You are on page 1of 15

Geothermics 79 (2019) 67–81

Contents lists available at ScienceDirect

Geothermics
journal homepage: www.elsevier.com/locate/geothermics

Geological structures controlling the placement and geometry of heat T


sources within the Menengai geothermal field, Kenya as evidenced by
gravity study

Isaack Kandaa, , Yasuhiro Fujimitsub, Jun Nishijimab
a
Graduate School of Engineering, Department of Earth Resources Engineering, Faculty of Engineering, Kyushu University, Fukuoka 819-0395, Japan
b
Department of Earth Resources Engineering, Faculty of Engineering, Kyushu University, Fukuoka, 819-0395, Japan

A R T I C LE I N FO A B S T R A C T

Keywords: Menengai volcano is one of the late Quaternary caldera volcanoes formed on a massive shield in the inner-trough
Bouguer anomaly of the Kenya rift valley, associated with a high thermal gradient resulting from shallow magmatic intrusion.
3-D gravity inversion While drilling of geothermal wells in Menengai area proved the existence of exploitable steam, the complexity of
Faults its geological setup has led to various technical setbacks where sited well targets turned out to be unproductive.
Menengai caldera
These challenges were attributed partly to the lack of adequate knowledge to delineate the depth and lateral
Kenyan rift
Geothermal system
extent of heat sources. The present work attempts to understand the geological structures that appear control the
placement and geometry of heat sources within the Menengai geothermal field. This study utilizes the sensitivity
of gravity data to deep structures resolution to determine a three-dimensional (3-D) inversion model. A total of
1823 gravity points were used in generating the Bouguer anomaly, using a Bouguer density of 2.23 g/cm3. The
model was constrained using estimated densities of drill hole cutting recovered from the geothermal wells. The
resultant model was then interpreted together with surface manifestations, geology, and geological structures.
Lineament structures were extracted from satellite images and used to enrich the geological structures. The
model showed a close relationship between faults and inferred geometry of subsurface volcanic complexes
displayed as discrete dike-like bodies in the near-surface environment. The structural system controlling the
extent of the dense trachytic body of Olrongai is distinct, but for the caldera, some inference can be made from
the results of this study. These faults are interpreted as the feeder dikes of the dense syenitic intrusives, that are
hosted within the trachytic formation, and believed to be the heat source for the geothermal system. These
structures act as conduits of magma to shallow levels, supplying the much-needed heat to the geothermal re-
servoir.

1. Introduction The Menengai geothermal field is located in the central segment of


the Kenya rift and is hosted partly within a major Quaternary volcano.
Exploration for geothermal resources in Kenya shows that the In this field, the Menengai caldera is the most central for its geothermal
Quaternary volcanic complexes of the Kenya rift valley offer the most resource and conspicuously the major geological feature in the area (see
promising prospects for geothermal exploration. This is attributed to Fig. 1). This area is characterized by a complex geological setting as-
their geodynamic setting and high-temperatures, which is indicative of sociated with an uplifted region of the so-called Kenya dome, a geo-
magmatic activity close to the surface (e.g., Mungania and Lagat, graphical upwelling province of anomalously high elevations. It is si-
2004). These thermally privileged zones are associated with local tuated at the confluence of structures that resulted from major tectonic
‘highs’ of the hot asthenosphere roof, from which the preferential events and covered for the most part by Cenozoic volcanic rocks (e.g.,
magmatic mass transfer towards the surface occurs (Geotermica Davis and Slack, 2002; Leat, 1991). This elliptical-shaped hoisted fea-
Italiana, 1987). Because of its encouraging indications of having an ture is a local culmination on the eastern rim of the East African plateau
exploitable geothermal resource, intensive exploration has so far fo- (e.g., Baker and Wohlenberg, 1971), conspicuously marked by three
cused on these areas (Clarke et al., 1990; Lagat et al., 2010; Mungania major central volcanoes; Mounts Kenya and Elgon in Kenya, and Mount
and Lagat, 2004; Omenda et al., 2014; Simiyu, 2010). Kilimanjaro in Tanzania.


Corresponding author.
E-mail address: isaac.kanda@mine.kyushu-u.ac.jp (I. Kanda).

https://doi.org/10.1016/j.geothermics.2018.12.012
Received 28 August 2018; Received in revised form 14 December 2018; Accepted 24 December 2018
0375-6505/ © 2018 Elsevier Ltd. All rights reserved.
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 1. Map of the study area. The high altitude (> 1500 m above sea level) area outlines part of the Kenya dome.

Several studies have been conducted on Kenya rift region and a post-caldera activity, the system represents the source for a significant
number of them on geophysics with focus on seismic and gravimetric thermal anomaly at shallow depth (Geotermica Italiana, 1987). Seismic
studies (e.g. Bullard, 1936; Fairhead, 1976; Fairhead and Reeves, 1977; studies conducted by Simiyu (2013) reveal the presence of a large, re-
Girdler, 1983; Girdler and Sowerbutts, 1970; Searle, 1970; Simiyu and silient and shallow (about 6 to 8 km) magma body extending from
Keller, 1998, 2001; Sippel et al., 2017; Smith and Andrew, 1962). Ex- within the caldera to the NE walls and Olbanita - Olrongai areas.
ceptional attention has been drawn to East African’ salient oval-shaped Several vents and active fumaroles are scattered on the caldera floor
negative Bouguer anomaly that covers the whole of east and west and Olrongai area (Fig. 2). At least the fumaroles at Olrongai show an
branches of the rift. The minimum of this negative anomaly is centered observable concurrence with tectonic structures. The fumaroles inside
over the East Africa rift system (EARS) near the equator (within Me- the caldera display a distinct pattern and perhaps they are sited on
nengai area) with the major axis striking NNE (e.g., Girdler, 1975, fractures formed during piecemeal caldera collapse (Leat, 1984). These
1983; Girdler and Sowerbutts, 1970; Swain, 1992). The negative structures provide a means of access to hydrothermal fluids to reach the
anomaly zone suggests an isostatic disequilibrium within the region surface (Geotermica Italiana, 1987). Similarly, diffuse soil degassing of
(McCall, 1967), considered to be produced either by the low density of magmatic Carbon Dioxide (CO2) and Radon 220 (220Rn) emanating
the surficial volcanics (Fairhead, 1976) or perceived asthenospheric from these structures were measured to be anomalously higher than the
intrusions (Chorowicz, 2005; Fairhead, 1972; Girdler et al., 1969). background (Kanda, 2010). This suggests that either these structure are
The Menengai caldera is distinguished by the existence of a graben deep-seated or the magma body is shallow enough for the gases to as-
basin confined within a fault segment, and its partial collapse features cend freely and escape naturally through the soil.
suggest that portions of an adjoining caldera wall also represent surface Preceding studies are part of concerted attempts to understand the
expressions of an embayed ring-fracture (Leat, 1984; see Fig. 2). Such regional lithospheric structure of the EARS (e.g., Chorowicz, 2005), as
faults are not common in the Kenya rift valley, which is a tectonic well as local occurrence and extent of geothermal systems (e.g.,
province characterized by the presence of numerous parallel normal Geotermica Italiana, 1987; Lagat et al., 2010; Mungania and Lagat,
faults (e.g., Fairhead et al., 1972; Leat, 1984; McCall, 1967). Leat 2004; Simiyu, 2013). The studies on geothermal resources have led to
(1984) observed that the ring-fault structure must have been controlled the drilling of producing wells within the caldera; consequently,
by a local crustal anomaly at Menengai as explained by a collapse into a 3 × 35 MW power plants are currently under construction (GDC, 2018).
magma chamber. The voluminous ash-flow tuff erupted as single flow Despite these encouraging results, development of geothermal in
unit implying the magma must have been stored, prior to eruption, in a Menengai has some challenges. The geothermal system of Menengai
vast, shallow reservoir. caldera is quite complex due to its intricate geological setting. This has
This magma chamber is estimated to have a minimum volume of led to various technical setbacks where some promising well targets
170 km3 and considering recent effusion of trachytic magma during the became cold and unproductive, while some experienced prolonged

68
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 2. Map showing elevation (m), faults, fracture and location of fumaroles in the study area.

