Journal of Hydrology 396 (2011) 233245

Contents lists available at ScienceDirect

Journal of Hydrology
journal homepage: www.elsevier.com/locate/jhydrol

Validation of a Vulnerability Estimator for Spring Protection Areas:

The VESPA index
Lorenzo Galleani a, Bartolomeo Vigna b, Cinzia Banzato b, Stefano Lo Russo b,
a
b

Department of Electronics (DELEN), Politecnico di Torino, C.so Duca degli Abruzzi 24, 10129 Torino, Italy
Department of Land, Environment and Geo-Engineering (DITAG), Politecnico di Torino, C.so Duca degli Abruzzi 24, 10129 Torino, Italy

a r t i c l e

i n f o

Article history:
Received 24 March 2010
Received in revised form 29 September 2010
Accepted 14 November 2010
This manuscript was handled by
Philippe Baveye, Editor-in-Chief
Keywords:
Groundwater protection zones
Springs
Vulnerability
VESPA index
Piemonte
Italy

s u m m a r y
The delineation of groundwater protection zones around springs requires the assessment of the spring
vulnerability, aquifer properties, and preferential inltration zones in the recharge area. We propose
the use of the Vulnerability Estimator for Spring Protection Areas (VESPA) index to quantify spring vulnerability. The VESPA index is based on a joint analysis of discharge, temperature, and electrical conductivity derived from hydrographs. The method requires 1 year of data measured on the spring, and
estimates the spring vulnerability for several hydrogeological contexts. No inltration data in the catchment area is required by the VESPA index. We validated the proposed method using experimental data
from 12 test springs, and found that the vulnerability estimated by the VESPA index corresponds to
the known vulnerability of the test springs. Four signicant test sites are presented in the paper.
2010 Elsevier B.V. All rights reserved.

1. Introduction
Although uncertain as a process, the delineation of protection
zones has increasingly become a fundamental component of
groundwater protection. Aquifers are recharged from precipitation
and surface waters that percolate through the land surface and
become part of the groundwater ow system. This water may
become contaminated as a result of land use practices (Powell
et al., 2003; Barry et al., 2009; Schijven et al., 2010). In identifying
the level of territorial protection required to preserve a spring from
polluting activities in the recharge area, the practical problem is
whether a signicant proportion of the freshly inltrated water
can quickly reach the spring (high spring vulnerability) or not
(low spring vulnerability) during or just after a recharge event.
Low spring vulnerability is assumed if the following conditions
are met (Pochon et al., 2008): (i) spring discharge varies only with
marked inertia and low amplitude due to recharge events, without
signicant relative variation of physicalchemical parameters; (ii)
no signicant turbidity is observed, even after intense recharge
events, and no sediment deposits accumulate in the spring pool;
and (iii) bacterial contamination is never detected. High spring vulnerability is concluded if any of the above conditions is not met.
Corresponding author. Tel.: +39 011 564 7648; fax: +39 011 564 7699.
E-mail address: stefano.lorusso@polito.it (S.L. Russo).
0022-1694/$ - see front matter 2010 Elsevier B.V. All rights reserved.
doi:10.1016/j.jhydrol.2010.11.012

Conditions (ii) and (iii) can be directly veried in a unique way

(i.e. water turbidity and/or bacterial contamination are clearly
detectable). However, condition (i) depends on a subjective evaluation of the hydrographs by the operator. Without evidence of spring
turbidity or contamination, the same hydrographs can be evaluated
in different ways by the operators, thus making the spring vulnerability estimation highly operator-dependent. This is particularly
true in intermediate situations, where typical hydrograph behaviors are not so clearly detectable by qualitative analysis.
To standardize spring vulnerability level assessment, we developed the Vulnerability Estimator for Spring Protection Areas
(VESPA) index, a vulnerability estimator based on the analysis of
spring hydrographs. The VESPA index uses the discharge (ow)
rate Q, groundwater temperature T, and electrical conductivity
EC, and requires 1 year of data derived from the spring monitoring.
Using the VESPA index, one can detect the spring vulnerability level or, alternatively, verify the vulnerability determined by means
of other hydrogeological investigation techniques. Consequently,
the appropriate procedure for identifying the spring protection
areas, which depend on the hydrogeological environment and local
regulatory framework, can be determined.
In the preparatory phase of this research the authors analyzed
12 springs located in the SW Alps in the Piedmont region (Italy)
to individuate and dene the VESPA index. The level of vulnerability has been qualitatively determined by means of conventional

234

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

hydrogeological analysis for each spring. The VESPA index was thus
calculated and compared with the level of vulnerability previously
detected. The results show that the VESPA index correctly identify
the vulnerability of the analyzed springs. Four signicant test sites
are presented in the paper. The protection areas were determined
according to local technical guidelines (Regione Piemonte, 2006).

R t0

Q trtdt
0
q

q q
Rt
R t0 2
Q tdt 00 r2 tdt
0
and u(q) is the Heaviside step function

uq

2. Methods
Analysis of the spring hydrograph responses with respect to the
inltrative events in the recharge area forms the basis for our
developed procedure to quantify spring vulnerability.

qP0
0; q < 0

1;

The parameter a is a scaling coefcient constrained by 0 6 a 6 1.

