Tectonic Setting of the Kenya Rift in the Nakuru Area, Based on Geophysical Prospecting

In this paper, we present results of tectonic and geophysical investigations in the Kenya Rift valley, in the Nakuru area. We compiled a detailed geological map of the area based on published earlier works, well data and satellite imagery. The map was then integrated with original fieldwork and cross sections were constructed. In key areas, we then performed geophysical survey using Electrical Resistivity Tomography (ERT), Hybrid Source Audio MagnetoTelluric (HSAMT), and single station passive seismic measurements (HVSR). In the study area, a volcano-sedimentary succession of the Neogene-Quaternary age characterized by basalts, trachytes, pyroclastic rocks, and tephra with intercalated lacustrine and fluvial deposits crops out. Faulting linked with rift development is evident and occurs throughout the area crosscutting all rock units. We show a rotation of the extension in this portion of the Kenya rift with the NE–SW extension direction of a Neogene-Middle Pleistocene age, followed by the E–W extension direction of an Upper Pleistocene-Present age. Geophysical investigations allowed to outline main lithostratigraphic units and tectonic features at depth and were also useful to infer main cataclasites and fractured rock bodies, the primary paths for water flow in rocks. These investigations are integrated in a larger EU H2020 Programme aimed to produce a geological and hydrogeological model of the area to develop a sustainable water management system.


Introduction
The East African Rift System is an active divergent boundary between the Nubian and Somalia plates ( Figure 1) and is the ideal place to investigate the mechanisms and dynamics of continental rifting and breakup. Thanks to these ideal conditions a large amount of geological, volcanological and geophysical investigations were carried out over the years and a large amount of data were acquired, especially about relationships between tectonics, magmatism, and sedimentation (see reviews in [1][2][3][4][5]). Tectonic investigations have greatly developed in recent times due to the availability of high resolution satellite imagery, that coupled with structural field investigations in the key areas led to better understanding the role of magmatism to initiate rifting, the role of fluids for wakening of the continental lithosphere and the role of preexisting tectonic features for breakup geometry and faults development [5][6][7][8][9][10][11].
In order to better understand the geometry and kinematics of rifting in the central Kenya rift, we carried out geological fieldwork in the Nakuru area (Kenya). To constrain the geometry of faults and lithostratigraphic units below widespread alluvial deposits, we extensively used geophysical prospecting: (a) electrical resistivity imaging methods, because of their effectiveness to distinguish main subsoil lithologies and locate faults, and (b) the horizontal-to-vertical spectral ratio technique (HVSR), a method widely used in seismic exploration as a cost-effective survey to assess seismic-amplification effects in terms of S-wave resonance frequency, applied to achieve information about the spatial variability of shallow deposits thickness.
Integration of geological and geophysical investigations led to better constrains on the geometry and kinematics of this section of the Kenya rift.  [1], background image by NOAA [12]. Present day plate movement after Saria et al. [4] (Nubian plate fixed). (b) Geological map, after Persits et al. [13], Geological Survey of Ethiopia [14], Geological Survey of Tanzania [15] and Schlüter [16].

Geological Setting
The East African Rift System is a series of half grabens (rift valleys) bounded by border faults that can range in length from 10 to up to 200 km and filled with volcanic rocks and sediments. Borders of rifts valleys are usually uplifted areas with mountain belts and plateaus (Figure 1a). The East African Rift System can be divided in two parts, the Eastern branch and the Western branch. A third southeastern branch includes basins offshore of the Africa coast and Madagascar. The Eastern branch runs from the Afar region (NE Ethiopia) through the main Ethiopian rifts, the Turkana lake rift, the Kenya (Gregory) rift, up to the North-Tanzanian divergence (for extensive discussion see [1,17,18] and references therein).
First evidences of Cenozoic magmatism in the area is linked with mantle plume upwelling and started at 45 Ma (see review in [19]) followed by diffuse flood basalts (trap series) that were emplaced at about 30 Ma [20,21]. These basalts overlie a Paleogene-Neogene and a late Paleozoic-Mesozoic sedimentary succession (Figure 1b), unconformably deposited above a metamorphic basement, belonging to the Arabian-Nubian shield (see [22] for review and references). Starting from the Miocene (Figure 2) plateau uplift occurred and extension affected the area, with the development of rift valleys and associated magmatism and sedimentation.
The study area is located in western Kenya (Nakuru County), in the Kenya rift ( Figure 1b). This part of the Kenya rift is in an advanced stage of the rifting processes with magmatism and tectonic events currently located only in the inner trough main rift [2,3,5,23].  [19], Baker [24], Leat [25], Roessner and Strecker [26]. Labels used in geological map of Figure 3 are indicated, in parentheses are formational names. Age of Oligocene trap series and Eocene basaltic volcanism outside of the study area are reported. Chronostratigraphic scale after Cohen et al. [27]. Extension directions are from this study.

