GROUNDWATER PROTECTION IN KARST ENVIRONMENT

GROUNDWATER PROTECTION IN KARST ENVIRONMENT

1TULIPANO LUIGI,

2FIDELIBUS M. DOLORES,

1SAPPA GIUSEPPE

1 Sapienza University, Dept. Hydraulics Transportation and Roads, Rome (Italy) 2 Bari Technical University, Geotechnical and Geoenvironmental Eng. Dept., Bari (Italy)

Key-words: karst, groundwater, pollution, natural tracers, nitrates

Abstract Protection of karstic groundwater quality can be successful only if it is based on reliable reconstructions of conceptual models of the karstic aquifers, which are normally of high

complexity: this requires the integration of classical hydrogeological information with that obtainable by using natural tracers, according to multi-tracing methods. The definition of the conceptual model includes elements as the identification of recharge areas, their connection

with discharge areas, the sequence of physical – chemical processes acting into the aquifer and the transport mechanisms. With reference to the identification of recharge areas, two case-

studies are illustrated. The first deals with a continental karstic aquifer (Monti Simbruini, Central Italy) discharging through springs used for drinking purposes: it has been studied by using the stable deuterium and oxygen-18 isotopes according to the “mass-center” method

coupled with the “inverse hydrogeological budget” method. The second regards a platform karstic aquifer (Murgia, Southern Italy), discharging through coastal brackish springs, which is exploited by wells for both drinking and agricultural purposes: for recognizing some elements of the conceptual model a multi-tracing approach has been adopted, which uses the cross-

verification of information coming from the interpretation of isotopic, chemical and physical tracers. Moreover, with the aim of outlining the factors that control the pollutant transport in karstic aquifers, two case-studies related to Murgia aquifer illustrate a first method for defining

the hazard due to direct injection of effluents from treatment plants and a second approach for defining, through monitoring data, the transport mechanisms of pollutant released at land

surface.

1. Introduction

It is well-known that natural karst groundwaters, representing in some regions the unique accessible water resource, have the best quality for human consumption. Especially in Mediterranean countries with high demographic pressure, karstic groundwater quality is threatened by natural and human factors: the constant increase of agricultural, civil and industrial activities involves resource over-exploitation and increase of pollution loads on

soils and belowground. Unluckily, karst aquifers are very vulnerable to pollution with respect to other types of aquifers, due to their peculiar structure and functioning (EC DG XII 1995;

Tulipano and Fidelibus 1995a; Drew and Hötzl 1999; Zwahlen 2003; Tulipano et al 2005). The main tool used in the practice of safeguard and protection of karstic water resources is the assessment of their intrinsic vulnerability, based on the origin-pathway-target model: the origin

is normally the land surface, the pathway is the flow-path through the protective cover and the target is the groundwater surface. When the protection of the sources is concerned, the model includes the path from the groundwater surface to the point to be protected as well. In the last decades many efforts were made to develop specific vulnerability maps for karstic

aquifers. COST 620 Final Report (Zwahlen 2003) illustrates the results of said efforts. However, the same Report, at the same time as asserts that vulnerability maps are “a vital tool with which to protect groundwater”, says as well that “they remain a simplification”.

Difficulties concerning extrapolation and interpretation of data set over large areas and the use of data of unknown quality can be only counterbalanced by validation processes. This is

especially true when dealing with protection at the scale of a specific source.

Each karstic aquifer, indeed, shows such peculiar characteristics in the structure (and consequently in its functioning) that researchers are obliged to cope with them through a

validation process, which has to start necessarily with the reliable reconstruction of the conceptual model of the aquifer. The flow system and the geochemical system are the two components of the conceptual model.

The flow system, that is the whole 3D aquifer with groundwater, includes the effects of geology, interaction between surface- and ground- waters, exploitation regime, and others natural and human factors that reflect on groundwater movement and, as a consequence, govern the convective transport of pollutants through unsaturated zone and in groundwater. The geochemical system includes the whole physical – chemical factors governing the entrance of a pollutant in underground systems, as well as the attenuation and the transport during both infiltration and flow. The recognition of both flow- and geochemical systems implies the

collection of information on physical local characteristics (climate, vegetation, geology, lithology, quality of superficial- and ground- waters, hydrology, etc.), on anthropic characteristics (urbanization, industrialization, presence of pollution sources, groundwater

exploitation) and their interaction. The definition of the conceptual model includes the outline of border conditions, geometry of

the dominion, geological-structural framework, permeability distribution, flow and transport processes (including transport mechanisms), identification of recharge areas, their connection with discharge areas, and sequence of physical – chemical processes acting into the aquifer.

The attainment of a reliable conceptual model of a karstic aquifer requires the integration of classical hydrogeological information with that obtainable from physical, chemical and isotope natural tracers, by using a multi-parametric approach: this is especially true when the scale of the aquifer prevents from the use of artificial tracers, useful at local scale.

Results can be reliable to the extent that the selected methods are trustworthy: moreover, the studies have to be based on a proper number of data of adequate quality. Classical hydrogeological methods lead to more than one possible conceptual model: the methods based

on natural tracers play a fundamental role in selecting the more consistent one.

2. Elements for the reconstruction of conceptual models of karstic aquifers

2.1. Stable isotope composition of spring waters in the evaluation of the recharge area

elevation In mountainous karstic areas where groundwater discharges through springs, the connection between each spring and the related recharge area can be established by using the stable isotope (Deuterium and Oxygen-18) composition of spring and precipitation waters. Many times, however, there is a lack of information about stable isotope composition of precipitation at

various altitudes. Within the studies for delineating the protection zones for some springs (Ventriglia 1990)

belonging to the Simbrivio carbonate (karstic) aquifer (Lazio, Central Italy), which feed one of the main aqueduct systems of the Roma Province, the “mass-center” and the “inverse hydrogeological budget” methods, which work independently from information on stable

isotope composition of local precipitation, were developed and coupled for recharge area elevation assessment (Sappa and Vitale 2007). The former method determines the mean elevation of the recharge areas of a basin: knowing the

δ18

O ‰ and δD�‰ gradients it allows estimating the recharge elevation for the springs. The latter (Civita et al. 1999) is an innovative, but well tested method, for evaluating the effective

infiltration to an aquifer: the elevations where the highest effective infiltration occurs coincide with the elevations of the main recharge areas. The results of the application of the second method are used for validating the results of the application of the first one. The carbonate formations constituting the Simbrivio aquifer outcrop in the upper part of Aniene river basin, located in the Central Apennines: granular limestones outcrop allover the Simbrivio hydrogeological basin, while dolomitic limestones outcrop in its east part. The geological

structure of the formations is a typical monoclinal, dipping N-NE with 40-45° inclination. Recharge occurs during autumn – winter and the aquifer discharges through numerous springs:

the most important are located in the geological map of Figure 1 and listed in Table 1, with their elevation and stable isotope composition.