Since all terms in Eq. (2) are non-negative, c(q) is also non-negative.
The key element of the correlation factor is the correlation coefcient, which can vary in the interval

2.1. Spring hydrograph responses to the inltrative process

1 6 q 6 1
Hydrograph analysis is one of the most diffuse and effective
way to evaluate the properties of an aquifer supplying a spring
(Sanz Prez, 1997; Szilagyi et al., 1998; Zecharias and Brutsaert,
1998; Halford and Mayer, 2000; Wicks and Hoke, 2000; Pinault
et al., 2001; Mendoza et al., 2003; Fiorillo, 2009). A decay exponential function can be applied to data observed from karst springs, to
evaluate the volume and recession of groundwater reservoirs with
different permeabilities (Baedke and Krothe, 2001; Grasso and
Jeannin, 2002; Ford and Williams, 2007; Civita, 2008; Doeriger
et al., 2009; Birk and Hergarten, 2010). More comprehensive approaches, including consideration of the physicalchemical parameters (temperature, EC, isotopes, chemical elements, turbidity)
(Dreiss, 1989; Sauter, 1992; Grasso et al., 2003) and tracer tests
(Kss, 1998) have proven useful for improving the characterization
of karst hydrogeological systems. Spring hydrographs in non-karst
fractured rocks have received less attention, but the same methods
can be applied to the characterization and comparison of crystalline aquifers (Gentry and Burbey, 2004).
Springs with different aquifer types display different hydrographs (Barnes, 1939; Brutsaert and Nieber, 1977; Mangin, 1982;
Amit et al., 2002; Malvicini et al., 2005). The aquifer drainage system can be characterized by an impulse function that transforms
the input (e.g., rainfall or snowmelt) into spring hydrograph
responses in terms of discharge, temperature, and EC variations.
The impulse functional analysis can then be related to the drainage
effectiveness (i.e. network connectivity) (Plagnes and Bakalowicz,
2001; Vigna, 2007; Kresic and Stevanovic, 2009).

The joint analysis of water ow discharges, temperature, and EC

potentially offers a useful and replicable way to identify the spring
vulnerability. To determine the vulnerability index, 1 year of data
sampled by automatic sensors every 1 or 2 h and stored in an
opportune data logger was considered. This minimum time interval helps limit the possible errors associated with the loss of information provided by the main ood events during the hydrological
year (spring and autumn events). The VESPA index is dened as

V cqbc

where c(q) is the correlation factor, b is the temperature variability,

and c is the discharge factor.
2.2.1. Correlation factor
The correlation factor is dened by

cq uq auqjqj

2.2.2. Temperature variability factor

The temperature variability is dened as


2
T max  T min
1 C

where q is the correlation coefcient between discharge and conductivity, computed on the reference time interval t0 = 1 year (one
hydrologic year) as

where Tmax and Tmin refer to the maximum and minimum values,
respectively, of the temperature T on the reference time interval
t0 (explored data set: 1 year). Division by 1 C is performed to ensure that b is dimensionless. Since temperature stability over time
indicates a high aquifer residence time and low vulnerability, the
maximum temperature variation is a fundamental parameter for
estimating the spring vulnerability. Hence, we use its squared value
to enhance the corresponding weight in the vulnerability index V.
2.2.3. Discharge factor
The discharge factor measures the variability of the discharge
time series, as according to

Q max  Q min
Qm

where Qmax and Qmin are the maximum and minimum values,
respectively, of the discharge Q on the reference time interval t0,
and Qm is the average discharge given by

Qm

2.2. Spring vulnerability estimator (VESPA index)

1
t0

t0

Q tdt

2.3. Selection of the test sites

The purpose of the study is to evaluate the VESPA index and apply the described methodology to the identication of protection
zones. In the preliminary phase of this research, several hydrographs taken almost continuously over the previous 58 years in
various hydrogeological environments within the Western Alps
(12 mountain springs) were analyzed. These unpublished data
were acquired directly through automatic monitoring systems.
To validate the VESPA index, springs were chosen for analysis
based on the following criteria: (i) knowledge of the geological features and hydrogeological setting of the spring recharge area, as
obtained by adequate geological investigations (e.g., survey, tracer
tests, etc.); (ii) availability of a complete year of discharge,
temperature, and EC data, continually collected by automatic
sensors; (iii) knowledge of the actual spring vulnerability level,
as derived by historical analytical measures of bacteriological contamination and/or increased turbidity episodes; (iv) signicance of
the spring yield (minimum value >5 L/s); and (v) diversity of the
supplying aquifer among the test sites.

235

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

Fig. 1. Hypothetical examples of spring hydrographic responses to the inltrative process related to drainage effectiveness. (A) High effectiveness, prevailing replacement. (B)
Moderate effectiveness, prevailing piston effect. (C) Low effectiveness, prevailing homogenization.

3. Results and discussion

Q M  Q l
 100
Qa

3.1. Behavioral models of the drainage network effectiveness

Qualitative analysis of the hydrographs and observed correlations between the ow rate, temperature, and EC as a function of
inltration input revealed three broad behavioral categories (types
AC), based on the drainage network effectiveness. The proposed
classication is not directly related to the vulnerability assessment,
and only identies the type of response to the inltrative input.
In the highly effective drainage system (type A) during high
water levels (e.g., ood or snow melting period), most of the
freshly inltrated water reaches the spring very quickly, due to
the presence of open fracture systems, well-developed karst conduits, or highly permeable horizons, according to the local hydrogeological situation. Generally, these systems show a thin
saturated zone (with limited total spring storage volume) and a
high degree of permeability. The quick and strong discharge rate
growth after an inltration event is rapidly followed by a fast
depletion, due to the end of the inltrative process. Normal discharge conditions are recovered swiftly, within hours or a few days
(Fig. 1A). The annual discharge variability index m (in percent)
(Meinzer, 1923) is determined by

Table 1
Proposed VESPA index intervals for the identication
of the spring vulnerability level.
Vulnerability

VESPA index

Very high
High
Medium
Low

V P 10
1 6 V < 10
0:1 6 V < 1
0 6 V < 0:1

Table 2
Intervals spanned by the correlation coefcient and correlation factor for the three
basic types of spring described in Section 3.1.
Spring type and prevailing
phenomena

Correlation
coefcient (q)

