- Email: [email protected]
GPR exploration for groundwater in a crystalline rock terrain
Journal of Applied Geophysics 55 (2004) 239 – 248 www.elsevier.com/locate/jappgeo
GPR exploration for groundwater in a crystalline rock terrain Jandyr de Menezes Travassos a,1, Paulo de Tarso Luiz Menezes b,* a b
CNPq-Observato´rio Nacional, Rua General Jose´ Cristino 77, 20921-400, Rio de Janeiro, Brazil GEFEX/FGEL/UERJ, Rua Sa˜o Francisco Xavier 524-4006A, 20550-013, Rio de Janeiro, Brazil Received 21 October 2002; accepted 19 January 2004
Abstract The Ground Penetrating Radar (GPR) method has been extensively used to map shallow subsurface features. Such kind of information is very important for different types of studies, ranging from archaeological to groundwater search. Almost all GPR surveys for groundwater exploration are usually conducted in sedimentary terrains. In this paper, we demonstrate the applicability of the GPR method in the exploration of underground water in a crystalline terrain. An example of fixed offset data collected at one known spring in the district of Petro´polis (Brazil) is given. Common Mid Point (CMP) data were also collected to estimate the velocity of the radar waves. The saturated region is clearly outlined in the GPR section as a zone of attenuation and inversion of the wavelet phase polarity. D 2004 Elsevier B.V. All rights reserved. Keywords: GPR; Groundwater; Crystalline rocks; Geophysical prospecting
1. Introduction Fresh water is a valuable resource for human needs, specially in Brazil, which has a large portion of its territory affected by a semiarid climate, and where droughts are frequent events. To supply the great demand, the Brazilian market for mineral water has increased in the last few years at a rate of 3% per year (DNPM, 1997). The main producing regions are located in the southeastern part of the country (Fig. 1a) in the States of Sa˜o Paulo, Minas Gerais and Rio de Janeiro (Martins et al., 1997).
* Corresponding author. Tel./fax: +55-21-587-7598. E-mail addresses: [email protected] (J. de Menezes Travassos), [email protected] (P. de Tarso Luiz Menezes). 1 Tel.: +55-21-5807081; fax: +55-21-5853782. 0926-9851/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jappgeo.2004.01.001
The available data indicates that the majority of the Brazilian groundwater resources are located in faults and fractures in crystalline rock (DNPM, 1997). It is important to mention here that about 60% of the Brazilian territory, or about 4,600,000 km2, consists of crystalline rocks. This indicates the great potential for mineral water exploration in this type of terrain. Traditionally, electrical and electromagnetic methods are the most popular geophysical tools for groundwater exploration in crystalline rocks (Meju, 2002, and references therein). The Ground Penetrating Radar (GPR) method has been used extensively in hydrogeological exploration (Van Overmeeren, 1994; Beres and Haeni, 1991) and soil studies (Aranha et al., 2002). In particular, it has demonstrated its effectiveness in mapping aquifers in sedimentary rocks (Cardimona et al., 1998). As far as we are aware, however, the literature provides no example of the GPR method
240
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
Fig. 1. (a) Map of Brazil with the main producing States (Minas Gerais (MG), Sa˜o Paulo (SP) and Rio de Janeiro (RJ) in the figure). (b) Location of the studied spring (black square in the center portion of the map) superimposed on a geologic map of the region (DNPM, 1998). ´ rga˜os Batholith. Prn—Rio Negro Complex. Sr—Serra dos O
being used for groundwater exploration in crystalline rocks. The imaging of 3D subsurface structures, as fractured reservoirs in crystalline environment, present a higher complexity than the interpretation of the traditional horizontal/sub-horizontal reflector of the water bearing level at sedimentary terrains. A similar situation occurs with the seismic method, that is widely used for exploring sedimentary basins, but still not for mineral and groundwater exploration in complex terrains. Recent advances in seismic imaging are overcoming these difficulties (Berrer et al., 2000; Drumond et al., 2000).
In the present work, we demonstrate that the GPR method can be highly effective in the groundwater exploration in crystalline rocks. A case study at a known spring is presented.
2. Geological setting The studied area is located within the district of Petro´polis in the State of Rio de Janeiro (Fig. 1b). The topography is uneven being on the scarp of Serra do Mar chain of mountains, reaching an altitude of 2000
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
m. The formation of the relief is related to Gondwana break-up, which began in Middle Cretaceous, with a strong increment at the Early Tertiary (Tupinamba´ et al., 2000). The main tectonic unit of the region is the Serra dos ´ rga˜os batholith (see Fig. 1b). That batholith is made O up of granite and gneiss with a granitic and granodioritic composition. The surrounding rocks belong to the Rio Negro Complex, an assemblage of migmatites, gneisses, gneiss-schist with interleaved quartzite layers (paragneiss). The main rock types in the area are the migmatites and high grade metamorphosed (Amphibolite facies) gneiss-schist, both classified as the Santo Aleixo Unit (DNPM, 1998). The migmatites frequently exhibit estromatic, flebitic and schollen structures and are mainly composed of biotite –hornblende gneiss/ amphibolite with interleaved pegmatite layers. The gneiss frequently exhibit structures and foliations in anti-form and syn-form with two independent folding phases. Due to the high metamorphism and intense folding it is very difficult to observe S0 (original sedimentary bedding). Low-angle dipping metamorphic foliation and several NW shear zones are also characteristic in the studied region (Tupinamba´ et al., 2000). Fig. 2 shows a migmatite outcrop of Santo Aleixo Unit in a nearby quarry. The hammer is positioned over a small leucossome boudin structure within a mafic layer. At the left corner of the picture a stromatic migmatitic structure, characterized by intercalation of felsic and mafic layers, is dominant. These
Fig. 2. Banded leucossome boudin (indicated by the hammer position) in a migmatite of Santo Aleixo Unit. A stromatic migmatite structure dominate in the left corner of the photo. These structures are typical of Santo Aleixo Unit (courtesy of Miguel Tupinamba´).
241
structures are very common in the studied region and occurs at all size scales, from centimeters to tens of meters. The occurrence of mineral water in the area is usually associated with zones of fractures in the banded and folded migmatite and paragneiss of the Santo Aleixo Unit (Fig. 2). The reader can find several examples of structures and rocks of Rio Negro Complex and Serra dos Orga˜os batholith elsewhere in the literature (Tupinamba´ et al., 2000).