Figure 2 shows the relation between δD ‰ and δ18

O ‰ for the same springs with respect to the Global and East Mediterranean Water Lines.

Figure 1. Geological map of Simbruini aquifer and location of springs: (q) alluvial deposits

(Pleistocene);(tr) travertine (Pleistocene); (cg) taluses (Pleistocene); (mar) sandstones (Miocene); (cr) granular limestones with dolomitic intercalations (Cretaceous); (dc) massive

limestones with dolomites and dolomitic limestones (Cretaceous); (c1) sandstones and

dolomitic limestones (Jurassic). ID of springs is in Table 1.

Table 1. Elevation and stable isotope composition of the springs of

the Simbruini Mountains. ID refers to locations in Figure 1.

Nome Sorgente ID

Quote

(m.s.l.m)

δδδδ18O

(SMOW)‰

δδδδD

(SMOW)‰

Cardellina Alta 1 1057 -8,67 -50,14

Cardellina media 2 989 -8,74 -52,71

Cardellina bassa 3 939 -8,82 -51,87

Cesa degli Angeli 4 940 -9,12 -52,24

Cornetto 5 945 -8,38 -47,55

Carpinetto 6 960 -8,74 -49,91

Pantano Alta 7 952 -8,84 -51,56

Pantano presa 8 830 -8,82 -50,84

Pantano Bassa 9 901 -8,90 -50,30

The “mass-center” in solid mechanics is an application of the weighed average: the mass center of a body is the point of the body itself where, from the mechanical point of view, we can

consider focused all the mass and its mechanical properties. The application of mass-center method to a hydrogeological basin defines a point of the basin where it is likely to assume that all precipitation concentrates in.

On the base of a mesh applied to a basin, the elevation (qav) of this point can be calculated as the average elevation of the basin weighed by the effective infiltration distribution, being each QFE

(Quadrate Finite Element) of the mesh characterized by its average elevation. On the same QF elements, indeed, the “inverse hydrogeological budget” gives the distribution of the effective

infiltration. Figures 3a and 3b show the effective infiltration for the Simbrivio Basin according to a 200 m mesh; Figure 3b shows the same effective infiltration as contour lines.

The equation translating the concept of “mass-center” is:

tot

nniiav

I

IqIqIqq

++++=

......11 (1)

where: qi is the average elevation of the ith QFE

Ii is the effective infiltration of the ith QFE

=n number of Quadrate Finite Elements

=totI total effective infiltration.

-9.5 -9 -8.5 -8 -7.5 -7

δ18Oo/oo (SMOW)

-60

-50

-40

-30

δD

o/ o

o (S

MO

W)

1

23

4

56

7

89

MWL

EMWL

ID Spring

1 Cardellina Alta

2 Cardellina Media

3 Cardellina Bassa

4 Cesa degli Angeli

5 Cornetto

6 Carpinetto

7 Pantano Alta

8 Pantano (presa)

9 Pantano Bassa

Figure 2. Relation between δD ‰ and δ18O ‰ for the springs of the Simbruini aquifer with

reference to the Global (MWL) and East Mediterranean Water (EMWL) Lines.

Figure 3. Distribution of effective infiltration (mm/y) in the Simbruini hydrogeological basin

according to the 200 m mesh (a) and as contour lines (b).

The effect of isotope fractionation due to topographic elevation drives the interpretation of the stable isotope composition of spring waters. The atmospheric temperature is lower at the highest

elevations than at the lowest ones, while the distance between the clouds and the ground increases in the same direction: consequently, precipitations are more and more depleted at increasing altitudes.

At global scale the ∆δ18O ‰/∆h and ∆δD‰/∆h gradients of precipitation for ∆h =100 m vary respectively in the range -0.15 ÷ -0.5 and -1 ÷ - 4. The values adopted for precipitation in Simbruini basin are from Bortolami et al. (1978)

.100

‰31.018

mh

O −=

∆

∆δ (2)

.100

‰5.2

mh

D −=

∆

∆δ (3)

The value of qav (1400 m) obtained by the application of the equation (1) has been adopted as the end member to be assigned to the equations (2) e (3) being coupled to the isotopic composition of the Cardellina Alta spring: this way, equations (4) and (5) allow for the other springs the evaluation of the corresponding recharge area elevation (Table 2):

DqDq avav δ×−= 0025,0)( (4) OqOq avav

1818 00031,0)( δ×−= (5)

For the identification of the main recharge areas feeding the concerned springs was used a second average elevation, calculated as arithmetic average of the elevations of the part of the basin placed above the spring of highest elevation (Cardellina Alta spring, Table 1). The

isotopic composition of this last spring was attributed to the second average elevation (1412 m):

the two couples of values (average elevation with respective δD and δ18O) were used as end members in the equations (2) and (3). The results obtained using the arithmetic average elevation (Table 3) are very similar to those obtained by using the “mass-center” method. The analysis of the distribution of the effective infiltration along with elevation intervals (Figure

4) shows that the 48% of the total effective infiltration occurs in the elevation interval 1300÷1500 m a.s.l., confirming the reliability of the above calculations. Taking into account all the above information, the connections between the spring of the

simbruini basin and the most probable elevation of the related recharge area can be drawn: they are shown in Figure 5, with respect to the distribution of the effective infiltration and the

contour lines of 1450 and 1330 m of elevation.

Table 2. Average elevation of recharge areas of springs by “mass-center” method

Spring Equation (2)

(δ(δ(δ(δ 18O)

Equation (3)

(δ(δ(δ(δD)

Average Elevation

(m.a.s.l.)

Cardellina Alta 1400 1400 1400

Cardellina media 1423 1503 1463

Cardellina bassa 1448 1469 1459

Cesa degli Angeli 1545 1484 1515

Cornetto 1306 1296 1301

Carpinetto 1423 1391 1407

Pantano Alta 1455 1457 1456

Pantano presa 1448 1428 1438

Pantano Bassa 1474 1406 1440

Table 3. Average elevation of recharge areas of springs by “arithmetic average” method

Spring Equation (2)

((((δδδδ 18

O)

Equation (3)

((((δδδδD) Average Elevation (m.a.s.l.)