recovery after drilling (GDC, 2018). Therefore, in order to understand tensional forces resulted in block faulting, which includes tilted blocks,
the geothermal system in a way that increases power production, ad- evidently seen in both the floor and rift scarps. The rift trough is
ditional inputs from more detailed studies are indispensable. These truncated by several normal faults, which represent persistent and
challenges can partly be attributed to the lack of adequate information wide-ranging tectonism under the rift floor. The geometry and sizes of
for defining the geothermal heat source. In this paper, we utilized the well preserved erosional surfaces of Miocene as well of Late Pliocene
results from 3-D gravity inversion model to infer the heat source geo- ages give evidence for three central phases of epeirogeny and a crustal
metry and delineate the geological structures controlling its extent flexure that consist of episodic up-arching of central Kenya and local
within the Menengai geothermal area. These results can be used to down-warping of the coastal zone (e.g., Baker and Wohlenberg, 1971;
locate target areas for drilling of promising production wells or at least Saggerson and Baker, 1965). Like the rest of the East African rift, the
delineate favorable areas for further studies. central Kenya rift has undergone a very complex geological evolution
and tectonic movements.
2. Geological and Tectonic Setting Volcanism in the Kenya rift occurred almost continuously from
Early Miocene to Holocene times. The succession in the central part of
The general geological background of the study area has been in- the rift consists of Pliocene and Pleistocene trachytes (with subordinate
corporated both in the quantitative and qualitative interpretation phase basalts) overlying Miocene phonolites, which in turn rest on the
of the observed Bouguer gravity anomaly. A clear understanding of the Precambrian metamorphic basement (e.g., Leat, 1984; Williams, 1978).
regional geology of the central Kenyan rift and the adjacent areas is, The domal uplift in the central rift during the Late Miocene led to
therefore, required for the interpretation of the gravity data. To define massive fissure phonolitic eruptions and emplacement (Baker et al.,
the spatial extent of major rock units, four geological maps with a scale 1971), locally referred to as Rumuruti phonolite (McCall, 1967). Plio-
of 1:125,000 were used to construct the structural and geological map cene phonolitic trachytes outcrop in the Olrongai area, whereas
of the study area (Fig. 3). The maps were drawn from the geological younger trachytic flows cover the caldera floor, and both are associated
reports of Nakuru and Thomson’s Falls - Lake Hannington areas with the occurrence of fumarolic activities.
(McCall, 1967), Kabarnet – Eldama Ravine area (Walsh, 1969), and Menengai geothermal area is located within an area characterized
Molo area (Jennings, 1971). by an intricate tectonic activity linked with the rift triple junction; the
The East African rift has long been recognized as a typical illustra- latter being a zone at which the failed rift arm of the Nyanza rift joins
tion of a continental rift (Chorowicz, 2005; Fairhead, 1976), which the central Kenyan rift valley. Apart from the Menengai caldera ring
forms part of the largest Tertiary–Quaternary rift system that extends structure, two rift floor tectono-volcanic axes (TVA) are important in
south-southwest from the junction of the Red Sea and Gulf of Aden to controlling local geothermal system; the Molo TVA which trends NNW-
the Zambezi river in Mozambique (McConnell, 1972). The Kenyan SSE across the Olrongai area and the Solai TVA which tends NNE-SSW
portion is characterized by extension tectonism where the E-W direction. These two structural features are characterized by long

69
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 3. Surface geology map of the study area, modified from previous work done by McCall (1967); Jennings (1971) and Walsh (1969).

narrow bands of a thermally anomalous zone, where faulting and vol- the rock formation is predominantly composed of trachyte, tuff, and
canic activity tend to be concentrated (Geotermica Italiana, 1987). pyroclastics (e.g., GDC, 2018; Mibei, 2012; Omondi, 2011). However,
The main trend of these tectonic structures in the study area is erratic lenses of syenitic intrusives and basaltic rocks have also been
predominantly of N-S, NE-SW, and NW-SE, and mostly lie to the north encountered in some of the wells. These syenite intrusives were inter-
of the caldera (Fig. 3). A similar structural trend is observed across the cepted intermittently in some wells, suggesting that magma pulses from
central Kenyan rift as shown in Fig. 1. The other important tectonic the main magma chamber were injected into the overlying formations
feature is the arcuate-shaped structure that cuts across the caldera from in the form of dikes and other intrusives. At least four of these wells
east to west. A thermal manifestation of numerous steaming grounds recorded bottom-hole temperatures close to 400 °C at depths of about
occurs along this structure and is believed to act as a magma chimney 2.1 km, and the presence of fresh glass encountered by some wells in-
where magma is believed to be shallower (GDC, 2018). It is also con- dicate a possible encounter with magmatic intrusive at these shallow
sidered to have been a dominant structure during the uplift of the post- depths (GDC, 2018). The inference is supported by drilling challenges
caldera dome and the pathway for magma ascent and effusion (Strecker that were experienced during offloading, where the torque, weight on
and Melnick, 2013). The post-caldera dome structure is characterized bit, pressure and temperature increased substantially.
by a high altitude summit (> 2 km), shown in Fig. 2 that resulted from
piling up of volcanics erupted either of a central vent or closely spaced
vents near the center of the caldera (Leat, 1984). 3. Methodology
The local surface geology, shown in Fig. 3, is composed mainly of
unconsolidated low-density volcanic sediments that cover an area of 3.1. Automatic extractions and analysis of lineaments
876 km2 (72% of the total area), and tuffs and ignimbrites covering
160 km2 (13% of the total area). Remaining formations consist of Properties of lineaments that are related to geological displacement
phonolites, trachytes, and basalts with a combined surficial area of have played an essential role in tectonic studies for the delineation of
15%. structural units or identification of geological boundaries (e.g., Abebe
Menengai is a considerably young caldera volcano (less than et al., 1993; Chandrasiri Ekneligoda and Henkel, 2010; Hashim et al.,
200,000 years), composed mostly of silica-oversaturated, peralkaline 2013; Kageyama et al., 2000; Marghany and Hashim, 2010; Mountrakis
trachyte, with smaller volumes of metaluminous trachyte (Leat, 1984). and Luo, 2011; Noltimier et al., 1998; Saadi et al., 2011; Salati et al.,
Two periods of caldera collapse have been recognized, and each of them 2011). Analysis of these structures from remotely sensed satellite data is
is accompanied by the eruption of an ash-flow tuff, preceded by air-fall widely used as the basis of information for geologists to map lineaments
pumices, and both of which were emplaced as single flow units (Leat, at local and regional scales and provides useful information for energy
1984, 1985; Macdonald et al., 1994). and mineral exploration. Density distribution in the study of lineaments
Recent lithostratigraphical studies for about 50 geothermal wells has been widely used as a tool in the delineation of fracture zones based
within the caldera with varying depths of up to 3 km have shown that on lineament indices extracted from satellite imagery (e.g., Hung et al.,
2005), and reveals the distribution of lineaments per unit area.