3. Hydrogeological setting The hydrogeological knowledge of Rio de Janeiro State is still not well established. At the whole state the government has only 527 producing wells registered. In a first attempt to characterize the hydrogeological potential, CPRM (Brazilian Geological Survey) recently published the hydrogeological map of Rio de Janeiro State in the 1:500,000 scale (Barreto et al., 2000). This regional study was based on a multi-criteria analysis integrated in a Geographic Information System (GIS). The main parameters analyzed, in descending importance order, were: slope, fracture density (interpreted from LANDSAT-TM images), soil type, soil management, lithology and drainage density. The employed methodology was partially based on the work developed by Lanvegin et al. (1991) when studying a fractured granite in France. Two mains aquifers systems, sedimentary and crystalline, were identified by Barreto et al. (2000). In the crystalline system, our target, groundwater circulate through fractures and fissures. The storage capacity of a crystalline aquifer is related to the number and connectivity of fractures. Therefore, one of the most important parameter to analyze is the fracture density. This parameter is defined as the total length (L) of the existing fractures at an area (L2) divided by this area (L/L2). Five density ranges were defined by Barreto et al. (2000), from the lowest to highest, they are (in km/km2): 0 –0.36; 0.36 –0.72; 0.72– 1.08; 1.08 – 1.44; >1.44. The district of Petro´polis is located in an area of high fracture density range (1.08 – 1.44), therefore being a high favorably area for groundwater occurrence. Indeed, just one mineral water company installed in that district, is responsible for about 27% of the production of mineral water of Rio de Janeiro
242
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
State (Martins et al., 1997). The production zone, composed by four springs, is located at 1000 m of altitude, by the scarp of Serra do Mar mountain chain. This occurrence has been explored since 1953, with average annual production of 28,000,000 l (Martins et al., 1997). Physical and chemical characteristics of the mineral water can be found in the literature (Martins et al., op. cit.). The water is classified as radioactive at the spring, probably due to the radionuclide present in the percolated rocks. The water has a pH of 6.1 at 25 jC, a conductivity of 3.5 mS/m and 34.8 mg/l of dissolved salts in its composition.
4. Data acquisition Virtually all the data were collected keeping the transmitter and the receiver in a fixed offset configuration. A GPR section is made up of many traces collected along a profile thus allowing the observer to locate targets. Each new trace is obtained while dragging the two antennas together along the profile. The wavelets reflected from two targets appear on the corresponding traces plotted as a function of known as two-way time (TWT). This is the time it takes the pulse to be emitted, to bounce back, and to be recorded at the receiver. The TWT can be converted to depth when velocity of radar waves in the subsurface is known. An estimate of velocity can be achieved through velocity analysis of Common Mid Point (CMP) sections. In the CMP field configuration, the transmitter and receiver are moved away each other up to a maximum distance. This distance is a compromise between achieving greater investigation depths and consequent increased absorption of the electromagnetic waves in the medium with increased transmitter –receiver distance. The data were collected with a Pulse Ekko IV GPR (Davies and Annan, 1989) with a 1000-V pulser. The transmitter antenna radiates a broadband wavelet that has an amplitude spectrum as wide as its center frequency: 100 MHz. The transmitting antenna has a broad radiation pattern reaching apertures between 90j and 180j. The bi-static antenna configuration allows a great flexibility of data collection strategies. Antennas were dragged along the profiles with a constant step of 0.20 and 0.25 m, keeping their mutual distance constant (fixed offset) or increasing that distance continuously (CMP).
The total time window used in the fixed-offset profiles was 250 ns, enough to reach depths of 10– 12 m. That time window, however, was increased to 350 ns for the CMP profiles. The whole survey amounted to more than 500 m of fixed-offset reflection profiles mostly deployed on uneven terrain. The spatial density of traces was 0.25 m/trace throughout. The antennae were kept 1 m apart. A section of a 120-m-long profile on a relatively flat terrain and crossing the main one of the four known springs in the area is used here to illustrate the effectiveness of the GPR to locate water in fractures. A producing well was drilled at the spring till 145 m depth. The hydrostatic level is confined between 4.9 (static) and 62 (dynamic) m depth (Barreto et al., 2000). Two 25-m CMP profiles were done at two locations in the survey area in order to estimate an average velocity to be used in the fixed-offset sections. Antennae separation was increased stepwise in increments of 0.1 m from an initial separation of 1 m. The two CMP profiles yielded similar results. We show here the CMP profile done at the well location (Fig. 3a).
5. Velocity analysis The propagation velocity of radar waves, Vr, is a function of the dielectric constant of the subsurface. The dielectric constant, in turn, is affected by its water content. The propagation velocity can be determined by CMP measurements. From a CMP section, it is possible to infer the velocity of the direct waves (in air as well as the ground wave), the velocity of the reflected waves, and of the refracted waves in some special circumstances. Here, we concentrate on the velocity of reflected waves. The reflections from the interfaces between layers with different dielectric constants appear as hyperbolas in CMP sections (Fig. 3a). This characteristic shape is based on the assumption that the arrival time for signals from reflectors varies hyperbolically with the separation between the transmitter and the receiver. This assumption is valid as long as the reflectors have small dip. The curvature of a given hyperbola depends on the velocity of the radar waves. Therefore, the velocity analysis of a given hyperbola will give an average velocity to the depth of the reflector.
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
243
Fig. 3. (a) CMP profile at the studied spring. The airwave is represented by the first straight line in time. The groundwave is represented by the second straight line, note that beyond 10 m, this wave is highly attenuated. Subsurface reflectors are represented by hyperbolas. (b) Velocity analysis for the CMP profile. Velocity is incremented in 0.001 m/ns steps from 0.01 to 0.15 m/ns.
The velocity analysis used here uses the concept of velocity stack in a constant velocity earth (Yilmaz, 1987). CMP traces are compensated for normal moveout assuming a constant velocity hyperbolic equation. Traces are then stacked. A range of velocity from 0.05 to 0.15 m/ns was covered with increments of
0.001 m/ns. When a given velocity in that range matches the normal moveout velocity, a reflector will stack coherently. This will result in larger amplitude in the stack. On the other hand, when a given velocity does not match that of the reflector, traces add together incoherently. This will result in smaller amplitudes.
244
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
The velocity analysis using the CMP profile yields a three-layered velocity model of 0.065, 0.076 and 0.072 m/ns (Fig. 3b). We adopt an average velocity of Vr = 0.07 m/ns for the whole section, which is close to the estimate given by the first important reflector at 60 ns, arrowed in Fig. 3b. That reflector is probably the interface between the saprolite and the granite-gneiss below. As a matter of fact, the shallower part of the saprolite has a higher velocity (0.07 m/ns) than the remainder of the section, as estimated directly from the groundwave signature, a straight line in Fig. 3a. The adopted velocity is 54% less than it is usually tabulated for granite-gneiss (0.13 m/ns). This can be attributed to a higher water content in soil and rock. There is one continuous reflector at 150 ns, i.e., below 5 m deep, yielding the almost the same velocity as it can be seen in Fig. 3b. Note that the static hydrostatic level measured at the well is 4.9 m, corroborating our interpretation.
