The study area consisted of Tarkwa, Nsuta, Aboso and Domeabra as shown in Figure 2. Tarkwa is the capital of Tarkwa-Nsuaem Municipal, a district in the Western Region southwest of Southern Ghana. Aboso is a town located in the Prestea-Huni-Valley District of the Western Region of Ghana. It is located between Tarkwa and Huni-Valley along the Tarkwa-Agyempoma road and approximately 9.9 km and 13.7 km away from these towns respectively. Domeabra is a village within the Tarkwa-Nsuaem municipality of the western region of Ghana. It is 9.7 km from Tarkwa and 4.9 km from Nsuta and is accessed by road from Tarkwa.

There are mainly two types of soil in the Tarkwa-Prestea area, the forest oxysols in the south and the forest ochrosol-oxysol integrates in the north [19] . The forest oxysols are porous, well drained, and generally loamy brown to orange. Due to the heavy and plentiful rainfalls in the south, a high degree of leaching and reduction of calcium, magnesium and other nutrients have occurred in the soil. This has made the soilacidic. The forest ochrosol-oxysol integrated is an intermediate between the forest oxysols and the forest orhrosol. It contains more of its nutrients and is therefore more alkaline then the forest oxysols in the south [19] .

The study area (Figure 3) falls within the BirimianMetavolcanic rocks. The Birimian rocks form an extensive southernmost approximately two-thirds of the estimated 4.5 × 10⁶ km² area of the Precambrian West African Craton (WAC) [20] . It is bounded to the west by the Pan-African mobile zone of the Mauritanides-Rokellides, to the north and the east by the transgressive cover of the Phanerozoic Taoudeni and Volta basins respectively, stretching over Côte d’ Ivoire, Ghana, Mali, Burkina Faso, Guinea, Liberia, Sierra Leone and west Niger. Attoh et al. [20] stated that these rocks record the most extensive episode of continental crust formation in the early part of the Proterozoic. The WAC furthermore comprises a western domain consisting essentially of Achaean rocks of Liberian age (3.0 - 2.5 Ga) [21] .

The Birimian rocks fall within Paleoproterozoic terrane which is characterised by narrow sedimentary basins and a dominantly linear arcuate bimodal volcanic and volcano-sedimentary with mostly felsic to intermediate protolith [22] . Available data suggest that the rocks formed during one of the most rapid (early Proterozoic) continental crust-forming events in the world [23] . The stratigraphic subdivision of the Birimian rocks has been a matter of considerable debate since the middle of the last century. The Birimian rocks were overlain dominantly by clastic Tarkwaian rocks and were, in addition, extensively intruded by granitoids during the Eburnian Orogeny [24] . The latter is predominantly mafic volcanics; with pyroclastics, hypabyssal intrusives, phyllites and greywackes as common constituents. The former is characterised by a predominance of immature or volcaniclastic sediments, including argillites, tuffs and greywackes [25] .

The term “Birimian” was introduced by Kitson [26] to describe rocks from the valley of the River Birim in Ghana. The stratigraphy of the Birimian is a subject of substantial debate with proposal of three distinctly different subdivisions. In numerous studies, the volcanic formations are considered to be younger than the sedimentary sequences [27] . Leubeet al. [28] suggested that the volcanic sequences and sedimentary basins are contemporaneous lateral facies equivalents, while others place the basalts and andesites at the base, overlain by flysch-type sedimentary sequences [29] . The latter view was also shared by Milesi et al. [30] . Leube et al. [28] are of the view that penecontemporaneous deposition of the tholeiitic flows and the various sedimentary rocks is indicated by a transition from volcanic rocks into sedimentary rocks along strike and interbedding between the two rock types.

The Birimian and Tarkwaian rocks that underlie the area are largely crystalline and inherently impermeable, unless fractured or weathered. Groundwater occurrence is therefore attributed with the development of secondary permeability and porosity. According to Kortatsi [31] , the zones of secondary permeability are often discrete and irregular and occur as fractures, faults, lithological contacts and zones of deep weathering. Groundwater in the area is acknowledged to occur in two distinct hydraulically connected aquifer system; an upper weathered zone aquifer and a deeper un-weathered aquifer or fractured zones and dyke contacts [25] .