The Kenya Rift in the Menengai-Longonot Area
The investigated area is located in the central part of the Kenya rift (Gregory rift). The morphology of the area is the result of its tectonic evolution, and the following main features can be recognized, from W to E (Figures 3a and 4):  [28], Thompson and Dodson [29], Thompson [30], McCall [31], Jennings [32], Clarke et al. [33], Williams [34] and original field work. See Figure 1b Figure 3, based on Baker et al. [35], Smith [36], Roessner and Strecker [26], Simiyu and Keller [37], and original field data.

1.
The western rift shoulder (Mau escarpment), bordered by the Mau Fault, a fault system (fault zone) later overprinted by younger faults and partially covered by younger deposits; 2.
The inner trough (Lake Nakuru, Lake Naivasha), bordered by the Mau Fault (to the west) and the Bahati and the Kinangop fault (to the east), this is the lowermost portion of the rift valley; 3.
The intra-rift plateau (Bahati-Kinangop plateau) bordered by the Bahati and the Kinangop fault (to the west) and the Sattima fault to the east; 4.
The eastern rift shoulder (Aberdare range), east of the Sattima fault.
The oldest rocks in the area are medium-to high-grade metamorphic rocks belonging to the Precambrian basement (labeled PC in Figure 3b), exposed in limited outcrops in the southwestern part of the area. Metamorphic rocks are represented by quartzites, micaschists, feldspathic and biotite gneisses. Above the basement, a thick Neogene-Quaternary succession of volcanic rocks occurs ( Figure 2).
The oldest volcanic rocks in the area are Lower Miocene basalts (Mb in Figure 3b) linked with a sub-aerial fissure and flood style volcanism, during crustal up-warping are most likely related to the presence of a mantle plume [36]. These basalts are present in the eastern part of the area, but are not reported in the western part of the area above the Precambrian basement.
A second large stage of flood volcanism in the same geodynamic context occurred during the Middle-Upper Miocene and is represented by phonolitic lavas with local basalt flows (Mp). This regional flooding event is the largest phase of volcanism in the central Kenya rift and phonolites covered the whole area at this time. In the study area these phonolites are found both in the western and in the eastern rift shoulders.
In the Upper Miocene, volcanism was restricted to the eastern portion of the rift and in the eastern rift shoulder are documented olivine basalts (Aberdare basalts, [35]) with intercalated trachytic tuffs (Ma). These basalts do not crop out in the western part of the study area.
During the Pliocene the entire area was covered by a thick succession of tuffs, trachytes and phonolites (Pb). It is a important phase of explosive volcanic activity with widespread pyroclastic deposits that extend from the Equator to south of Nairobi. Eruptive sources are not preserved, but facies and thickness analysis suggest they were erupted from the Nakuru-Naivasha area [31,34,35]. A significant thickness of the Pliocene tuffs in the central part of the Kenya rift is therefore expected. This major ignimbritic event can be correlated with an important rifting stage due to penetration of the asthenospheric diapir in the lower crust [36].
Ongoing extension in the central part of the Kenya rift during Lower-Middle Pleistocene led to voluminous volcanism, represented by basalts, trachytes, phonolites, basanites and tuffs (Qb in Figure 3b). These rocks crop out extensively in the Mau plateau and in the main rift, but are missing in the east rift shoulder. This volcanic succession is documented to extend across the Kenya rift south of our study area [35,39].
Younger Quaternary volcanism is limited to the inner trough of the main rift. In the southern part of the area, the Longonot, Olkaria and Eburru volcanic complexes are developed [33], while the Elmenteita and the Sirrkon volcanic complexes are located in the rift center. In the northern part, the main rift is covered by the volcanic rocks of the Menengai volcano. All these volcanic complexes (Q1, Q2, Q3 and Q4 in Figure 3b) are characterized by basaltic and trachytic lava flows, and thick pyroclastic deposits. Contemporaneous with the Quaternary volcanism sedimentation occurred in the area. The sediments (Aq) are represented by conglomerates, sandstones, silts, clays and diatomites of lacustrine and fluvial origin, intercalated with tuffs and volcanic rocks.