Cardellina Alta 1412 1412 1412

Cardellina media 1435 1515 1475

Cardellina bassa 1460 1481 1471

Cesa degli Angeli 1557 1496 1527

Cornetto 1318 1308 1313

Carpinetto 1435 1403 1419

Pantano Alta 1467 1469 1468

Pantano presa 1460 1440 1450

Pantano Bassa 1486 1418 1452

Figure 4. Percent of effective infiltration with respect to the total one vs. elevation interval

(Simbruini basin).

C onnections between recharge area and springs

# S # S # S

# S

# S

# S

# S

# S # S

0 2 0 0 0 4 0 0 0 6 0 0 0 M e t e r s

N

E W S

6

1 4 0 - 3 4 0

3 4 0 - 5 4 0

5 4 0 - 7 4 0

7 4 0 - 9 4 0

9 4 0 - 1 1 4 0

1 1 4 0 - 1 3 4 0

1 3 4 0 - 1 4 4 0

1 4 4 0 - 1 6 4 0

1 6 4 0 - 1 8 4 0

1 3 0 0 m . s . l . m . 4 5 0 m . s . l . m . 1

E ffecti ve infiltration

mm/y

5

75

85

95

15 2

5 35

4 5

Figure 5. Connections between the springs of the Simbruini basin and the most probable

elevation of related recharge area. ID of springs is in Table 1.

3. An example of multitracing approach in the reconstruction of the conceptual model: the

Murgia aquifer (Apulia - Southern Italy)

The Murgia region is a part of the carbonate sedimentary cover of the Apulian Foreland; it is characterized by monotonous, well bedded restricted carbonate facies (Figure 6). Different fields of tectonic stresses produced various superposing patterns of deformations (folds) and of

ruptures (faults and fissures). The aquifer is very anisotropic, due to an irregular distribution of fracture system and karstic channels. Groundwater is recharged only by rainfall, with an amount of 1,500 Mm3/year; the piezometric heads reach maximum values of 200 m a.s.l. and decrease toward the Ionian and Adriatic coasts, where groundwater discharges through coastal springs (no continental springs are present). The hydraulic gradient varies between 1.5 and 8 ‰.

Figure 6. Geological map (schematic) of Murgia - Mesozoic lithofacies distribution: 1)

dolomite and calcareous dolomite; 2) limestone and laminated dolomite; 3) limestone with

pelitic intercalation; 4) micrite biostromal and calcarenite successions: a - prevalent

interbedded biostromal limestone; b - both types present in regular alternation; b’ - local

concentration of rudistis; c - calcarenite intercalation; 5) post-Cretaceous formations; 6)

doline; 7) hypogean karst form; 8) surface hydrography (from Zezza, 1975, modified)

In the Murgia aquifer, the isotopic stable composition (δD‰ and δ18

O‰) of fresh groundwaters

was used to establish the associated recharge areas (Tulipano et al. 1990; Tulipano and Fidelibus 1996). Figure 7 shows that the points representative of isotopic stable composition of

groundwaters are grouped along three parallel trends: the resultant sample groups correspond to three zones of Murgia region, different for climatic and topographic conditions (Figure 8): going from the A to the C zone the mean elevation decreases and the mean yearly temperature

increases. Hypothesising that the most depleted groundwaters originate from infiltration of precipitation falling at the highest elevation of the region (and vice versa), the location of each sampling point can be directly connected to an elevation interval within each zone (Figure 8).

Connections do not represent the real pathways, but only the link between the recharge area and the groundwater sample. The connections were examined on the light of the chemical composition of groundwater samples, with the aim of establishing a scale of relative residence

time.

The evolution of groundwaters flowing in a carbonate aquifer, indeed, results mainly from water-rock interaction. In particular, the sequence of dissolution and precipitation processes

causes continuous change of Ca2+, Mg

2+ and Sr

2+ concentrations: hence, in such a context, these

ions can be considered as tracers of groundwater evolution.

δδδδ

18O ‰

δD‰

Figure 7. δD ‰ vs. δ�18

O‰ for groundwater samples of Murgia aquifer. The three parallel lines correspond to three zones (A, B, and C in Figure 7) of the Murgia territory

Figure 8. Partition of the Murgia territory in three zones derived from data in Figure 7. The

arrows indicate the connections between each sampling point and the most probable recharge elevation.

Due to incongruent dissolution of carbonate minerals, after several cycles of dissolution and re-precipitation, while calcium and magnesium concentrations result higher or lower than those

characterizing the water at the origin of its path, strontium concentrations result normally higher. Thus, the occurrence of high strontium concentrations should indicate that groundwaters have been subject to important evolution (i.e. they have spent a relatively long time into the aquifer). Figure 9a shows the relationship between the sum of calcium and magnesium concentrations and the strontium concentration: compared to the low concentrations of both parameters of the sample no. 17, which represents the starting point of the chemical evolution, according to above incongruent dissolution processes, the other groundwater samples deviate

due to the increase of the residence time. For each water, the increasing values of the ratio (R)

(ratio between the % increment of Sr++

and the % variation of (Ca++

+ Mg++

) with respect to the concentrations in the reference groundwater sample) characterize groundwaters subject to

increasing number of dissolution and re-precipitation cycles (Figure 9b). The lowest values of the ratio mean that groundwaters have been principally subject to dissolution (low evolution grade); higher ratios involve the action of precipitation processes

(high evolution grade). Groundwaters having R > 2 (medium or high evolution grade) belong to slow circuits; groundwaters with R < 2 (low evolution grade) relate to fast circuits, and are distinguished in three sub-classes. Most of the fast circuits originate from the two main recharge areas recognised in the region. For groundwaters, which chemical evolution is dominated only by dissolution (R < 2), the increase of the total concentration of calcium and magnesium provides an evaluation, in relative terms, of the residence time of the same groundwaters (Figure 10).

The information obtained from the interpretation of chemical data allows differentiating the connections of Figure 8, established by the interpretation of stable isotope composition, according to a relative scale of flow velocity (from slow to very fast) (Figure 11), easily

translatable in a relative scale of residence time.

Figure 9. (a) Relationship between the sum of calcium and magnesium concentrations and the

strontium concentration; (b) percent variation of strontium concentration in relation with the

percent variation of the sum of calcium and magnesium, both calculated with respect to the

concentrations characterizing the sample no. 17 shown in (a); lines indicate different values of

the ratio R.