Correlation
factorc(q)

Type A replacement
Type B piston
Type C homogenization

1 6 q 6 0:2
0:2 6 q 6 1
0:2 6 q 6 0:2

0:2 6 cq 6 1
0:1 6 cq 6 0:5
0 6 cq 6 0:2

where QM, Ql, and Qa are the maximum, minimum, and average
ow, respectively. The m is generally quite high and can reach values
of >100%.
Freshly inltrated water of low salinity tends to replace the
groundwater supplying the spring during the baseow. Therefore,
the water chemistry response is usually characterized by a fast and
intense reduction in mineralization, highlighted by decreased EC
values corresponding to the ood peaks. The behavior of the
groundwater temperature is relatively similar to that of the EC:
its intense variability is almost synchronous with the ood peaks,
and it recovers rapidly after the end of the inltrative processes.
The extent and geometry of the peak temperature detectable by
a hydrograph is a function of the difference between the groundwater temperature during undisturbed (baseow) conditions and
the temperature of the freshly inltrated water, which varies
seasonally.
In the moderately effective drainage system (type B), the spring
hydrodynamic response can display impulse behaviors (Fig. 1B).
Generally, these systems are characterized by fractured or slightly
karstied carbonate aquifers, a thick saturated zone, and a signicant total spring storage volume. The discharge variability indices
are generally lower than those in type A systems. Freshly inltrated water increases the hydraulic head in the saturated zone,
and induces a pressure increase in the saturated fractured portions
of the rock mass (fractures and/or karst conduits). This pressure
increase and the corresponding pushing effect tend to mobilize
the resident groundwater. The groundwater is near thermal equilibrium with the aquifer and is characterized by a higher salinity
than the freshly inltrated water. Therefore, a ow discharge and
increased EC and temperature are observed by monitoring the
spring (piston effect).
After the inltrative peak, the system is dominated by mixing
between the resident pre-event groundwater and the freshly inltrated water. The recovery times of the ood discharge rate, temperature, and EC values surveyed at the spring are delayed, and
have a relatively longer duration than those observed in type A systems. Generally, the peak temperature value decreases with increases in the elevation above sea level of the recharge area. This
phenomenon is due to the small difference between the inltration
water temperature (from rainfall or snowmelt) and the resident
groundwater temperature of the more elevated aquifers.

Table 3
Differentiation of allowed land uses in the groundwater protection zones according to the Italian and Piemonte region water regulations (Repubblica Italiana, 2006; Regione
Piemonte, 2006).
Zone

Allowed anthropogenic land uses

Total protection zone (TPZ)

Inner protection zone (IPZ)
Outer protection zone (OPZ)

None. This zone should be fully preserved, enclosed, and with limited access for authorized personnel only
Strongly limited. No excavation and subsurface work is allowed. Hazardous activities should be re-located if present
Limited. Only minor activities are allowed, and safeguard measures against pollution are necessary for new buildings

236

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

Fig. 2. Hydrogeological map of the Piemonte region (modied after Civita et al., 2004).

In a low effective drainage system (type C), the hydrodynamic

impulsional response to the inltrative processes is almost absent

(Fig. 1C). The discharge ow displays slow and modest uctuations

that are delayed up to several months relative to the main rainfall

237

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

The VESPA index is dened as the product of three factors, such

that each factor can increase or decrease the spring vulnerability.
These factors include the correlation between the discharge and
conductivity, discharge variability, and temperature variability. If
the temperature of a spring varies by 0.1 C in 1 year, then the corresponding variability factor b equals 0.01 and the spring vulnerability is strongly reduced by the temperature stability, as
intuitively expected. Calculation of the VESPA index is independent
of the relationship between recharge (precipitation, snowmelt,
stream loss in the recharge area) and discharge hydrographs. Therefore, the vulnerability assessment performed by the VESPA index
does not require recharge measures (which can be difcult to perform in mountainous areas), and only requires measurements collected at the spring tapping.
The VESPA approach can enable identication of the effectiveness of the spring drainage systems. Based on the experience gained
on-site, when disturbances due to snow melting or minor ood
events during the year are negligible and annual peaks are clearly
detectable on the discharge hydrographs, the effectiveness of the
basic spring drainage system types (Section 3.1) can be identied
by analyzing the value of the correlation coefcient q. Three types
of springs are identiable using q: replacement (type A), piston
(type B), and homogenization (type C) types. Type A systems are
identied by q < 0, since an increased discharge implies a decreased
EC. Eq. (2) shows that type A corresponds to c(q) = |q|. Type B springs
are identied by q > 0, since an increased discharge corresponds to

events. Chemical parameters (salinity) and the discharge water

temperature usually display a similar trend, with slow and minor
variations. Type C systems generally occur in crystalline and dolomitic intensely fracturated rock aquifers or porous media with an
abundance of ne matrix. These systems are usually characterized
by an extended saturated zone, with a very low groundwater ow
velocity due to limited permeability. Freshly inltrated water
moves slowly in the unsaturated and saturated zones, thereby
reaching equilibrium with the aquifer and resident pre-event circulating groundwater. Homogenization occurs due to the aquifer
characteristics, and the external output due to the inltrative process (low salinity and cold water) is strongly reduced.
3.2. Spring vulnerability and V index

50

Discharge [l/s]

0
0
15

EC [ s/cm]

Rainfall [mm/day ]

Table 1 displays a proposed classication scheme for spring vulnerability levels based on the VESPA index V. Such classication
arises from the experimental application of the VESPA method to
the 12 springs analyzed in the preparatory phase of this work. Four
of these test sites are presented in Section 3.4. The intervals
between the degrees of vulnerability were chosen based on the
four levels of vulnerability required by local regulations (Regione
Piemonte, 2006). If a smaller number of classes is required (i.e.
three), it may be appropriate to join the High and Very high levels
(Table 1), and thus consider V P 1 as the marker for a highly vulnerable spring.