The weathered zone aquifer is generally phreatic, and the principal groundwater flow occurs where relic’ s quartz veins are more abundant. The regolith is generally dominated by clay and silt, rendering the aquifer highly porous, with high storage, but low permeability. Thus, the aquifers here are either un-confined or semi-confined, depending on the clay and silt proportion. Aquifers are recharged by direct infiltration of precipitation through brecciated zones and the weathered outcrop [31] . Groundwater recharge and actual evapotranspiration have been estimated at between 11% - 17% and 54% respectively of annual rainfall [32] with the average borehole yield ranging from 2.7 - 12.7 m³/h [33] .

A site visit was undertaken to ascertain the current state of the existing open landfill at Aboso. During the visit, the factors such as the topography of the study area, closeness of the landfill to surface waters, settlement and general accessibility of the site were studied.

Geographic Information Systems (GIS) has proven to be a useful tool in site selection, thus, over the years many researchers have employed GIS in landfill site selection [34] [35] [36] [37] . Landfill siting is an extremely difficult task to accomplish because the site selection process depends on different factors and regulations [38] . Many of the attributes involved in the process of selection of sanitary landfill sites have a spatial representation, which in the last few years has motivated the predominance of geographical approaches that allow for the integration of multiple attributes using geographic information systems [39] - [44] . The process of assessing these factors involves comparing the actualconditions of the alternative sites with desirable characteristics and is usually referred to as capability/suitability evaluation [45] . The general procedure used by the decision maker to perform suitability analysis using GIS usually entails 1) selecting important factors and defining evaluation criteria, 2) comparing the attributes of the alternative sites on the desirability criteria and generating commensurate suitability values/ratings for each of the factors, 3) aggregating ratings of individual factors into a combined suitability map to identify suitable area(s) [46] . The important factors representing site characteristics under consideration are usually available as attributes of vector or raster layers in a GIS [47] . Each attribute is represented by a criterion map which displays the spatial distribution of an attribute that measures the degree to which its associated objective is achieved [48] .

Multi-Criteria Evaluation (MCE) is used to deal with the difficulties that decision makers encounter in handling large amounts of complex information [38] . The principle of the method is to divide the decision problems into smaller understandable parts, analyse each part separately, and then integrate the parts in a logical manner [49] . The integration of GIS and MCE is a powerful tool to solve the landfill site selection problem because GIS provide efficient manipulation and presentation of the data and MCE supplies consistent ranking of the potential landfill areas based on a variety of criteria [41] . Higgs [50] reported the potential of integrating multi-criteria techniques with GIS in waste facility location and documented through a review of the existing literature to highlight the opportunities and challenges facing decision makers at different stages of the waste facility management process.

The flow chart in Figure 4 shows the summary of the various stages and procedures which were employed in the selection of a suitable landfill site in Tarkwa. The criteria for the selection of the landfill are given in the Table 1.

Secondary data from Ghana Geological Survey was used for this research. They included boundary, road, surface water, wetlands, railway and built-up areas shapefiles of Tarkwa and its environs. The data was processed using ArcGIS platform. The buffer zones were created based on the concept of proximity using EPA guideline for landfill in Table 1. The resultant feature class maps

from the buffer generation were vector-based data. The vector-based data were converted to raster-based data before they were reclassified. The binary model approach was used to reclassify all raster-based maps. The Boolean approach was used to the overlay the individual criterion binary maps to generate a final suitability map. The Boolean expression contains two operands and one logical operator. In the expression (Wetlands) = 1, wetlands and 1 are the operands and “=” is the logic operator. Five expressions ((Wetlands) = 1 (Roads) = 1 (Railway) = 1 (Built-up areas) = 1 (Surface water) = 1) have been developed from the five criteria considered. In merging the expressions, the Boolean AND connector was used to separate the cells that satisfies suitability.

Seismic-refraction methods measure the time taken for compressional sound wave generated by a sound source to travel down through the subsurface and back up to detectors (geophones) placed on the land surface. The field data acquired consist of measured geophone distances and seismic travel times. From the time-distance information, depths and velocity variations to individual layers can be calculated and modelled [51] . The propagation of seismic (energy) disturbance through a heterogeneous medium is extremely complex although it can be expressed in some equation, where the velocity, travel time, distance can be used to predict the elastic properties of the region through which they pass through [52] . To generate a seismic refraction tomographic model of the subsurface, a tomographic inversion method must be done as stated by [53] [54] [55] [56] [57] after picking the first arrivals and assigning layers to the data.