Figure 10. Relationship between the sum of calcium and magnesium concentrations and total concentration of cations and anions.

Figure 11. Classification of the connections of Figure 7 in terms of relative flow velocity. Lined

areas roughly outline the main recharge areas.

The true pathways followed by groundwaters from the recharge areas to the sampling points, as

well as the location of the recharge areas themselves, can be better outlined by interpreting the trend of the convective thermal field reconstructed through the interpolation of temperature logs carried out along wells of the region. Two horizontal distributions of the groundwater temperature related to the Murgia aquifer at 200 and 600 m b.s.l. are shown in Figure 12a; Figure 12b shows a vertical section of the convective thermal field. Groundwater temperature can play, indeed, the role of tracer of groundwater mobility and might be used to infer some qualitative characteristics of groundwater flow systems. In practice, the

interpretation of thermal conductive fields disturbed by forced advection, defined via above correlation, allows recognizing main recharge and discharge areas and, in aquifers characterised by high anisotropy, allows delineating the main groundwater flow pathways. Thermal gradients

give qualitative information on flow velocity and residence times of groundwaters: main flow directions coincide with the directions of the lowest thermal gradients (Domenico and

Palciauskas 1973; Cotecchia et al. 1978; Tulipano 1988; Tulipano and Fidelibus 1989; Fidelibus and Tulipano 2005). The information gained by groundwater convective thermal field can be integrated with that

obtained by the interpretation of chemical and isotope data: Figure 13 outlines the final result of the data integration. Isotherm horizontal trend allows identifying two main recharge areas,

located at the highest elevations of the Murgia. The trend of δD‰ contour lines outlines the same directions for the preferential flow pathways, confirming the indications obtained by

interpretation of whole data set. δD‰ trend delineates flow pathways smoother than those

defined by the straight connections previously outlined (Figure 8), allowing distinguishing main and secondary flow pathways: an important flow pathway, approximately parallel to the

Adriatic coast (following the Apennine tectonic direction, along which karst processes mainly developed), indicates that Murgia aquifer provides also a lateral recharge to the bordering Salento aquifer.

The multi-tracing approach proves to be a powerful tool in the construction of the conceptual model of anisotropic karstic aquifers: the reliability of such reconstructions depends only on the number and depth of available wells where to accomplish log profiles and sampling for geochemical and isotope analyses.

Figure 12. (a) Horizontal sections of convective thermal field of Murgia aquifer and (b) vertical

section (trace of the section in (a).

Figure 13. δD ‰ contour lines and isotherms (200 m b.s.l.) for Murgia groundwater.

4. Groundwater pollution in karstic aquifers The impact of pollution on groundwater quality depends, besides on intrinsic factors, as time distribution, intensity and duration of precipitation, and, overall, on aquifer structure, which

(a)

(b)

determines the transport mechanisms, on external factors, as the pollutant loads and their distribution on land surface or underground.

The complexity of the aquifer structure of karstic aquifers represents the main challenge to deal with; an additional difficulty originates from the entering mode of pollutants into aquifers, that is if pollutants come from surface by leaching, seepage and infiltration of pollutant loads

released at surface, or derive from direct injection or leaks underground. In the studies concerning karstic groundwater pollution, in order to recognize pathways of pollutants through unsaturated and saturated zones, the conceptual model and its elements reveal essential as well as in the studies of the natural conditions.

4.1 Nitrates as tracers of underground waste water injection: hazard and validation Before the entering in force of very restrictive national regulations, the main source of karstic

groundwater pollution in the Apulia region (Southern Italy) was the release on the ground or underground (by injection wells or sinkholes) of effluents from treatment plants. The pollution potential method was used with the aim of delineating the hazard with respect to

this type of pollution for the Murgia (see Chapter 3) groundwater (Tulipano and Fidelibus 1995b).

The total N load (transformed in the final nitrate oxidized form) arriving at the wastewater treatment plants was estimated on the base of the value of N production for inhabitant of 2250 g/y (defined by the Italian Regulation on statistical base), considering all the production

conveyed in the municipal sewer systems. According to the possible presence of biological oxidation and denitrification in the treatment, the N loads were correspondingly reduced in the single effluents; moreover, on the base of the disposal mode of effluents, only those destined to soil or subsoil (the others being discharged into the sea) were included in the pollution potential

evaluation. The official water endowment per capita (varying among the municipalities of the region according to the population, and being in the average 250 l/inh per day) was used for estimating,

for each treatment plant, the effluent discharge rate: this was made under the hypothesis that the water for civil use distributed to each municipality arrived at the plants by the city sewer

systems. Afterwards, the nitrate concentration in the effluents was calculated on the base of the already evaluated nitrate loads. The comparison among the effluents characteristics (discharge rates with related concentrations)

was achieved by the calculation of a theoric discharge rate (dilution rate: QDi) of unpolluted water required to dilute the effluent NO3 concentration down to 5 mg/l (natural background of unpolluted groundwaters). In a karstic environment, indeed, the main self-depuration process is the dilution operated by groundwater flow at the effluent discharge point. Assuming that the total volume available for dilution coincided with the annual recharge, the effect of dilution was

calculated considering, for each area (i) delimitated by Thyessen polygons, that all related recharge rate (QAi) dilutes the effluents discharging within the area of the (i) polygon. For

characterizing the effluent discharge points as to the nitrate pollution hazard, each point was marked by a value (hazard index, Ip) corresponding to the ratio between the QDi and the ratio between QAi and the number of effluent discharge points (corresponding to the number of

municipalities existing in the area of each i-polygon). Figure 14 shows the result of the procedure. The contour lines interpolate the value of the hazard index calculated for each effluent discharge point. For validating the hazard map, samples of groundwaters from 297 wells and from 53 coastal

springs (mainly brackish) were analyzed for nitrates. Nitrate concentrations of spring waters were corrected for the dilution effect operated by saltwater of marine origin (with zero concentration of nitrates) mixed with freshwater before the outflow into the sea: thus, the entire

nitrate load is attributed to the freshwater components. Figure 15 shows the location of sampling points and the contour lines of nitrate concentration. Moreover, along the coasts, the mean

nitrate concentration of freshwater components of springs is shown in bands.

The main result of the comparison between the maps of Figure 14 and Figure 15 is that the areas characterized by high nitrate concentrations correspond to those characterized by the high

hazard indexes: this validates the procedure followed for defining the pollution hazard deriving form the disposal of effluents of treatment plants.

Figure 14. Nitrate hazard index map due to treatment plant effluent injection or discharge on land surface (1989 - Murgia karstic aquifer).