6. Groundwater signature The water in the area appears in fractures indicating that our target is not a long and continuous reflector as would be the case of a water table. Due to the fractures,
we expect that reflection sections will be cluttered with diffractions. Diffractions are hyperbolic events caused by the discontinuities in the rock that may interfere with each other obscuring the sections. Diffractions and interference patterns can be dealt with by using migration, an imaging technique that focuses the energy along hyperbolas to the true spatial position from which the energy originated. Migration moves dipping reflections to their true subsurface position and collapses diffractions, thus increasing spatial resolution. The goal of migration is to make the fixed-offset section appear similar to the geologic cross-section in depth along a GPR profile. The section presented in this paper (Fig. 4) is migrated assuming that the rocks below the saprolitic layer are more or less homogeneous. In this work, we use the f– k migration approach of Stolt (Yilmaz, 1987). It is assumed that the GPR data can be considered as zero offset and that the average velocity estimated through the velocity analysis above can be taken as a constant background velocity along the profile: Vr = 0.07 m/ns. We have also chosen a relatively flat profile to apply the migration. The data is dewowed, clipped, resampled and tapered before migration. A limitation in this procedure lies in the fact that our data is 2D and out-of-plane diffractions will be not
Fig. 4. (a) Raw GPR time section on crystalline rock, note the several diffraction patterns (bowties) below 100 ns, specially between traces 40 and 50 in the profile. (b) Same GPR section f – k migrated and converted to depth, an uniform velocity of Vr = 0.07 m/ns is assumed throughout. Compare with (a), migration untied the bowties on the section and turn them into dipping reflectors. A producing well, with static hydrostatic level at 4.9 m depth is located at trace 32 m.
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
correctly dealt with the migration process. That will harm our section smearing out the data at the fracture zone. Such effect can only be dealt with in a true 3D survey, where the inline and cross-line spatial sampling are compatible. In that case, one could possibly perform a two-step migration that is satisfactory for a simple subsurface. For a more involved structure, one would require a full 3D migration (Jakubowicz and Levin, 1983). Fig. 4a shows a 70-m raw data section clipped from the 120-m-long profile crossing the studied spring (trace 32 m). A spreading and exponential compensation gain assuming an attenuation of 0.1 dB/m was applied to the section. Data were low-pass filtered with a cutoff at 250 MHz to reduce highfrequency noise and clipped in time to chop off values earlier than the first break. The section was migrated assuming a constant velocity throughout (0.07 m/ns), the final result is shown in Fig. 4b. Water-filled fractured rock appears in the section as a region of lower amplitude reflections beyond trace 25 m and below 5 m depth. That region gets deeper towards the end of the section, extending to trace 60 m. Fracture induced diffraction were virtually eliminated by the energy focusing provided by the migration. The wavelet phase polarity can also be used to determine the presence of water. The phase of the wavelet is defined in this paper as the sequence of phase polarities as seen along a given trace. Fig. 5 shows the end portion of the trace obtained at 35 m
Fig. 5. End section of trace 35 m. The arrow shows the first of a series of phase inversions, occurring at the top section of the waterfilled fractured rock. Amplitudes are clipped at a maximum of 3.2. The vertical scale is arbitrary.
245
along the section shown in Fig. 4. The trace was AGC gained and appears clipped at an arbitrary maximum. The AGC attempts to equalize the signals by applying a gain which is inversely proportional to signal strength, and therefore does not preserve relative amplitude information. The first of a series of phase polarity inversions occurs at 178 ns as seen in Fig. 5. This gives the top of the water-filled fractured rock at 6.23 m, less than 1 m deeper than it appears to be in Fig. 4b. This discrepancy may be due to the uncertainty in recognizing the phase polarity sequence properly, because the leading edge of the wavelet cannot be ascertained due to closed spaced reflections, as may be the case on the top of the fractured rock.
7. Interpretation A great deal of effort in interpreting radar profiles goes into not only understanding reflections and diffractions but also into deciphering of interference patterns. The focusing of energy provided by migration make that task easier. The expected product is to recognize changes in reflector characteristics such as configuration, continuity, frequency and amplitude so to characterize radar facies. Radar facies are 3D regions representing particular combinations of physical properties like lithology, stratification, fracturing and fluid contents. Recovering the full geometry of such regions is not an issue here as we are dealing with a 2D profile, but with the data we can have an idea of the layering of the gneiss and recover the contact between less and more fractured rock. The section in Fig. 4 gives a good image of the subsurface, revealing features such as the top portion of the water-filled fractured rock and the structure of the gneiss. The interpreted section is shown in Fig. 6. The most prominent feature is the top of the fractured rock delimiting an extensive saturated zone (35 m wide), which is responsible for the high drainage rate of the spring. This interpretation is corroborated by the existence of the producing well at trace 32 m. In the radar section of Fig. 4b, it is possible to identify a series of semi-parallel folded reflectors all long the profile and small structures such as a lens shape between 0 and 30 m in the profile that is very similar to a boudin structure (B in Fig. 6). As said before these structures (Fig. 2) are very common in
246
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
Fig. 6. Interpretation of the radar section shown in Fig. 4. The F’s are interpreted as fractures and B is interpreted as a boudin structure, very common in the region, refer to Fig. 2.
the region. The overall pattern of the reflectors along the profile resembles a migmatitic texture, with alternated veined/folded mafic (biotite– hornblende) and felsic layers (quartz). Another important feature identified are the sub-vertical fractures (F in Fig. 6) that occur all long the section. These fractures are associated to the main conduits to water ascension in the area. Independent geophysical data is in accordance with the GPR data. 1D Interpretation of one VES positioned at trace 42 m along the profile indicate a four-layer earth (S. Berrino, personal communication, 1997). Schlumberger soundings, due to its low costs and simplicity of use, are still traditionally employed in many regions to enhance the odds of finding groundwater in crystalline terrains (Carrasquilla et al., 1997). The main idea behind these surveys is, after identifying a favorable fracture in aerial photo-interpretation, to try to find lower resistivity zones within high resistive rocks. That lowering in the resistivity pattern should be associated to groundwater. The success rate of that type of approach reaches 85 – 90% of the drilled boreholes. To further reduce the ambiguity of the geophysical interpretation of the shallow subsurface we incorporated the geological knowledge of the weathering profile in the region (Barreto et al., 2000). Descriptions of the typical weathering profiles of tropical and subtropical basement complexes are found elsewhere in the literature (refer to Fig. 4, p. 141 of Meju, 2002). The 1D model at the spring starts from top to bottom with a 2-m saprolite layer of 1500 V m, followed by a 5-m fractured layer of 3000 V m and
then to a fractured saturated zone of 300 V m that may extend down to 50 m, well beyond the penetration depth of the GPR data. This 10 times lowering in the resistivity values indicates a high degree of fracturing and connectivity. A half-space of highly resistive fresh rock terminates the model. The resistivity of the saturated zone is within the range of common fractured aquifers at igneous and metamorphic rocks (refer to Fig. 5, p. 142 of Meju, 2002). It is interesting to note that this simple 1D interpretation presented a reasonable estimate of the top of the saturated zone (7 m) when compared to the borehole log and the radar section (5.7 m). The main discrepancy, about 20% error, is associated to the bottom of the saturated zone, that was estimated in 50 m by the 1D inversion and is 62 m at the borehole. This is comprehensive as we are interpreting with a 1D method a complex 3D earth. But as stated before, for exploration purposes, the main interest is just identify a zone of low resistivity within crystalline rocks that could associated to groundwater, so enhancing the chances of a successful drilling.
