gravel
320.81
0.66
6.59
7.70
Arable land
30
selected
observation
wells.
of
either
PROMET
or
the
WTF
approaches.
In
695
Q/Alluvium
390.36
1.20
1.76
9.86
Arable land
In the following,
wells have arable
been land
sorted into four
where the calculated
702 cases
Q/Alluvium
382.68Sy values
4.95 fall
1.58
3.30
groups
to specic characteristics
of the results.
724 within
Q/Alluvium
1.65 according
1.82
Arable land
the range of the365.42
Sy values3.40
provided in
907 the regional
Q/Alluviumliterature 406.24
2.00
2.17
5.45
Arable
land
Group
1:
(0.010.25, with a
909
Te/Sandstone
390.70
0.82
3.63
Complex cultivation patterns
median
of around 0.08),
it can at9.07
least be
936
Q/Alluvium
378.66
7.05
1.52
8.33
Arable land
that neither of
these methods
is
4.16
Reasonable simulated
Sy values
1129 assumed
Q/Alluvium
409.49
2.75
10.60
Arable(0.01
land < Sy < 0.25,
However, 2.10
in cases where assuming
that 4.60
0.25 is an upper
boundaryurban
corresponding
1162 fundamentally
Q/Alluvium wrong. 463.62
2.47
Discontinuous
fabric
1412 Sy values
Q/Alluvium
lie outside the
467.35
range found
5.30 in the
3.72
18.50
Discontinuous
urban
fabric
to
naturally occurring
geological
media and
0.01
is a
1450 literature,
Q/Alluvium
431.55
3.56one of the
24.62 boundary
25.68
Complex
cultivation patterns
it must be assumed
that
lower
corresponding
to unconned,
uncon1494 methods
Q/Alluvium
534.60
4.26
13.80
is awed. This
step can,5.58
therefore, be solidated
aquifer
material Arable land
1514
Q/Alluvium
538.43
5.34
13.41
17.60
Discontinuous urban fabric
usedP/Gravel
for invalidation (though
not0.09
necessarily in 2.44
Positive linear -groundwater-level/groundwater-decline
1614
424.14
Water courses
rather more for
the particular
observation
relationship
1924 general;
Q/Alluvium
723.84
0.66
2.08
4.49
Pastures
under consideration).
3.35
High correlation
coefcientsConiferous
(> 0.4) between
calculated
1945 well P/Moraines
798.62
9.16
15.00
forest
2407
Q/Alluvium
442.60
0.36 was used to recharge
0.86
9.20
Arable land
Step 6. Post-processing:
Cross-correlation
and percolation
2439 analyze
Q/Alluvium
590.50
0.14 the time lag 5.00
31.31
Pastures
the linear dependency
and
2564
Unknown
472.70
0.43
9.97
11.70
Discontinuous urban fabric
between percolation and the calculated recharge Group 2:
2575
Unknown
448.10
0.18
4.45
6.35
Discontinuous urban fabric
series. Further, an
analysis0.14
of the ground- 2.01
2719 time
Unknown
500.48
9.00
Pastures
level/groundwater-decline
relationship
4417 water
Unknown
861.87
0.08
5.30
Pastures
2.34
Reasonable simulated
Sy values
(0.01<Sy <0.25; see
a
derived
for each
provided
a better underthe preceding)
See Wolf
et al. (2008)
for awell
detailed
description
standing of
factors that determine drainage (Almost) linear and positive groundwater level/groundbNo irrigationrain
fedthe
agriculture
Q Quaternary, P Pleistocene, Te Tertiary, Cr Cretaceous, J Jurassic,water-decline
Tr Triassic.
relationships
Hydrogeology Journal (2011) 19:
19:14871502
14871502