400

100

150

200
t [days]

250

300

350

50

100

150

200
t [days]

250

300

350

50

100

150

200
t [days]

250

300

350

50

100

150

200
t [days]

250

300

350

10
5
0

350
0

Temperature [C]

50

11
10
9
0

Fig. 3. Monitoring data of the Ray test site. Automatic acquisition interval of 1 h.

Table 4
Summary of VESPA calculated parameters in the four test sites.
Test site
spring

Vulnerability
index V

q Lower
condence
interval
(95%)

q Upper
condence
interval
(95%)

0.11605

0.0953

0.1367

0.14482
0.47704
0.93917

0.1157
0.4931
0.9416

0.1737
0.4607
0.9366

Correlation
coefcient

q
Ray
Dragonera
Fuse
Balmetta

0.0024788
0.29237
14.2771
28.8469

Correlation
factorc(q)

Temperature
variability
factor b

0.058023

0.0841

0.07241
0.47704
0.93917

0.3364
3.7249
14.44

Discharge
factor c

0.50798
12.0027
8.0348
2.1271

Estimated
vulnerability

Aquifer
homogeneity

Effectiveness
of drainage
network type

Low

High

Medium
very high
Very high

Medium
Low
High

Type C
homogenization
Type B piston
Type A substitution
Type A substitution

238

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

Temperature [C]

EC [s/cm]

Discharge [l/s] Rainfall [mm/day]

Fig. 4. Delineation of groundwater protection zones for the Ray spring with the distance method.

50
0
0

50

100

150

200
t [days]

250

300

350

0
0

50

100

150

200
t [days]

250

300

350

240
220
200
180
160
0

50

100

150

200
t [days]

250

300

350

50

100

150

200
t [days]

250

300

350

3000
2000
1000

8
7
6
0

Fig. 5. Monitoring data of the Dragonera test site. Automatic acquisition interval of 2 h.

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

239

Fig. 6. Delineation of groundwater protection zones with the distance method for the Dragonera spring. Differentiation of the IPZ and OPZ is opportune.

an increased EC. Eq. (2) indicates that type B corresponds to

c(q) = a|q| with a < 1, since type A springs are generally more vulnerable than type B springs. For the same absolute value of the correlation coefcient q, the correlation factor c(q) is smaller for type B
than for type A. The experimental results veried that a = 0.5 is a
reasonable choice to individuate type B. Type C springs are identi-

ed by |q| 0, since the discharge and EC generally show independent and moderate variations with time. Eq. (2) indicates that this
type corresponds to c(q) 0, and the experimental results indicated
that jqj 6 0:2 is a suitable interval for a homogenization-type spring.
Table 2 summarizes the values of c(q) for the three basic spring
types. While the correlation coefcient intervals are disjointed, the

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

Temperature [C]

EC [s/cm]

Discharge [l/s] Rainfall [mm/day]

240

100
0
0

50

100

150

200
t [days]

250

300

350

0
0

50

100

150

200
t [days]

250

300

350

240
220
200
180
160
0

50

100

150

200
t [days]

250

300

350

50

100

150

200
t [days]

250

300

350

1000
500

6
5
4
0

Fig. 7. Monitoring data of the Fuse test site. Automatic acquisition interval of 1 h.

correlation factor c(q) intervals overlap. The reason for such overlap is that two springs belonging to different drainage effectiveness
categories can have the same degree of vulnerability.
3.3. Delineation the spring protection area and vulnerability
To delineate the spring protection areas, additional hydrogeological investigation is needed after numerical assessment of the
vulnerability. Particular attention should be paid to the aquifer
homogeneity and to the presence of preferential inltration zones
in the catchment area. For homogeneous or slightly heterogeneous
aquifers, the calculated (isochrone) or xed distance methods are
convenient. Vulnerability mapping methods are suitable for delineating protection areas for heterogeneous media that supply highly
vulnerable springs, and when preferential inltration zones are detected in the catchment area (Adams and Foster, 1992; DoELG/EPA/
GSI, 1999; Doeriger et al., 1999; SAEFL, 2000; Daly et al., 2002;
Zwahlen, 2004; Pochon et al., 2008). The technical procedures for
individuating a spring protection area vary depending on the local
regulatory framework. Following the regulations of the Piemonte
region (Regione Piemonte, 2006), the spring protection area can
be divided into two sub-areas, the inner (IPZ) and outer (OPZ) protection zones, which are characterized by their different allowed
land uses (see Table 3). A total protection zone (TPZ) around the
spring tapping is always essential.
3.4. Test sites
3.4.1. Hydrogeological characteristics of the Piemonte region
The Piemonte region is characterized by the arcuate orogenic
belt of the Western Alps (crystalline and carbonate rocks) on its
west side, morphologically connected by extensive alluvial and
morainic fans to the continental plain area in the central and eastern parts (see Fig. 2). The alpine range continues NE with the central segment of the chain, and SE with the Ligurian Apennines
(Debelmas, 1986; Cadoppi et al., 2007). The mountain area, which
is carved by valleys transverse to the direction of the main struc-

tures and convergent towards the barycentre region, displays

increasing relief energy from SW to NE. In the southern sector of
the Piemonte, the transition from the plain to the Apennines is
gradual, characterized by the intermediate presence of hilly terrigenous sectors (Monferrato and Langhe) belonging to the Piedmontese Tertiary basin. Crystalline and terrigenous rocks show low to
very low primary permeability, with no considerable aquifer. Some
signicant water circulation is found only in the main fractured
zones along the tectonic discontinuities, and in carbonates (karstic
circulation). Most springs in the mountain sector are generally supplied by detritus aquifers covering the crystalline bedrock. These
diffuse, high quality water resources typically provide moderate
to low yields of 130 L/s, and are employed for human consumption in most mountain villages. Signicant carbonate structures
and correlated karstic springs are diffuse only in the southern part
of the region (Cuneese), and in a limited portion of the northern alpine chain.