70
I. Kanda et al. Geothermics 79 (2019) 67–81

A quantitative study of lineaments was carried out to determine the The stations were positioned by Differential Global Positioning System
length and directional patterns of lineament sets in relation to geolo- (D-GPS), whose vertical uncertainty was lower than 3 cm. The mea-
gical structures. In this study, a band combination of cloud-free Landsat surement stations focused on areas with unreliable and insufficient
7 Enhanced Thematic Mapper Plus (ETM+) of path/row 169/60 ac- data, along with readily accessible areas. However, areas falling within
quired on May 10, 2016, was integrated with the geological map protected zones, densely populated urban settlements, heavily forested
(Fig. 3) and analyzed alongside a shaded relief map created from Digital regions, private firms, and rugged terrains were not accessible.
Elevation Model (DEM). The DEM was extracted from Shuttle Radar To be able to evaluate the Bouguer anomaly, a priori information
Topography Mission (SRTM) data retrieved from the online Data Pool, about gravity reductions is required. According to Hay et al. (1995), the
courtesy of the NASA Land Processes Distributed Active Archive Center rift infill consists of a lower layer, up to 500 m thick, of volcanic rocks
(LP DAAC), USGS/Earth Resources Observation and Science (EROS) and sediments with a density of 2.45 g/cm3 expressive of pre-
Center. The preprocessing of Landsat image included conversion from dominantly down-faulted Plateau phonolites overlain by later volcanic
calibrated digital numbers (DNs) to physical units, such as sensor ra- rocks and sediments with a density of about 2.2 g/cm3, ranging up to
diance and surface reflectance (SR), using Landsat calibration and 2500 m thick, representing a significant proportion of low-density tuffs
FLAASH (Fast Line-of-Sight Atmospheric Analysis of Spectral Hy- and alluvium. Mungania and Lagat (2004) prepared plots for density
percubes) algorithm of ENVI software (e.g., Gawad et al., 2016; values ranging between 1.9–2.5 g/cm3 using Nettleton's method
Matthew et al., 2002). (Nettleton, 1939, 1976). They found that to get the best fit for the re-
The edge detection algorithm of Canny (Canny, 1986) was used to gion of the rift shoulders a Bouguer density of 2.5 g/cm3 had to be used,
retrieve the lineaments. This algorithm is famously known for its suit- while for the rift floor, a density of 2.0 g/cm3 was the most appropriate.
ability based on the three criteria of proper localization, good detection Simiyu and Keller (2001) assessed density measurement of more than
and single response to an edge (e.g., Ding and Goshtasby, 2001; Heath 2000 samples from drill holes to depths of about 3000 m in the geo-
et al., 1996). The Canny algorithm uses a filter based on the first de- thermal prospects. Shallow drill holes outside the rift reveal densities of
rivative of a Gaussian for the reason that it is prone to noise present on less than 2.4 g/cm3 down to depths of 2 km, increasing with depth to
the unprocessed image data (Marghany and Hashim, 2010). The op- more than 2.5 g/cm3. The mean drill core density value is 2.5 − 2.6 g/
eration of retrieving lineaments was carried out using arguably the cm3 and in-situ density measurements carried out for geothermal ex-
most widely used software for the automatic lineament extraction, the ploration in the volcanic terrains show basement density values that
LINE module of the PCI Geomatica (PCI Geomatics, 2015). The linear range from 2.58 to 2.88 g/cm3.
density of lineaments was then calculated as their total length per unit In this study, some methods were selected and used to ensure a
of the area using Line Density tool built in ESRI ArcGIS© software better approximation of Bouguer density. The G-H method (Rikitake
which computes the density (in units of length per unit of area) of linear et al., 1965), F-H method (Parasnis, 1952, 1986), Extended F-H (Fukao
features in the neighborhood of each output raster cell. et al., 1981), the Akaike's Bayesian Information Criterion (ABIC)
Lineaments obtained from the edge detection and line extraction method (Murata, 1990, 1993), and Comparison with Variance of the
algorithms can be differentiated according to spatial variations in Upward-Continuation (CVUR) method (Komazawa, 1995), gave com-
length, density, and trend. These attributes were analyzed using sta- parable Bouguer densities of 2.10 g/cm3, 2.22 g/cm3, 2.33 g/cm3,
tistical methods and presented in the form of a density map and rose 2.24 g/cm3, and 2.25 g/cm3, respectively. An average value of 2.23 g/
diagrams, which are some of the conventional methods generally used cm3 was selected and deemed sufficient to produce a smooth anomaly
(e.g., Hung et al., 2005; Karnieli et al., 1996; Zakir et al., 1999). In this considering the local geology, which reflects the predominance of un-
study area, a total of 8,838 lineaments (with lengths ranging from 30 m consolidated formations.
to 24,338 m) were extracted. These lineaments consist of both natural Bouguer anomaly (BA) can be loosely defined as the measured
relief features owing to geology-related structures and artificial ele- gravity corrected for the known or modeled gravity effects at a plane-
ments that arise due to human-made features such as roads, farm tary scale (Karcol et al., 2017), and written in the form of a simplified
boundaries, among others. Eq. (1) (e.g., Yamamoto, 1999):
To separate the natural from non-natural lineaments, ancillary in-
BA = g − γ + βh − 2πGρh + ρT (1)
formation was used. Two shaded relief images with multiple sun azi-
muth directions created from the DEM map were used to classify and Where g is the observed gravity value, γ is the normal gravity value, β is
filter the lineaments to exclude features that originate from human the elevation-dependent free-air term, and T is terrain correction per
activities or those that do not result from topographic features. The first unit density. G, ρ and h, are the universal gravitational constant, den-
image had light azimuth angles of 0°, 45°, 90°, and 135°, and the second sity and elevation terms, respectively. Both Bouguer correction and
image had the azimuth angles of the light sources that are 180°, 225°, terrain correction were made using a spherical cap crust formula in a
270°, and 315°. After excluding non-natural features from the data, a circular region within a radius of 50 km distance. Terrain correction
lineament density map was generated (Fig. 4). Lineament density was was also made by modeling the actual topographic relief using a col-
calculated based on the number of lineaments per unit area, with the lection of rectangular prisms (Banerjee and Das Gupta, 1977;
unit area taken as a square kilometer and the resulting map was then Komazawa, 1988). The digital elevation model for terrain correction
superimposed over the DEM map to ease interpretation. The geological used was derived from ASTER GDEM v2 Worldwide Elevation Data (1-
map was used to corroborate the density distribution and registered arc-second Resolution) obtained from the NASA Land Processes Dis-
faults and lithological units in the study area. tributed Active Archive Center (LP DAAC), USGS/Earth Resources
Observation and Science (EROS) Center. The complete Bouguer
3.2. Gravity data collection and reduction anomaly was then interpolated to a 250-m grid with the minimum
curvature method with tension (Smith and Wessel, 1990), using Surface
The gravity dataset used covers an area of approximately 880 km2 module implemented in the Generic Mapping Tools system (Wessel
and a total of 1823 gravity stations. The data consist of those we ac- et al., 2013).
quired during a field survey conducted in 2016–2017, combined with
previous data collected by Mungania and Lagat (2004); Geotermica 3.3. 3-D gravity inversion
Italiana Srl (1987) and Bureau Gravimétrique International – BGI
(Drewes et al., 2016). In the course of the fieldwork, measurements The implementation of 3-D modeling of gravity has become an es-
were done using a Scintrex CG-5 which has a reading resolution of sential tool in the exploration of geological resources (Li and
0.001 mGal, and repeatability accuracy of 0.005 mGal (Scintrex, 2012). Oldenburg, 1998). However, the inversion of the field gravity data is

71
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 4. Lineament density map of the study area superimposed on a relief map. The rose diagrams show the number and orientation of lineaments, with their radial
coordinate indicating the number of lineaments.

one of the most ambiguous problems in exploration geophysics studies Table 1


(Oldenburg, 1974). Inversion of gravity data is usually fraught with Different rock types and their depths of occurrence from the initial 10 geo-
problems due to data noise and inhomogeneity of geological bodies thermal wells drilled in Menengai geothermal field (adapted from GDC, 2018).
even with a reasonably accurate gravity measurement and data re- Well Total depth (m) Depths (m) of occurrence
duction. Therefore, carrying out gravity inversion of field data and
generating geologically realistic results may not be easy because the Tuff Trachyte Syenite
inverse problem in potential field modeling suffers from ambiguity and
M01 2195 – 1050 1900
non-uniqueness of its solutions. M02 3189 520, 2000 1100, 2350 2700
To remedy the non-uniqueness problem and improve the inversion M03 2196 650 1350 2100
performance, one may opt for regularization or instead addition of a M04 2106 800 1300 2080
priori knowledge as constraints to reduce the ambiguity of the inverse M05A 2085 – 1100 2085
M06 2171 680 1350 2150
problem. The application of a priori data such as drilling or geological M07 2136 580 850, 1600 220
information to constrain gravity inversions has been increasing in the M08 2345 1090 2000 –
last two decades (e.g., Dahlin et al., 2002; Fullagar et al., 2008; Li and M09 2077 940 1200 –
Oldenburg, 1996, 2000; Represas et al., 2012). M10A 2150 1620 1000, 2100 1945
In this study, a priori information from borehole geology from the
initial ten geothermal wells drilled in Menengai geothermal field has
cm3), phonolite (2.53 g/cm3), basalt (2.82 g/cm3) and syenite (2.75 g/
been integrated to produce a reasonable source model approximation.
cm3) rock formations were estimated from the literature (Daly, 1935;
These wells are located within the caldera and reveal the geological
Schön, 2015; Tenzer et al., 2011). These approximations gave a density
formation intercepted down to a minimum depth of 2077 m and a
contrast of 0.97 g/cm3 from an average of 2.32 g/cm3, which is
maximum of 3189 m (Table 1). The lithostratigraphy is predominantly
somehow higher than the computed density of 2.23 g/cm3 that was
composed of trachytes, tuffs, and syenitic intrusives, with the trachyte
used to reduce the gravity data. We attribute the discrepancy largely to
being the most dominant rock-type. Tuffs occur as thin intercalations
the dominance of the low-density sediments and tuffs, which tends to
between the thick trachyte lavas, and in some wells, syenitic intrusives
lower the calculated Bouguer density. Despite this, the Bouguer density
occur intermittently. The lithology of these wells generally shows very
value was removed for each of the rock types, which corresponds to the
close similarity, but the lack of a clear marker horizon makes it difficult
removal of a uniform background from the observed data. As a result, a
to correlate the various stratigraphic units (GDC, 2018).
density contrast of between -0.38 g/cm3 and 0.59 g/cm3 was obtained.
The actual measurements of rock densities from these wells (or of
The resulting density contrast was then empirically gridded to a dis-
surface geology within the study area) were not available, and as a
tance of 50 m using Kriging method to obtain a reference model.
result, representative densities were approximated. The densities for
To determine the density distribution of the subsurface structure,
volcanic sediments (1.85 g/cm3), tuffs (2.11 g/cm3), trachyte (2.63 g/
we carried out a 3-D inversion of the Bouguer gravity anomaly using the

72
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 5. A 3-D Gravity inversion grid. The cell size is 620 m × 650 m × 50 m with the height component set to increase by a factor of 1.08 for each successively deeper
layer. The colored map represents complete Bouguer anomaly (CBA) with values ranging from −189.4 to −134.7.

