|
|
 |
|||||||||||||||||
|
|||||||||||||||||
 |
|||||||||||||||||
| JOURNAL HOME | HELP | CONTACT PUBLISHER | SUBSCRIBE | ARCHIVE | SEARCH | TABLE OF CONTENTS |
SPECIAL SECTION: Hydrogeophysics |
Leibniz Institute for Applied Geophysics
Berlin University of Technology
Corresponding author: ugur.yaramanci{at}liag-hannover.de
Surface nuclear magnetic resonance (surface NMR) is the only geophysical exploration method that nondestructively provides direct information on subsurface aquifer properties (i.e., geometry, water content, hydraulic conductivity, and, partially, resistivity). The method combines the information accessible via nuclear magnetic resonance (NMR) measurements with a nondestructive surface acquisition approach to derive subsurface water information. These characteristics have made surface NMR a useful tool for hydrogeophysics during the last decade.
NMR allows a glimpse at the structure of matter on the scale of atoms. To capture different kinds of information, a number of different measurement schemes exist, all of which measure NMR relaxation signals. Even though the technique is far better known in medical, chemical, and physical applications, NMR is also used in laboratories and borehole petrophysical investigations to measure water content, pore-size distributions, and permeabilities.
All NMR-based techniques share a common basic procedure in which an alternating magnetic field (the excitation or secondary field) at Larmor (resonance) frequency forces reorientation of the macroscopic magnetic moments of protons from their thermal equilibrium. After the excitation field is extinguished, the orientation of magnetization returns to the equilibrium state. This relaxation process generates a weak magnetic field that is measured and analyzed to determine properties of the materials.
Surface NMR adopts the basic principles of NMR measurements, emitting an excitation pulse and recording the relaxation signals (free induction decays or FID), but it uses large surface coils to measure the induced decay. This differs from laboratory NMR, in which a sample is placed inside a strong artificial primary magnetic field; for surface NMR the object of investigation is outside the loop, and the Earth field acts as the primary field. The electromagnetic-wave propagation of both, the excitation field emitted by the surface loop and the relaxation signal originating from the illuminated subsurface volume, determines the spatial location and measurability of the NMR signal. Consequently, the spatial sensitivity is based on changing excitation intensities q that are products of the loop's current and duration of the excitation field. Thus, surface NMR data sets are NMR signals dependent on the pulse moment q (Figure 1). Surface NMR data include complex values, both due to the subsurface resistivity and separate loops for transmitting the excitation field and receiving the FIDs.
|
Surface NMR has recently been extended successfully to 2D and is referred to as magnetic resonance tomography (MRT). In this paper, we will step through three case studies of MRS, each highlighting different aspects of using MRS in the field, and one field case of MRT.
Figure 2 summarizes the key properties and their (inter) relationships to the measured data. Keep in mind that field parameters (such as loop sizes and shapes, Earth field inclination, and Earth field strength) have to be taken into account beforehand and integrated into the forward operator.
|
The first field example will demonstrate a standard case. Total water content can be derived from the PWC distribution summed over all decay times. Unlike laboratory NMR, surface NMR is unable to measure relaxation signals arising from water associated with very small decay times—i.e., very small pore spaces or bounded water such as clayey material (decay times less than 30 ms). Therefore, very small pore structures with very high water content such as clay do appear from surface NMR data to possess very low or zero water content. This effect is due to the 40-ms "dead time" usually observed before signal is recorded. Consequently, the water content detectable by surface NMR is referred to as extractable or mobile water content.
A mean decay-time depth distribution is calculated as a logarithmic mean of the PWC. By combining water content (or porosities if full saturation is assumed) and decay time, hydraulic permeability can be calculated. Since surface NMR measures the T2* decay time, several other parameters (e.g., magnetic field gradient) influencing permeability estimations might lead to misinterpretation.
Standard applications use only amplitude data to estimate the partial water content distribution. Since the phase data depend on the PWC distribution (beside dependency on subsurface resistivity and separate loops), inversion of the complex data allows increased penetration depth and resolution. But, often, the phase data appear corrupt and thus complex inversion fails.
Since electromagnetic field propagation is integral to the method, subsurface resistivity influences the measured data, especially the measured phase data. All applicable inversion schemes incorporate the subsurface resistivity distribution into the forward operator. With the influence of resistivity on the measured data, one can take a further step and try to extract resistivity from the inversion based on this dependency. Again, due to the rather erratic measured phases commonly observed, this approach could be complicated.