3.4.2. Test site: Ray spring

The Ray test site is situated in the Lurisia valley (Roccaforte
Mondov municipality) at 620 m amsl (geographical coordinates
44180 540 N, 7430 0200 E, see Fig. 2 for location). The supplying aquifer
consists of intensely fractured carbonate without any developing
karst. The bottom of the carbonate dolomitic aquifer (Trias) is limited by a crystalline basement constituted by quartzites (Permian
Trias). The carbonate aquifer is laterally in tectonic contact with the
crystalline basement through a series of sub-vertical normal faults
mainly oriented NS and NWSE. The saturated zone is quite extended, while the unsaturated zone is almost absent. In the overall
catchment area, an extensive protective cover constituted by residual clays is present above the carbonate unconned aquifer.
Land use in the catchment area is predominantly forest. Locally,
there are small areas of grassland used for pasture. Small farms and
minor roads are also present. Water quality has historically been
good over time. No bacteriological contamination episodes or
turbidity increase has been shown, and no signicant variation in

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

241

Fig. 8. Delineation of groundwater protection zones for the Fuse spring. The protection zone coincides with the overall catchment area. IPZ land use restrictions are suitable.

discharge, temperature, or EC has been observed during heavy

rainfall. These observations suggest low spring vulnerability.
The monitoring data set starts on June 4th, 2007. The data set
revealed a yearly maximum temperature variation of <0.3 C (aver-

age 10.4 C), EC variation of 15 lS (average 366 lS) (Fig. 3),
and average discharge of 10 L/s with limited and slow variations.
The VESPA parameters for the Ray spring are reported in Table 4.
The vulnerability index suggests a low vulnerability for the spring,

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

Temperature [C]

EC [s/cm]

Discharge [l/s] Rainfall [mm/day]

242

100
50
0
0

50

100

150

200
t [days]

250

300

350

0
0

50

100

150

200
t [days]

250

300

350

120
100
80
60
40
0

50

100

150

200
t [days]

250

300

350

50

100

150

200
t [days]

250

300

350

40
20

8
6
4
0

Fig. 9. Monitoring data of the Balmetta test site. Automatic acquisition interval of 1 h.

while the correlation factor identies a prevailing homogenization

phenomenon (type C system), consistent with the qualitative
hydrogeological features detected during eld investigations. A
suitable proposal for delineating the protection areas is shown in
Fig. 4. Due to the lithological features and the intense, uniform
fracture distribution, the carbonate aquifer can be considered
homogeneous overall in the catchment area. A xed distance approach to delineate spring protection areas is suitable (Regione
Piemonte, 2006). Due to the low degree of vulnerability, to avoid
excessive land use restrictions the protection area will be dened
by the minimum proposed by local regulations (200 m upstream
from the spring tapping). No differentiation in the protection area
is required; therefore, the OPZ coincides with the IPZ.
3.4.3. Test site: Dragonera spring
The Dragonera test site is situated in the Roaschia valley (Roaschia municipality) at 840 m amsl (geographical coordinates
44150 590 N, 7270 1600 E, see Fig. 2 for location). The supplying carbonate aquifer (TriasCretaceous) is characterized by an intense
fracture network and limited karst. It is stratigraphically overlaid
by a sequence of carbonate sandstones (Eocene) that highlights a
strongly reduced hydraulic permeability. Both the saturated and
unsaturated zones are extended. In the catchment area, extensive
outcrops of carbonate rocks are present, with no protective soil
cover. In contrast, soil covers are found over almost all of the detrital quaternary deposits. The area is mostly covered by pasture used
for summer grazing. There are no anthropic settlements (farms),
and the viability is only represented by tracks and paths. The
spring discharge responses to inltration events are usually quick
during the increment period to the ood peak, while the discharge
occurs for a longer time. However, the uctuations in the spring
discharge rate during the year are signicant.
The monitoring data set starts on April 29th, 2007. The data
highlights a yearly maximum temperature variation of 0.4 C
(average 7.2 C) and EC variation of 30 lS (average 210 lS)
(Fig. 5). Normally, increases in the discharge are quickly followed
by corresponding increases in the EC and temperature, thus highlighting a prevailing piston phenomenon. However, melting snow

limits the evidence of this phenomenon on the hydrographs. The

VESPA parameters for the Dragonera spring are reported in Table
4. The vulnerability index conrms a medium vulnerability for
the spring, while the correlation factor is in the transition region
between a piston process (type B system) and a homogenization
process (type C system), as expected from a qualitative analysis
of the hydrographs. This is probably due to the mixing effect and
the consequent noise induced by snowmelt, which masks the piston behavior of the hydrographs.
Considering the relative homogeneity of the aquifer supplying
the spring and the medium level of vulnerability, a xed distance
approach was used to identify the protection zones, according to
the indications provided by local regulations Regione Piemonte,
2006. The protection zone can be conveniently subdivided into
an IPZ and OPZ, to avoid excessive land use restriction in the upper
part far from the withdrawal point (Fig. 6). Minimum distances of
200 m and 400 m from the spring tapping for the outer boundary
of the IPZ and the OPZ, respectively, can be considered suitable.
3.4.4. Test site: Fuse spring
The Fuse test site is situated in the Tanaro valley (Ormea municipality) at 1475 m amsl (geographical coordinates 44090 0100 N,
7450 050 0 E, see Fig. 2 for location). The supplying carbonate aquifer
(TriasCretaceous) is intensely fractured with well-developed
karst. The crystalline basement is comprised of quartzites and porphyroids (PermianTrias). The spring is located on the sub-horizontal contact between the carbonate aquifer and the impermeable
basement. The unsaturated zone is quite extended, while the saturated zone is strongly limited. The whole catchment area is characterized by extensive outcrops of carbonate rocks and a complete
absence of soil and vegetation. Signicant uctuations in spring discharge, water temperature, and EC, as well as indications of bacteriological contamination (even in winter) have been observed.
Consequently, the spring appears to be highly vulnerable to
contamination.
The monitoring data set starts on March 7th, 2004, and highlights a yearly maximum temperature variation of 2.5 C (average
5.2 C) and EC variation of 130 lS (average 200 lS) (Fig. 7).