VOXI Earth Modelling tool, a voxel-based tool that is built within the density variation.
Geosoft platform. This platform is a cloud-based computing service, and The inversion will improve the model iteratively until the predicted
this makes it possible to calculate a large-scale 3-D model at high speed data satisfy an assigned tolerance criterion. In this study, a maximum
without depending on the local computer specification. It applies a absolute error was initially set at 0.05 mGal but then adjusted until an
Cartesian Cut Cell method developed by Ingram et al. (2003), which acceptable fit tolerance of 0.02 mGal was achieved. The simulation ran
allows accurate geometric representation of geological surfaces (Ellis for 21 iterations and resulted in a 3-D model with density contrasts that
and MacLeod, 2013). The solution search method uses the iterative r E- range from -0.35 g/cm3 to 0.58 g/cm3 and lie within a geologically
W eighting least squares (IRLS) algorithm initially developed by reasonable density variation for this study area. The Bouguer-derived
Lawson (1961). The IRLS approach solves certain optimization pro- density was then added to each cell to obtain the density distribution.
blems using weighted least squares calculations, iteratively. In order to Finally, the results were gridded to a higher resolution of
obtain a stable solution, the objective function is regularized by the 120 × 120 × 120 rectangular cells using an isotropic inverse distance
misfit between observed and calculated values. The reduced chi- weighting interpolator (Shepard, 1968) to enhance the presentation of
squared misfit (Taylor, 1997), is then used to measure the degree to the model recovered from the inversion.
which the predicted response agrees with the observed data.
The subsurface of the study area was discretized into 58 × 60 × 49
rectangular cells in the x, y and z-direction, respectively (Fig. 5). The 4. Results
size of a rectangular cell was set to 620 m and 650 m along respective x
and y directions. The height of the cells increases from a top layer of The density, trend, and lengths of lineaments mapped from satellite
50 m by a factor of 1.08 for each successive deeper layer. The inversion imagery are illustrated in Fig. 4. The lineament distribution displayed
process involves varying the trade-off regularization parameter in an on the density map puts into consideration the sum of all of the
attempt to fit the data, and the inversion improves the model iteratively lineament line lengths that occur within each grid cell unit. The map
until the predicted data satisfy the residual chi-squared criterion to shows that the high-density values (i.e., density values > 1.0) are
obtain the optimal model. concentrated mostly in areas occupied by phonolites and trachytes in
The reference model generated from the density contrasts was used Fig. 3, and more so to the NE area where these rocks are reasonably
to constrain the inversion. To ensure the results are within an accep- well exposed. Most of the faults are located adjacent to zones with
table range, lower-bound and upper-bound constraints of -0.38 g/cm3 average to high-density values. On the contrary, the results obtained
and 0.59 g/cm3, respectively, were set. These values are based on the from terrains masked by volcanic sediments, as one might have ex-
density variation derived from the estimated densities of local geology. pected, show low to moderate density values (i.e., density values <
Also, an Iterative Reweighting Inversion Focusing (IRIF) constraint was 0.8).
applied to enhance the model and produce a more geologically plau- Another way to assess the extracted lineaments is by use of rose
sible inversion result (Ellis, 2012). IRIF uses the output of the inversion diagrams to analyze their orientation. The use of rose diagrams to
as a weighting function in successive iterations and this results in a analyze lineaments helps in ascertaining the most common directions of
more focused model. The weighting function makes sharpening of the lineaments, which then can be compared with directions of mapped
contrasts possible by enhancing boundaries around distinctive anom- faults. The resulting lineament density map was subdivided into four
alous features (MacLeod and Ellis, 2015). In the present case, we ap- equally large zones (namely NE, SE, SW, and NW) for the sake of in-
plied two passes of IRIF to emphasize both positive and negative ends of terpretation. The lineaments were then grouped with an angular spa-
cing of 10° and presented in the form of four rose diagrams each

73
I. Kanda et al. Geothermics 79 (2019) 67–81

representing the partitioned zones. within the caldera and to the north-northwest of the caldera. The high
The rose diagrams shown in Fig. 4 indicate that the most general Bouguer anomaly appears to be confined within the caldera and the
orientation of lineaments extracted from the satellite image is N 0° - N Olrongai area and is associated with the occurrence of trachytic rocks as
30°, N 330° - N 10°, N 0° - N 40° and N 350° - N 20°, in the NE, SE, SW well as fumarolic activities (Fig. 3). The low Bouguer anomalous region
and NW zones, respectively. This angular tendency corresponds to that is associated with the Solai graben and the Rongai plain situated on the
of the faults, particularly in the NW and NE zones where the fault north-east and south-west areas, respectively. As seen from the local
system is more pronounced. Despite the limited number of faults in the geology (Fig. 3), these low anomaly regions are covered by volcanic
SW and SE zones, the general trend from lineaments matches with the sediments.
regional trend of faults. Such conformity underscores the importance of The planar slices at depths of 1 km (Fig. 7), 2 km (Fig. 8) and 3 km
lineaments in the mapping of geological structures in inaccessible areas. (Fig. 9) below the surface were extracted from the inversion model,
The presence of a dense pattern of lineaments is a positive indicator then overlain on a shaded relief map and plotted alongside faults. High-
of fractured or faulted formations. The densities in the NE area is density areas in shallower levels appear as distributed bodies but con-
greatest and coincides with the intensely faulted and high relief surface verge with depth to form two main bodies; lying within the caldera and
expression of Bahati platform. This platform is part of the eastern es- the Olrongai area. These two main bodies coincide with the presence of
carpment of the central Kenya rift and apparently with the highest trachytes while the dense bodies north of Olbanita and also north and
elevation (up to > 2400 m above sea level) in the study area (Fig. 2). east of Solai seem to be associated with the presence of phonolites
Bahati platform shows a complex form of step-faulting and its succes- (Fig. 3).
sive steps is a structural feature that may be numbered among the most A striking relationship is also observed between the occurrence of
ideally developed structures in the Kenya rift (McCall, 1967). The the high-density anomalous structures and the existence of tectonic
concurrence of faults and high lineament density is also evident in the faults. The two appear to coexist, and this association becomes even
NW and SW sectors. more distinct with increasing depth. The high-density anomalous
Interpretation of the Bouguer anomaly was limited to the distribu- structure beneath the caldera appears to be confined mainly within the
tion of survey stations shown in Fig. 6. Areas with no stations were caldera walls while NNE-SSE and N-S trending structures enclose the
removed, and structures located in sparsely distributed stations were anomaly beneath Olrongai..
interpreted with caution. Nevertheless, the areas considered relevant to The isosurface of 2.45 g/cm3 in Fig. 10 infers the geometry and
this study appear to have been fairly covered during the survey. positions of shallow-crustal high-density volumes located beneath the
The Bouguer anomaly shown in Fig. 6 has high values generally caldera and Olrongai hill. The isosurface used displays the geometry of
trending NNW-SSE direction, with a dominant twofold anomaly located the high-density structure of the study area, and similar isosurface was

Fig. 6. Bouguer anomaly map calculated for a Bouguer density of 2.23 g/cm3. Black dots represent the gravity stations.