Our second field example shows a successful implementation and therefore proof of concept.
Finally, the last field example directly compares surface NMR with borehole NMR at a location close to Abu Dhabi in the United Arab Emirates.
Shallow aquifer characterization
Schillerslage, close to Hannover, Germany, is one of the Leibniz Institute of Applied Geophysics test sites. Glacial sediments down to about 23 m, underlain by marine bedrock of Cretaceous age, characterize the geology. The sediment package itself is structured and the object of this investigation.
Besides a very detailed knowledge of the subsurface structure from test drilling (Figure 3), geophysical measurements such as a common-midpoint georadar measurements and geo-electric measurements exist. Therefore, the test site is a useful place to compare MRS-derived results with drilling results and other geophysical measurements.
|
The MRS measurements were carried out using a circular loop with a diameter of 50 m and 2 turns. The Numis Lite equipment from Iris Instruments provided a maximum pulse moment of 2.9 As. The data (Figure 4a) quality was excellent due to very low noise conditions. Figure 4 also shows the inverted PWC distribution (4b), permeability estimation normalized to its maximum (4c), and the total water content (4d). The inversion was carried out using the QT scheme. Basically, all geologic units are well represented.
|
Salt water intrusion
The area around Muravera in Sardinia is characterized by the delta of the Flumendosa River and intensive agricultural use. During the last few decades, problems caused by salt water intrusion into the primary aquifer have increased. This development has triggered the use of plants more resistant to salt water (e.g., rice). But at the same time, the amount of groundwater needed by these new agricultural products compared to historical varieties has increased the demand for groundwater.
MRS measurements were carried out to determine the aquifer's thickness and subsurface structure (especially the lower boundary with the clayey aquiclude). Direct current geoelectric surveys had been carried out prior to the MRS field campaign. Due to the salt water's effects on the groundwater, the resistivity contrast between the aquifer and aquiclude was not sufficient to allow a reliable identification of that boundary.
The assumed aquifer has resistivities of only about 1.5 
-m while the clayey aquiclude apparently has about 10–15 -m (Figure 5a). Even the top layer of dry sand and soil shows very low resistivities.
|
-m are significant enough to detect. Thus, the Muravera soundings provided excellent conditions both for the collection of data with high signal-to-noise ratio and resistivity-affected phases. Even though several soundings were measured, only a few provided phase data fit for inversion (Figure 5b). The aquifer could be clearly identified on surface NMR data by its water content of about 20% and its calculated depth, which matches the second layer of the dc-geoelectrics quite well. The clay aquiclude appears as dry due to its immeasurable short decay times. The MRS-estimated resistivity distribution coincides very well with the dc-geoelectrics both in layer depths and resistivity values. Finally, the approximately 120-ms decay times correspond to fine sand and are consistent with the known geology.
With the high-conductive subsurface, we were fortunate to have measured some invertible phase data, although the penetration depth is strongly limited. Again, if resistivity can be neglected, as a rule of thumb the MRS penetration depth can be estimated as 1.5 times the loop diameter. In the previous case, the maximum depth reached was around 30 m (notably less than 1.5 times loop diameter).
Comparison to borehole NMR
The main objectives of a joint research project between the Berlin University of Technology and Schlumberger Water Services (SWS) was to compare borehole NMR measurements and surface NMR measurements and to assess how high-resolution borehole NMR can enhance the interpretation of MRS measurements.
At a groundwater test site in the desert near Abu Dhabi, many measurements were collected along a profile in close proximity to several boreholes. The geology is dominated by a tightly folded and thrust-faulted regime striking NNE-SSW. The near-surface sediments are divided into (from top to bottom): unsaturated aquifer, unconsolidated quartz-rich sand dunes (
30-m high), saturated aquifer, quaternary unconsolidated aeolian sands and silt clays, the Upper Fars unit, and claystone.