In the study area, geophysical data was gathered using Geometric ES-3000 Seismodule Controller with a 12-channel seismograph. Four (4) seismic refraction profiles were randomly acquired across the study area. Nine (9) offsets were carried out over each profile line, to have maximum coverage of the subsurface for the tomographic analysis. Stack of three (3) shots were used at various shot locations on a profile to minimize background noise effect and to increase signal to noise ratio. A sampling rate of 62.50 µs with recording length of 0.25s was used. Additionally, a low-cut filter of 15 Hz was used to filter noise frequency from traffic and a notch filter of 60 Hz was used to filter frequency noise from power lines.

Data processing started with the picking of the first arrivals using Pickwinsoftware. The next step was to use Plotrefasoftware to assign layers to the various traveltimes. After the layer assignment, an initial velocity model was estimated using Time-term inversion. In this case, two types of inversions, namely time-term inversion and, tomographic inversion, must be performed. The time-term inversion was used to generate the initial velocity model. The depths to the top of the underlying layers are calculated under each point on the traveltime versus offset distance plot. 2D subsurface Seismic Refraction Tomography (SRT) model for the various profiles were generated using tomographic inversion method.

After the initial velocity model, a tomographic inversion was generated after some number of iterations were completed. After each iteration, ray tracing was initiated. The ray tracing produced a calculated travel time curve. The difference between the calculated and the observed travel times is referred to as the Root Mean Square (RMS) error; the smaller the RMS error, the higher the accuracy of the data. The iterations for the tomographic inversions were stopped when the RMS error reduced to a minimum value, i.e. where further iterations results in no change in the RMS error.

According to Sukiman [58] , the selection of a landfill site put a major concern in the geotechnical site investigation regarding to the fundamental problem in waste disposal sites related with water and soil characteristics in the proposed site. He stressed that aspects to be considered in geotechnical site investigations include hydrology, geology, geochemistry, and engineering properties of soil. Sukiman [58] suggested that for ground investigation up to a depth of 5 m, involving unconsolidated materials, hand auger drill is suitable. Soil engineering properties such as Atterberg’ s limit, bulk density, moisture content, particle size and permeability should be determined.

Both field and laboratory tests were undertaken to obtain data for the research. The field work involved visit to the site and sampling. This was followed by laboratory tests to establish the engineering properties of the soil. Laboratory equipment and materials at the soil and rock laboratory of the University of Mines and Technology Tarkwa were used for the laboratory works. These included the Cassagrande apparatus, sieves, wax, mould, hammer, and moisture cans.

The field work involved visits to the study area to take samples for the laboratory work. Other relevant information that was obtained during the field work was the topography and drainage of the area, nature and condition of the soil present. The samples were taken from eight different points (Figure 5) with an interval of 200 m from each point using an auger drill. Each point was drilled to a depth of 3 m of which a sample was taken at each meter (a total of (3) three samples were taken from each point), except for point seven where only one-meter sample was taken due to an underneath bed rock. The samples were carefully sealed to prevent loss of some soil properties such as moisture content. Moisture content and bulk density test were conducted for all the sealed samples. Afterwards, the samples were sun dried for two days and the following tests were conducted; particle size analysis, specific gravity, Atterberg limit test andcompaction test. The laboratory test works were conducted according to the British Standards 1377 [59] . The hydraulic conductivity (K) of the soils were calculated from the effective diameter (D₁₀) values using Hazen [60] empirical formula:

K = C H ∗ ( D 10 ) 2 (1)

where:

C_H = Empirical constant (0.01157)

D₁₀ = the particle size for which 10% of the material is finer (mm)

Buffer maps (Figure 6) were generated for wetlands, roads, railway, built-up areas and surface water. The buffer maps serve as constraint maps that restrain unsuitable areas from suitable areas. The Binary model gives two results for each map generated. These results are displayed with two different colour codes with values assigned to them. The colours which have been assigned 1 (true) indicate areas which are suitable for the landfill and the colours which have been assigned 0 (False) indicate areas which are not suitable for the landfill. The resultant Binary maps for the individual criteria considered for the siting of the landfill in the study area are shown in Figure 6.

The final suitability map (Figure 7) shows the suitable areas in blue and unsuitable areas in red after the overlay of individual criterion-based maps.