74
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 7. A planar view of density distribution at a depth of 1 km below the surface.

also used in Figs. 11 and 12. The two masses appear as discrete bodies, caldera. The occurrence of active fumaroles (with temperatures be-
but they probably may be connected at depth. These two high-density tween ∼62 and ∼88 °C), steaming ground and hot water boreholes
features underlie Menengai caldera and Olrongai areas that have been (e.g., Mungania and Lagat, 2004), implies that the area overlies an
explored for geothermal generation. Menengai caldera and Olrongai are active hydrothermal system. The occurrence of such hydrothermal
known by their surface manifestation of the high-temperature geo- systems inside the caldera has already been confirmed by drilling.
thermal system particularly fumarolic activities.. The results of the 3-D gravity inversion presented in this study re-
The orientation of tectonic faults shown in Fig. 11 appears to delimit veal a close relationship between the local surface tectonic structures
the lateral extent of the body beneath Olrongai. The relationship is still and shape of the high-density anomalies in the subsurface. This re-
evident despite the depth difference between surface structures and the lationship is more prominent in the Olrongai area where the surface
dense body. These structures are relatively younger and are composed structures appear to delineate the geometry of the dense body as de-
of both new faults as well as reactivated old fractures (McCall, 1967). scribed by a sharp density gradient. The two features appear to coexist
The presence of fumaroles at boiling temperatures (Mungania and implying that these structures acted as conduits during the ascent and
Lagat, 2004) are located adjacent to these faults. emplacement of at least the phonolites and trachytes.
Unlike the Olrongai region, the caldera has no clear structures other The distribution of high-density anomaly bodies occurs as separate
than the E-W arcuate structure. The pattern of this structure coincides features in shallower levels (Fig. 7) but converge with depth (Fig. 9) to
with a high-density anomaly shown in Fig. 12. The eastern end of this form two main bodies. The flow of magma through and along the faults
structural feature seems to be connected with a series of NNE trending in the shallow parts of the system seems to have caused these variations
normal faults. These faults consequentially render the caldera wall in- as evidently seen to the north of Menengai caldera, although, this might
complete as it cuts through the northeastern rim by north-south generally not be the case. The high-density anomalies in the shallow
trending segment graben (McCall, 1967), marking the southern end of levels (< 2 km), particularly beneath the Rongai basin, may have re-
the Solai graben. Some inferred faults are presented in Fig. 12 based on sulted either from the emplacement of pockets of lava originating from
the shape of the dense body and caldera rim, and the pattern displayed the western flanks of the Kenya rift or from neighboring eruption
by the distribution of fumaroles. centers (Menengai and Kilombe, (McCall, 1964, 1967; Walsh, 1969)),
into the tuffaceous and volcanic sediment host formations.
The occurrence of fumarolic activity in Olrongai seems to be con-
5. Discussion
trolled by a swarm of NNE and NE trending faults. These fault systems
may act not only as conduits for hydrothermal upwelling that taps the
The Menengai area is characterised by signatures of high-tempera-
heat source but also for magmatic intrusions from sources beneath the
ture geothermal reservoirs that include relatively young volcanism re-
hydrothermal system. The presence of fumaroles at boiling
presented by the numerous recent eruptions within and outside the

75
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 8. A planar view of density distribution at a depth of 2 km below the surface.

temperatures adjacent to these surface tectonic structures suggests that fault created during the post-caldera collapse activity. Such arcuate
the geothermal reservoir is positioned at a relatively shallow depth and fracture systems are usually related to the local intensification of the
most likely beneath the fumaroles. The surface manifestation of a high- tension resulting from inflation of underlying magma bodies commonly
temperature geothermal system within the dense bodies of Menengai seen or inferred at volcanoes (e.g., Anderson, 1936). Magma emplaced
caldera and Olrongai suggest that their heat sources are hosted within at shallow levels beneath the caldera may cause localized but sig-
the dense structures. nificant deformation that might have led to the growth of the dome-
The Menengai caldera is formed on a large exposed joint at the shaped forced structure (GDC, 2018; Mungania and Lagat, 2004), and
tailcrack culmination of a northerly striking dextral normal oblique possible fracturing.
slip-fault region, described in Chorowicz (2005). This E-W arcuate A closer look at the caldera wall, the distribution of fumaroles and
structure is a large open crack that curves across the caldera from east the shape of the dense body appears to propose an alternative view of
to west (Fig. 2), considered responsible for the emplacement of the NE and NW striking network of faults (Fig. 12). This supports a possible
younger eruptives (GDC, 2018; Strecker et al., 2013). The results of the extension of post-caldera faults across the caldera superficially ob-
inversion shown in Fig. 12 reveals the presence of an underlying dense scured by younger deposits. The fumaroles are found to be located right
body whose orientation coincides with the tailcrack feature and cor- on top of the dense trachytic formation but at shallower levels (Fig. 7),
roborates with the earlier studies that this structureaided the ascend of they appear close to the edges. This is probably because the hydro-
younger trachytic magma to the surface. thermal system occurs at a lithological contact zone where the
Some kinematic assessment inside the caldera relates these faults permeable host formation allows circulation of water that interacts with
with strike-slip and oblique-slip transfer faults, commonly seen across and is heated by the intrusives embayed within the dense formation.
the central Kenyan rift (Riedl et al., 2015). We suggest that the acti- The distribution of these fumaroles describes zones of highly positive
vation of these regional faults (Strecker et al., 1990), and the formation thermal anomalies and is generally located close to or on permeable
of the arcuate fracture allowed trachytic magma to ascend to shallower structures. We suggest that these structures act as the pathways for fluid
levels, which then flows laterally being accommodated by permeable to rise to the surface implying that beneath these fumaroles is a zone of
faults and fractures. These structures might have acted as feeder dikes a hydrothermal system.
of syenitic intrusives, hosted within the trachytic formation, and be- The generation of a network of faults or a ring fracture and spatially
lieved to be the heat source for the geothermal system. related uprising of fumarolic fluids along these openings could indicate
The loop-shaped pattern portrayed by the locations of the fumaroles a possible coupling of the hydrothermal and magmatic systems. This
inside the Menengai caldera and geometry of the high-density anomaly inference takes into consideration the presence of magma pockets or
body in Fig. 7 and Fig. 12 seems to suggest the presence of an arcuate dikes interpreted from borehole data to be as shallow as 2 km. The
fracture system. This can be interpreted as surface-expression of a ring- presence of such a shallow heat source is supported by the recorded

76
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 9. A planar view of density distribution at a depth of 3 km below the surface.

bottom hole temperatures close to 400 °C at about 2.1 km in some of the useful in adding value to the model. Nonetheless, the imaging of the
wells that intercepted syenitic intrusives (GDC, 2018). The interception dense body beneath the E-W arcuate structure confirms the validity of
of cold wells during drilling of geothermal wells within the caldera the 3-D inversion model.
suggests that although the heat source is quite shallow, its lateral extent Lastly, we constrained the model in this study by including density
is structurally bounded. The structural system restraining the breadth of estimates of lithologies interpreted from drill hole cuttings. However,
the dense trachytic body of Olrongai is distinct, but for the caldera, an intermittent losses of circulation affected the recovery process and led
inference can be made from the results of this study. The results of this to incomplete lithological logs (GDC, 2018). Consequently, the depths
study suggest that the high-density anomalous body provides more fa- affected by these losses were avoided and this narrowed the scope of
vorable sites for geothermal detailed exploration and drilling, at least available geology. To improve the robustness and reduce uncertainty on
from the heat source perspective. the inversion model, a rigorous characterization of the rock unit den-
Although the results obtained from the inversion model are en- sities from measurements of downhole and surface samples collected
couraging, there are some limitations. First, a significant part of the from the study area is necessary.
caldera floor is known by its rugged terrain that is mostly inaccessible
(Mungania and Lagat, 2004). Therefore, the gravity data coverage is
6. Conclusions
somewhat sparse (Fig. 6), and may not be conclusively used to infer
small changes in density or explicitly define lithological contacts.
The primary focus of this study was to assess the relationship be-
Hence, this study could not image these pockets of magma since they
tween geological structures and a geothermal heat source in the
are confined within the trachytic and syenitic formations. The problem
Menengai geothermal area. An improved Bouguer anomaly map of the
with data sparsity might also be the case for the Rongai plain where the
Menengai area based on an updated gravity dataset presented here
inversion process appears not to have resolved the nature of the isolated
reveals a dominant two-fold high Bouguer anomaly underlying the
anomalies appropriately, and this makes it difficult to infer causative
caldera and Olrongai hill associated with the occurrence of trachytic
agent at least from the inversion results.
rocks and fumarolic activities. Away from this northerly trending zone,
Secondly, except for the E-W arcuate structure that is readily ob-
low-density sectors extend over the volcanic sediments terrains of
servable, the presence of faults inside the caldera floor is poorly known
Rongai plain and Solai graben.
although some significant attempts have been made to uncover them
A 3-D gravity inversion for predicting the geothermal heat source
(e.g., Riedl et al., 2015; Strecker and Melnick, 2013). The caldera floor
architecture within the Menengai caldera and Olrongai hill area is also
is covered with more than 70 post-caldera lava flows, pumice deposits
presented. The model provides an in-depth insight into the subsurface
and strombolian cinder cones (Macdonald et al., 2011). This makes it
structure from potential field perspective by constraining deep bodies
difficult to establish any existence of geological faults that would be
in 3-D space, which is significant for geological interpretations in such a

77
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 10. A planar profile and isosurface (2.45 g/cm3) inferred from the density model.

