Seepage Investigation on an Existing Dam Using Very Low Frequency Electromagnetic (VLF-EM) Methods: A Case Study of Shagari Earth Dam, Sokoto, North Western Nigeria

Very low frequency (VLF-EM) was used to assess variations in overburden composition, bedrock lithology and the hidden Sedimentary structures within the foundation of Shagari Dam of the sedimentary basin of Northwestern Nigeria. Four VLF-electromagnetic (EM) traverses were occupied at 10 m intervals. The VLF normal and filtered real component irregularities identify major geological interfaces suspected to be faults/fractured zones. The points of crossover between the real and imaginary components delineate the fractured zones, which were identified as areas of possible seepage (sloughing and piping). The fractured zones are suspected to be present at all traverses. In total, 38 fractured zones were identified along the dam embankment and canal site, while 17 major fractures occurrence dippers along the Traverses at a point, Traverse 1 (F3, F4, F5, F7, F10, F11), Traverse 2 (F12, F13, F21, F22), Traverse 3 (F23, F24, F29) and Traverse 4 (F32, F33, F34, F38), and coincide with cross over point at 4 traverses. These seepage zones cause heterogeneity in the subsurface structure that could be prime to dam failure which in turn leads to the flooding, decreases in irrigations activities of the peoples leaving around the dam and loss of several hundreds of life when care not taken. The result of the study suggests that VLF is suitable for observing seepages in embankment dams.

Dams are among the largest and most important projects in civil engineering [1]. For geologic, hydrologic and topographic reasons, there are limited numbers of ideal sites for dam It is therefore very important to carry out geotechnical and geophysical investigations on any proposed dam site. The information provided by the study is expected to aid dam site investigation, dams mainly intended for water supplies require a low tolerance of seepage losses. Besides, the design of dam structures must be adapted to the existing site conditions [2], to minimize the losses. Failure to do any of these may invariably result in unplanned seepage and/or total collapse of the structure. [3] Examined causes of dam failures worldwide and discovered that 25% of the failures were due to geotechnical problems associated with seep-age, inadequate seepage cut-off, faults, settlements and land-slides.
Seepage through a dam is a slow discharge or escape of liquid that is supposed to be retained in the reservoir, through conduits (fracture/fault lines) to the flank of the dam. It reduces the expected quantity of water, weaken dam construction and damage dam gates [4]. Certain soil can swell if they get saturated with the seeping water and when there is loss of water in them, they shrink drastically.
These expansions and shrinkages of soils can result in cracks and collapse of the dam. Seepage is one of the major reasons for embankment dam failures and may constitute a significant potential social dislocation (displacement of people, loss of valuable life and properties) if not examined. Cost of reconstruction of a failed dam can be worrisome and shortage of water supply may hinder the objective of dam construction. The purposes of this study is to identify and describe structural defects such as faults, joints, solution channels in the rocks, to determine Figure 1. Location of the study area [11].

Location
The place described its origin in his book titled "Beckoned to serve". He mentioned that Dan Fodio reportedly said "you can remain and sha (enjoy) your gari (millet flour)"; it is a valley located in Shagari Local Government Area, Nigeria [5]. The dam falls between latitudes 12˚4357'5''N and 12˚7326'30''N, Longitudes 5˚07'070''E and 5˚118''615''E. It is located south east of Sokoto city. Shagari falls into the typically tropical climate which has distinct wet and dry season. The study area itself is situated in the arid and Semi-arid areas of Northern Nigeria [6] ( Figure 1). The climate of the study area is tropical continental and is dominated by two opposing air masses, the dry tropical continental blowing from the Sahara Desert and moist Tropical maritime blowing from the North Easterly direction, the later comes from South-Westerly direction [7]. These two major air masses result in two major seasons, the wet and dry season. While the rainy season commences from mid-April and lasts to the end of September, the dry season extends from early October to May. The Harmattan period is characterized by a dry, cold and dusty wind experienced in the state between November and February. During this period, the weather is usually cold at night and in the morning with temperature often dropping to less than 200˚C in some cases [7]. Temperatures are highest from late March to May, ranging from 350˚C to 450˚C. Mean annual rainfall is about 600 mm with most of it falling in July and August. Over the last fifty years, the rainy season has been characterized by late arrivals of rains, long spells of aridity of up to 21 days and early cessations; this is a suggestive of climate change [8].

