This paper investigates the relationship between the two parameters, aquifer media and hydraulic conductiv-ity, and assesses their reliability for aquifer vulnerability assessment. To evaluate the possible non-inde-pendence of the parameters in question, sensitivity analysis procedure was used in a GIS environment. The procedure was tested in three areas where aquifer vulnerability was assessed, corresponding to three different hydrogeological settings. The applications demonstrate that non-independence is confirmed and is more evident in alluvial and volcano-pyroclastic aquifers.
Groundwater protection is a prior environmental concern in many countries, such as in Europe, where more than 50% of the water supply is obtained from groundwater [
Pollution vulnerability assessment requires in-depth knowledge of the hydrogeological, hydrodynamic and hydrochemical characteristics of aquifers. Several methods and simulation models have been developed since the 1970s. In the last 15 years the advent of GIS, permitting the inventory, archival, retrieval and display of spatial data and the link to numerical rating systems, has resulted in the widespread use of parametric methods, based on the hydrogeological setting: e.g. DRASTIC [
In parametric methods all the parameters have to be independent. This study tests the possibility of the nonindependence of two parameters—aquifer media and hydraulic conductivity—by application in three sample areas corresponding to three different hydrogeological environments: karstic, volcano-pyroclastic and alluvial aquifers.
The western sector of the Matese Mts. encompasses approximately 170 km2 and is located between the regions of Campania and Molise (southern Italy). The area where carbonate rocks (limestone and dolomite) crop out is approximately 130 km2, with a mean elevation of 950 m a.s.l. Large sectors are characterized by several forms of karstic origin, which influence the groundwater flow. The groundwater body reaches the springs at the foot of the mountains with a total discharge of about 2 m3/s.
On the basis of geological, structural and hydrochemical data, [
Roccamonfina, the oldest quaternary volcano in Campania, is a stratovolcano with a smooth morphology (400 km2), reaching a maximum height of 1006 m a.s.l. (
The piezometric setting (1996) shows a basal radial flow that partly feeds the contiguous aquifers by underflows, and partly increases the flow of some peripheral streams. Only a small amount of groundwater reaches the numerous springs located on the volcano. The recharge volume of the aquifer, calculated by the hydrogeological balance, is nearly 113 ´ 106 m3/a and its source is wholly dependent on rainfall [
The study area, with its almost flat morphology, covers approximately 68 km2. The Venafro Plain is crossed by the upper part of the Volturno River. It is delimited by the Matese Mountains to the east and by the Venafro ridge to the west. The Venafro Plain consists of talus and alluvial deposits (thickness 0-200 m). The bedrock is constituted by Mesozoic limestones (outcropping in the proximity) and Miocenic areanaceous-marly turbiditic deposits (
Along the boundaries of the plain, where a significantly thick layer of talus deposits occurs, the sedimentary deposits have a relatively coarse grain size. In the SE part the alluvial sediments are interbedded with pyroclastic deposits.
The aquifer in the plain can be considered a single continuous body. Water circulation takes place mainly in the permeable layers, with sand and gravel-sized grains. The aquifer is unconfined; locally, pyroclastic or clayey layers generate confined sectors. It Errore. L'origine riferimento non è stata trovata.is recharged by seepage waters from limestone mountains, and global flow is in an SE direction.
Aquifer pollution vulnerability of the sample study areas has been assessed elsewhere using the SINTACS method [8-10]. The SINTACS method [4,5], which originally derived from DRASTIC [
ISINTACS = SP(1,7)∙W(1,n)
where P(1,7) is the score of the 7 parameters used and W(1,n) is the weight in each class.
The five weight classes used by SINTACS depend on the hydrogeological features (porous, fissured or karstified media) and on impact condition (uncontaminated or urbanized environment). In the same map, different weight classes in different sectors can be used.
The aquifer pollution vulnerability maps of the sample areas are shown in Figures 4, 5 and 6. The range variation of each parameter and the rating assigned are shown in
In the Matese area (
In the Roccamonfina volcano area, vulnerability was evaluated only for the southern part (
In the Venafro Plain area (
This study verifies the possible non-independence of two parameters—the Aquifer media and hydraulic conductivity. Often, on applying the DRASTIC or SINTACS method the same input data and basic information are used to define the aquifer media and hydraulic conductivity parameters: hydrogeological structure, lithology, fissuretion and karst conditions of the aquifer, especially when pumping test data are lacking (see
Aquifer media reflects the attenuation characteristics of the aquifer material, taking account of the mobility of the contaminant through the aquifer material. Characterization of aquifer media, like those of the vadose zone, is one of the most difficult aspects of SINTACS: problems arise as to how in practice to assign scores, deal with vertical and horizontal lithologic variability, and how to interpolate point data (well data) to generate surfaces.
M. Civita and M. de Maio [4,5] suggested calculating the weighted mean of the scores of the hydrogeologic units in order to manage vertical variability.
