Geostatistic application for the study of contaminated groundwater in broad industrial areas

Transcription

1 Geostatistic application for the study of contaminated groundwater in broad industrial areas C. Cherubini & C. I. Giasi Polytechnical University of Bari, Italy Abstract A geostatistic application is proposed for the realization of a hydraulic conductivity map in an abandoned industrial area, characterized by the extension of about 300 ha, in which a characterization study is in progress. The study has made it possible to define the distribution of the grade of vulnerability of the area and, in relation to the knowledge of the centres of danger, has given further elements for the individuation of hot spots and the paths of contamination. In order to obtain a map of the hydraulic conductivity, through investigation of a direct nature, experimental values of permeability and trasmissivity have been subjected to specific experimental checks, and to calibration made through theoretic hydrogeologic approaches in order to establish the reliability of the data that have been given. The results, which are the object of verification, have been treated afterwards through a geostatistic approach in order to get a punctual valuation of the hydraulic conductivity, also in non-sampled points which are inside the examined area. Keywords: geostatistic application, hydraulic conductivity, calibration. 1 Introduction The knowledge of the grade of vulnerability of the soil in broad industrial areas could give useful indications, not only for the understanding of the episodes of pollution, but also in order to work out strategies of utilization of the soil, and to plan anthropic activities aimed at eliminating the risk of pollution of the aquifer, and for evaluations concerning the possibility of remediation. For this purpose the aim of the study carried out is that of contributing, even if partially, to the study and the hydrogeologic characterization of an abandoned

2 32 Brownfield Sites II industrial area, with the aid of a valid instrument represented by the geostatistics. This one in particular, combined with a competent knowledge, makes it possible to undertake decisions about the delimitation: of potentially contaminated areas, towards which measures of remediation could be taken or however a valid risk analysis could be made. of areas characterized by soils of good quality towards which specific management plans could be developed. The possibility that groundwater could be polluted depends on the velocity with which the transfer of the substance from the ground level to the surface of the groundwater takes place, on the magnitude of infiltration, on the path run by the particles, and on the physical-chemical-biologic mechanisms that operate selectively in relation to the type of soil and of substance. This study in particular, has been referred to a real industrial area on which the studies of characterization are in progress; thus for discretion s reasons, no specific reference about the area could be made. On this area, investigations of a direct nature have been carried out, aimed at estimating the experimental values of the principal hydrogeologic groundwater properties, such as permeability, transmissivity, and aquifer storage. Principally focusing our attention on the hydraulic conductivity, the in-situ practicable tests could be distinguished in pumping aquifer tests and borehole tests. In any case, the experimental determination of transmissivity is affected by great variability; thus its measured value could draw away from the real ones of the deposit by orders of magnitude. For this reason it would be useful to submit the experimental results to check processes. It would be better if the interpolation of data affected by such variability would be carried out by following geostatistic procedures so as to extend conveniently the hydrogeologic analysis also to the non-sampled points and to obtain punctual estimations of the hydraulic conductivity inside the examined area. 2 In situ investigation In the examined area, 46 permeability tests have been executed in order to determine the principal hydrogeologic parameters of the aquifer. Among those tests, 30 are Lefranc and 17 are Slug tests with survey of the absorption measures through automatic acquirer (data logger) and pressure transducer. The Lefranc permeability tests have been executed in advancement in the boreholes and at variable head, by saturating the well and by measuring the consequent drawdown for no longer than 30 minutes. The hydraulic conductivity has been obtained in this case by relation: k=(a/c 1 (t 1 -t 2 ))ln(h 1 /h 2 ) where A is the area of the borehole, h 1 and h 2 are the water levels as function of t 1 and t 2 and C 1 is a shape coefficient. The Slug test consists of measuring the static water level (head) in the well, then introducing a near instantaneous change in water level by adding or