DOI 10.1007/s10040-011-0772-8

1496
Table 2 Summary of results related to the application of the maximum decline procedure (MDP)
Well ID []

Precipitation Percolation
Recharge R Ratio
N [mm/11a]a P [mm/11a]b (Sy =0.08) R/N []
[mm/ 11a]c

a [m/m]d

R2
[]e

CCPR
[]f

Time lag Sy calculated ()h Group


[weeks]g

88
95
182
246
320
529
563
576
663
695
702
724
907
909
936
1129
1162
1412
1450
1494
1514
1614
1924
1945
2407
2439
2564
2575
2719
4417

9,120
9,560
8,740
7,370
7,260
8,040
7,970
7,230
9,430
8,010
7,790
8,090
8,160
8,150
8,210
8,280
7,760
9,400
10,200
10,500
10,900
11,800
17,600
19,700
8,170
10,900
16,600
16,600
13,960
7,480

0.32
0.00
0.03
0.07
0.01
0.20
0.22
0.04
0.05
0.18
0.13
0.19
0.58
0.04
0.06
0.12
0.09
0.03
0.02
0.03
0.00
0.00
0.37
0.08
0.09
0.43
0.24
0.01
0.09
0.10

0.94
0.02
0.46
0.88
0.00
0.95
0.16
0.67
0.43
0.15
0.31
0.40
0.99
0.63
0.55
0.95
0.60
0.34
0.32
0.46
0.06
0.00
0.23
0.57
0.66
0.93
0.24
0.24
0.19
0.95

0.61
0.00
0.60
0.57
0.24
0.76
0.33
0.26
0.58
0.44
0.54
0.75
0.65
0.57
0.51
0.64
0.22
0.09
0.19
0.08
0.69
0.26
0.51
0.48
0.63
0.33
0.44
0.36
0.29
0.15

1
0
1
1
9
0
0
1
1
1
1
0
1
1
1
1
2
10
5
7
1
10
0
1
1
4
1
1
0
1

4,320
0.00
6,140
4,630
1,360
3,630
1,790
210
4,330
3,180
4,180
3,490
3,590
4,420
4,230
4,830
0
30
2,690
2
6,660
2,450
13,320
31,360
5,450
130
11,110
2,510
14,360
3,620

5,190
37,300
2,440
4,530
2,740
4,590
4,250
2,980
1,570
4,630
3,450
4,300
6,500
3,040
4,170
3,060
2,100
3,580
940
3,300
2,130
10,020
11,690
14,360
6,120
14,300
11,610
5,290
7,210
5,980

0.57
3.90
0.28
0.61
0.38
0.57
0.53
0.41
0.17
0.58
0.44
0.53
0.80
0.37
0.51
0.37
0.27
0.38
0.09
0.31
0.20
0.85
0.66
0.72
0.75
1.31
0.70
0.32
0.52
0.80

0.07
0.00
0.20
0.08
0.04
0.01
0.03
0.01
0.22
0.05
0.10
0.06
0.04
0.12
0.08
0.13
0.00
0.00
0.23
0.00
0.25
0.02
0.09
0.17
0.07
0.00
0.08
0.04
0.16
0.05

1
4
1
1
3
1
2
4
1
1
1
1
1
1
1
1
4
4
2
4
1
3
1
1
1
4
1
2
3
2

11-year sum of precipitation at the observation well locations (spatially interpolated by PROMET from station

b
data)

11-year sum of percolation, simulated by PROMET


sum of recharge, calculated using MDP with a constant Sy of 0.08
da: slope of the linear regression between R and P
eCorrelation coefcient of the bin-average point/groundwater-level relation (see Fig. 7 b, d, f)
fCross-correlation coefcients between R and P
gTime lag identied in the cross-correlation analysis
c11-year