8. Conclusions This work presents the results from a GPR survey done for water exploration on crystalline terrain (migmatized gneiss). Reflection data were collected with the fixed-offset configuration along a profile 70 m long. The velocity was estimated doing velocity analysis on CMP data. The estimated velocity value was 0.07 m/ns. This value is 54% less than the tabulated velocity for granite-gneiss and is attributed to high water content in rock.
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
The raw GPR section was cluttered with diffractions. A migration scheme was applied assuming the constant value of 0.07 m/ns throughout. The energy focusing provided by migration can be used to remove/attenuate interference patterns and make the GPR section appear similar to the geologic crosssection in depth. As expected, the fracture induced diffractions were virtually eliminated in the migrated section, enhancing the visualization of the geologic structures. In the migrated radar profile the water filled fractured zone is clearly outlined by a wide (35 m) region of lower amplitude values. Those are result of the scattering of the incident energy as well as due to the smearing out resultant from the process of migrating out-of-plane diffractions. The depth to the top of this saturated zone is variable along the profile (4– 10 m). The presence of water in the fractured zone can also be mapped from inversion on the polarity of the wavelet phase at the radar traces. To further reduce the inherent ambiguity of the geophysical interpretation a priori geological knowledge of a borehole located at trace 32 m was incorporated. In this borehole the water producing zone is confined within 4.9 and 62 m depth. There is a very good agreement between the GPR data and the available borehole information. A priori knowledge of the weathering profile and migmatite structures of Santo Aleixo Unit were also incorporated to help in the geological interpretation of the radar profile. The main limitation of the method is the relative shallow depth of investigation. The GPR data provided a detailed characterization of shape and extension of the shallow saturated zone. Depth of investigation was limited to 10 m, insufficient to image the whole saturated zone (till 62 m depth). The imaging of greater depths can be achieved by integrating the GPR data with another geophysical method. In our case, independent geophysical data from a VES Schlumberger sounding are in accordance with the GPR results and the available geological knowledge.
Acknowledgements We would like to thank the editor-in-chief A. Hoerdt, S. Hautot and an anonymous reviewer for
247
their constructive comments that helped to improve the text. We also thank Miguel Tupinamba´ for the photo of Santo Aleixo migmatite. PTLM was supported by an UERJ-Procieˆncia scholarship.
References Aranha, P.R.A., Augustin, C.H.R.R., Sobreira, F.G., 2002. The use of GPR for characterizing underground weathered profiles in the sub-humid tropics. J. Appl. Geophys. 49, 195 – 210. Barreto, A.B.C., Monsores, A.L.M., Leal, A.S., Pimentel, J., 2000. Hidrogeologia do Estado do Rio de Janeiro. Mapa de Favorabilidade Hidrogeolo´gica na escala 1:500000 e texto explicativo. CPRM-Servicßo Geolo´gico do Brasil, Rio de Janeiro, CD-ROM. Beres, M., Haeni, F.P., 1991. Application of ground-penetratingradar methods in hydrogeologic studies. Ground Water 29, 375 – 386. Berrer, E.K., King, A.R., Watts, A.H., Roberts, B., Adam, E., Eaton, D.W., Wu, J., Salisbury, M.H., 2000. Development of 3-D seismic exploration technology for deep nickel – copper deposits: a case history from the Sudbury basin, Canada. Geophysics 65, 1890 – 1899. Cardimona, S., Clement, W.P., Kadinsky-Cade, K., 1998. Seismic reflection and ground-penetrating radar imaging of a shallow aquifer. Geophysics 63, 1310 – 1317. Carrasquilla, A., Porsani, M.J., Tavares, A., 1997. Prospeccß a˜o de a´guas subterraˆneas no Alto Xigu-Para com me´todos geofı´sicos eletromagne´ticos. Rev. Bras. Geocieˆnc. 27, 221 – 228. Davies, J.L., Annan, A.P., 1989. Ground-penetrating radar for highresolution mapping of soil and rock stratigraphy. Geophys. Prospect. 37, 531 – 551. DNPM, 1997. Principais depo´sitos minerais do Brasil. v4—Rochas e minerais industriais. In: Schobbenhaus, C, Queiro´z, E.T., Coelho, C.E.S. (Coordenadores), Coordenador. Departamento Nacional de Producßa˜o Mineral (DNPM). Brası´lia. pp. 9 – 38. DNPM, 1998. Mapa geolo´gico do Estado do Rio de Janeiro, escala 1:400000. In: Fonseca, M.J.G. (Coordenador), Departamento Nacional de Producßa˜o Mineral. 141 pp. Drumond, B.J., Goleby, B.R., Owen, A.J., Yeates, A.N., Swager, C., Zhang, Y., Jackson, J.K., 2000. Seismic reflection imaging of mineral systems: three case histories. Geophysics 65, 1852 – 1861. Jakubowicz, H., Levin, S., 1983. A simple exact method of 3-D migration—theory. Geophys. Prospect. 31, 34 – 56. Lanvegin, C., Pernel, F., Pointet, T., 1991. Aide a` la de´cision en matie`re de prospection hydroge´ologique—L’analyse muticrite`re au service de l’e´valuation du potenciel aquife`re, en milieu fissure´ (granite de Huelgoat, Finiste`re, France). Hydroge´ologie 1 (64 pp.). Martins, A.M., Maurı´cio, R.C., Pereira Filho, J.C., Caetano, L.C., ´ guas Minerais do Estado do Rio de Janeiro. Departa1997. A mento de Recursos Minerais (DRM), Nitero´i. 90 pp. Meju, M.A., 2002. Geoelectomagnetic exploration for natural resources: models, cases studies and challenges. Surv. Geophys. 23, 133 – 205.
248
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
Tupinamba´, M., Penha, H.M., Bustamante Junho, M.C., 2000. Arcrelated to post-collisional magmatism at Serra dos Orga˜os region, Rio de Janeiro State, Brazil: products of Gondwana assembly, during the Brasiliano – Pan African orogeny. Field Trip guide of 31st International Geological Congress, Rio de Janeiro, Brazil. 30 pp.
Van Overmeeren, R.A., 1994. Georadar for hydrogeology. First Break 12 (8), 401 – 408. Yilmaz, O., 1987. Seismic data processing. In: Doherty, S.M. (Ed.), SEG Series: Investigations in Geophysics, vol. 2. SEG, Tulsa, OK. 526 pp.