Geology of the Nakuru Area
In this section, we present the stratigraphy and the main tectonic features of the volcano-sedimentary succession in the Nakuru area. The geological map of the area is shown in Figure 5. More tectonic inferences are discussed in the final section after the results of geophysical investigations. The oldest rocks of the area crop out east of the Nakuru Lake and form the Bahati platform, between the Bahati fault and the easternmost Sattima fault (Figure 3). Limited outcrops are also located SW of Lake Nakuru. They are trachytic ignimbrites with interbedded trachytic lavas and tuffs, with a thickness at least of 250 m [25]. No evidence of volcanic centers related to this succession are documented and it is possible that sources where located west of the Bahati fault, and now covered by more recent deposits.
This succession it is not directly dated, but from relationships with underlying and overlying rocks, a Pliocene age is presumed [25,31].  [25], McCall [31], Leat [45] and original field work.

Meroronyi Tuffs (Rt)
At about 2 Ma (lower Pleistocene), trachytic lavas were emplaced in this part of the Kenya rift [35]. In the study area, these rocks outcrop near the Bahati fault, above the Bahati Trachytes and Tuffs, and are defined as Meroronyi Tuffs by Leat [25]. The succession is formed by trachytic tuffs, intercalated sediments and welded ignimbrites, with a total thickness of about 30 m.

Mbaruk Basalts and Trachytes (Bb)
The succession is represented by trachytic lavas, followed by thicker basaltic lavas, an upper trachytic lava flow and upper fluviatile sediments, with a total thickness of about 40 m. These are the youngest rocks to crop out all across the rift valley floor and in the study area are exposed in the western part of the Bahati platform (along the Bahati fault), between Lake Nakuru and Lake Elmenteita and also west of Lake Nakuru. West of Lake Nakuru trachytic air-fall tephra are also present (West Lake Trachytes of [25]).

Sirrkon Trachyte Lava and Tuffs (St)
East of Lake Nakuru a single volcanic edifice is present, the Sirrkon (or Lion) Hill. This volcanic edifice is cut by N-S striking faults, and now only remnants of a crater are still recognizable. The volcanic succession is represented by trachytic lava flows, pyroclastic and fall deposits. Trachytic lava is also documented north of Sirrkon (Plaat hill and Hyrax hill) and are interpreted by Leat [25] as strombolian cone deposits. The age inferred is late Pleistocene, based on correlation with nearby trachytic rocks outcropping in the area (Menengai volcanics).

Menengai Volcanics (Mp, Ms, Ml, Mt)
The Menengai Caldera is the main topographic feature in northern part of the Nakuru area that developed during the evolution of the Late Quaternary Menengai volcano. The Menengai Caldera is one of the major high-temperature geothermal fields in Kenya and numerous studies have been carried out in the area (see [48][49][50][51][52] for references).
The evolution of the Menengai volcano encompasses three stages: 1.
Formation of a shield volcano with low-angle slopes (pre-caldera stage); 2.
Collapse of the volcanic edifice and formation of the caldera (syn-caldera stage); 3.
Eruption of lavas onto the caldera floor (post-caldera stage).

Pre-Caldera Volcanics (Mp)
Remnants of the shield volcano that developed during the first stage of evolution are still recognizable and form low hills around the Menengai Caldera. The approximate extension of the shield volcano is indicated in Figure 5. This early shield is composed of trachytic lavas and interbedded fall deposits, dated at 0.18 Ma [45] with a maximum thickness of about 300 m. This succession can be observed at the cliffs of the caldera borders or in limited outcrops NW of the Menengai Caldera.

Syn-Caldera Volcanics (Ms)
Between 29,000 and 8500 B.P. [45,53,54], eruption and collapse of the caldera occurred, accompanied by the development of two cycles of trachytic ignimbrites and intercalated air fall deposits, locally separated by sediments. Ignimbrites and air fall deposits cover an area of about 1350 km 2 and can be found up to 30 km from the caldera. Ignimbrites and air fall deposits are found also in the nearby Lake Nakuru and Lake Elmenteita area, interbedded with the ongoing lacustrine sedimentation.

Post-Caldera Volcanics (Ml, Mt)
Post-caldera activity is represented by trachytic lava flows with blocky and ropy morphologies inside the caldera (Ml), as tuffs and air-fall deposits separated by soil horizons outside of the caldera (Mt), and strombolian pumice cones. Two trachytic tuff cones at Honeymoon and Crescent Hill, between Nakuru and Lake Nakuru, are considered satellite vents of the post-caldera series [25].

Bahati Sediments (Bs)
On the Bahati rift shoulders a few meters of sand and silt deposits are locally preserved, derived from erosion of soils and tuffs. Similar deposits are also reported from the Mau plateau, with a thickness of up to 200 m [55].