Figure 15. Nitrate concentration contour lines for the Murgia karstic aquifer (1989). The bands

parallel to the coastline show the mean nitrate concentration in the fresh water component of

the coastal spring waters.

In addition, the reconstruction of the flow pathways made by multi-tracing approach (Figure 13) explains the concentrations at the coastal discharge: the springs work as vectors of the pollution

produced inland according to the preferential flow pathways that feed them.

4.2 Transport mechanisms in karstic aquifers of the pollution derived from diffuse sources

When dealing with diffuse sources dispersed at the land surface, owing to the nature of karstic aquifers, the assessment of intrinsic vulnerability (which refers to the protection of the resource) and the distribution of potential pollution loads are not enough to define the pollution hazard (and consequently the risk, connected to the valuable uses of the resource). The complexity of the karst aquifer structure affects the definition of the intrinsic vulnerability. The definition of source vulnerability (aimed at the protection from pollution of water points to be exploited, i.e. wells, springs, exploitation works) is affected in turn: for the source

vulnerability the additional pathway to be studied with respect to the pathway pertinent to intrinsic vulnerability is that from the groundwater surface to the source(s) (targets). Therefore, the most useful way to cope with resource (intrinsic) and source vulnerability, and gain

elements for correcting the usual intrinsic vulnerability maps and define the vulnerability of the sources is to recognize the transport mechanisms, which, all things being equal, within a same

karstic aquifer, can differ from place to place in a very complicated way. The main feature of a karst aquifer is its organised heterogeneity (Kiraly 1998), which may be outlined as a high permeability channel network (which spatial distribution is generally

unknown) with kilometres meshes, immersed in a low permeability fractured limestone volume. This network is associated to a local discharge area: in the case of the karstic aquifer of Murgia, discharge occurs into the sea, as diffuse flow or through focused outlets (coastal and/or submarine springs). As a consequence of the organized heterogeneity, karst shows duality of

infiltration processes, groundwater flow field and discharge conditions. The conceptual models of karst systems consider normally four sub-systems: the soil zone, the epikarst (subcutaneous zone), the unsaturated zone, and the saturated zone, each behaving

differently with respect to flow and transport. The soil and the epikarst zone contribute in large part to groundwater storage. The epikarst (the

uppermost 4 to 15 m about of weathered limestone) is very important in karstic aquifers because of its high secondary permeability due to karstic processes: karstic enlargements diminish with depth, causing a decrease in permeability, except for down widened master joints and faults.

The epikarst, because of its storage, contributes to base flow towards conduits and groundwater during low precipitation periods (Dorfliger et al. 1999). It concentrates the flow in the upper part of unsaturated zone when the infiltration rate overcomes the percolation rate through the vadose zone (Klimchouk 2004): thus, it is a key element in explaining the nervous hydraulic response of classic karstic springs.

In epikarst waters, due to the residence times of the order of weeks or months, the parameters indicating dissolution (as Total Dissolved Solids, TDS, and magnesium), and those related to

conservative pollution (as nitrates) have high values, while the values of pollution indicators related to organic matter content (as. TOC, Total Organic Carbon) (Batiot et al. 2003) decrease due to attenuation processes of stored organic loads.

The unsaturated (transition) zone is a zone of low storage, associated with fractures, joints and inter-granular seepage: it connects the epikarst to the saturated zone, where waters coming from the upper reservoirs mix and are drained towards the discharge area: a non-linear-mixing of tributaries occurs during the flood events.

Aquilina et al. (2005, 2006) propose a transfer scheme of precipitation to some karstic springs of southern France under the effect of consecutive important precipitation events occurred in the winter 1997-1998. Chemical and hydrological data indicate the existence of a “piston flow”

mechanism, induced by an “n” event that displaces towards the springs from the sub-system of the unsaturated zone the waters of the “n-1” event, occurred formerly. The epikarst reacts to the

“n” precipitation event transferring to the conduits and towards the saturated zone water volumes and pollution loads stored in the period preceding the “n” event. Waters coming from

the different sub-systems can be recognized at due to their particular chemical and isotope characteristics: the proportion of the different tributaries arriving at springs depends on total

infiltration volume and changes during flood, varying the shape of chemograms (trend of chemical and isotopic parameters over time) at springs. Authors evidence that the direct contribution of precipitation water can be rarely observed during monitoring: this direct transfer

is noticed at springs very soon after the precipitation event, indicating the direct infiltration along a main drainage axis. Precipitations do not seem to reach directly the saturated zone: the transfer occurs mainly through the mediation of the sub-systems of the unsaturated zone. The effect of removal from the sub-systems of the unsaturated zone should be more evident at the outlet of the karstic systems if a period without effective infiltration occurs before an important (extreme) precipitation event: during said periods, pollutants can accumulate and can degrade into the sub-systems. Later, under the pressure caused by the extreme event, the sub-

systems should release, in different times and with different, but evident, chemical imprints, what is accumulated in the previous periods. In conclusion, it is suggested that soil and epikarst play an important role in delaying the

transfer of pollutant towards the saturated zone, due to their high storage capability and water residence times.

Given that a direct reconstruction of the structure of a karstic system is very difficult, to obtain information concerning infiltration, karst evolution, distribution of porosity and field of hydraulic parameters, presence and role of different recharge mechanisms, and, consequently,

pollutant transport, researchers use an indirect approach by studying the “global response” of the whole above factors with respect to precipitation. This “global response” can be evaluated observing the variations of chemical and hydraulic behaviour of groundwater at springs: the interpretation of both, hydrograph and variation of chemical characteristics, allows recognizing

the contribution of the waters from different reservoirs, because of their distinctive chemical/isotopic imprint (Fidelibus, 2008). Unluckily, in the Murgia aquifer there is a total lack of inland freshwater springs: the type of

monitoring (and the consequent interpretation methodologies of hydrographs and chemograms outlined for karstic springs that a wide scientific literature deals with) is not easily adjustable to

wells (that represent the only measure points apart from brackish coastal springs), especially if the monitoring concerns a regional scale. The lack of a unique point for measurements (as the outlet of a spring), increases the number of variables to be considered: if at a spring the study

involves the measure of parameter variability in the time, in a well this variability, at a fixed time, regards the space as well (variability of water characteristics along the saturated thickness of the aquifer). Moreover, sampling at springs can be as frequent as required by the evolution of discharge, i.e. sampling follows the rising and recession limbs of the hydrograph: water quality can vary dramatically over short time periods, and weekly or even daily sampling may be