SWS has completed an Aquifer Storage and Recovery Project that included the drilling and logging of more than 20 wells spread across the whole test site (Figure 6). Here we will not focus on borehole interpretations but use the combinable magnetic resonance (CMR) porosities and the decay time distribution.
|
The measurements, taken approximately 70 m west of well SWS16 using a 100-m square loop, had higher pulse moments and thus larger penetration depths. The local Larmor frequency was 1850 Hz, Earth field inclination was 35° N, and (for a 40-ms pulse duration) the maximum pulse moment was 12.5 As. The noise conditions were acceptable but based on the expected, very small decay times the signal-to-noise ratio was quite unfavorable even after 256 stacks. Most recorded NMR signal vanished into the background noise after 50 ms. In fact, we were fortunate to measure reliable signals in the few tens of nanovolt range.
The inversion (Figure 7) of surface NMR data reconstructs the subsurface very well in comparison to the borehole NMR logs. The upper boundary of the aquifer estimated by surface NMR is close to the boundary determined by the logging tool and the decay times allow a confident estimation of the water content. Furthermore, even at greater depths, the total water content is correctly estimated. Only part of the borehole NMR-detected water content is measurable via surface NMR due to the instrument's dead time. Only water content associated with decay times larger than 20 ms is invertible (decay times lower than 30 ms were detected). Keep in mind that the borehole NMR measures T2 while MRS measures T2* at a different resonance frequency.
|
Two-dimensional investigation
One of the very first magnetic resonance tomography (MRT) surveys was conducted at the Berlin University of Technology's Nauen, Germany, test site. Quaternary sediments down to more than 100 m characterize the site. After previous geophysical investigations (direct current geoelectrics and georadar) and a test well, the very near surface was determined to include a shallow aquifer confined by a till layer at a depth of about 20 m.
The MRT survey consisted of four positions (P1-P4) spaced over a 200-m profile (Figure 8). In addition to the conventional coincident loop configuration at individual positions, MRT also uses separate transmitter and receiver loops to increase near-surface and lateral resolution, like transmitting at Px and receiving at Py. Separations larger than a loop diameter are generally not used due to decreasing signal-to-noise ratio with increasing loop separation. Taking all combinations that provided useful data led to incorporating 16 soundings into the analysis. A loop with a diameter of 50 m was used for all soundings.
|
Part of current research activities is to extend the 2D water content distribution to a 2D partial water content distribution (i.e., include the decay time dependency to map the hydraulic conductivities).
Summary and outlook
Using four field cases, we have shown that surface NMR has already passed the experimental stage and is evolving into a useful tool for hydrogeophysics. It provides unique access to aquifer structures, due to its direct sensitivity to hydrogen protons. This makes surface NMR a powerful method for solving several routinely requested hydrogeological tasks.
Some of these tasks might be solving ambiguities when interpreting geoelectrical measurements or determining aquifer properties such as porosities and—with restrictions—hydraulic permeabilities. The latter is of significant interest but still needs intensive research to determine all the parameters that influence the surface NMR-measured decay time. Although we have successfully inverted resistivities directly from MRS measurements, understanding the measured phase is still in an early stage of research.
A key factor for further increasing the use of surface NMR will be developments that improve the signal-to-noise ratio, i.e., develop enhanced noise-cancellation techniques. A limitation is the range of applicable sites, which is limited to places with low electromagnetic noise. Fortunately, some more recent developments concerning new measurement devices seem to address this limitation.
Suggested reading
Some comprehensive resumes were given in four special issues following international workshops on MRS: Journal of Applied Geophysics (2002), Near Surface Geophysics (2005), Journal of Applied Geophysics (2008), and Bulletin Geologico y Minero (2007). "Surface nuclear magnetic resonance tomography" by Hertrich et al. (IEEE Transactions on Geoscience and Remote Sensing, 2007). On Inversion of Magnetic Resonance Sounding (MRS) and Magnetic Resonance Tomography (MRT) by Müller-Petke (PhD thesis, Berlin University of Technology, 2009). "Study on complex inversion on magnetic resonance sounding signals" by Braun et al. (Near Surface Geophysics, 2005). "Inversion of resistivity in magnetic resonance sounding" by Braun and Yaramanci (Journal of Applied Geophysics, 2008). "Aquifer characterization using surface NMR jointly with other geophysical techniques at the Nauen/Berlin test site" by Yaramanci et al. (Journal of Applied Geophysics, 2002).
| ||||||||||||||||||||||||||||||||||||||||||||||||||||
| JOURNAL HOME | HELP | CONTACT PUBLISHER | SUBSCRIBE | ARCHIVE | SEARCH | TABLE OF CONTENTS |