GPR exploration for groundwater in a crystalline rock terrain Jandyr de Menezes Travassos a,1, Paulo de Tarso Luiz Menezes b,* a b
CNPq-Observato´rio Nacional, Rua General Jose´ Cristino 77, 20921-400, Rio de Janeiro, Brazil GEFEX/FGEL/UERJ, Rua Sa˜o Francisco Xavier 524-4006A, 20550-013, Rio de Janeiro, Brazil Received 21 October 2002; accepted 19 January 2004
Abstract The Ground Penetrating Radar (GPR) method has been extensively used to map shallow subsurface features. Such kind of information is very important for different types of studies, ranging from archaeological to groundwater search. Almost all GPR surveys for groundwater exploration are usually conducted in sedimentary terrains. In this paper, we demonstrate the applicability of the GPR method in the exploration of underground water in a crystalline terrain. An example of fixed offset data collected at one known spring in the district of Petro´polis (Brazil) is given. Common Mid Point (CMP) data were also collected to estimate the velocity of the radar waves. The saturated region is clearly outlined in the GPR section as a zone of attenuation and inversion of the wavelet phase polarity. D 2004 Elsevier B.V. All rights reserved. Keywords: GPR; Groundwater; Crystalline rocks; Geophysical prospecting
1. Introduction Fresh water is a valuable resource for human needs, specially in Brazil, which has a large portion of its territory affected by a semiarid climate, and where droughts are frequent events. To supply the great demand, the Brazilian market for mineral water has increased in the last few years at a rate of 3% per year (DNPM, 1997). The main producing regions are located in the southeastern part of the country (Fig. 1a) in the States of Sa˜o Paulo, Minas Gerais and Rio de Janeiro (Martins et al., 1997).
* Corresponding author. Tel./fax: +55-21-587-7598. E-mail addresses: [email protected] (J. de Menezes Travassos), [email protected] (P. de Tarso Luiz Menezes). 1 Tel.: +55-21-5807081; fax: +55-21-5853782. 0926-9851/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.jappgeo.2004.01.001
The available data indicates that the majority of the Brazilian groundwater resources are located in faults and fractures in crystalline rock (DNPM, 1997). It is important to mention here that about 60% of the Brazilian territory, or about 4,600,000 km2, consists of crystalline rocks. This indicates the great potential for mineral water exploration in this type of terrain. Traditionally, electrical and electromagnetic methods are the most popular geophysical tools for groundwater exploration in crystalline rocks (Meju, 2002, and references therein). The Ground Penetrating Radar (GPR) method has been used extensively in hydrogeological exploration (Van Overmeeren, 1994; Beres and Haeni, 1991) and soil studies (Aranha et al., 2002). In particular, it has demonstrated its effectiveness in mapping aquifers in sedimentary rocks (Cardimona et al., 1998). As far as we are aware, however, the literature provides no example of the GPR method
240
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
Fig. 1. (a) Map of Brazil with the main producing States (Minas Gerais (MG), Sa˜o Paulo (SP) and Rio de Janeiro (RJ) in the figure). (b) Location of the studied spring (black square in the center portion of the map) superimposed on a geologic map of the region (DNPM, 1998). ´ rga˜os Batholith. Prn—Rio Negro Complex. Sr—Serra dos O
being used for groundwater exploration in crystalline rocks. The imaging of 3D subsurface structures, as fractured reservoirs in crystalline environment, present a higher complexity than the interpretation of the traditional horizontal/sub-horizontal reflector of the water bearing level at sedimentary terrains. A similar situation occurs with the seismic method, that is widely used for exploring sedimentary basins, but still not for mineral and groundwater exploration in complex terrains. Recent advances in seismic imaging are overcoming these difficulties (Berrer et al., 2000; Drumond et al., 2000).
In the present work, we demonstrate that the GPR method can be highly effective in the groundwater exploration in crystalline rocks. A case study at a known spring is presented.
2. Geological setting The studied area is located within the district of Petro´polis in the State of Rio de Janeiro (Fig. 1b). The topography is uneven being on the scarp of Serra do Mar chain of mountains, reaching an altitude of 2000
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
m. The formation of the relief is related to Gondwana break-up, which began in Middle Cretaceous, with a strong increment at the Early Tertiary (Tupinamba´ et al., 2000). The main tectonic unit of the region is the Serra dos ´ rga˜os batholith (see Fig. 1b). That batholith is made O up of granite and gneiss with a granitic and granodioritic composition. The surrounding rocks belong to the Rio Negro Complex, an assemblage of migmatites, gneisses, gneiss-schist with interleaved quartzite layers (paragneiss). The main rock types in the area are the migmatites and high grade metamorphosed (Amphibolite facies) gneiss-schist, both classified as the Santo Aleixo Unit (DNPM, 1998). The migmatites frequently exhibit estromatic, flebitic and schollen structures and are mainly composed of biotite –hornblende gneiss/ amphibolite with interleaved pegmatite layers. The gneiss frequently exhibit structures and foliations in anti-form and syn-form with two independent folding phases. Due to the high metamorphism and intense folding it is very difficult to observe S0 (original sedimentary bedding). Low-angle dipping metamorphic foliation and several NW shear zones are also characteristic in the studied region (Tupinamba´ et al., 2000). Fig. 2 shows a migmatite outcrop of Santo Aleixo Unit in a nearby quarry. The hammer is positioned over a small leucossome boudin structure within a mafic layer. At the left corner of the picture a stromatic migmatitic structure, characterized by intercalation of felsic and mafic layers, is dominant. These
Fig. 2. Banded leucossome boudin (indicated by the hammer position) in a migmatite of Santo Aleixo Unit. A stromatic migmatite structure dominate in the left corner of the photo. These structures are typical of Santo Aleixo Unit (courtesy of Miguel Tupinamba´).
241
structures are very common in the studied region and occurs at all size scales, from centimeters to tens of meters. The occurrence of mineral water in the area is usually associated with zones of fractures in the banded and folded migmatite and paragneiss of the Santo Aleixo Unit (Fig. 2). The reader can find several examples of structures and rocks of Rio Negro Complex and Serra dos Orga˜os batholith elsewhere in the literature (Tupinamba´ et al., 2000).