inadequate to describe pollution events. In the Monitoring Network of Murgia (Figure 16), including 65 wells, due to both the number

of wells and their length, the monitoring frequency was of about three-four months. Given the high inertia of Murgia aquifer with respect to that of the classical karstic aquifers, which literature mainly refers to, one month frequency should be appropriate: however, specific

studies on this topic for the Murgia region are presently lacking. Notwithstanding above limits, chemical analyses of groundwater samples from the Murgia Monitoring Net can help in outlining, at regional scale, the transport mechanisms acting in the karstic aquifer. The tri-monthly frequency sampling was effected from the end of 1994 to the

beginning of 1997, with many gaps concerning the number of tested wells each time. The sampling was made in static conditions, close to the water table (when groundwater was in unconfined condition) or below the top of the permeable formation (when groundwater was

locally confined): related samples gave information on the dynamic reserve, normally more subject to pollution coming from the surface than the perennial one. In some cases sampling

extended to higher depths. Analyses included pH, dissolved oxygen, electrical conductivity, redox potential, temperature, major constituents, nitrates, nitrites, ammonium, biochemical

oxygen demand, chemical oxygen demand, silica, plumb, mercury, iron, bacterial charge. Luckily, most of pollution parameters were under the detection limits, except nitrates and

organic carbon. Concerning the trend of nitrate concentrations over time, only some considerations can be proposed on the base of the entire available data-set (location of all measure points in Figure

16), which includes surveys carried out by a variety of public institutions in the period 1987-2003. Figure 17 shows the nitrate concentrations plotted according to the date of sampling. From the 1987 up to the 2008, the concentration range amplifies significantly. Waters from pumping wells used for drinking purposes show nitrate concentrations in the range 1- 60 mg/l, independently from the date; a few samples from pumping wells (authorized for agricultural use) show sometime concentrations higher than 100 mg/l. The samples collected in static

condition from the Monitoring Net show nitrate concentrations in a larger range, from less than 1 mg/l to maximum values of about 55 mg/l. The results of a research survey of 2008 show concentrations in the same range of previous surveys, even though some values are higher than

200 mg/l. In the whole, Figure 17 does not allow a definition of the time evolution of the nitrate concentrations, even if it indicates that a clear diffuse pollution exists from almost two decades,

especially within freshwaters, which are of interest especially because of their use for drinking purposes. Figure 18, based as well on the data collected from the Regional Monitoring Net, shows the

distribution of nitrates according to the elevation of sampling. Even if the number of samples at high depth is lower than the number of samples taken at shallow depth, figure 18 suggests that the shallowest horizon (up to -50 m a.sl) is the most polluted: nitrate contents reach the 60 mg/l, with a few exceptions. At higher depths the upper limit of the nitrate concentration range

progressively decreases. Thus, the upper part of the groundwater shows the largest variations, which attenuate with depth: nitrate pollution likely comes from the surface, mainly due to leaching of fertilizers percolating into the ground.

Figure 16. Location of wells belonging to different institutions or research nets surveyed in

different dates; related data on groundwater nitrate concentration are shown in Figure 17.

Figure 17. Nitrate concentrations of Murgia groundwaters according to the date of sampling.

0 20 40 60

Nitrates (mg/l)

-400

-300

-200

-100

0

100

Ele

vati

on

(m

a.s

.l.)

Figure 18. Nitrate concentrations with depth in Murgia groundwaters: samples come from the

wells of the Monitoring net of Murgia Region and refer to the period between the end of 1994

and the beginning of 1997.

In the August 1995, during the running period of the Regional Monitoring Net, an extreme rain

event interested large part of the Apulia Region: the pluviometric station of Cassano Murge in the Murgia registered 225 mm between the 16th and the 27th of August, with the daily maximum

of about 88 mm the 17th of August for the entire region. Figure 19 shows the total monthly

precipitation recorded at the Murgia pluviometric stations from January ‘95 to December ’96.

A sampling was made at most of the monitoring wells before the event (mainly in May ‘95) and another sampling was carried out after the event (October ‘95): the sampling of May is preceded by a low effective infiltration period.

Figure 20 shows the time trend of the nitrate concentrations in the monitoring period: only a part of the 65 monitored wells is included in the plot (wells of the Northern area) and they are grouped in three main classes, according to the progressive distance from the NW main recharge area of the region. The nitrates show peaks up to about 60 mg/L after a period of about two months from the extreme rain event, starting from concentrations typical of the natural background (about 5 mg/l), measured before the event: thus, an enrichment in nitrates of about one order of magnitude occurs with a time lag of about two months from the event. After

September ’95 most of analyses are incomplete: thus monitoring fails in recognizing the effects of another extreme rainfall event (during October 1996), which was preceded by a dry period as the previous extreme event of ‘95. In Figure 20 the lack of information following the second

sampling is marked by question marks. However, nitrate concentrations in December 1996 as well are higher than concentrations before the first event. The end of monitoring, in December

1996, prevents us from significant additional observations. The time lag of about two months between the date of the extreme precipitation event of August ’95 and the recognition of the high nitrate concentrations has to be considered with caution,

because the samplings could have encountered indifferently the ascending or the descending part of a peak, i.e., the true peak could have appeared before or after the measures: consequently the real lag could be shorter or longer than two months. The increase of nitrate in the post-event phase in any case remains a piece of evidence.

Figure 20 shows also the Total Organic Carbon (TOC) trends for the same wells. The TOC trend is specular to the nitrate trend: post-event values are lower of about one order of magnitude than those characterizing the pre-event phase.

The organic content of soils can reach 300 mg/l and normally does not overcome the 100 mg/l in rivers: in rapid infiltration waters, TOC reaches maximum values of 10 mg/l, while in

groundwaters TOC is normally lower than 1 mg/l (Bakalowicz 2003; Batiot et al. 2003). In the case of karstic systems not polluted from specific human sources and not recharged directly by river waters, the Total Organic Carbon (TOC) can be considered to come from biological

activity in the soil: this makes TOC an interesting tracer of rapid infiltration. During floods, the arrival at a karstic spring of low residence time waters is revealed by a TOC increase; during spring recession, spring waters are of higher residence time and show TOC concentrations significantly lower than previous ones. TOC concentrations in groundwaters are thus inversely proportional to residence time, because of the oxidation of organic matter, with mediation of

bacteria. The low concentrations of TOC in groundwater samples after the extreme rainfall event of August ’95 cannot be linked to waters coming from the surface, which should have

higher concentrations: the low concentrations maybe indicate that waters are of high residence time, that is, they come from a reservoir in-between the topographic surface and the saturated zone, where they spent enough time to allow degradation of organic matter.