Geology of the Area
The Rima Group (Figure 1), consists of a marine transgressive series of fine-grained sand and friable sandstone, mudstone, and some marly limestone and shale. North of the River Sokoto in the northern part of the Sokoto Basin, the group is divided into three formations, the Taloka at the base, the Dukamaje in the middle, and the Wurno at the top [9]. The Taloka Formation, where exposed in the northern part of the Sokoto Basin, attains a thickness of 400 feet and consists of white fine-grained friable sandstone containing some thin intercalated beds of carbonaceous mudstone or shale [10]. The Dukamaje Formation, between the Taloka and the Wurno, crops out only north of the River Rima. Originally named the "Mososaurus Shales", it consists at Dukamaji of 70 feet of shale, thin limes-International Journal of Geosciences tone, and mudstone, and some gypsum and an assemblage of invertebrate and vertebrate fossils at the base. The uppermost formation of the Rima Group, the Wurno, consists of 75 feet of pale fine sand and some silt. According to [9], the Dukamaje Formation has not been found south of Rabah on the River Sokoto, and hence south of this point the Wurno and the T'aloka are mapped together as a unit. In the northern part of the Sokoto Basin, samples fron 1 boreholes closely resemble those taken from surface outcrops of the Rima Group. Thick beds of well-sorted fine white sand and dark gray lignitic clay containing iron sulfide (pyrite) are found in boreholes near Sokoto and also at Rabah and Wurno. Individual beds may exceed several hundred feet in thickness and grade laterally from clay to fine sand within short distances. Coarse sand in thin beds occurs near the base of the Taloka Formation at Wurno, Dange, and Sokoto. Toward the southern part of the Sokoto Basin, the Wurno Taloka lithology departs once again from the typical fine-grained lithology observed in outcrop. Clay beds become less abundant than sandy beds, and the sand changes from the characteristic fine texture of the north to medium and coarse texture in the South. The total thickness of the Rima Group ranges from zero at the eastern limit of its outcrop to more than 1000 feet in it's down dip extensions near the Niger frontier. The regional dip of the Rima near Sokoto is about 20 feet per mile.

Very Low Frequency Electromagnetic (VLF-EM) Technique
The instrument used for the VLF profiling is the ABEM WADI, the dam profile cover a 2.9 km, 4 transverses with total of 2900 VLF stations were established in all. The reading was taken at an interval of 10 m. VLF-EM is an inductive exploration technique that is suitable for mapping shallow subsurface structural features in which the primary EM waves induces current flow, [12]. Although both the real and quadrature components of the VLF-EM were measured, the real component data which are usually more diagnostic of linear features [13], were processed for qualitative interpretation. The VLF data that is, real and filtered real components of the EM fields measured was subjected to Fraser filtering techniques to increase the signal to noise ratio of the data set and enhance the anomaly signature. Using the [14], techniques, the raw real and filtered real components were plotted against station positions. A 2D inversion of real component data was also carried out using the same software.

VLF-EM Survey
VLF surveys may be run on airborne or ground platforms. GEM's GSM-90 AV is an example of an airborne version and the GSM-19 V and GSM-19 TV are GEM's ground versions. The airborne version is equipped with two sensors so it can be tuned to two VLF transmitting stations concurrently. The ground version is new and is called walking VLF. With this system, you can use it either with a magnetometer or standalone (two VLF receivers). Using the walking version, it is possible to survey 10 km or more of profile a day. GEM's VLF systems meas- i.e. out-of-phase/quadrature as a percentage of the total field. This is so called, also, because the component is 90˚ to the primary signal; and field strength in Pico-Tesla [15]. This is a powerful combination of parameters designed to maximize the effectiveness of VLF survey. GEM has also made advanced technological innovations in its software and hardware-giving easy-to-use interface increase accuracy. The development of a chip-based tilt sensor to replace commonly used a fluid-filled sensor which means that the detector can operate much faster.