3. Hydrogeological setting The hydrogeological knowledge of Rio de Janeiro State is still not well established. At the whole state the government has only 527 producing wells registered. In a first attempt to characterize the hydrogeological potential, CPRM (Brazilian Geological Survey) recently published the hydrogeological map of Rio de Janeiro State in the 1:500,000 scale (Barreto et al., 2000). This regional study was based on a multi-criteria analysis integrated in a Geographic Information System (GIS). The main parameters analyzed, in descending importance order, were: slope, fracture density (interpreted from LANDSAT-TM images), soil type, soil management, lithology and drainage density. The employed methodology was partially based on the work developed by Lanvegin et al. (1991) when studying a fractured granite in France. Two mains aquifers systems, sedimentary and crystalline, were identified by Barreto et al. (2000). In the crystalline system, our target, groundwater circulate through fractures and fissures. The storage capacity of a crystalline aquifer is related to the number and connectivity of fractures. Therefore, one of the most important parameter to analyze is the fracture density. This parameter is defined as the total length (L) of the existing fractures at an area (L2) divided by this area (L/L2). Five density ranges were defined by Barreto et al. (2000), from the lowest to highest, they are (in km/km2): 0 –0.36; 0.36 –0.72; 0.72– 1.08; 1.08 – 1.44; >1.44. The district of Petro´polis is located in an area of high fracture density range (1.08 – 1.44), therefore being a high favorably area for groundwater occurrence. Indeed, just one mineral water company installed in that district, is responsible for about 27% of the production of mineral water of Rio de Janeiro
242
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
State (Martins et al., 1997). The production zone, composed by four springs, is located at 1000 m of altitude, by the scarp of Serra do Mar mountain chain. This occurrence has been explored since 1953, with average annual production of 28,000,000 l (Martins et al., 1997). Physical and chemical characteristics of the mineral water can be found in the literature (Martins et al., op. cit.). The water is classified as radioactive at the spring, probably due to the radionuclide present in the percolated rocks. The water has a pH of 6.1 at 25 jC, a conductivity of 3.5 mS/m and 34.8 mg/l of dissolved salts in its composition.
4. Data acquisition Virtually all the data were collected keeping the transmitter and the receiver in a fixed offset configuration. A GPR section is made up of many traces collected along a profile thus allowing the observer to locate targets. Each new trace is obtained while dragging the two antennas together along the profile. The wavelets reflected from two targets appear on the corresponding traces plotted as a function of known as two-way time (TWT). This is the time it takes the pulse to be emitted, to bounce back, and to be recorded at the receiver. The TWT can be converted to depth when velocity of radar waves in the subsurface is known. An estimate of velocity can be achieved through velocity analysis of Common Mid Point (CMP) sections. In the CMP field configuration, the transmitter and receiver are moved away each other up to a maximum distance. This distance is a compromise between achieving greater investigation depths and consequent increased absorption of the electromagnetic waves in the medium with increased transmitter –receiver distance. The data were collected with a Pulse Ekko IV GPR (Davies and Annan, 1989) with a 1000-V pulser. The transmitter antenna radiates a broadband wavelet that has an amplitude spectrum as wide as its center frequency: 100 MHz. The transmitting antenna has a broad radiation pattern reaching apertures between 90j and 180j. The bi-static antenna configuration allows a great flexibility of data collection strategies. Antennas were dragged along the profiles with a constant step of 0.20 and 0.25 m, keeping their mutual distance constant (fixed offset) or increasing that distance continuously (CMP).
The total time window used in the fixed-offset profiles was 250 ns, enough to reach depths of 10– 12 m. That time window, however, was increased to 350 ns for the CMP profiles. The whole survey amounted to more than 500 m of fixed-offset reflection profiles mostly deployed on uneven terrain. The spatial density of traces was 0.25 m/trace throughout. The antennae were kept 1 m apart. A section of a 120-m-long profile on a relatively flat terrain and crossing the main one of the four known springs in the area is used here to illustrate the effectiveness of the GPR to locate water in fractures. A producing well was drilled at the spring till 145 m depth. The hydrostatic level is confined between 4.9 (static) and 62 (dynamic) m depth (Barreto et al., 2000). Two 25-m CMP profiles were done at two locations in the survey area in order to estimate an average velocity to be used in the fixed-offset sections. Antennae separation was increased stepwise in increments of 0.1 m from an initial separation of 1 m. The two CMP profiles yielded similar results. We show here the CMP profile done at the well location (Fig. 3a).
5. Velocity analysis The propagation velocity of radar waves, Vr, is a function of the dielectric constant of the subsurface. The dielectric constant, in turn, is affected by its water content. The propagation velocity can be determined by CMP measurements. From a CMP section, it is possible to infer the velocity of the direct waves (in air as well as the ground wave), the velocity of the reflected waves, and of the refracted waves in some special circumstances. Here, we concentrate on the velocity of reflected waves. The reflections from the interfaces between layers with different dielectric constants appear as hyperbolas in CMP sections (Fig. 3a). This characteristic shape is based on the assumption that the arrival time for signals from reflectors varies hyperbolically with the separation between the transmitter and the receiver. This assumption is valid as long as the reflectors have small dip. The curvature of a given hyperbola depends on the velocity of the radar waves. Therefore, the velocity analysis of a given hyperbola will give an average velocity to the depth of the reflector.
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
243
Fig. 3. (a) CMP profile at the studied spring. The airwave is represented by the first straight line in time. The groundwave is represented by the second straight line, note that beyond 10 m, this wave is highly attenuated. Subsurface reflectors are represented by hyperbolas. (b) Velocity analysis for the CMP profile. Velocity is incremented in 0.001 m/ns steps from 0.01 to 0.15 m/ns.
The velocity analysis used here uses the concept of velocity stack in a constant velocity earth (Yilmaz, 1987). CMP traces are compensated for normal moveout assuming a constant velocity hyperbolic equation. Traces are then stacked. A range of velocity from 0.05 to 0.15 m/ns was covered with increments of
0.001 m/ns. When a given velocity in that range matches the normal moveout velocity, a reflector will stack coherently. This will result in larger amplitude in the stack. On the other hand, when a given velocity does not match that of the reflector, traces add together incoherently. This will result in smaller amplitudes.
244
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
The velocity analysis using the CMP profile yields a three-layered velocity model of 0.065, 0.076 and 0.072 m/ns (Fig. 3b). We adopt an average velocity of Vr = 0.07 m/ns for the whole section, which is close to the estimate given by the first important reflector at 60 ns, arrowed in Fig. 3b. That reflector is probably the interface between the saprolite and the granite-gneiss below. As a matter of fact, the shallower part of the saprolite has a higher velocity (0.07 m/ns) than the remainder of the section, as estimated directly from the groundwave signature, a straight line in Fig. 3a. The adopted velocity is 54% less than it is usually tabulated for granite-gneiss (0.13 m/ns). This can be attributed to a higher water content in soil and rock. There is one continuous reflector at 150 ns, i.e., below 5 m deep, yielding the almost the same velocity as it can be seen in Fig. 3b. Note that the static hydrostatic level measured at the well is 4.9 m, corroborating our interpretation.