Low correlation coefcients (< 0.4) between calculated


A discussion of the characteristics of each group is
recharge and percolation
provided in the following. Additional criteria used for a
qualitative assessment of the results, but not for
Group 3:
grouping,
are the recharge/precipitation ratio (R/N) and the RMSE
Reasonable simulated Sy values (0.01<Sy <0.25; see(root mean square error) of the bin-average point/groundthe preceding)
water-level relationship (see Fig. 7b, d, f). Results of the
Poorly dened groundwater level/groundwater-decline
recharge calculation (R) are considered reasonable if
relationships
the
Low correlation coefcients (< 0.4) between calculated
R/N ratio lies between 0.3 and 0.6 when using Sy =0.08,
recharge and percolation
as
this ratio is typical for the UDC region according to
previous studies (local reports and maps, not cited).
Group 4:
Deviations from the Sy =0.08 assumption are very likely,
Unreasonable simulated Sy values (<0.01, see the however, and this criterion is therefore a rather
qualitative
preceding) due to very little or no percolation
No distinct groundwater level/groundwater-decline one that cannot be used without taking other results into
account. The RMSE is a semi-qualitative indicator for the
relationships
linear groundwater level/groundwater-decline relationship
Variety of correlation coefcients
Hydrogeology Journal (2011) 19:14871502

DOI 10.1007/s10040-011-0772-8

1498

1497

level/groundwater-decline relationships, declines become


relationship can have various causes including strong
greater with higher corresponding groundwater levels.heterogeneities and anisotropies of the aquifer and
The
unsaturated zone, a short distance to surface waters,
slope of the regression of the selected maximum decline
and nearby pumping or injection wells, to name only a
values (parameter a in Table 2) is greater than zero. The
few. It should be noted that of the 3,700 wells
selected maximum decline points form an upper
available in the study area, roughly 25% show obvious
boundary within the plot of all groundwater level/
anthropogenic inuence which is not surprising in a
groundwater-decline points, as shown in Fig. 7b. The densely populated catchment where observation wells
resulting recharge time series mimic the shape of thewere more often built to monitor human actions than
percolation and precipitation time series (see Fig. 7a).to observe natural processes. The quality of results
The simulated Sy values range from 0.01 to 0.25,
could be improved by using preprocessing to eliminate
which is reasonable according to the assumptions laidthese outliers, but it is questionable if reliable results
out in Eq. 1; section Application, step 5 or literature could be obtained without a very detailed analysis of
the specic situation at each location. In general,
however, the recharge calculations for such wells need
further analysis.
Group 2
For wells in group 2, groundwater declines for each
bin are linearly related to groundwater levels, but the
resulting recharge time series show a weaker (or
Group 4
negative) correlation with percolation time series
The percolation values derived from PROMET for the wells
(correlation coefcients <0.4) than those characterizing
in group 4 are very low or close to zero. As is shown in
group 1. As shown in Fig. 7c, groundwater levels
Tables 1 and 2, three of the wells are located in urban
follow the precipitation (and percolation) inputs only areas
very generally. The groundwater-level time series are which are considered sealed in the parameterization
dominated by low-frequency behaviour and vary
utilized
weakly due to individual precipitation events. These by PROMET. The other wells in this group are located in
results indicate that the reaction of the water table tolowland areas close to rivers (wetlands). If the water table
