3D seismic data for shallow aquifers characterisation
Michela Giustiniani, a, , Flavio Accainoa, Stefano Picottia and Umberta Tinivellaa
aIstituto Nazionale di Oceanografia e Geofisica Sperimentale — OGS, Borgo Grotta Gigante 42c,
34010 Sgonico, Trieste, Italy
Received 25 June 2008;
accepted 12 March 2009.
Available online 24 March 2009.
We present the results obtained from conventional and non-conventional analysis of 3D high-resolution seismic data acquired nearby the water spring line, which separates the upper from the lower Friuli–Venezia Giulia plain (Italy), in order to characterise an important multilayered confined aquifer. The main targets of this study were two shallow aquifers located at about 30 m and 180 m depth, respectively. The aquifer structures were reconstructed by adopting a technique consisting an iterative updating procedure, for refining and improving an initial model in depth. The method includes pre-stack depth migration, residual move-out analysis and seismic reflection tomography. In the final 3D migrated cubes, two high velocity layers were identified at about 270 m and 480 m respectively, which correspond to unknown deep aquifers, as confirmed by recent well data (stratigraphies and down-hole velocity measurements). Travel-time tomography and Amplitude Versus Offset analysis evidence that seasonal variation in the seismic response of the aquifers are not detectable. However, in this case, aquifers are well detectable by lithological changes.
Keywords: Aquifer; Amplitude versus offset; 3D pre-stack depth migration; Depth modelling Article Outline
Seismic data acquisition
2.1. Processing of seismic data
Depth modeling of 3D seismic data
Correlation between seismic and well data
Amplitude versus offset
Discussions and conclusions
Fresh water is becoming one of the most important natural resource and many nations have carried out long term sustainable policies for the protection and public use of groundwater. The technological development has permitted the application of the seismic method to shallow hydrogeological studies starting from the late 1980s. In the last 20 years, in fact, the seismic
reflection technique has been applied to many hydrogeological problems purposes with varying degrees of success ([Steeples and Miller, 1990], [Bachrach and Nur, 1998], [Bradford, 1998], [Whiteley et al., 1998], [Cardimona et al., 1998] and [Bradford and Sawyer, 2002]). Bachrach and Nur (1998) have shown clearly the relation between a deep water table and its seismic image, as well as the seismic response of the subsurface under different wetting and draining conditions. Whiteley et al. (1998) and Cardimona et al. (1998) have shown good examples of mapping shallow aquifers by reflection seismic data. Bradford and Sawyer (2002) have illustrated that pre-stack depth migration can significantly improve image quality and accuracy of hydrogeologic data.
In this paper, we present the results of conventional and non-conventional processing of 3D high-resolution seismic data in order to characterise a multilayered confined aquifer system, located near the spring water line of the Friuli–Venezia Giulia plain (North-East Italy; Fig. 1). The
main targets of this study were the identification of two shallow aquifers located at about 30 m and 180 m depth, respectively. The site is particularly suitable for the experiment described in the following because of the presence of highly permeable sedimentary layers, which host an important aquifer system.
Full-size image (56K)
Fig. 1. Location map of the 2D and 3D seismic surveys.
View Within Article
The Friuli Plain is divided in two parts by the water spring line (Fig. 1): the upper and lower Friuli Plain. The upper part is characterised mainly by the presence of gravel deposits, which host unconfined aquifers. Instead, alternating layers of sand, gravel and clay are present in the lower part hosting important multilayered confined aquifers. In the study area, the Plio-quaternary sediments lie on the miocenic sediments, which are directly in contact with mesozoic carbonates. The latter were deposited in a shallow marine environment during an extensional phase. Cenozoic compressional phase caused the deposition of turbiditic sediments (Nicolich et al., 2004).
During the winter 2005, three 2D seismic lines were acquired to define a preliminary subsurface geometry (Giustiniani et al., 2008), to analyze the seismic response of the aquifers and to optimise a subsequent 3D survey, which is the object of this paper (Fig. 1). Four 3D seismic datasets were acquired in March 2006 and July 2006, providing information about the 3D geometry of the aquifers and about possible variations between different seasons. The data processing was focused on the signal-to-noise ratio increase and vertical resolution enhancement. A ‘preserving-amplitude’ processing was adopted in order to allow a successive
Amplitude Versus Offset (AVO) analysis (Yilmaz, 2001).
An iterative updating procedure involving pre-stack depth migration, residual move-out analysis and seismic reflection tomography to obtain the velocity model and a realistic depth imaging of the investigated area, was applied.
Finally, AVO analysis was performed to evaluate the Poisson's ratio contrast (Accaino et al., 2005), which provides useful information about the petro-physical properties of the shallow layers, such as the fluid content.
2. Seismic data acquisition
Seismic prospecting is an effective tool for underground exploration, but the costs limited so far the use of this technology for hydrological applications. Nowadays, however, the increasing water request and the technological improvements with the decreasing costs of the seismic acquisition are making the seismic prospecting suitable in environmental problems.
In the studied area, the 2D geophysical investigations and well data showed strong lateral variations in petro-physical underground characteristics. In particular, the results obtained from a previous 2D seismic data analysis (Giustiniani et al., 2008) have pointed out the presence of lateral variation of the seismic velocity that could be related to lithological and/or to pressure variation of the water filling the pores. The catchment wells drilled by the Acquedotto Basso Livenza reached the shallow aquifers at 30 m and 180 m. In particular, hydrogeological investigations performed in the Southern part of the studied area pointed out the absence of the aquifer located at 180 m, as confirmed by the previous 2D seismic investigation (Fig. 2; Giustiniani et al., 2008). The strong lateral velocity variation is evident in the 3D velocity field obtained from the interpolation of the 2D sections (Fig. 2), which is adopted in the present work as starting model for the updating procedure described in the following sections.
Full-size image (65K)
Fig. 2. 3D velocity model, obtained from the interpolation of three 2D velocity fields.
View Within Article
The results obtained with the previous 2D seismic profiles were utilized to plan the subsequent 3D seismic acquisitions in two different areas called A (about 0.4 km2) and B (about 0.6 km2), as shown in Fig. 1. In each area, we carried out two surveys: the first one in March 2006 acquiring two cubes called respectively CubeA-spring and CubeB-spring, and the second in July 2006 obtaining the CubeA-summer and CubeB-summer. It is important to underline that the two cubes show differences in the coverage because the shot and receiver displacements are different. In
fact, environmental problems strongly influenced the seismic acquisition. In conclusion, two 4D prospections surveyed the areas A and B.
Preliminary field tests were performed to define the best sweep lengths and frequency windows of the vibroseis and to analyze the strong ground-roll and the environmental noise. We used the DMT Summit recording instruments with 260 active channels. The MiniVib IVI T-2500 was the seismic source and 10 Hz single geophones were deployed.
Both the receiver interval and the receiver lines spacing (in-line) were 18 m. The source interval and the shot line spacing (cross-line) were 18 m and 36, respectively. Considering the receiver and shot intervals and the receiver and the shot line spacing, we adopted a binning equal to 9 × 9 m. The signal sweep of 15 s length and a frequency band of 20–200 Hz were used to provide
2 s length records after cross-correlation of recorded data with the pilot trace with a time sampling interval of 1 ms.
2.1. Processing of seismic data
The quality of the raw data was satisfactory, despite of the ground-roll (see Fig. 3). The reflections of interest are evident from 0.1 s to about 1 s, while the refractions are not always detectable because of velocity inversions caused by the alternation of impermeable and water filled permeable layers (Bourbié et al., 1987).