The amount of water percolating to the groundwater through the vadose zone is influenced by the hydraulic conductivity of the soil media. It relates to fractures, bedding planes and voids between grains which become pathways for fluid movement. Clearly, high hydraulic conductivity indicates great movement once a contaminant has entered the aquifer. Accordingly, a high DRASTIC and SINTACS score (
Hydraulic conductivity can be calculated from the transmissivity data from withdrawal, pumping and slugtests (m2/s), and aquifer thickness (m). Lacking field data, the rating can be assigned according to the characteristics of the hydrogeological units, as shown in the horizontal bar chart in
First of all, the correlation between the two parameters was evaluated; the horizontal bar chart in
Using this modification, the ratings of conductivity and aquifer media (
strates: a low correlation for glacial deposits, lavas, dolomites; a moderate correlation for pyroclastic sediments, plutonic rocks, sands and limestones; a good correlation for metamorphic rocks, clay and silt (fine-grained alluvial sediments), peat, sandstone, gravel (coarse-grained alluvial sediments). The correlations, with reference only to the hydrogeological units most commonly found in the study areas, are displayed in
The classified maps of the two parameters were then compared for each study area. Correlation analysis between conductivity and aquifer media maps was carried out using GIS, allowing spatial statistics [
The last procedure performed to evaluate the problems related to “non-independence” in vulnerability assessment parametric methods was based on sensitivity analysis [
This contribution deals with the aspects related to the influence of ratings and weights assigned to the single parameters on the final overlay map. This method was preferred to uncertainty analysis [
Quantitative comparison of the vulnerability maps (pixel size 30 m × 30 m) involved a normalization procedure to obtain comparable values of vulnerability degree. In fact, the possible minimum and maximum values
of SINTACS vulnerability index have to be considered: regarding a normal index range of 26-260 for the Roccamonfina and Venafro areas, the range without aquifer media or without conductivity varies between 23 and 230; for the Matese area, where the weight of the karst milieu is 5 for both parameters, the range stretches between 21 and 210.
Final interpretation is based on the analysis and comparison of the three vulnerability maps (in degree) for each area. In the Matese area the more diffused moderatehigh vulnerability degree changes into a low-moderate degree in the map without aquifer media, and into a moderate degree in the map without conductivity. The crossing operation between these raster maps (≈ 530,000 pixels) reveals a fitting of 47% between the “full” map and the map without aquifer media; of 70% between the “full” map and the map without conductivity; 42% between the map without aquifer media and the map without conductivity. Construction of a map of differences where the variation degree ranges from 0 to 2 generally shows values ≤ 1.
In the Roccamonfina volcano area the prevailing moderate vulnerability degree changes into a low-moderate degree in the maps without aquifer media and in the map without conductivity. The crossing operation between these raster maps (≈ 550,000 pixels) reveals a coincidence of 81% between the “full” map and the map without aquifer media; 75% between the “full” map and the map without conductivity; 80% between the map without aquifer media and the map without conductivity. Due to these high coincidences the map of differences is generally 0, and rarely 1.
In the Venafro Plain area the prevailing high vulnerability degree is unchanged and the maps are very similar. Crossing of the raster maps (≈ 170,000 pixels) shows a coincidence of 90% between the “full” map and the map without aquifer media; 84% between the “full” map and the map without conductivity; 89% between the map without aquifer media and the map without conductivity. The map difference is 0 almost everywhere.
Interpretation of the results is based on the analysis and comparison of the vulnerability maps, representing the two layers, and the crossing tables with resulting statistics. In the Matese area the low correspondence between the maps was expected on the basis of the diagram in
A general consideration on this analysis is that the removal of aquifer media and conductivity from the seven parameters generates a significant variation in the resulting vulnerability map only in the Matese area, where their removal tends to decrease the vulnerability degree. This confirms the assumption that the two parameters have to be considered non-independent. Otherwise, these layers could be both non-critical for vulnerability analysis, as in the Roccamonfina area where their importance, compared with the values of the other parameters, is low. These considerations are in contrast with [
In this paper, the non-independence of the aquifer media and conductivity parameters in the SINTACS aquifer vulnerability assessment method was evaluated following three steps: first, the two parameters were correlated, the parameter maps of three sample areas were then compared, and finally the contribution of each of the two parameters in the final vulnerability map by map removal sensitivity analysis was assessed.
The results of the analysis indicate that aquifer media and conductivity are quite non-independent parameters in the aquifer vulnerability assessment methods, especially with the lack of field conductivity data from pumping tests . In addition, the application to three sample areas demonstrated that this dependency is greater in porous media aquifers than in fractured and karstified aquifers. Moreover, sensitivity analysis, performed using a GIS, reveals the real influence that each parameter exerts in relation to the value of the other parameters.
This confirms the validity of the SINTACS and DRASTIC methods and suggests they should be applied and developed in terms of higher accuracy in parameter definition (use of experimental data). Alternatively, fewer well-defined representative parameters should be used to assess pollution vulnerability.