3 Brownfield Sites II 33 removing a volume of water or solid from the well, and measuring the change in water level over time until it returns to the original static water level. The measurement of the variations of water level in the well have been executed continuously by means of a pressure transducer which has a resolution of 0,1 cm and an interval of measurement of 1 sec. The data obtained has been elaborated in graphs, and have been placed so that the variations in water level from the beginning of the test were in the axis of ordinates in arithmetic scale, and the time in the axis of abscissas in logarithmic scale. The graphs obtained have been superimposed to datum standard curves [7], up to the definition of the best interpolation. 3 Data analysis Determination of the hydraulic properties of aquifers and confining units is critical to our understanding of groundwater flow and the development of groundwater flow models. The slug test is currently the most common method for the in situ estimation of hydraulic conductivity at sites of suspected groundwater contamination. However, inappropriate procedures in one or more phases of a slug test can introduce considerable error into the resulting parameter estimates. For this reason, the results of the slug test have been subjected to a specific preliminary check, through theoretic approaches, in order to establish the reliability of the order of magnitude of the obtained data. The aim of the investigation is that of quantifying the permeability values with analytical methods, in order to compare the magnitude order of those results with the one obtained by the experimental tests made in the site. In order to realize a preliminary check on the experimental values, a reference has initially been made to Theis theoretical treatment [1], whose relation valid for confined acquifers is extensible for determinate conditions for water bearing ones. That relation represents a traditional approach for the interpretation of the pumping tests; in this specific case, this solution has been used for a preliminary check of the interpretation of the Slug Tests. The data necessary for those elaborations have been taken from the technical sheets concerning the single tests. Those data, through the implementation of a calculation spreadsheet for each single test conducted, have been opportunely graphed in order to facilitate the comparison with the theoretical ones obtained by Theis expression. Through an iterative method for each test it has been possible to obtain a value T in correspondence of which the theoretical curve coincides, or at least has a very close trend to the curve obtained, by the experimental values. (Fig.1a). The calibrations have been repeated for the 17 conducted slug tests. It has been possible to check that in one case, the result of the test was not very significant (because there has been a probable depletion of the slug in a part of the test that was placed in the unsaturated zone), circumstance that doesn t allow a valid estimation of the conductivity (Fig 1b). From the comparison of the

4 34 Brownfield Sites II results of the experimental values of permeability with the ones obtained theoretically through the above mentioned treatment, it results evident that, except for some piezometers, the experimental permeability values draw away from the theoretical ones by orders of magnitude; for this reason a check of the results of the Slug Tests has been carried out by means of relations taken from literature [2], [3]. Figure 1: Trend of the curves s(t): a)calibration of a test considered valid; b) not representative result of the test. By means of a spreadsheet [5] it has been possible to obtain the values of the aquifer storage and of the hydraulic conductivity for each piezometer [2]. In Fig. 2(a) an example of this comparison is reported: the results are the more reliable, the more the theoretic curves approximate the points representing the experimental estimations. Also in this case, as far as the estimation of conductivity is concerned, the scarce significativity of some tests could be pointed out: Fig. 2(b) is an example of test whose results are to be rejected, so their respective K value will not be considered in the following calculations. By the look of the order of magnitude obtained for hydraulic conductivity, we would consider valid those calibrations carried out by using the relations of Cooper and Greene, and surely more reliable compared to the ones made through the theory of Theis. For the calibration of Lefranc tests, the relations proposed by Theis [6] and by Jacob [4] have been used, which are valid in the pumping tests for confined aquifers, but extensible for determinate conditions for water bearing ones. Nevertheless, as the part of soil interested by the tests is unsaturated, it hasn t been possible to compare the experimental points with the curves obtained analytically, so the phenomenon has been interpreted with the Nasberg-

5 Brownfield Sites II 35 Terletskata method; the value of the coefficient of permeability (K) comes from the application of the following formula: K 1 = ( 0,423/ h 2 )*Q*Log(4 h/d) where: h = water level (in m) in the absorbent part of borehole; Q = discharge injected (m 3 /sec); d = diameter of the absorbent part of borehole (in m). The results obtained for all the tests are very close to the ones given by the experimental tests; as a consequence of that all the values obtained experimentally through Lefranc tests have been assumed valid. 1,00 1,00 0,75 0,75 y/y0 0,50 y/y0 0,50 0,25 0,25 0, , TIME, Minute:Second TIME, Minute:Second Figure 2: Values of S and K obtained for a specific test a) considered valid b) considered not valid. 4 The map of hydraulic conductivity In the environmental field, in which the information is generally constituted by few observations irregularly distributed in the examined domain, the description of the spatial continuity (in other words the analysis of the similarity or dissimilarity among couples of values in function of their distance), represents the first fundamental phase in order to reach the estimation and thus the construction of the spatial variable object of study on the entire domain. In Geostatistics (science that permits the statistical treatment of data coming from surveys) the spatial correlation is described by the variogram.