Elmenteita Basalts (Eb)
In the southern part of the study area ( Figure 5) west of Lake Elmenteita, olivine basalts are present [31,33,38]. Basalts are not affected by faulting, so a recent Quaternary age can be inferred. Many cinder cones are present and probably represent the vents from which the basalts were erupted.
These deposits represent shallow lacustrine (testified by diatomite occurrence) and fluviatile deposits with intercalation of pyroclastic material. Volcanic activity is documented to be continuous throughout the sequence.

Alluvium, Trona Impregnated Silt (Al)
Recent alluvial deposits occur in swamp area near the Nakuru and Elmenteita lakes. They are usually represented by clay and silt deposits. Silts are often impregnated by trona, a hydrous acid sodium carbonate.

Tectonics
The study area ( Figure 5) shows a flat topography and outcrops with small scale tectonic features exposed are scarce. More considerations can be put forward from analysis of the geological map and the digital elevation model (DEM), presented in Figure 3a.
The main tectonic features of the area are N-S striking normal faults that lowered the central part of the rift. The N-S faults do not cut either the Elmenteita basalts or the Menengai Caldera, of late Holocene age, so their activity before late Holocene can be put inferred. The N-S faults are prevalent in the central part of the area, while more NW-SE striking faults are present in the Bahati-Kinangop plateau. It is important to note that although the main normal faults of the Nakuru area, i.e., the central part of the main rift, are N-S oriented, the main direction of the Kenya rift here is NW-SE, as evident in Figure 3 (strike of the Mau and Bahati escarpment) and in Figure 1.
More discussion about tectonics of the area is presented after presentation of the results of the geophysical investigations.

Geophysics
Extensive geophysical fieldwork was performed in the Nakuru area during October and November 2018, aimed to integrate the geological model and observations derived from surface outcrops with deep information. We performed electrical resistivity surveys at two different scales by using Electrical Resistivity Tomography (ERT) and Hybrid Source Audio MagnetoTelluric (HSAMT). Single station passive seismic measurements (HVSR) were also performed, at 55 locations to achieve information about the spatial variability of superficial deposits thickness across the study area.
The surveys were planned taking into account main geological features and logistical constrains, i.e., the technical suitability of survey areas according to geophysical methods applied. Resistivity measurements (i.e., ERT and HSAMT lines) were located across the Rift Valley east of the Nakuru Lake along three different lines, two orthogonal to the main geological features, here N-S oriented, and one oriented NW-SE crossing the entire valley and reaching the eastern part of the rift. HVSR were spread across all the study area in order to cover as much area as possible. Location of the ERT and HSAMT lines, as HVSR points are shown in Figure 5. The purpose of a electrical resistivity survey is to individuate the spatial resistivity distribution in the subsurface. The resistivity ρ of a medium is defined as: where R is the electrical resistance, L is the length of a cylinder and S is the area of its cross section. Conductivity is the inverse of resistivity. The electrical resistance R is defined by Ohm's law as: where V is the potential and I is the current. In a homogeneous half-space, with point source of current, the current flow is radial from this point and the electrical equipotential surfaces are hemispherical. In this configuration the potential can be expressed through the resistivity and the current: In an elementary quadripolar configuration, on homogeneous subsoil and with four electrodes called C 1 , C 2 , P 1 , P 2 ( Figure 6), the feeding of current from C 1 -C 2 dipole, produces current flow and a correspondent potential field with equipotential surfaces, which can be measured through P 1 -P 2 dipole. The potential difference measurement between P 1 -P 2 dipole can be expressed in analytical way by the equation: where C 1 P 1 , C 2 P 1 , C 1 P 2 and C 2 P 2 represent distances among every couple of electrodes C 1 , C 2 , P 1 , P 2 . The electrical resistivity ρ can then be calculated as follows: where: a geometrical factor that depends on the relative disposition of the four electrodes. In non-homogeneous subsoil, the current flow distribution depends on the resistivity of the medium, and it is concentrated into the more conductive volume [57].
The electrical resistivity of rocks and soils is a function of solid part (grain size and mineralogy), pore size distribution and connectivity, fluids content on pore and their resistivity properties and temperature, gas content [57][58][59][60].
Electrical resistivity for rocks and soils, range from few Ω·m to several tens thousands Ω·m (Figure 7). Most rocks near the Earth's surface are composed by insulating grain minerals and have low conductivity, conduction of electricity is mostly through groundwater contained in the pores of the rocks and along surface layers at the contact of rocks and solution. Soils can have clayey conductive minerals and, rarely, metallic conductive minerals. In the case of clayey soils, resistivity depends on electrical charge density at the surface and is related to the cations exchange capacity. When massive metallic minerals are present without discontinuities the current flow is due to the free electron displacement.