Fig. 11. Geological faults as seen from the surface and fumaroles location superimposed against a 2 km deep planar layer and an isosurface of 2.45 g/cm3 to highlight
the dense body of Olrongai.

78
I. Kanda et al. Geothermics 79 (2019) 67–81

Fig. 12. Geological faults/caldera rim ring-fracture as seen from the surface and fumaroles location superimposed against a 2 km deep planar layer and an isosurface
of 2.45 g/cm3 to highlight the dense body of Menengai caldera. The inferred faults are presented as broken lines.

complex geothermal system. The host-rock boundary is highlighted by a Company of Kenya (GDC), for sharing field equipment and existing
strong density gradient that suggests a precipitous contact between two data, and by generously availing their staff during data collection. The
lithologies delineated by tectonic faults, more so evident in Olrongai sacrifices made by the individuals involved in the field survey is highly
area. appreciated. We would also like to thank the Japan International
The complex nature of post-caldera emplacement of the trachytic Cooperation Agency (JICA) for supporting this research. We are
formation and ash-flow tuffs is linked to many eruptions, the products grateful to the laboratory of Geothermics, Graduate School of
of which cover the caldera floors and make potential data analysis one Engineering, Kyushu University, for their recommendations that sig-
of the best method for inferring buried faults. The inversion model nificantly improved this manuscript. Finally, the authors wish to thank
reveals information about the undulation of the dense bodies beneath the editor and the three anonymous reviewers for their valuable com-
Olrongai hill and Menengai caldera. The locations of the two main high- ments and recommendations.
density structures in the shallower portions of the model correspond
with eruption centers, manifested by the existence of younger volcanic References
rocks and fumarolic activities. The geomorphology of these dense
bodies at depth is interpreted to embay the heat source, which is crucial Abebe, T., Mazzarini, F., Innocenti, F., Manetti, P., 1993. The yerer-tullu wellel exten-
for the existence of a geothermal system. sional structure (Central Ethiopia): evidences from remote sensing, petrologic and
geochronologic data. J. Afr. Earth Sci. 17, 145–165. https://doi.org/10.1109/
These results show the significance of gravity inversion methods to IGARSS.1995.520285. IEEE.
provide information of subsurface structure beneath the Menengai Anderson, E., 1936. The dynamics of the formation of cone sheets, ring dykes, and
geothermal field. In addition, it underscores the significance of these cauldron subsidences. Proceedings of the Royal Society of Edinburgh, Session 1935-
1936.
structures (i.e., ring structures and faults) in guiding and restraining the Baker, B.H., Wohlenberg, J., 1971. Structure and evolution of the Kenya Rift Valley.
up flow of magma and also the hydrothermal system. The subsurface Nature 229 (5286), 538–542. https://doi.org/10.1038/229538a0.
model obtained adds to the existing volcanological and geological in- Baker, B.H., Williams, L.A.J., Miller, J.A., Fitch, F.J., 1971. Sequence and geochronology
of the Kenya rift volcanics. Tectonophysics 11 (3), 191–215. https://doi.org/10.
formation, giving valuable details about heat source geometry and
1016/0040-1951(71)90030-8.
geological structures controlling its lateral extent and can be used to Banerjee, B., Das Gupta, S.P., 1977. Gravitational attraction of a rectangular parallele-
isolate target areas for more detailed geothermal investigations. piped. Geophysics 42 (5), 1053–1055. https://doi.org/10.1190/1.1440766.
Bullard, E.C., 1936. Gravity measurements in East Africa. Philos. Trans. Math. Phys. Eng.
Following up on our approach will potentially lead to more successful
Sci. 235 (757), 445–531. https://doi.org/10.1098/rsta.1936.0008.
drilling of productive wells, facilitate a reduction in drilling risk, and Canny, J., 1986. A computational approach to edge detection. IEEE Transactions on
serve to improve the geologic model necessary for accurate numerical Pattern Analysis and Machine Intelligence, PAMI 8 (6), 679–698. https://doi.org/10.
reservoir simulations. 1109/TPAMI.1986.4767851.
Chandrasiri Ekneligoda, T., Henkel, H., 2010. Interactive spatial analysis of lineaments.
Comput. Geosci. 36 (8), 1081–1090. https://doi.org/10.1016/j.cageo.2010.01.009.
Chorowicz, J., 2005. The East African rift system. J. Afr. Earth Sci. 43 (1–3), 379–410.
Acknowledgments https://doi.org/10.1016/j.jafrearsci.2005.07.019.
Clarke, M., Woodhall, D., Allen, D., Darling, G., 1990. Geological, volcanological and
hydrogeological controls on the occurrence of geothermal activity in the area
The first author is indebted to the Geothermal Development