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

243

Fig. 10. Delineation of groundwater protection zones for the Balmetta spring. The protection zone coincides with the overall catchment area. IPZ land use restrictions are
suitable.

The yield is strongly variable, with an average discharge rate of

130 L/s. Various structural discontinuities at the groundwater
catchment and outcrop levels have been observed. Geomorphological depressions constitute potential locations of concentrated
inltration. Three uoroscein tracer tests (see Fig. 8 for location
of injection points) have shown highly variable ow velocities for
different injection points in the spring catchment (380800 m/
day). These data indicate the presence of rapid ow along interconnected networks of highly permeable joints (karstic conduits and
fractures). Rapid connections are possible between the springs
and surfaces that may be distributed across the whole spring
catchment area. Consequently, the groundwater residence time
does not increase with the distance from the spring in a uniform
way, and assimilation of the karstied aquifer to a continuous
medium is inappropriate.
Under these conditions, only the use of a multi-parameter
groundwater vulnerability mapping method over the whole catchment area enables the effective delineation of protection zones, by

considering the large degree of heterogeneity within the aquifer.

However, with reference to the delineation of protection zones at
the Fuse test site, the presence of a preferential inltration zone
(doline, sinkhole, interconnected fractures) nearly in the overall recharge zone led us to consider the entire recharge zone as a protection area, without differentiation (Fig. 8). Land use permissions
must be stringent, and thus the application of IPZ constraints to
the overall catchment area appears appropriate.
The VESPA parameters for the Fuse site are reported in Table 4.
The VESPA index conrms a very high vulnerability for the spring,
while the correlation factor indicates a substitution prevailing phenomenon (type A system).
3.4.5. Test site: Balmetta spring
The Balmetta test site is situated in the Ellero valley (Roccaforte
Mondov municipality) at 960 m amsl (geographical coordinates
44150 0600 N, 7430 000 0 E, see Fig. 2 for location). The supplying
aquifer consists of a thin coarse quaternary detritus with high

244

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

permeability, without a signicant protective soil cover. The aquifer

overlays a crystalline substrate constituted by impermeable
porphyroids (PermianTrias) that outcrop extensively in the catchment area. The saturated and unsaturated zones are limited in
width. The spring catchment area is characterized by a lack of vegetative cover. Only in the upper part are summer grazing pastures
are present. Signicant uctuations in spring discharge, water
temperature, and EC, as well as indications of bacteriological contamination have been observed, even in winter. Flood peaks correspond to quick and signicant depletions of temperature and EC,
highlighting a prevailing replacement phenomenon, operated by
the freshly inltrated water in the aquifer. Consequently, the Balmetta spring appears to be highly vulnerable to contamination.
The monitoring data set starts on June 8th, 2006, and highlights
a yearly maximum temperature variation of 4 C (average 6.8 C)
and EC variation of 80 lS (average 90 lS) (Fig. 9). The yield is
strongly variable, with an average discharge rate close to 15 L/s.
According to local regulations (Regione Piemonte, 2006), due to
the high spring vulnerability, aquifer homogeneity (coarse detritus), and lack of effective protective soil cover in the overall catchment area, the entire recharge zone is identied as the protection
area, without differentiation (Fig. 10). Land use permissions are
consequently stringent, and thus the application of the IPZ appears
appropriate.
The VESPA parameters for the Balmetta site are reported in
Table 4. The VESPA index conrms a very high vulnerability for
the spring, while the correlation factor conrms a replacement
prevailing phenomena (type A system).
3.5. Critical aspects of the VESPA index
Two physical situations can negatively inuence the VESPA index. First, a spring can have a rst phase of piston ow, followed by
a replacement generated by freshly inltrated water. If the replacement phenomenon prevails, the correlation coefcient between
discharge and conductivity could approach the limit value of
q = 1. Consequently, the estimated vulnerability V could be higher than its true value. Second, due to snow melting events, a slow
trend could be present in the measured time series. This trend
could inuence the correlation coefcient in an unpredictable
way. A possible improvement of the VESPA index is the detection
and removal of such slow trends, without affecting method performance. Both of these scenarios can affect the correlation factor
c(q), but have a minor effect on the variability of the temperature
and discharge. Evaluation of the VESPA factor from three coefcients improves method robustness.
4. Conclusions
Without a suitable method for identifying spring protection
zones, it is vital that the operator subjectivity be limited in spring
vulnerability estimations. Analysis of the linked behavior of the
discharge, temperature, and EC using hydrographs seems to provide a good quantitative estimate of the spring vulnerability. These
parameters are also useful for identifying the drainage network
effectiveness and the main phenomena occurring in the aquifer
after the ood peak (prevailing replacement, piston, or homogenization). Vulnerability assessment can be used to dene the groundwater protection zones, according to the hydrogeological situation
and local regulatory framework.
The goal of the VESPA index is to provide a numerical assessment of the spring vulnerability. Subsequently, the protection area
can be dened by considering the hydrogeological setting as well
as the local regulations. However, we observe that a correct delineation of the spring protection areas must always rely on intensive