Data Acquisition
We distributed two ERT lines across the study area ( Figure 5), the two profiles have a length of 7670 m and 3830 m, for profile ERT1 and ERT2, respectively. The high-resolution geoelectrical resistivity measurements were performed using the georesistivimeter Syscal-Pro Switch from IRIS Instruments. In order to reach a maximum depth and extend laterally along the profile, the roll-along technique was applied and the measurements were performed using the georesistivimeter Syscal-Pro Switch from IRIS Instruments. The base configuration adopted was with 96 electrodes (10 m spaced) and the array was moved forward each time with an overlap of 48 electrodes (half of a single profile). In order to explore the subsurface and highlight the lithological variations both in vertical and horizontal direction, we preliminary tested over a portion of the profile ERT1 four quadripolar configurations: Dipole-Dipole (DD) and Schlumberger Reciprocal (SR) optimized for multichannel measurements, Wenner (W) and Wenner-Schlumberger (WS) for single channel measurements [62]. The Schlumberger reciprocal is a manipulation of the classical Schlumberger with the current electrodes in the inner part of the quadripole.
As Figure 8 shows, the DD configuration produces the poorest signal among the applied configurations, where the 65% of the measures have a ∆V lower than 1 mV. As expected, W and WS have the strongest signals with a mode equal to 10 mV. The higher frequencies of V p > 10 mV, obtained by the W configuration compared to the WS configuration are probably due to the lower n factor. In terms of root-mean squared error (RMS) between observed and calculated apparent resistivity we detect the best result for the Wenner-Schlumberger and the worst for the Dipole-Dipole configuration (Table 1).  Based on all the above considerations, the Wenner-Schlumberger was chosen as most suitable configuration to perform the ETR survey.

Data Processing
In order to process the electrical resistivity data, we applied RES2DINV V. 4.03.13 [63] inversion routine. The data processing can be described in two main steps. In the first step bad data were removed by using both threshold values on each measurement (st-dev > 5%, abs (∆V) < 1 mV, apparent resistivity < 0 Ω·m) and identification of the outliers. In the second step, we carried out the tomographic inversion starting from apparent resistivity values. The robust constraint (L1-norm) on inversion method [62] was applied for the processing of the resistivity model, since the algorithm attempt to minimize the absolute changes in the resistivity values and, therefore, to produce models with sharp interfaces between different regions with a high contrast of resistivity.
The optimization method used by the inversion routine of RES2DINV basically tries to reduce the difference between the calculated and measured apparent resistivity values by changing the resistivity of the model blocks. This difference can be assessed by the RMS. The routine stops when a defined number of iterations is reached or the variation in terms of RMS is lower than a threshold value and successive iterations do not improve the results. However, the model with the lowest possible RMS error can sometimes show large and unrealistic variations in the modelled resistivity and might not always be the "best" model from a geological perspective. In general, the most prudent approach is to choose the model at the iteration after which the RMS error does not change significantly. This usually occurs between the 3rd and 5th iterations [63].
Finally, RMS values were used to remove noisy data, where the percentage difference between the observed and calculated apparent resistivity was far and well separated from the average values.

Theory and Basic Principles
The magnetotelluric method or magnetotellurics (MT) is a passive exploration technique to infer the electrical properties of the Earth's subsurface [64][65][66]. Interaction of the solar wind with the magnetosphere and worldwide thunderstorm activity produce variations of the Earth Magnetic Field, inducing electric currents (telluric currents) in the Earth's subsurface. These telluric currents are the energy source used by magnetotellurics. Measurements of naturally occurring variations of the magnetic field at the Earth's surface (B t ) and the induced electric field (E t ) provide information about subsurface electrical resistivity and therefore about rocks and geological structures in the Earth's interior. The main advantage of this method is its capability for exploration from shallow to very great depths (from a few tens of meters to several hundreds of kilometers), without the use of an artificial power source. For more information about the method refer to Campbell [67], Simpson and Bahr [68], Chave and Jones [69], and references therein. In the case of HSAMT, a combination of natural MT signals and human-made (hybrid sources) signals are measured, to overcome the problem of weak-amplitude of the signal in the frequency range from 0.8 to 64 kHz.