79
I. Kanda et al. Geothermics 79 (2019) 67–81

surrounding lake naivasha, kenya Kenya: with coloured 1:250,000 geological maps. transform. Photogramm. Eng. Remote Sensing 62 (5), 525–531.
Ministry of Energy, Nairobi, pp. 138. Komazawa, M., 1988. A gravimetric terrain correction method by assuming annular
Dahlin, T., Bernstone, C., Loke, M.H., 2002. Case History A 3-D resistivity investigation of prism. Geod. Soc. Japan 34, 11–23 (in Japanese).
a contaminated site at Lernacken, Sweden. Geophysics 67 (6), 1692–1700. Komazawa, M., 1995. Gravimetric analysis of Aso Volcano and its interpretation. Journal
Daly, R.A., 1935. Densities of rocks calculated from their chemical analyses. Proc. Natl. of the Geodetic Society of Japan 41 (1), 17–45.
Acad. Sci. U.S.A. 21 (12), 657–663. https://doi.org/10.1073/PNAS.21.12.657. Lagat, J., Mbia, P., Muturia, C., Njue, L., Kanda, I., Mutonga, M., et al., 2010. Menengai
Davis, P.M., Slack, P.D., 2002. The uppermost mantle beneath the Kenya dome and re- Geothermal Prospect: a Geothermal Resource Assessment Project Report Update.
lation to melting, rifting and uplift in East Africa. Geophys. Res. Lett. 29 (7), 1117. Nakuru, Kenya.
https://doi.org/10.1029/2001GL013676. Lawson, C., 1961. Contribution to the Theory of Linear Least Maximum Approximation.
Ding, L., Goshtasby, A., 2001. On the Canny edge detector. Pattern Recognit. 34 (3), Ph. D. Dissertation. Univ. Calif.
721–725. https://doi.org/10.1016/S0031-3203(00)00023-6. Leat, P.T., 1984. Geological evolution of the trachytic caldera volcano Menengai, Kenya
Drewes, H., Kuglitsch, F., Adám, J., Rózsa, S., 2016. The International Gravimetric Rift Valley. J. Geol. Soc. 141 (6), 1057–1069. https://doi.org/10.1144/gsjgs.141.6.
Bureau. In “The Geodesist’s Handbook 2016”. J. Geod. 90 (10), 907–1205. https:// 1057.
doi.org/10.1007/s00190-016-0948-z. Leat, P.T., 1985. Interaction of a rheomorphic peralkaline ash-flow tuff and underlying
Ellis, R., 2012. Iterative reweighted inversion. Geosoft Technical Note. Retrieved from. deposits, Menengai volcano, Kenya. J. Volcanol. Geotherm. Res. 26 (1–2), 131–145.
http://www.geosoft.com/products/voxi-earth-modelling/Fiterative-reweighting- https://doi.org/10.1016/0377-0273(85)90049-6.
inversion. Leat, P.T., 1991. Volcanological development of the Nakuru area of the Kenya rift valley.
Ellis, R., MacLeod, I., 2013. Constrained voxel inversion using the Cartesian Cut Cell J. Afr. Earth Sci. 13 (3–4), 483–498. https://doi.org/10.1016/0899-5362(91)
method. Aseg Ext. Abstr. 1, 1–4. 90111-B.
Fairhead, J.D., Girdler, R.W., 1972. The seismicity of the East African rift system. Li, Y., Oldenburg, D.W., 1996. 3-D inversion of magnetic data. Geophysics 61 (2),
Tectonophysics Vol. 15, 115–122. https://doi.org/10.1016/B978-0-444-41087-0. 394–408.
50017-8. Li, Y., Oldenburg, D.W., 1998. 3-D inversion of gravity data. Geophysics 63 (1), 109–119.
Fairhead, J.D., 1976. The structure of the lithosphere beneath the Eastern rift, East Africa, Li, Y., Oldenburg, D.W., 2000. Incorporating geological dip information into geophysical
deduced from gravity studies. Tectonophysics 30 (3–4), 269–298. https://doi.org/10. inversions. Geophysics 65 (1), 148–157.
1016/0040-1951(76)90190-6. Macdonald, R., Navarro, J.M., Upton, B.G.J., Davies, G.R., 1994. Strong compositional
Fairhead, J.D., Reeves, C.V., 1977. Teleseismic delay times, Bouguer anomalies and in- zonation in peralkaline magma: Menengai, Kenya Rift Valley. J. Volcanol. Geotherm.
ferred thickness of the African lithosphere. Earth Planet. Sci. Lett. 36 (1), 63–76. Res. 60 (3–4), 301–325. https://doi.org/10.1016/0377-0273(94)90057-4.
https://doi.org/10.1016/0012-821X(77)90188-1. Macdonald, R., Baginski, B., Leat, P.T., White, J.C., Dzierzanowski, P., 2011. Mineral
Fairhead, J.D., Mitchell, J.G., Williams, L.A.J., 1972. New K/Ar determinations on rift stability in peralkaline silicic rocks: information from trachytes of the Menengai
volcanics of S. Kenya and their bearing on age of rift faulting. Nat. Phys. Sci. 238 volcano, Kenya. Lithos 125 (1–2), 553–568. https://doi.org/10.1016/j.lithos.2011.
(83), 66–69. https://doi.org/10.1038/physci238066a0. 03.011.
Fukao, Y., Yamamoto, A., Nozaki, K., 1981. A method of density determination for gravity MacLeod, I., Ellis, R., 2015. Quantitative magnetization vector inversion. Extended
correction. Letter J. Phys. Earth 29, 163–166. Abstracts of 14th SAGA Biennial Technical Meeting and Exhibition 2015. pp. 1–5.
Fullagar, P.K., Pears, G.A., McMonnies, B., 2008. Constrained inversion of geologic sur- Marghany, M., Hashim, M., 2010. Lineament mapping using multispectral remote sensing
faces — pushing the boundaries. Lead. Edge 27 (1), 95–105. satellite data. Res. J. Appl. Sci. 5 (2), 126–130. https://doi.org/10.3923/rjasci.2010.
Gawad, A.E.A., Donia, A.M.A., Elsaid, M., 2016. Processing of Landsat 8 Imagery and 126.130.
Ground Gamma-Ray Spectrometry for Geologic Mapping and Dose-Rate Assessment, Matthew, M.W., Adler-Golden, S.M., Berk, A., Felde, G., Anderson, G.P., Gorodetzky, D.,
Wadi Diit along the Red Sea Coast, Egypt. Open J. Geol. 06 (08), 911–930. https:// et al., 2002. Atmospheric correction of spectral imagery: evaluation of the FLAASH
doi.org/10.4236/ojg.2016.68069. algorithm with AVIRIS data. Proceedings of Applied Imagery Pattern Recognition
GDC, 2018. Steam Status and Resource Assessment of Menengai Geothermal Project, Workshop. (Pp. 157–163). IEEE Comput. Soc. https://doi.org/10.1109/AIPR.2002.
Kenya. Internal Report. 1182270.
Geotermica Italiana, 1987. Geothermal Reconnaissance Survey in the Menengai Bogoria McCall, G.J.H., 1964. Kilombe caldera, kenya. Proc. Geol. Assoc. 75 (4), 563–572.
Area of the Kenya Rift Valley. https://doi.org/10.1016/S0016-7878(64)80028-6.
Girdler, R.W., 1975. The great negative Bouguer gravity anomaly over Africa. Eos Trans. McCall, G.J.H., 1967. Geology of the Nakuru-Thomson’s Falls-Lake Hannington Area:
Am. Geophys. Union 56 (8), 516. https://doi.org/10.1029/EO056i008p00516. Degree Sheet No. 35 SW Quarter and 43 NW Quarter (No. 78). Geological Survey of
Girdler, R.W., 1983. Processes of planetary rifting as seen in the rifting and break up of the Kenya Republic (Vol. 78). Geological Survey of Kenya.
Africa. Dev. Geotecton. 19 (C), 241–252. https://doi.org/10.1016/B978-0-444- McConnell, R.B., 1972. Geological development of the rift system of Eastern Africa. GSA
42198-2.50021-3. Bulletin 83 (9), 2549–2572. https://doi.org/10.1130/0016-7606(1972)
Girdler, R.W., Sowerbutts, W.T.C., 1970. Some recent geophysical studies of the rift 83[2549:gdotrs]2.0.co;2.
system in East Africa. J. Geomagn. Geoelectr. 22 (1–2), 153–163. https://doi.org/10. Mibei, G., 2012. Geology and Hydrothermal Alteration of Menengai Geothermal Field.
5636/jgg.22.153. Case Study: Wells MW-04 and MW-05. Geothermal Training in Iceland, Technical
Girdler, R.W., Fairhead, J.D., Searle, R.C., Sowerbutts, W.T.C., 1969. Evolution of rifting Report 21. pp. 437–465.
in Africa. Nature 224 (5225), 1178–1182. https://doi.org/10.1038/2241178a0. Mountrakis, G., Luo, L., 2011. Enhancing and replacing spectral information with inter-
Hashim, M., Ahmad, S., Johari, M.A.M., Pour, A.B., 2013. Automatic lineament extraction mediate structural inputs: a case study on impervious surface detection. Remote Sens.
in a heavily vegetated region using Landsat enhanced Thematic Mapper (ETM+) Environ. 115 (5), 1162–1170. https://doi.org/10.1016/j.rse.2010.12.018.
imagery. Adv. Space Res. 51 (5), 874–890. https://doi.org/10.1016/j.asr.2012.10. Mungania, J., Lagat, J., 2004. Menengai Volcano: Investigations for Its Geothermal
004. Potential.
Hay, D.E., Wendlandt, R.F., Keller, G.R., 1995. Origin of Kenya Rift Plateau-type flood Murata, Y., 1990. Estimation of Bouguer reduction density using ABIC minimization
phonolites: integrated petrologic and geophysical constraints on the evolution of the method. Journal of Seismology Society of Japan 43, 327–339.
crust and upper mantle beneath the Kenya Rift. J. Geophys. Res. Solid Earth 100 (B6), Murata, Y., 1993. Estimation of optimum average surficial density from gravity data: an
10549–10557. https://doi.org/10.1029/94JB03036. Objective Bayesian Approach. J. Geophys. Res. 98 (1), 12 097-12,109.
Heath, M., Sarkar, S., Sanocki, T., Bowyer, K., 1996. Comparison of edge detectors: a Nettleton, L.L., 1939. Determination of density for reduction of gravimeter observations.
methodology and initial study. Proceedings CVPR IEEE Computer Society Conference Geophysics 4 (3), 176–183. https://doi.org/10.1190/1.0403176.
on Computer Vision and Pattern Recognition 143–148. https://doi.org/10.1109/ Nettleton, L.L., 1976. Gravity and Magnetics in Oil Prospecting. McGraw-Hill, New York.
CVPR.1996.517066. IEEE. Noltimier, K.F., Jezek, K.C., Wilson, T.J., Johnson, A.C., 1998. Evidence for the tectonic
Hung, L.Q., Batelaan, O., De Smedt, F., 2005. Lineament extraction and analysis, com- segmentation of the Antarctic peninsula from integrated ERS-1 SAR mosaic and
parison of LANDSAT ETM and ASTER imagery. Case study: suoimuoi tropical karst aeromagnetic anomaly data. In IGARSS’ 98. Sensing and managing the environment.
catchment, Vietnam. In: In: Ehlers, M., Michel, U. (Eds.), Remote Sensing for 1998 IEEE International Geoscience and Remote Sensing. Symposium Proceedings.
Environmental Monitoring, GIS Applications, and Geology V Vol. 5983 International (Cat. No.98CH36174) (Pp. 1439–1441 Vol.3). IEEE. https://doi.org/10.1109/
Society for Optics and Photonics. https://doi.org/10.1117/12.627699. p. 59830T. IGARSS.1998.691486.
Ingram, D.M., Causon, D.M., Mingham, C.G., 2003. Developments in Cartesian cut cell Oldenburg, D.W., 1974. The inversion and interpretation of gravity anomalies.
methods. Math. Comput. Simul. 61 (3–6), 561–572. https://doi.org/10.1016/S0378- Geophysics 39 (4), 526–536. https://doi.org/10.1190/1.1440444.
4754(02)00107-6. Omenda, P., Simiyu, S., Muchemi, G., 2014. Geothermal Country Update Report for
Jennings, D.J., 1971. Geology of the Molo Area (No. 86). Kenya: 2014. 001374011. pp. 29–31.
Kageyama, Y., Nishida, M., Oi, T., 2000. Analysis of the segments extracted by automated Omondi, C., 2011. Borehole geology and hydrothermal mineralisation of wells Mw-01
lineament detection. International Geoscience and Remote Sensing Symposium and Mw-02, Menengai Geothermal Field, Central Kenya Rift Valley. Geothermal
(IGARSS) Vol. 1. pp. 289–291 IEEE. Trainning Programme 30, 54–58.
Kanda, I., 2010. Structural controls on thermal anomalies and diffuse soil degassing at Parasnis, D.S., 1952. A study of rock densities in the English Midlands. Geophys. Suppl.
menengai geothermal prospect, the Kenyan rift valley. Third African Rift Geothermal Mon. Not. R. Astron. Soc. 6 (5), 252–271. https://doi.org/10.1111/j.1365-246X.
Conference. pp. 67–72 ARGEO - C3. 1952.tb03013.x.
Karcol, R., Mikuška, J., Marušiak, I., Karcol, R., Mikuška, J., Marušiak, I., 2017. Normal Parasnis, D.S., 1986. Principles of Applied Geophysics. Chapman and Hall.
earth gravity field versus gravity effect of layered ellipsoidal model. Understanding PCI Geomatics, 2015. PCI Geomatica User’s Guide Version. Ontario, Canada.
the Bouguer Anomaly. Elsevier, pp. 63–77. https://doi.org/10.1016/B978-0-12- Represas, P., Catalão, J., Montesinos, F.G., Madeira, J., Mata, J., Antunes, C., Moreira, M.,
812913-5.00003-8. 2012. Constraints on the structure of Maio Island (Cape Verde) by a three-dimen-
Karnieli, A., Meisels, A., Fisher, L., Arkin, Y., 1996. Automatic extraction and evaluation sional gravity model: imaging partially exhumed magma chambers. Geophys. J. Int.
of geological linear features from digital remote sensing data using a Hough 190 (2), 931–940. https://doi.org/10.1111/j.1365-246X.2012.05536.x.