After the event, nitrates, differently from TOC, increase notably. These high concentrations, according to the indications of TOC, should characterize waters of reservoirs overlaying saturated zone, where evidently nitrates can accumulate during low effective infiltration periods. The time lag between the extreme event and the recognition in groundwater of a nitrate

concentration peak (for the examined case it can only be considered to coincide with measurement) should refer to an average time of transfer of the water volumes stored in the soil, the epikarst and the conduits to the groundwater. The significant and contemporary decrease of

TOC supports the hypothesis that waters come from the reservoirs of the unsaturated zone, in particular from the epikarst.

Major constituent analyses of the same water samples have been carried out with continuity: unluckily, in some cases they are affected by high analysis errors, which prevent us from using

data with complete reliability. However, magnesium, chlorides and sulphates, which give indication about residence times (water maturity), increase in concentration after the event. This

fact supports the previous indication of the arrival at groundwater surface, in the post-event period, of waters that have spent some time into the aquifer. Afterwards, the diluting effect of the ’95-’96 winter recharge causes the decrease of magnesium, chlorides and sulphates; they

increase again after the extreme event of October ’96. When nitrates increase and TOC decreases, the hydraulic heads do not increase significantly. This fact may indicate that the observed chemical variations in the post-event period are caused by modest water volumes, unable to modify sensibly the hydraulic heads, but able to temporarily modify the quality of groundwaters. Hydraulic heads increase only in March ‘96, following the winter recharge: correspondingly, major constituent concentrations dilute. In conclusion, the sequence of mechanisms working on the Murgia karstic system can be

outlined as follows: � the extreme event of August ’95 pushes by piston flow the waters residing in conduits, soil

and epikarst;

� the epikarst becomes a perched aquifer, where horizontal flow activates towards conduits, sinkholes or master joints;

� the waters coming form the reservoir of the unsaturated zone convey towards groundwater the conservative pollutant loads accumulated in the same reservoirs during the period of low effective infiltration (April ’95 - July ’95) preceding the extreme rainfall event;

� in the post-event period, nitrate concentrations increase in groundwater without appreciable hydraulic head increase, while TOC values decrease, indicating the arrival of modest water volumes having spent enough time in the reservoirs of the unsaturated zone;

� the epikarst, under the prolonged precipitation is progressively leached and waters into the

epikarst and unsaturated zone become progressively younger due to the contribution of waters of recent infiltration;

� during winter, the waters of effective infiltration wash completely the epikarst and the

unsaturated zone, which at this point are almost completely free from contaminants; in the same period, groundwaters show much diluted concentrations of major constituents;

� hydraulic heads increase only after winter precipitation, with a time lag, with respect to nitrate peaks, ranging from two to four months.

All above findings suggest that nitrate pollution depends on the structure of karstic system, but

also strictly from precipitation dynamics, especially when it produces striking alternation of dry and wet periods. The qualitative groundwater status, under this dynamics, cannot be defined

from one only survey, even if ample and dense as to measurements: the above trends suggest, as expected for a karstic aquifer, that different and diversely worrying pollution scenarios can be determined in the different seasons of under different precipitation dynamics.

With the aim of synthesizing the effects of above dynamics, Figure 21 shows four maps, obtained interpolating nitrate data referred to the shallowest part of groundwater according to

quadrimonthly periods. The first two maps show nitrate distributions before the extreme rainfall event of August ’95; the third map shows the nitrate distribution in the four months following the event, and the fourth map is the nitrate distribution during winter recharge. Apart from the

uncertainty about statistical significance of such maps, due to low density area distribution of the monitoring net wells, certainly it can be asserted that the situation greatly modifies during the sixteen months period. The post-event period shows the highest nitrate concentrations: it is preceded by another four-month period where only an initial increase of concentrations is

outlined, probably due to the inclusion of August in the period itself. Thus, it could be concluded that modest rain events, even closer, but of total volume lower than that of waters resident in the different reservoirs of the unsaturated zone, cause, even after

periods of low effective infiltration, only minimum variations of the groundwater quality. The risk of exposition to peak pollutant concentrations is, for karstic groundwater, dependent firstly

on the return time of extreme events: however, such events have the potential of cause peak concentrations in groundwater only if preceded by periods of low effective infiltration.

Au

g-9

4

De

c-9

4

Ap

r-95

Au

g-9

5

De

c-9

5

Ap

r-96

Au

g-9

6

De

c-9

6

Ap

r-97

Au

g-9

7

0

40

80

120

160

200

Pre

cip

ita

tio

n (

mo

nth

ly t

ota

l, m

m)

MurgiaPluviometric

Stations

Gioia del Colle

Ruvo di Puglia

Cassano Murge

Altamura

Grumo Appula

Corato

Noci

Figure 19. Time trend of total monthly precipitation registered at the Murgia pluviometric

stations.

Au

g-9

4

Dec

-94

Ap

r-95

Au

g-9

5

Dec

-95

Ap

r-96

Au

g-9

6

De

c-9

6

Ap

r-97

Au

g-9

7

0

20

40

60

Nit

rate

s (

mg

/l)

Au

g-9

4

De

c-9

4

Ap

r-95

Au

g-9

5

De

c-9

5

Ap

r-96

Au

g-9

6

De

c-9

6

Ap

r-97

Au

g-9

7

0

1

2

3

4

To

tal O

rgan

ic C

arb

on

(m

g/l

)

Wells of the Regional

Monitoring Net4AIM

LS7BA

LS11BA

1M

LS17BA

7B

30AIM

31AIM

LS4BA

P-CBAL

LS18BA

LS20BA

LS26BA

P1CASMEZ

16 PROG

21AIM

??

??

Figure 20. Time trend of nitrate and TOC concentrations in groundwater sampled from the

Murgia monitoring net in the period March 95 – January 1996.

Figure 21. Nitrate concentration in the shallowest part of Murgia groundwater: each map

gathers the samples collected in the indicated fourth-month period of monitoring.