Data Acquisition
During this study, we used the Stratagem EH4 Conductivity Imaging System (by Geometrics, Inc. and Electromagnetic Instruments Inc., ver. 2.19) to perform the HSAMT survey. The Stratagem EH4 system consists of two basic components: a receiver and a standard transmitter. The latter is used to provide additional high-frequency signals in the range from 0.8 to 64 kHz, where natural signals are weak and, therefore, improve data quality. Electric dipoles and magnetometers are laid out in perpendicular directions (i.e., E x , E y , B x and B y ) and both natural and transmitted frequencies are recorded from distant and non-polarized sources.
In this study, we carried out a total of 35 soundings along three profiles ( Figure 5) within the frequency range from 10 Hz to 100 kHz. The dipole length was fixed at 25 m, and the coils followed the right-hand rule, with the z axis directed to increasing depth. The antenna was located at an average distance of 250 m from the acquisition station, far enough to avoid near-field effects due to the low resistivity subsurface values at the site. For each sounding an oversaturated NaCl solution was used to improve the ground coupling with the electrodes. Along the MT_L1 line, we carried out 18 soundings with X azimuthal direction N120 • ; along the MT_L2 line, we carried out nine soundings with X azimuthal direction N250 • ; and along the MT_L3 line, we carried out eight soundings with azimuthal direction N2100 • .

Data Processing
After the data recording by Stratagem EH4, data geometry was first corrected according to the coordinates obtained on the site by GPS measurements. After the geometry correction, the first step of the data processing is the mesh building. The mesh was discretized in the x-direction using a regular spacing and in the z-direction using an incremental spacing from one layer to another, with a factor of 1.1. The maximum depth of the grid was chosen according to an average depth obtained by the one-dimensional inversions, and correspond to 700 m, as an initial resistivity value of 10 Ωm was assigned to the whole mesh.
Aiming to obtain the most correct data performance and reduce the data noise level, two smoothing operations were applied on the data: one on the transversal magnetic (TM) mode and one on the transversal electric (TE) mode. In order to remove possible distortion of resistivity images due to superficial inhomogeneity of the geological structures present in the investigated subsoil, the static shift correction was applied to the TM mode (the TE mode is not considerably affected by this error).
The inversion was then computed using the effective impedance and the "smoothness constrain". The "smoothness constrain" uses a procedure based on the Gauss-Newton least-squares method and returns a stable parameter distribution. The resistivity limits of the model were set at 1-5000 Ωm and the inversions where terminated after a fixed number of iterations (i.e., 34). The first four inversions were used to correct the galvanic effects due to the connections between the electrodes and the earth. The remaining 30 inversions were sufficient to obtain a satisfactory image with a low RMS.

Theory and Basic Principles
The horizontal-to-vertical spectral ratio (HVSR or H/V) technique is used to retrieve information about the shallow-subsoil seismic properties by single-station measurements performed on the soil surface [70][71][72][73]. This method, widely used as a cost-effective survey to assess seismic amplification effects, is therefore useful to evaluate the presence of impedance contrast, its magnitude and estimate the thickness of cover deposits.
The HVSR is a curve that represents the value of the ratio between the average spectral amplitudes of the environmental vibrations as a function of frequency. This is an experimental evaluation of the spectral amplitude ratios between the horizontal components (N-S, E-W) and the vertical component of the environmental vibrations on the ground surface measured at a point with a special three-axial seismometer [70,74].
In the field of environmental vibrations, there are both volume waves (P and S) and surface waves (Love and Rayleigh) and their contribution to the total produced energy varies from case to case and depends on the frequency range considered. Seismic noise can be linked both to a natural origin (low frequencies) or to human activity (high frequency) [72].
The fundamental assumption of the microtremor analysis is that the multiple reflections of the seismic waves, which remain trapped in the sedimentary cover lying on the stiffer material, give rise to constructive interferences that are the cause of resonance. The frequency of the peaks identified in the HVSR curves presents a non-linear and inversely proportional relationship with the thickness of soft shallow layers. This relationship is expressed by Equation (7) where f is the fundamental resonance frequency, V s is the average velocity of S waves of the layer and H is the thickness of the soft layer [75,76].
Equation (7) shows that H/V survey can also provide indications of stratigraphic nature: starting from a measure of environmental vibrations that limit the frequency value and the shear velocity of the coverage, the depth of the main seismic reflectors can be estimated or vice versa.