The results of the seismic tomographic sections are shown in Figure 8. Inferences about the condition of the subsurface lithology are made from the velocity gradient of the tomographic section. The geological field data of the study

area coupled with the velocity gradient from the tomographic inversion as stated by Raghava et al. [61] aided in the interpretation of the tomographic sections as shown in Table 2. The velocity models are shown in shades of different colours between pink (low 0.30 km/s) and blue (high 3.0 km/s). The model depicts a velocity distribution that generally increases with depth. The models in Figure 8 are displayed with horizontal distance along the profiles in metres and with a vertical distance which represent the depth of probe measured in metres. The velocities, average thickness and subsurface layer properties for the various profiles (SD1 to SD4) are presented in Table 2.

The multicriteria GIS modelling generated suitable and unsuitable areas for the construction of an engineered landfill. The suitable areas are those which are beyond 500 m buffer distance and the unsuitable areas below the 500 m buffer distance. The suitability map (Figure 6) shows fourteen (14) areas within Tarkwa and its environs to be suitable, the Domeabra site was selected because of its proximity to the Tarkwa area.

From the Seismic Refraction Tomography (SRT) analysis, it can be inferred that the subsurface is made up of four layers. The first layer is the top soil with thickness ranging from 0.5 m to 2 m. The second layer is the weathered or loose material with thickness ranging from 5 m to 15 m. The third layer is the saturated material with a thickness ranging from 10 m to 15 m. The fourth layer is the fresh bedrock which is vertically extensive. Figure 9 shows the subsurface profile of the study area with the respective average thicknesses and velocities as interpreted from the SRT models. The analysis reveals that the aquifer is in the third layer with seismic velocities ranging from 1450 m/s to 2400 m/s. The depth to the aquifer ranges from 7 m to 16 m whiles the thickness ranges from 10 m to 15 m.

The moisture content values indicate that the area has good moisture content depicting well drained soils. Das [62] , proposed that sandy and gravelly soils may have moisture content of about 15% to 20%, fine-grained (silty or clayey)

Cu is coefficient of uniformity.

soils may also have moisture content up to about 50% to 80%. The soils in the proposed area have varying proportion of gravel and sand with low proportion of fine grains at 0 - 1 m depth. Higher values of moisture content are obtained at 1 - 2 m depth as that of fine-grained soils.

Specific gravity values from Table 5 depicts the usual values of soils suitable for landfill construction. Soils which have specific gravity values less than 2 are organic in nature and they may not be suitable for landfill site construction. The values from Table 6 fall within the usual bulk density and dry density values for soils. The samples with high clay content exhibit high bulk density but low dry density due to their ability to hold more water. The clay content in the area, thus, increases with increasing depth.

The plasticity and plasticity index of the various samples increases with depth due to the increasing clay content with depth (Table 7). According to Marcin et al. [63] , clay materials of high plasticity index approximately greater than 30%, allow very good sealing capabilities but show high shrinkage potentials. They also show horizontal deformation during shrinkage as well as susceptibility for desiccation, triggering the increase in permeability and are unable to regain their initial hydraulic conductivity after rewetting. Usually recommended soil properties to achieve hydraulic permeabilities of order 10⁻⁹ m/s by compaction are: percentage of fines (<0.075 mm) ≥30%, plasticity index between 20 and 30 and percentage of gravel (5 to 50 mm) ≤20% [64] . The soils at depth 2.0 - 3.0 m has high clay content and high plasticity. However, the total percentage fines for all the samples is less than 15 and are not good lining materials.

Compacting soils at the study area at water contents above 25.01% will result in a relatively dispersed soil structure that is weaker, more ductile, less pervious, softer, less susceptible to swelling and more susceptible to shrinking. Conversely, when the soil on the proposed landfill site is compacted at a water content lower than 25.01%, it will result in a flocculated soil structure with random orientation that has opposite characteristics of when it is compacted wet at the optimum water content. The samples with high clay content exhibit low MDD values with high OMC whiles samples with low clay content exhibits high MDD values with low OMC values. The hydraulic conductivity values of the in-situ soils are greater than 10⁻⁹ m/s proposed by Roehl et al. [64] . The soils at the proposed area at Domeabra should be compacted to achieve the required hydraulic conductivity. This will increase the residence time for leachates to interact with the clay liner thereby degrading the leachate plume.