6. Groundwater signature The water in the area appears in fractures indicating that our target is not a long and continuous reflector as would be the case of a water table. Due to the fractures,
we expect that reflection sections will be cluttered with diffractions. Diffractions are hyperbolic events caused by the discontinuities in the rock that may interfere with each other obscuring the sections. Diffractions and interference patterns can be dealt with by using migration, an imaging technique that focuses the energy along hyperbolas to the true spatial position from which the energy originated. Migration moves dipping reflections to their true subsurface position and collapses diffractions, thus increasing spatial resolution. The goal of migration is to make the fixed-offset section appear similar to the geologic cross-section in depth along a GPR profile. The section presented in this paper (Fig. 4) is migrated assuming that the rocks below the saprolitic layer are more or less homogeneous. In this work, we use the f– k migration approach of Stolt (Yilmaz, 1987). It is assumed that the GPR data can be considered as zero offset and that the average velocity estimated through the velocity analysis above can be taken as a constant background velocity along the profile: Vr = 0.07 m/ns. We have also chosen a relatively flat profile to apply the migration. The data is dewowed, clipped, resampled and tapered before migration. A limitation in this procedure lies in the fact that our data is 2D and out-of-plane diffractions will be not
Fig. 4. (a) Raw GPR time section on crystalline rock, note the several diffraction patterns (bowties) below 100 ns, specially between traces 40 and 50 in the profile. (b) Same GPR section f – k migrated and converted to depth, an uniform velocity of Vr = 0.07 m/ns is assumed throughout. Compare with (a), migration untied the bowties on the section and turn them into dipping reflectors. A producing well, with static hydrostatic level at 4.9 m depth is located at trace 32 m.
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
correctly dealt with the migration process. That will harm our section smearing out the data at the fracture zone. Such effect can only be dealt with in a true 3D survey, where the inline and cross-line spatial sampling are compatible. In that case, one could possibly perform a two-step migration that is satisfactory for a simple subsurface. For a more involved structure, one would require a full 3D migration (Jakubowicz and Levin, 1983). Fig. 4a shows a 70-m raw data section clipped from the 120-m-long profile crossing the studied spring (trace 32 m). A spreading and exponential compensation gain assuming an attenuation of 0.1 dB/m was applied to the section. Data were low-pass filtered with a cutoff at 250 MHz to reduce highfrequency noise and clipped in time to chop off values earlier than the first break. The section was migrated assuming a constant velocity throughout (0.07 m/ns), the final result is shown in Fig. 4b. Water-filled fractured rock appears in the section as a region of lower amplitude reflections beyond trace 25 m and below 5 m depth. That region gets deeper towards the end of the section, extending to trace 60 m. Fracture induced diffraction were virtually eliminated by the energy focusing provided by the migration. The wavelet phase polarity can also be used to determine the presence of water. The phase of the wavelet is defined in this paper as the sequence of phase polarities as seen along a given trace. Fig. 5 shows the end portion of the trace obtained at 35 m
Fig. 5. End section of trace 35 m. The arrow shows the first of a series of phase inversions, occurring at the top section of the waterfilled fractured rock. Amplitudes are clipped at a maximum of 3.2. The vertical scale is arbitrary.
245
along the section shown in Fig. 4. The trace was AGC gained and appears clipped at an arbitrary maximum. The AGC attempts to equalize the signals by applying a gain which is inversely proportional to signal strength, and therefore does not preserve relative amplitude information. The first of a series of phase polarity inversions occurs at 178 ns as seen in Fig. 5. This gives the top of the water-filled fractured rock at 6.23 m, less than 1 m deeper than it appears to be in Fig. 4b. This discrepancy may be due to the uncertainty in recognizing the phase polarity sequence properly, because the leading edge of the wavelet cannot be ascertained due to closed spaced reflections, as may be the case on the top of the fractured rock.
7. Interpretation A great deal of effort in interpreting radar profiles goes into not only understanding reflections and diffractions but also into deciphering of interference patterns. The focusing of energy provided by migration make that task easier. The expected product is to recognize changes in reflector characteristics such as configuration, continuity, frequency and amplitude so to characterize radar facies. Radar facies are 3D regions representing particular combinations of physical properties like lithology, stratification, fracturing and fluid contents. Recovering the full geometry of such regions is not an issue here as we are dealing with a 2D profile, but with the data we can have an idea of the layering of the gneiss and recover the contact between less and more fractured rock. The section in Fig. 4 gives a good image of the subsurface, revealing features such as the top portion of the water-filled fractured rock and the structure of the gneiss. The interpreted section is shown in Fig. 6. The most prominent feature is the top of the fractured rock delimiting an extensive saturated zone (35 m wide), which is responsible for the high drainage rate of the spring. This interpretation is corroborated by the existence of the producing well at trace 32 m. In the radar section of Fig. 4b, it is possible to identify a series of semi-parallel folded reflectors all long the profile and small structures such as a lens shape between 0 and 30 m in the profile that is very similar to a boudin structure (B in Fig. 6). As said before these structures (Fig. 2) are very common in
246
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
Fig. 6. Interpretation of the radar section shown in Fig. 4. The F’s are interpreted as fractures and B is interpreted as a boudin structure, very common in the region, refer to Fig. 2.
the region. The overall pattern of the reflectors along the profile resembles a migmatitic texture, with alternated veined/folded mafic (biotite– hornblende) and felsic layers (quartz). Another important feature identified are the sub-vertical fractures (F in Fig. 6) that occur all long the section. These fractures are associated to the main conduits to water ascension in the area. Independent geophysical data is in accordance with the GPR data. 1D Interpretation of one VES positioned at trace 42 m along the profile indicate a four-layer earth (S. Berrino, personal communication, 1997). Schlumberger soundings, due to its low costs and simplicity of use, are still traditionally employed in many regions to enhance the odds of finding groundwater in crystalline terrains (Carrasquilla et al., 1997). The main idea behind these surveys is, after identifying a favorable fracture in aerial photo-interpretation, to try to find lower resistivity zones within high resistive rocks. That lowering in the resistivity pattern should be associated to groundwater. The success rate of that type of approach reaches 85 – 90% of the drilled boreholes. To further reduce the ambiguity of the geophysical interpretation of the shallow subsurface we incorporated the geological knowledge of the weathering profile in the region (Barreto et al., 2000). Descriptions of the typical weathering profiles of tropical and subtropical basement complexes are found elsewhere in the literature (refer to Fig. 4, p. 141 of Meju, 2002). The 1D model at the spring starts from top to bottom with a 2-m saprolite layer of 1500 V m, followed by a 5-m fractured layer of 3000 V m and
then to a fractured saturated zone of 300 V m that may extend down to 50 m, well beyond the penetration depth of the GPR data. This 10 times lowering in the resistivity values indicates a high degree of fracturing and connectivity. A half-space of highly resistive fresh rock terminates the model. The resistivity of the saturated zone is within the range of common fractured aquifers at igneous and metamorphic rocks (refer to Fig. 5, p. 142 of Meju, 2002). It is interesting to note that this simple 1D interpretation presented a reasonable estimate of the top of the saturated zone (7 m) when compared to the borehole log and the radar section (5.7 m). The main discrepancy, about 20% error, is associated to the bottom of the saturated zone, that was estimated in 50 m by the 1D inversion and is 62 m at the borehole. This is comprehensive as we are interpreting with a 1D method a complex 3D earth. But as stated before, for exploration purposes, the main interest is just identify a zone of low resistivity within crystalline rocks that could associated to groundwater, so enhancing the chances of a successful drilling.