References Aquilina L., Ladouche B. and Dörfliger N. 2005. Recharge processes in karstic systems investigated through the correlation of chemical and isotopic composition of rain and spring-

waters. Appl. Geochem. 20: 2189-2206. Aquilina L., Ladouche B. and Dörfliger N. 2006. Water storage and transfer in the epikarst of karstic systems during high flow periods. Journal of Hydrology 327: 472-485. Bakalowicz M. 2003. Natural organic carbon in groundwater. C. R. Geoscience 335: 423–424. Batiot C., Emblanch C. and Blavoux B. 2003. Total Organic Carbon (TOC) and magnesium (Mg2+): two complementary tracers of residence time in karstic systems. C.R. Geoscience 335: 205–214.

Bortolami, G.C., Ricci, B., Susella, G.F. and Zuppi, G.M., 1979. Isotope hydrology of Val Corsaglia, Maritime Alps, Piedmont, Italy. Isotope Hydrology 1978, IAEA, Vienna, 1:327-350. Civita M., De Maio M. and Vigna B. 1999. Una metodologia GIS per la valutazione della

ricarica attiva degli acquiferi. Atti 3° Conv. Naz. sulla protezione e gestione delle acque sotterranee per il III Millennio, Parma, 13-15 ottobre 1999. 1: 291-303.

Cotecchia, V., Tadolini, T. and Tulipano L. 1978. Groundwater temperature in the Murgia karst aquifer ((Puglia - Southern Italy). Proc. Int. Symp. on Karst Hydrology, Budapest. European Commission, DG XII 1995. Final report of COST Action 65, Karst groundwater

protection, Office for Official Publications of the European Communities Commission, Luxembourg, EUR 16547 EN, ISBN 92-827-4682-8, 446 pp. Domenico P.A. and Palciauskas V. V. 1973. Theoretical analysis of forced convective heat transfer in regional ground-water flow. Geol. Soc. Am. Bull. 84, 3803–3813.

Dörfliger N., Jeannin P.Y. and Zwahlen F. 1999. Water vulnerability assessment in karst environments: a new method of defining protection areas using a multi-attribute approach and GIS tools (EPIK method). Environ. Geol. 39: 165–176.

Drew D. and Hötzl H. (Eds) 1999. Karst hydrogeology and human activities - Impacts, consequences and implications. IAH Karst Commission, International Contributions to

Hydrogeology, Taylor & Francis, ISBN-10: 9054104643, 20: 251-255. Fidelibus M.D. 2008. Meccanismi di trasporto degli inquinanti in acquiferi carsici sotto eventi estremi (Alta Murgia, Puglia), Ordine dei Geologi di Puglia, Convegno “Cambiamenti climatici

e rischi geologici in Puglia”, in press. Fidelibus M.D. and Tulipano L. 2005. Groundwater temperature as environmental tracer. In Stournaras G., Pavlopoulos K., Bellos Th (Eds.), Proc. 7th Hellenic Hydr. Conference, Athens, Geol. Soc. of Greece, ISBN 960-88816-2-5, II: 211-218 Kiraly L. 1998. Modelling karst aquifers by the combined discrete channel and continuum approach. Bulletin d'Hydrogéologie 16: 77-98. Klimchouk A. 2004. Towards defining, delimiting and classifying epikarst: Its origin, processes

and variants of geomorphic evolution. In Jones W.K. et al (Eds.) Proc. Symp. On Epikarst, Sheperdstown, West Virginia, USA. Karst Water Inst. Spec.Publ. 9: 23-35. Sappa G. and Vitale S. 2007. Approccio multisistema alla individuazione delle aree di ricarica

di un acquifero. Ingegneria e Geologia degli Acquiferi, Suppl. GEAM 2:79-92. Tulipano, L., 1988. Temperature logs interpretation for the identification of preferential flow

pathways in the coastal Carbonatic and Karstic Aquifer of the Salento Peninsula (Southern Italy). Proc. 21

th I.A.H. Congress, Karst Hydrogeology and Karst Environment Protection,

Guilin (China), 1988. Geological Publishing House, Bejing, China, XXI (2): 955-961.

Tulipano L., Cotecchia V. and Fidelibus M. D. 1990. An example of multitracing approach in the studies of karstic and coastal aquifers. In Gunay G., Johnson I. & Back W. (Eds.), Proc. Int. Symp. and Field Seminar on Hydrogeologic Processes in Karst Terranes, Antalya (Turkey), 1990. IAHS Publ. 207: 381-389.

Tulipano, L. and Fidelibus M. D. 1989. Temperature of ground waters in coastal aquifers: some aspects concerning saltwater intrusion. In De Breuck W. (Ed.), Procs. 10th SWIM, Ghent, Belgium, 1988. Natuurwetenschappelijk Tijdschrift, 70: 308-316

Tulipano L. and Fidelibus M.D. 1995a. Italian National Report. In Final Report of COST Action 65, Hydrogeological aspects of groundwater protection in karstic areas, Office for Official

Publications of the European Communities Communities, Luxembourg, Report EUR 16547, Part I, National Reports 171-201. Tulipano L. and Fidelibus M. D. 1995b. Metodologie per la valutazione degli effetti del rilascio

di reflui urbani sulla distribuzione dei nitrati nelle acque sotterranee delle unità idrogeologiche Murgia e Salento (Italia Meridionale). Quaderni di Tecniche di Protezione Ambientale 49: 167-179. Tulipano L. and Fidelibus M.D. 1996. Multitracing approach for the identification of main hydrogeological pathways feeding coastal springs of a karstic aquifer. in R.W. Buddemeier

(Ed.), Groundwater Discharges to the Coastal Zone: Proceedings of an International Symposium. LOICZ (Land - Ocean Interaction in the Coastal Zone), Moscow (Russia), 1996,

LOICZ Reports & Studies 8: 113-121. Tulipano L., Fidelibus M.D. and Panagopoulos A. (Eds.) 2005. Final Report of COST Action 621, Groundwater management of coastal karstic aquifers. Office for Official Publications of

the European Communities Commission, Luxembourg, EUR 21366, ISBN 92-898-0015-1, 2nd

Ed., 366 pp. Ventriglia U. 1990. Idrogeologia della provincia di Roma IV Volume Regione Orientale. Amministrazione Provinciale di Roma, Assessorato LL.PP. Viabilità e trasporti

Zezza F. 1975. Le facies carbonatiche della Puglia ed il fenomeno carsico ipogeo. Geol. Appl. e Idrogeol. X: 1-54. Zwahlen F. (Ed.) 2003. Final Report of COST action 620, Vulnerability and risk mapping for

the protection of carbonate (karst) aquifers. Office for Official Publications of the European Communities, Luxembourg, EUR 20912, ISBN 92-894-6416-X, 297 pp.