Data Acquisition
The digital tromograph TROMINO Engineering by Micromed, with nine channels and a built-in GPS receiver was used to perform the HVSR survey. The choice of location of the acquisition points was identified in the areas far from anthropic or natural structures (fences, buildings, roads, trees) that could interfere with the survey. Four preliminary measurements of 20 min were performed to evaluate the signal quality. Following the results of the preliminary measurements, the recording length was extended to 30 min in order to compensate for the presence of noisy transient events. The sampling frequency was set at 128 Hz for all the measurements. For a common evaluation of the signal directionality, the instrument was oriented to the magnetic north. Before data recording, the area around the Tromino was cleaned up of the elements that could generate transient events such as grass and stones.

Data Processing
The analysis of entire passive seismic survey was performed through the software Grilla ver. 7.5 (MoHo s.r.l. release 2018) which accompanies TROMINO. In Grilla, the three components of the seismic signal are represented in terms of spectral powers (square of the acceleration normalized to the frequency) from which the ratio trend between the horizontal spectral components and the vertical component is extrapolated as a function of frequency.
The recordings were divided into time intervals (i.e., windows) of 20 s and then a spectral analysis of the three components was performed and the spectral ratio for each of the windows obtained was calculated. All the HVSR data have been obtained by smoothing the spectra with a triangular smoothing function with a width equal to 10% of the central frequency. The final H/V curve was obtained as the average of the spectral ratios on all segments. In addition, for each data was generated a directional graphic which helped on the evaluation of the observed peak and a temporal stationary graph of the H/V spectral ratio.
For each observed peak, its validity was assessed by the SESAME criteria [77]. These criteria allow to evaluate the quality of the signal processing and the significance of the H/V peaks from a statistical point of view. In case the of failure to pass the SESAME criteria, it does not mean that there are no significant resonances but only that there are no well-defined individual peaks.
Finally, the frequency peaks individuated by HVSR were interpolated with the aim to produce a frequency map representative of the spatial variability of the impedance contrast detected (Figure 9).

Results and Discussion
Most rocks near the Earth's surface have high resistivity (Figure 7). Conduction of electricity negligibly occurs through the silicate crystalline lattice, but mostly through movement of ions in groundwater (solution) contained in the interconnected pore spaces, along the interface between pore space and pore walls, along grain boundaries and along fractures. Since the ions which conduct current are responsible for conductivity/resistivity of rocks, solutions with large number of ions (e.g., rocks with saline water in pores) will have the higher conductivity (lower resistivity). In general, the main factors that influence resistivity in rocks are porosity, temperature, rock texture, rock type, presence of clay minerals, water saturation, salinity of water and fluid permeability, as indicated in Table 2.
Geophysical explorations carried out so far in the study area were mainly oriented to investigate the geothermal potential in this portion of the Kenya Rift valley [50,78,79]. Sosi [80] and Sosi et al. [81] performed combined studies based on vertical electrical soundings and pumping tests, for groundwater flow assessment at the Kabatini well field (northern portion of our study area), highlighting layered models with large resistivity variations. Figure 10 shows two geological cross section across the Nakuru area that we constructed on the background geology of Figure 5, with resistivity values from our geophysical surveys. The following resistivity values were recorded: The high values of resistivity observed in the western portion of section ERT2 (Figure 10a) is assumed to be related to occurrence of predominantly lava flows, probably of the Sirrkon trachyte lava and tuffs (St) (500-1000 Ωm). From both sections, we can infer that the very first meters of soils have low resistivity, probably linked with soil moisture and water content. The underlying tuffs, lacustrine and fluvial deposits (Ql in Figure 10) show higher resistivity due to induration and carbonate precipitation. Pyroclastic deposits and pyroclastic rocks show resistivity values far below the typical values for igneous rocks (see Figure 7), this can be related to their high porosity. Underlying basalts and trachytes (St and Bt in Figure 10) have higher resistivity than the pyroclastic deposits. Figure 11 presents three cross-sections based on HSAMT that reach a deeper depth, and incorporate Figure 10 sections. Due to the smaller scale, the Quaternary tuff, lacustrine and fluvial deposits, the Menengai volcanics and the Bahati sediments are rendered all together and labeled "At". In these sections, the resistivity variations are larger than in Figure 10, ranging from a few Ωm to values of about 3000 Ωm. In the upper portion of the MT-L01 section lower resistivity rocks and deposits are present (mainly unconsolidated pyroclastic rocks, labeled as "At"), that overlay rocks with higher resistivity represented by predominantly trachytes and basalts (St and Bt). In the central and eastern part of the section significant lower resistivity values are recorded, that we interpret are related to presence of cataclastic and fractured rocks with the increased porosity and water content. Significant contribution to lowered resistivity of rocks in this area can be also linked to alteration of volcanic rocks along a fault zone, with production of alteration minerals as smectite and zeolites that usually have high cation exchange capacity and therefore high conductivity [82], as pointed out in the Menengai geothermal area [83].  Figure 5 for location. Sections include data from ERT1 and ERT2 lines ( Figure 10). Table 2. Effect of geological processes on resistivity, after Ward [84]. Effect of rock type is illustrated in Figure 7. The Bahati fault is the main tectonic feature in the surveyed area ( Figures 10 and 11), a west-dipping normal fault that led to uplift of the footwall Pliocene trachytes, tuffs and phonolites of the Bahati-Kinangop plateau. This fault is evident in profiles MT-L01 and MT-L02 of Figure 11 with low resistivity rocks in its footwall, probably due to widespread fracturing. This is not evident in the northern profile MT-L03, where high resistivity rocks are present throughout at depth. The Bahati fault has a strong geomorphological expression in the Mbaruk area, but is less evident to the the north in the Bahati and Kariandusi-Gilgil-Eburru areas (Figure 3), where it is overprinted by N-S striking faults. Consequently, we can infer that NW-SE striking normal faults are overprinted by N-S normal faults. This is also evident from Figure 3, where the NW-SE striking Mau Fault, responsible for the uplift of the Mau escarpment at the footwall, is not overprinted by N-S striking faults south of Njoro. Therefore, we argue that a change in extension direction occured in this part of the Kenya rift, from a NE-SW extension to E-W extension, as also documented at a larger regional scale [42,85,86]. Age constrains for this deformation can be inferred from geological map of Figure 5, where it is evident that the volcanic succession of the central part of the map (Sirrkon trachytes and tuffs and younger deposits) are affected by N-S striking faults only. Based on this evidence, we can argue that E-W extension predominated starting from the upper Pleistocene, i.e., after the emplacement of the Sirrkon trachytes and tuffs. NE-SW extension prevailed during the Miocene-Upper Pleistocene time span and it was responsible of the present-day large-scale NE-SW orientation of the Kenya rift. Time intervals of extension phases are also shown in Figure 2.