6 36 Brownfield Sites II The variogram represents an experimental diagram that measures the diversity of the values of the considered variable the more the distance among the couples of points increases. In particular the ordinates of such diagrams represent the mean differences among the couples of data γ(h), separated by a h vector; that value is equal to the square mean difference among the components of the couples of data. The primary unit, the Lag, which represents the ray of the ideal circular surface in analysis inside which the mean difference γ(h) among the couples of points contained in it, has been varied in an interval between mm and mm, so that the results obtained will result significant. Figure 3: Grid of points located inside the site. It is possible to visualize the contour of the contaminated area and the localization of the effectuated tests. Among the different variograms obtained the one relating to the Lag = 200 m has been considered the most significant. On this variogram an operation of

7 Brownfield Sites II 37 fitting has been carried out, so that it has been possible to obtain the theoretical curve that better interpolates the experimental data; the value of the range assumed is about mm. In reality from the operation of fitting two theoretical curves have been determined, the former Gaussian and the latter exponential. For the subsequent operation of Kriging the latter has been used, taking into consideration that the Gaussian model is generally unreal and leads to instable and adulterated Kriging systems in the estimated maps. In order to realize graphically, and thus make visible a reliable reconstruction of the map of permeability of our site, it has been necessary to create a grid (Fig.3) of 5692 points of which only 45 have a known value of permeability. By using the program it has been possible to solve the equations of Kriging. The so obtained interpolation could be defined exact, in fact it is easy to demonstrate that, if the localization in which the value of hydraulic conductivity has to be estimated coincides with one of the sampled locations, the Kriging system gives an estimation which is equal to the observed value in that location and a variance equal to zero. On the other end the accuracy of this method is assured by the fact that it takes into account of the number of observations, the quality of data, the position of the points inside the examined domain, the distance among the points and the regularity of the spatial variable (K) object of study. 5 Obtained results The result of our punctual estimation permits us to associate, for each point of the realized grid, a particular value of conductivity. Fig. 4 represents graphically the so obtained map of hydraulic conductivity. A first interpretation of the map makes it possible to assert that the site shows a zone upwards characterized by low values of conductivity, thus potentially less vulnerable to pollution. Here, the possible release of polluting substances would cause a relatively slow contamination and a propagation that takes place at molecular level, for diffusion or absorption. The zone in the lower part to the left appears characterized by higher values of permeability, especially in a localized area. This area represents a potential risk for groundwater because the aquifer is not confined. In this area, an accidental release could cause the propagation of the contamination even at microscopic level, for convection or hydrodynamic dispersion. The preventive localization of these potentially dangerous areas could require the setting up of specific intervention finalized to guarantee the safeguard of the environmental quality and the safety of human health. The results of Kriging are also compatible to the geological conditions of the underground. In fact, the industrial area is also interested by a morphologic drop in which it is possible to determine two levels, the superior one, less vulnerable and the inferior one, more permeable, characterized by much lower altitude.

8 38 Brownfield Sites II In the aquifer, constituted of sandy silt and silty sand, the amount of groundwater is scarce and is substained by a wide impermeable level. Figure 4: Map of hydraulic conductivities K (cm/sec). The analysis carried out could represent the basis for a hydrogeologic modelling, through which it is possible to realize more precise studies useful for the implementation of the hydrodynamic model of the aquifer. References [1] Beretta G.P., (1992) Idrogeologia per il disinquinamento delle acque sotterranee. Pitagora Editrice Bologna. [2] Cooper H, H., Jr., Bredehoeft, J.D., and Papadopulos, I.S., (1967), Response of a finite-diameter well to an instantaneous charge of water. Water Resources Research v. 3, [3] Greene E.A., Shapiro A.M., (1995). Methods of conducting airpressurized slug tests and computation of type curves for estimating transmissivity and storativity: U.S. Geological Survey open- File Report [4] Jacob C.E.(1946): Radial flow in a leaky artesian aquifer, Trans. Amer. Geoph. Union, vol.27.

9 Brownfield Sites II 39 [5] Keith J. Halford and Eve L. Ku niansky (2002) Documentation of spreadsheets for the analysis of aquifer-test and slug-test data U.S. Geological Survey. Report Nevada. [6] Theis C. V. (1935): The relation between the lowering of the piezometric surface and the rate and duration of discaharge of a well using groundwater storage, Trans. Amer.Geoph. Union, Nat. Res. Council, 16 th ann. Meeting, Washington. [7] Papadopoulos, I.S., Bredehoeft J.D., and Cooper H.H. (1973) On the analysis of slug test data. Water Resources Res., 9, pp