climatic changes is strongly dampened. Possible rea- is
sons for such behavior could be a thick or lowclose to the land surface, PROMET assumes that capillary
permeability unsaturated zone. For example, deep rise and root-water uptake from groundwater occur,
aquifers may not display sharp increases in ground- resultwater levels because wetting fronts tend to be
ing in direct evapotranspiration from the saturated zone
dispersed over long distances (Healy and Cook
and,
2002). Surface waters, in particular managed reser- thus, low or negative percolation. Based on the
voirs, can also buffer water-table uctuations. Even if assumption
the calculated recharge and Sy values for wells in
that PROMET provides correct results for the soil water
group 2 fall within a reasonable range, these results balance and percolation equals recharge, very low
will remain more uncertain than those of group 1 untilsimulated
an additional, more detailed analysis of the specic Sy values are calculated according to step 5. However, the
situation at each location is carried out.
observations recorded for these wells show that (1)
In summary, results for group 1 are straightforward and
reasonable. They are in accordance with the percolation
The example from group 3 depicted in Fig. 7e and f results given by the distributed soil-water balance model
exhibits seasonal water-table behavior (i.e., shape) PROMET. For wells in group 2, some uncertainty
similar to that of group 2s example (Fig. 7c and d), regarding the validity of the results remains and should
be looked at more critically, nevertheless the results are
but the groundwater level/groundwater-decline relationreasonable. Wells in group 3 (10% of all wells) proved
ships are not as well dened in this case as they are in
unsuitable for the application
of the
because
groups
Fig.
7 Examples
1 and of
2. recharge
In this specic
time series
case
and(GLOWA
groundwater-level
well
(GWL)/groundwater-decline
relationships
forWTF
1990,method
as calculated
using
thethe declines for the lowest two bins are
they show behavior contrary to the assumptions upon
1641),
maximum
decline
procedure
for selected GLOWA
529 from
group method
1, (cd) well
4417 from
group 2,be
and
(ef) well
which
the WTF
is based.
It should
noted
that
exceptionally
high,
thus undercutting
the wells:
linear(ab)
rela-well
1614
the
selection
of
the
30
wells
under
consideration
excluded
tionship.
A
closer
analysis
shows
that
the
exceptional
from group 3. Left-hand panel: time series for weekly measurements of groundwater level (GWL), interpolated precipitation,
declines for this well are caused by exceptional dropsany wells with obvious anthropogenic inuences. It can,
therefore, be inferred that the exceptional behavior
of unknown origin in the water table in September and
observed
December
1990.
Only
a
detailed
analysis
of
the
water-level relationship is often well-dened, but because
Group 1 was actually detected using the developed
does
necessarily
imply 1
specic
wellnot
location
the nature of
such events
it
is usually
linear,and
corresponding
RMSEs
are low. approaches.
Results showThis,
that however,
around 60%
of not
all wells
are in group
that
this
exceptional
behavior
was
caused
by
anthropocan
reveal
whether
natural
phenomena
or
articial
This criterion should therefore be treated cautiously as
and are characterized by recharge time series that are
genic
inuences.
Wellscorrelated
in group 4with
require
intensivetime
siteprocesses
are responsible
this
behavior.
In general,
well. It is, however,
a goodfor
sign
that
most wells
assigned
positively
and linearly
percolation
specic
analysis.
a
weak
groundwater-level/groundwater-level
decline
to group 1 (see the following) satisfy both criteria.
series (correlation coefcients >0.4). For all groundwater
Group 3