Conclusions and Final Remarks
Our investigations in the central part of the Kenya rift in the Nakuru area, Kenya (Gregory rift) included geological field work, together with geophysical prospecting. A careful analysis of geological maps available and original field work led first to compilation of a regional geological map ( Figure 3) and a detailed geological map for the Nakuru area ( Figure 5). To unveil the main tectonic features of the area and their prolongation at depth, geophysical investigations were carried out in the central part of the main rift zone, west of the Lake Nakuru. We studied electrical resistivity of rocks in the area first with Electrical Resistivity Tomography (ERT) methods and then with Audio MagnetoTelluric (HSAMT) methods. All these data were integrated with single station passive seismic measurements (HSVR).
All information was used to construct geological cross-sections that illustrate the shallow tectonic structure of the Kenya rift in this area (Figures 10 and 11). The central part of the main rift is characterised by a close spaced system of normal faults, striking N-S, well developed in the inner trough of the Rift valley, W and especially E of Lake Nakuru.
These normal faults crosscut almost all the volcanic succession of the area and a Holocene age for their activity can be inferred. Therefore, the most recent extension direction in this portion of the Kenya rift is E-W. Magnetotelluric and HVSR data allow us to trace the main tectonic features at depth and to construct geological cross-sections from the Bahati plateau to Nakuru Lake ( Figure 11). In these sections, the relevant thickness of the upper Pleistocene-Holocene pyroclastic rocks (and less developed trachytes and basalts) is shown. Faults can be traced at depth and cataclastic and fractured zones developed along the faults, as it can be observed from lateral variation of rock resistivity gathered from geophysical investigations.
At a smaller scale (Figure 3), the main faults affecting the Bahati-Kinangop plateau, the Mau plateau and developing the Bahati and Mau escarpment are NW-SE oriented. This is also the orientation of the Kenya rift at a larger regional scale. These faults affected the Miocene-middle Pleistocene succession and are absent from the Holocene volcanosedimentary succession of the inner trough in the main rift. Therefore, we can infer for these NW-SE oriented faults Miocene-middle Pleistocene activity with the main extension direction oriented NE-SW. Our investigations point out to a change in the extension direction in the Gregory rift, from NE-SW to E-W, that occurred in the upper Pleistocene. Funding: This research was funded by the European Union's H2020 Research and Innovation Programme "FLOWERED Project" (Grant Agreement n. 690378). The FLOWERED Project is aimed to contribute to the development of a sustainable water management system in areas affected by fluoride contamination in water, soils and food in the African Rift Valley countries (Ethiopia, Kenya, Tanzania).