geological and hydrogeological eld investigations of the recharge

area, mainly aimed at identifying the preferential inltration areas,
if present. We have carried out such investigation in the presented
test sites.
Application of the VESPA method seemed straightforward for
several reasons. First, the VESPA index allows estimation of the level of vulnerability in a wide range of hydrogeological environments. Second, the spring vulnerability evaluation is based solely
on the analysis of the spring monitoring data, and does not involve
any other parameter. Inltration data of the catchment area connected to rainfall events, snowmelt, or stream loss, which are difcult to evaluate, are not required to compute the V value.
Finally, the VESPA method provides reproducible results with a
moderate nancial effort.
Recently, the Piemonte regional Water Protection Plan imposed
the development of extensive automatic systems for monitoring
the discharge, temperature, and EC of springs supplying human
communities (Regione Piemonte, 2007). This provided several data
sets for testing and improving the methodology. One year of data
seems to be the minimum time interval necessary for a reasonable
estimate of the spring vulnerability using the VESPA index. We expect that our ongoing research will provide signicant input for
further renements.
Acknowledgements
The authors wish to thank the environmental authority of the
Piemonte region for its assistance during the eld survey and for
granting publication of the data, as well as the anonymous reviewers for their useful suggestions and comments. This study was
partially supported by the Piemonte region government.
References
Adams, B., Foster, S., 1992. Land-surface zoning for groundwater protection. J. Inst.
Water Environ. Manage. 6, 312320.
Amit, H., Lyakhovsky, V., Katz, A., Starinsky, A., Burg, A., 2002. Interpretation of
spring recession curves. Ground Water 40 (5), 543551. doi:10.1111/j.17456584.2002.tb02539.x.
Baedke, S.J., Krothe, N.C., 2001. Derivation of effective hydraulic parameters of a
karst aquifer from discharge hydrograph analysis. Water Resour. Res. 37 (1),
1319. doi:10.1029/2000WR900247.
Barnes, B.S., 1939. The structure of discharge recession curves. Trans. Am. Geophys.
Union 20, 721725.
Barry, F., Ophori, D., Hoffman, J., Canace, R., 2009. Groundwater ow and capture
zone analysis of the Central Passaic River Basin, New Jersey. Environ. Geol. 56,
15931603. doi:10.1007/s00254-008-1257-5.
Birk, S., Hergarten, S., 2010. Early recession behaviour of spring hydrographs. J.
Hydrol.. doi:10.1016/j.jhydrol.2010.03.026.
Brutsaert, W., Nieber, J.L., 1977. Regionalized drought ow hydrographs from a
mature glaciated plateau. Water Resour. Res. 13 (3), 637643.
Cadoppi, P., Giardino, M., Perrone, G., Tallone, S., 2007. Litho-structural control,
morphotectonics, and deep-seated gravitational deformations in the evolution
of Alpine relief: a case study in the lower Susa Valley (ItalianWestern Alps).
Quatern. Int. 171172, 143159.
Civita, M., 2008. An improved method for delineating source protection zones for
karst springs based on analysis of recession curve data. Hydrol. J. 16, 855869.
doi:10.1007/s10040-008-0283-4.
Civita, M., Lo Russo, S., Vigna, B., 2004. Hydrogeological sketch map of Piemonte
(NW Italy) 1:250.000. In: Map Presented at the 32nd International Geological
Congress (32IGC), Florence, Italy, August 2128.
Daly, D., Dassargues, A., Drew, D., Dunne, S., Goldscheider, N., Neale, S., Popescu, C.,
Zwahlen, F., 2002. Main concepts of the European approach for (karst)
groundwater vulnerability assessment and mapping. Hydrol. J. 10 (2), 340345.
doi:10.1007/s10040-001-0185-1.
Debelmas, J., 1986. Intracontinental subduction and mountain uplift: the example
of Western Alps. Geologie Alpine 62, 110.
DoELG/EPA/GSI, 1999. Groundwater Protection Schemes. Department of the
Environment and Local Government, EPA, Geol. Surv. of Ireland, Dublin, 24 pp.
Doeriger, N., Jeannin, P.Y., Zwahlen, F., 1999. Water vulnerability assessment in
karst environments: a new method of dening protection areas using multiattribute approach and GIS tool (EPIK method). Environ. Geol. 39 (2), 165172.
doi:10.1007/s002540050446.
Doeriger, N., Fleury, P., Ladouche, B., 2009. Inverse modeling approach to allogenic
karst system characterization. Ground Water 47 (3), 414426. doi:10.1111/
j.1745-6584.2008.00517.x.