8. Conclusions This work presents the results from a GPR survey done for water exploration on crystalline terrain (migmatized gneiss). Reflection data were collected with the fixed-offset configuration along a profile 70 m long. The velocity was estimated doing velocity analysis on CMP data. The estimated velocity value was 0.07 m/ns. This value is 54% less than the tabulated velocity for granite-gneiss and is attributed to high water content in rock.
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
The raw GPR section was cluttered with diffractions. A migration scheme was applied assuming the constant value of 0.07 m/ns throughout. The energy focusing provided by migration can be used to remove/attenuate interference patterns and make the GPR section appear similar to the geologic crosssection in depth. As expected, the fracture induced diffractions were virtually eliminated in the migrated section, enhancing the visualization of the geologic structures. In the migrated radar profile the water filled fractured zone is clearly outlined by a wide (35 m) region of lower amplitude values. Those are result of the scattering of the incident energy as well as due to the smearing out resultant from the process of migrating out-of-plane diffractions. The depth to the top of this saturated zone is variable along the profile (4– 10 m). The presence of water in the fractured zone can also be mapped from inversion on the polarity of the wavelet phase at the radar traces. To further reduce the inherent ambiguity of the geophysical interpretation a priori geological knowledge of a borehole located at trace 32 m was incorporated. In this borehole the water producing zone is confined within 4.9 and 62 m depth. There is a very good agreement between the GPR data and the available borehole information. A priori knowledge of the weathering profile and migmatite structures of Santo Aleixo Unit were also incorporated to help in the geological interpretation of the radar profile. The main limitation of the method is the relative shallow depth of investigation. The GPR data provided a detailed characterization of shape and extension of the shallow saturated zone. Depth of investigation was limited to 10 m, insufficient to image the whole saturated zone (till 62 m depth). The imaging of greater depths can be achieved by integrating the GPR data with another geophysical method. In our case, independent geophysical data from a VES Schlumberger sounding are in accordance with the GPR results and the available geological knowledge.
Acknowledgements We would like to thank the editor-in-chief A. Hoerdt, S. Hautot and an anonymous reviewer for
247
their constructive comments that helped to improve the text. We also thank Miguel Tupinamba´ for the photo of Santo Aleixo migmatite. PTLM was supported by an UERJ-Procieˆncia scholarship.
References Aranha, P.R.A., Augustin, C.H.R.R., Sobreira, F.G., 2002. The use of GPR for characterizing underground weathered profiles in the sub-humid tropics. J. Appl. Geophys. 49, 195 – 210. Barreto, A.B.C., Monsores, A.L.M., Leal, A.S., Pimentel, J., 2000. Hidrogeologia do Estado do Rio de Janeiro. Mapa de Favorabilidade Hidrogeolo´gica na escala 1:500000 e texto explicativo. CPRM-Servicßo Geolo´gico do Brasil, Rio de Janeiro, CD-ROM. Beres, M., Haeni, F.P., 1991. Application of ground-penetratingradar methods in hydrogeologic studies. Ground Water 29, 375 – 386. Berrer, E.K., King, A.R., Watts, A.H., Roberts, B., Adam, E., Eaton, D.W., Wu, J., Salisbury, M.H., 2000. Development of 3-D seismic exploration technology for deep nickel – copper deposits: a case history from the Sudbury basin, Canada. Geophysics 65, 1890 – 1899. Cardimona, S., Clement, W.P., Kadinsky-Cade, K., 1998. Seismic reflection and ground-penetrating radar imaging of a shallow aquifer. Geophysics 63, 1310 – 1317. Carrasquilla, A., Porsani, M.J., Tavares, A., 1997. Prospeccß a˜o de a´guas subterraˆneas no Alto Xigu-Para com me´todos geofı´sicos eletromagne´ticos. Rev. Bras. Geocieˆnc. 27, 221 – 228. Davies, J.L., Annan, A.P., 1989. Ground-penetrating radar for highresolution mapping of soil and rock stratigraphy. Geophys. Prospect. 37, 531 – 551. DNPM, 1997. Principais depo´sitos minerais do Brasil. v4—Rochas e minerais industriais. In: Schobbenhaus, C, Queiro´z, E.T., Coelho, C.E.S. (Coordenadores), Coordenador. Departamento Nacional de Producßa˜o Mineral (DNPM). Brası´lia. pp. 9 – 38. DNPM, 1998. Mapa geolo´gico do Estado do Rio de Janeiro, escala 1:400000. In: Fonseca, M.J.G. (Coordenador), Departamento Nacional de Producßa˜o Mineral. 141 pp. Drumond, B.J., Goleby, B.R., Owen, A.J., Yeates, A.N., Swager, C., Zhang, Y., Jackson, J.K., 2000. Seismic reflection imaging of mineral systems: three case histories. Geophysics 65, 1852 – 1861. Jakubowicz, H., Levin, S., 1983. A simple exact method of 3-D migration—theory. Geophys. Prospect. 31, 34 – 56. Lanvegin, C., Pernel, F., Pointet, T., 1991. Aide a` la de´cision en matie`re de prospection hydroge´ologique—L’analyse muticrite`re au service de l’e´valuation du potenciel aquife`re, en milieu fissure´ (granite de Huelgoat, Finiste`re, France). Hydroge´ologie 1 (64 pp.). Martins, A.M., Maurı´cio, R.C., Pereira Filho, J.C., Caetano, L.C., ´ guas Minerais do Estado do Rio de Janeiro. Departa1997. A mento de Recursos Minerais (DRM), Nitero´i. 90 pp. Meju, M.A., 2002. Geoelectomagnetic exploration for natural resources: models, cases studies and challenges. Surv. Geophys. 23, 133 – 205.
248
J. de Menezes Travassos, P. de Tarso Luiz Menezes / Journal of Applied Geophysics 55 (2004) 239–248
Tupinamba´, M., Penha, H.M., Bustamante Junho, M.C., 2000. Arcrelated to post-collisional magmatism at Serra dos Orga˜os region, Rio de Janeiro State, Brazil: products of Gondwana assembly, during the Brasiliano – Pan African orogeny. Field Trip guide of 31st International Geological Congress, Rio de Janeiro, Brazil. 30 pp.
Van Overmeeren, R.A., 1994. Georadar for hydrogeology. First Break 12 (8), 401 – 408. Yilmaz, O., 1987. Seismic data processing. In: Doherty, S.M. (Ed.), SEG Series: Investigations in Geophysics, vol. 2. SEG, Tulsa, OK. 526 pp.