Hydrogeology Journal (2011) 19:14871502


19: 14871502

DOI 10.1007/s10040-011-0772-8

1500

1499

water levels decline more slowly. The MDP developed difcult


in
but also interesting because the two methods
this study yields the highest amount of recharge in employ different scales. Each method focuses on different
periods
aspects of recharge, and considering the two in tandem
of generally declining groundwater levels. In that sense,
does
it give a more complete picture. This is an improveis more sensitive than the MRC approach by Heppner
ment over the uncritical acceptance and use of calculated
values which often occurs when only one method is
applied.

Conclusions
This paper describes existing procedures and presentsObjective: improve the conceptual model of WTF
methods
new
methods related to the water-table uctuation (WTF) The MRC approach by Heppner and Nimmo (2005)
evaluates the relationship between groundwater levels
method. It compares the results generated from
and measured groundwater declines taking all groundapplication
of these methods to results output from PROMET, a water declines into account, including those that do not
physically based soil-water balance model with high necessarily take place in the absence of recharge. This
can lead to an underestimation of groundwater
recharge. Three methods developed in this paper, the
1. To determine whether or not recharge results calculated
through the WTF Method and recharge results calcu-continuous decline procedure (CDP), the no precipitalated using PROMET could be combined to cross- tion/percolation procedure (NPP, and the maximum
decline procedure (MDP), suggest three different ways
validate and improve either of the methods.
Fig.
8 ab
Climate
data
(weekly), percolation
groundwater
(weekly) atgroundwater
observation well
576. Note
thatmost
percolation
of identifying
declines
that
likely (as
2. To
improve
the
conceptual
model onand
which
WTF levels
calculated by PROMET) occurs only if the fourth soil layer is saturated
to eld
capacity
and can
thus be signicantly
delayed
with
happen
in the
absence
of recharge.
Even though
the
approaches are based by eliminating the potential risk
regard to
enhancements
to
the
MRC
approach
by
Heppner
and
of recharge
underestimation
which results
precipitation
(1999).
a Recharge as calculated
with a from
xed Sthe
y value of 0.08 (median of all calculated Sy values) is shown. Recharge at
Nimmo (2005) seem small, they are signicant as they
consideration of all groundwater declines.
improve the conceptual model and underlying assumptions by not taking all declines into account; therefore,
The following paragraphs discuss the degree to which
Comparison
of MDP
other approaches
Comparison
of calculated
recharge time series
avoiding recharge
underestimation.
these objectives
weretomet.
Figure
9 shows
that
for the
1990, thebased
calculated
Four other methods were implemented as well, namely
Generally,
with
respect
toyear
the methods
on the
recharge
time series
using
the ve WTF
the
MRC approach
(here: determined
MDP) and the
bin-average
approach-based
in (1)room
total for
volume
of
CDP, the NPP, the MRC approach by Heppner and
technique, there methods
might bediffer
additional
improveObjective:
use and
WTFthe
approaches
for cross-validation
and (2)
sensitivity
to groundwater
level changes.
Nimmo (2005),
RISE approach.
A selection of recharge
ment by using
non-linear
relationships
and dynamic
of
other
methods
As
expected
frombin
the
aforementioned
theoretical
discusseven
wells
that fall into group 1 (previous section) has
methods
for the
assignment
(i.e. based
on decline
Based
on
hypothesis
percolation (PROMET)
the methods rather
applied
behave
differently
times of
been usedan
forinitial
further
analysis that
and comparison.
This sions,
value distributions
than
on xed
declineininteris
equal to recharge
long periods
of time,
percolation
fast
groundwater
rises and
and Nimmo
in times(2005)
when groundcomparison
is basedover
on simulated
Sy values
and
a visual
vals).
However, aslevel
Heppner
have
values can be used to calculate specic yield (Sy) using
pointed out already, the decision to use a particular
the
MRC type should be based on an examination of the
equation governing all WTF approaches (Eq. 1;
extracted data and knowledge of the site. Results
section Application, step 5). In cases where calculateddeviating
Sy
from normal should always be checked for
values lie outside a reasonable range (as determined non-recharge related inuences on groundwater levels
based
and inappropriate hydrogeological settings or well
on observed data and physical limits), it must be assumed
design. The methods developed and presented in this
that one of the methods (WTF or PROMET) is awed orstudy, in particular the comparison with the soil-water
the initial hypothesis is wrong. If this is the case, a more
balance method provide a good means for semidetailed analysis of the specic conditions of the obserautomatic quality checks in this respect.
vation well and model parameters is recommended.
Other cross-validation steps carried out in the study
but not explicitly discussed in this article include: (1)
checking whether percolation is happening as predicted
General conclusions
by PROMET in the case where no rise in water table The specic conclusions presented previously still
occurs, (2) checking whether rises in the water table leave several questions unanswered. It is not possible
occur during periods when PROMET does not predict to prove whether or not the enhancements made to
any percolation, (3) checking whether there are
existing methods do in fact lead to a more reliable or
signicant delays between percolation and groundwater
better estimation of recharge because, as Coes et al.
level responses or if groundwater-level rises occur
(2007) states, no single method can be termed best
before percolation events; (4) checking if negative
at estimating rechargesimilar statements can be
percolation rates (direct evaporation from groundwater,
found in Delin et al. (2007), Schilling (2009), and
e.g., in wetlands) correlate with groundwater level
Risser et al. (2009). Groundwater recharge, in particuctuations.
Fig. 9 Comparison of calculated recharge time series with percolation
andthe
precipitation
series
for wella 88
using uncertain
ve WTF
ular on
regional time
scale,
remains
highly
approach-

Hydrogeology Journal (2011) 19:


19:14871502
14871502

DOI 10.1007/s10040-011-0772-8

1501

subject. One has to face the fact that recharge is the Eagleson PS (1978) Climate, soil, and vegetation. Water Resour Res
04(5):722730
result of regionally scaled drivers (e.g., climate) and
Fetter CW (2001) Applied hydrogeology. Prentice-Hall, Upper
very local conditions at and below land surface (e.g., Saddle River, NJ
land use, soil, hydrogeology). The uniqueness of localHealy RW, Cook PG (2002) Using groundwater levels to estimate
conditions hinders the generalization of local calcula- recharge. Hydrogeol J 10:91109
Heppner CS, Nimmo JR (2005) A computer program for predicting
tions (WTF). On the other hand, regionally based
recharge with a master recession curve. US Geol Surv Sci Invest
calculations (e.g., those based on a catchment scale
Rep 20055172, 8 pp
using discharge time series) neglect local particularities
Heppner CS, Nimmo JR, Folmar GJ, Gburek WJ, Risser DW
and are thus of limited use in many practical
(2007) Multiple-methods investigation of recharge at a humidapplications (see related discussion in Barthel et al.
region fractured rock site, Pennsylvania, USA. Hydrogeol J
15:915927
2008). Distributed soil-water balance approaches
including climate, land use, and soil properties (such Hiscock KM (2005) Hydrogeology principles and practice. Blackas PROMET) provide locally scaled results but are not well, Oxford, UK
Johansson PO (1987) Estimation of groundwater recharge in sandy
usually validated using groundwater observations. The till with two different methods using groundwater level
reliability of recharge estimates can be improved
uctuations. J Hydrol 90:183198
through combination and cross-comparison of severalJohnson A (1967) Specic yield: compilation of specic yields for
various materials. US Geol Surv Water Suppl Pap 1662-D, 74
contrasting approaches.
pp
Ketchum N, Donovan J, Avery W (2000) Recharge characteristics
of a phreatic aquifer as determined by storage accumulation.
Acknowledgements GLOWA-Danube was funded by the BMBF
Hydrogeol J 8:579593
(German Federal Ministry of Education and Research). The authors
Lorenz DW, Delin GN (2007) Regional estimation of groundwater
would like to thank all governmental organizations, private
recharge in Minnesota using a multiple regression model.
companies and others who supported our work by providing data, Ground Water 45(2):374382
models, advice or additional funding. The authors would like to Ludwig R, Mauser W, Niemeyer S, Colgan A, Stolz R, Escherthank their colleagues from partner projects within GLOWA-Danube Vetter H, Kuhn M, Reichstein M, Tenhunen J, Kraus A,
for their cooperation over the last 10 years. Special thanks go to the
Ludwig M, Barth M, Hennicker R (2003) Web-based
research group of Prof. Dr. Wolfram Mauser, Department of
modelling of energy, water and matter uxes to support
Geography, LM-University Munich, Germany, who provided the
decision making in mesoscale catchments: the integrative
PROMET-results for the UDC project used in this study as well as perspective of Glowa-Danube. Phys Chem Earth 28:621
tools for the post-processing of these data.
634
Mauser W, Bach H (2009) PROMET: large scale distributed
hydrological modelling to study the impact of climate change
on the water ows of mountain watersheds. J Hydrol 376:362
References
377
Meinzer OE (1923) The occurrence of groundwater in the United
Barthel R (2006) Common problematic aspects of coupling hydroStates with a discussion of principles. US Geol Surv Water
logical models with groundwater ow models on the river
Suppl Pap 489, 321 pp
catchment scale. AdGeo 9:6371
Meinzer O, Stearns N (1929) A study of groundwater in the
Barthel R (2011) An indicator approach to assessing and predicting
Pomperaug Basin, Connecticut with special reference to