Filtering Procedures
The measured dip angle response needs to be filtered to reduce the random noise and long spatial wavelengths. However, for practical purposes it is not generally necessary to eliminate the noise, and long spatial wavelength as VLF ensures fairly accurate measurements with random errors well below an appreciable level. To overcome some of these problems [16], devised a simple numerical filter, which converts crossovers of current polarity into peaks by differencing successive values of tilt angle (the in phase component) along the profile. The Fraser filter shifts the measured dip angle data by 90˚. Thus, the Fraser filtering transforms the anomaly such that parts with maximum slope will appear as maximum amplitude. In a sequence of consecutive readings of tilt angle, say Q 1 , Q 2 , Q 3 and Q 4 ; the term (Q 2 − Q 1 ) not only shifts the dip angle but also attenuates the spatial wavelengths.
Numerically averaging weighted values of the adjacent sets of such differences i.e. (Q 2 − Q 1 )/4 + (Q 3 − Q 2 )/2 + (Q 4 − Q 3 )/4 which reduces to (Q 3 + Q 4 ) − (Q 1 + Q 2 ) reduces noise. [16] used linear filtering for the analysis of VLF dip angle data in an extension of the Fraser filter. They describe the magnetic field arising from a subsurface 2D current distribution assumed in a horizontal sheet of varying current density situated everywhere at a depth equal to the distance between the measurement stations.

VLF-EM Data Processing
VLF field data were filtered using Fraser Filtering method. [17] Suggested a simple numerical technique aimed at eliminating erroneous interpretation of very low frequency (VLF) data caused by large geological noise component generated from the transmitted frequency. According to [17], this filtering method improves the resolution of the anomalies and eases their recognition.

VLF Data Acquisitions
On the field, a total of 2900 VLF stations were established following these procedures:  An ABEM WADI was used for the acquisition of data; International Journal of Geosciences

VLF-EM Profile Presentation
The

Traverse 1
In Figure 2(a), fracture zones were located at station some positions, .the fracture zones identified as (F 1 -F 11 ), with some identified fracture F 1 , F 2 , F 6 , F 8 and

Traverse 2
In Figure 3(a), fracture zones were located at station some positions, the frac-   (Figure 3(b)), indicating a highly conductive body at subsurface that are probable seepage zones.

Traverse 3
In Figure 4(a) as shown in the traverse 3 it can be deduced that seven fractured zones were delineated, among which three fractured zones were located at station some positions, the fracture zones identified as (F 23 -F 29 ), with some identified fracture F 25 , F 26 , F 27 , and F 28 at station positions are the shallowest zone, while conductors (fracture zone) at F 23 , F 24 , and F 29 show abnormal feature with an amplitude of 10% -20% signal responses. Diverse features of conductivity fluctuating in different directions were delineated that coincide with crossover ( Figure 4(a)) with their corresponding depth, major fracture identified are extended beyond a depth of 40 m as shown in the KH Pseudo section ( Figure  4(b)), indicating a highly conductive body at subsurface that are probable seepage zones.

Traverse 4
In (Figure 5(a)) as shown in the traverse 3 it can be deduced that seven fractured zones were delineated, among which three fractured zones were located at station some positions, the fracture zones identified as (F 30 -F 38 ), with some identified fracture F 30 , F 31 , F 35 , F 36 , and F 37 at station positions are the shallowest zone, while conductors (fracture zone) at F 32 , F 33 , F 34 , and F 38 show abnormal feature with an amplitude of 25% -40% signal responses. Other features identified with varying degree of conductivity appear in different directions that coincide with crossover ( Figure 5(a)) with their corresponding depth, major fracture identified are extended beyond a depth of 60 m are shown in the KH Pseudo section ( Figure 5(b)), indicating a highly conductive body at subsurface that are probable possible seepage zones.

Conclusion
Geophysical investigation was carried out on Shagari Dam. The VLF-EM data were acquired along four traverses. The VLF data effectively identified relatively wide areas of conductivity by noticing the boundaries of conductive regions. Both the maximum differences and points of crossover between the real and imaginary curves were used to delineate the sheared/fractured and weathered zones, which were identified as areas of fracture and probable areas of internal destruction. The results of the study suggest that the EM VLF method is an adequate method for monitoring seepages in dam embankments. The point of crossover between the real and imaginary components was used to delineate the fractured zones, which were identified as areas of possible leakage. The fractured zones were identified on all four traverses as shown in Figure 2(a) and Figure 5(

Recommendation
We recommend additional detailed characterization of the Shagari Dam, using other complementary geophysical methods such as ground-penetrating radar, 2D and 3D electrical resistivity, Aeromagnetic method and seismic refraction tomography in order to confirm the fracture zones delineated using the EM method and to elucidate the shape of the fractures.