80
I. Kanda et al. Geothermics 79 (2019) 67–81

Riedl, S., Melnick, D., Mibei, G.K., Njue, L., Strecker, M.R., 2015. Post-caldera faulting of the southern Kenya rift. Geophys. J. Int. 147 (3), 543–561. https://doi.org/10.1046/
the late Quaternary Menengai caldera, Central Kenya rift (0.20°S, 36.07°E). EGU j.0956-540x.2001.01542.x.
General Assembly 2015, Held 12-17 April, 2015 in Vienna, Austria. pp. 17. Sippel, J., Meeßen, C., Cacace, M., Mechie, J., Fishwick, S., Heine, C., Strecker, M.R.,
Rikitake, T., Tajima, H., Izutuya, S., Hagiwara, Y., Kawada, K., Sasai, Y., 1965. 2017. The Kenya rift revisited: insights into lithospheric strength through data-driven
Gravimetric and geornagnetic studies of Onikobe area. Bulletin of the Earthquake 3-D gravity and thermal modelling. Solid Earth 8, 45–81. https://doi.org/10.5194/
Research Institute 241–267. se-8-45-2017.
Saadi, N.M., Abdel Zaher, M., El-Baz, F., Watanabe, K., 2011. Integrated remote sensing Smith, D.M., Andrew, E.M., 1962. Gravity meter primary station net in East and Central
data utilization for investigating structural and tectonic history of the Ghadames Africa. Geophys. J. Int. 7 (1), 65–86. https://doi.org/10.1111/j.1365-246X.1962.
Basin, Libya. Int. J. Appl. Earth Obs. Geoinf. 13 (5), 778–791. https://doi.org/10. tb02253.x.
1016/j.jag.2011.05.016. Smith, W.H.F., Wessel, P., 1990. Gridding with continuous curvature splines in tension.
Saggerson, E.P., Baker, B.H., 1965. Post-Jurassic erosion-surfaces in eastern Kenya and Geophysics 55 (3), 293–305. https://doi.org/10.1190/1.1442837.
their deformation in relation to rift structure. Q. J. Geol. Soc. 121 (1–4), 51–68. Strecker, M.R., Melnick, D., 2013. Structural Characteristics of Menengai Caldera and
https://doi.org/10.1144/gsjgs.121.1.0051. Regions Farther North, Central Kenya Rift.
Salati, S., Van Ruitenbeek, F.J.A., Van der Meer, F.D., Tangestani, M.H., Van der Werff, Strecker, M.R., Blisniuk, P.M., Eisbacher, G.H., 1990. Rotation of extension direction in
H., 2011. Lithological mapping and fuzzy set theory: automated extraction of litho- the central Kenya Rift. Geology 18 (4), 299. https://doi.org/10.1130/0091-
logical boundary from ASTER imagery by template matching and spatial accuracy 7613(1990)018<0299:ROEDIT>2.3.CO;2.
assessment. Int. J. Appl. Earth Obs. Geoinf. 13 (5), 753–765. https://doi.org/10. Swain, C.J., 1992. The Kenya rift axial gravity high: a re-interpretation. Tectonophysics
1016/j.jag.2011.05.004. 204 (1–2), 59–70. https://doi.org/10.1016/0040-1951(92)90269-C.
Schön, J., 2015. In: Cubitt, J., W. Holt (Eds.), Physical Properties of Rocks: Fundamentals Taylor, J., 1997. Introduction to Error Analysis, the Study of Uncertainties in Physical
and Principles of Petrophysics., 2nd edition. Elsevier. Measurements. University Science Books.
Scintrex, 2012. CG-5 Scintrex Autograv TM System Operation Manual. SCINTREX Tenzer, R., Sirguey, P., Rattenbury, M., Nicolson, J., 2011. A digital rock density map of
Limited. New Zealand. Comput. Geosci. 37 (8), 1181–1191. https://doi.org/10.1016/j.cageo.
Searle, R.C., 1970. Evidence from gravity anomalies for thinning of the lithosphere be- 2010.07.010.
neath the Rift Valley in Kenya. Geophys. J. Int. 21 (1), 13–31. https://doi.org/10. Walsh, J., 1969. Geology of the eldama Ravine-kabarnet Area: degree sheet 34 SE
1111/j.1365-246X.1970.tb01764.x. Quarter. Geological Survey of the Kenya. Nairobi.
Shepard, D., 1968. A two-dimensional interpolation function for irregularly-spaced data. Wessel, P., Smith, W.H.F., Scharroo, R., Luis, J., Wobbe, F., 2013. Generic mapping tools:
Proceedings of the 1968 23rd ACM National Conference on 517–524. https://doi. improved version released. Eos Trans. Am. Geophys. Union 94 (45), 409–410.
org/10.1145/800186.810616. https://doi.org/10.1002/2013EO450001.
Simiyu, S.M., 2010. Status of geothermal exploration in Kenya and future plans for its Williams, L.A.J., 1978. The volcanological development of the Kenya rift. Petrology and
development. World Geothermal Congress 2010. pp. 25–29. Geochemistry of Continental Rifts. Dordrecht: Springer, Netherlands, pp. 101–121.
Simiyu, S.M., 2013. Application of micro-seismic methods to geothermal exploration: https://doi.org/10.1007/978-94-009-9803-2_10.
examples from the Kenya rift. Short Course VIII on Exploration for Geothermal Yamamoto, A. (1999). Estimating the optimum reduction density for gravity anomaly. A
Resources (p. 27). United Nations University - Geothermal Training Programme. theoretical Overview. J. Faculty Sci. Hokkaido Univ. Series 7, Geophys., 11(3),
Simiyu, S.M., Keller, G.R., 1998. Upper crustal structure in the vicinity of Lake Magadi in 577–599.
the Kenya Rift Valley region. J. Afr. Earth Sci. 27 (3–4), 359–371. https://doi.org/10. Zakir, F.A., Qari, M.H.T., Mostafa, M.E., 1999. Technical note a new optimizing technique
1016/S0899-5362(98)00068-2. for preparing lineament density maps. Int. J. Remote Sens. 20 (6), 1073–1085.
Simiyu, S.M., Keller, G.R., 2001. An integrated geophysical analysis of the upper crust of https://doi.org/10.1080/014311699212858.

81

You might also like