L. Galleani et al. / Journal of Hydrology 396 (2011) 233245

Dreiss, S.J., 1989. Regional-scale transport in a karst aquifer, part I: component
separation of springow hydrographs. Water Resour. Res. 25 (1), 117125.
Fiorillo, F., 2009. Spring hydrographs as indicators of droughts in a karst
environment. J. Hydrol. 373, 290301. doi:10.1016/j.jhydrol.2009.04.034.
Ford, D.C., Williams, P., 2007. Karst Hydrogeology and Geomorphology. Wiley, New
York. 576 pp.
Gentry, W.M., Burbey, T.J., 2004. Characterization of groundwater ow from spring
discharge in a crystalline rock environment. J. Am. Water Resour. Assoc. 40 (5),
12051217. doi:10.1111/j.1752-1688.2004.tb01580.x.
Grasso, D.A., Jeannin, P.Y., 2002. A global experimental system approach of karst
springs hydrographs and chemographs. Ground Water 40 (6), 608618.
doi:10.1111/j.1745-6584.2002.tb02547.x.
Grasso, D.A., Jeannin, P.Y., Zwahlen, F., 2003. A deterministic approach to the
coupled analysis of karst springs hydrographs and chemographs. J. Hydrol. 271,
6576.
Halford, K.J., Mayer, G.C., 2000. Problems associated with estimating ground water
discharge and recharge from stream-discharge records. Ground Water 38 (3),
331342. doi:10.1111/j.1745-6584.2000.tb00218.x.
Kss, W., 1998. Tracing Technique in Geohydrology. Balkema, Rotterdam, The
Netherlands. 600 pp.
Kresic, N., Stevanovic, Z. (Eds.), 2009. Groundwater Hydrology of Springs. Elsevier
Inc., p. 565. ISBN: 978-1-85617-502-9.
Malvicini, C.F., Tammo, S.S., Walter, M.T., Parlange, J., Walter, M.F., 2005. Evaluation
of spring ow in the uplands of Matalom, Leyte, Philippines. Adv. Water Resour.
28, 10831090. doi:10.1016/j.advwatres.2004.12.006.
Mangin, A., 1982- Lapproche systmique du karst, consequences conceptuelles
mthodologiques [Systemic approach of the karst, conceptual and
methodological consequences]. In: Proc. Reun. Monogr. sobre el karst, Larra,
France, November 1982, pp. 141157.
Meinzer, O.E., 1923. Outline of Ground-water Hydrology. US Geol. Survey Watersupply Paper 494, Washington DC.
Mendoza, G., Steenhuis, T.S., Walter, M.T., Parlange, J.-Y., 2003. Estimating basinwide hydraulic parameters of a semi-arid mountainous watershed by
recession-ow analysis. J. Hydrol. 279, 5769.
Pinault, J.L., Plagnes, V., Aquilina, L., Bakalowicz, M., 2001. Inverse modelling of the
hydrological and the hydrochemical behaviour of hydrosystem: characterization
of karst system functioning. Water Resour. Res. 37 (8), 21912204.
Plagnes V., Bakalowicz M., 2001. May it propose a unique interpretation for karstic
spring chemographs? In: Proceedings of the 7th Conference on Limestone
Hydrology and Fissured Media, Besanon, France (2022 September 2001), pp.
293298.
Pochon, A., Tripet, J., Kozel, R., Meylan, B., Sinreich, M., Zwahlen, F., 2008.
Groundwater protection in fractured media: a vulnerability-based approach

245

for delineating protection zones in Switzerland. Hydrol. J. 16, 12671281.

doi:10.1007/s10040-008-0323-0.
Powell, K.L., Taylor, R.G., Cronin, A.A., Barrett, M.H., Pedley, S., Sellwood, J.,
Trowsdale, S.A., Lerner, D.N., 2003. Microbial contamination of two urban
sandstone aquifers in the UK. Water Res. 37 (2), 339352.
Regione Piemonte, 2006. Regolamento Regionale 11 dicembre 2006 n. 15/R recante:
Disciplina delle aree di salvaguardia delle acque destinate al consumo umano
(legge regionale 29 dicembre 2000, n. 61). Bollettino Ufciale della Regione
Piemonte n. 50 del 14 Dicembre 2006 Supplemento Ordinario No. 1. <http://
www.regione.piemonte.it/acqua/cons_umano.htm> (in Italia).
Regione Piemonte, 2007. Piano di Tutela Delle Acque. D.C.R. n. 117-10731, Turin, Italy.
<http://www.regione.piemonte.it/acqua/pianoditutela/pianoditutela.htm>.
Repubblica Italiana, 2006. Decreto Legislativo 3 aprile 2006, n. 152 Norme in
materia ambientale. Gazzetta Ufciale n. 88 del 14 aprile 2006 Supplemento
Ordinario No. 96. <http://www.camera.it/parlam/leggi/deleghe/06152dl.htm>
(in Italian).
SAEFL, 2000. Groundwater Vulnerability Mapping in Karstic Regions (EPIK).
Practical Guide. Environment in Practice. SAEFL, Berne, Switzerland, 56 pp.
Sanz Prez, E., 1997. Estimation of basin-wide recharge rates using spring ow,
precipitation, and temperature data. Ground Water 35 (6), 10581065.
doi:10.1111/j.1745-6584.1997.tb00178.x.
Sauter M., 1992. Quantication and Forecasting of Regional Groundwater Flow and
Transport in Karst Aquifers (Gallusquelle, Malm, SW Germany). PhD Thesis,
Univ. of Tbingen, Germany, 149 pp.
Schijven, J.F., Hassanizadeh, S.M., de Roda Husman, A.M., 2010. Vulnerability of
unconned aquifers to virus contamination. Water Res. 44, 11701181.
doi:10.1016/j.watres.2010.01.002.
Szilagyi, J., Parlange, M.B., Albertson, J.D., 1998. Recession ow analysis for aquifer
parameter determination. Water Resour. Res. 34 (7), 18511857.
Vigna, B., 2007. Classication and operation of aquifer systems in carbonate rocks.
Memorie dellIstituto Italiano di Speleologia XIX, 2126. ISBN: 978-88-8989703-4 (in Italian).
Wicks, C.M., Hoke, J.A., 2000. Prediction of the quality and quantity of maramec
spring water. Ground Water 38 (2), 218225. doi:10.1111/j.17456584.2000.tb00333.x.
Zecharias, Y.B., Brutsaert, W., 1998. Recession characteristics of groundwater
outow and base-ow from mountainous watersheds. Water Resour. Res. 24,
16511658.
Zwahlen, F. (Ed.), 2004. Vulnerability and Risk Mapping for the Protection of
Carbonate (karst) Aquifers, COST Action 620, Final Report. European
Commission Directorate-general XII, Science, Research and Development,
Report EUR 20912, COST Action 620, Luxembourg, Belgium.