the quantitative state of groundwater bodies on the river basin
intake and discharge. US Geol Surv Water Suppl Pap
scale with a special focus on the impacts of climate change.
597:73146
Hydrogeol J 19(3):525546. doi:10.1007/s10040-010-0693-y
Moon SK, Woo NC, Lee KS (2004) Statistical analysis of
Barthel R, Rojanschi V, Wolf J, Braun J (2005) Large-scale water
hydrographs and water-table uctuation to estimate groundresources management within the framework of GLOWA-Danube,
water recharge. J Hydrol 292:198209
part A: the groundwater model. Phys Chem Earth 30(67):372
Nickel D, Barthel R, Braun J (2005) Large-scale water resources
382
management within the framework of GLOWA-Danube: the
Barthel R, Jagelke J, Gtzinger J, Gaiser T, Printz A (2008) Aspects
water supply model. Phys Chem Earth 30:383388
of choosing appropriate concepts for modelling groundwater
Risser DW, Gburek WJ, Folmar GJ (2009) Comparison of recharge
resources in regional integrated water resources management:
estimates at a small watershed in east-central Pennsylvania,
examples from the Neckar (Germany) and Oueme Catchment
USA. Hydrogeol J 17:287298
(Benin). Phys Chem Earth 33(12):92114. doi:10.1016/j.
Rushton KR, Redshaw SC (1979) Seepage and groundwater ow.
pce.2007.04.013
Wiley, Chichester, UK, 332 pp
Barthel R, Janisch S, Nickel D, Trifkovic A (2010) Using the
Rutledge AT (1998) Computer programs for describing the
multiactor-approach in GLOWA-Danube to simulate decisions
recession of groundwater discharge and for estimating mean
for the water supply sector under conditions of global climate
groundwater recharge and discharge from streamow
change. Water Resour Manage 24:239275. doi:10.1007/
records: update. US Geol Surv Water Resour Invest Rep
s11269-009-9445-y
98-4148
Coes A, Spruill T, Thomasson M (2007) Multiple-method estimaSanford W (2002) Recharge and groundwater models: an overview.
tion of recharge rates at diverse locations in the North Carolina
Hydrogeol J 10:110120
Coastal Plain, USA. Hydrogeol J 15:773788
Scanlon B, Healy R, Cook P (2002) Choosing appropriate
Crosbie RS, Binning P, Kalma JD (2005) A time series approach to
techniques for quantifying groundwater recharge. Hydrogeol J
inferring groundwater recharge using the water table uctuation
10:1839
method. Water Resour Res 41
Schilling K (2009) Investigating local variation in groundwater
De Vries J, Simmers I (2002) Groundwater recharge: an overview of
recharge along a topographic gradient, Walnut Creek, Iowa,
processes and challenges. Hydrogeol J 10:517
USA. Hydrogeol J 17:397407
Delin GN, Healy RW, Lorenz DL, Nimmo JR (2007) Comparison
Schirmer M (2008) Zuknftige Grundwasserforschung: Was sind
of local- to regional-scale estimates of groundwater recharge in
unsere Aufgaben? [Future groundwater research: What are our
Minnesota, USA. J Hydrol 334:231249
Hydrogeology Journal (2011) 19: 14871502

DOI 10.1007/s10040-011-0772-8

1502
Schwartz FW, Zhang H (2003) Fundamentals of ground water. Voss C (2005) The future of hydrogeology. Hydrogeol J 13:16
Wiley, New York
Wolf J, Barthel R, Braun J (2008) Modeling ground water ow in
Shah N, Ross M (2009) Variability in specic yield under shallow alluvial mountainous catchments on a watershed scale. Ground
water table conditions. J Hydrol Eng 14:12901298
Water 46:695705. doi:10.1111/j.1745-6584.2008.00456.x
Sophocleous MA (1991) Combining the soilwater balance and Zhang J (2009) Estimation of groundwater recharge from groundwater-level uctuation methods to estimate natural groundwater
water level time series in the Upper Danube Catchment. MSc
recharge: practical aspects. J Hydrol 124:229241
Thesis, Karlsruhe University of Applied Sciences, Germany

Hydrogeology Journal (2011) 19:14871502

DOI 10.1007/s10040-011-0772-8

Você também pode gostar