Hazards from Pyroclastic Density Currents at Mount Bamenda (West-Cameroon, Cameroon Volcanic Line): Mapping and Assessment by AMS Method

1Department of Life and Earth Sciences, Higher Teachers’ Training College, University of Maroua, P.O. Box 55, Maroua, Cameroon 2Department of Earth Sciences, University of Yaoundé I, P.O. Box 812, Yaoundé, Cameroon 3Laboratory of Geology, Higher Teachers’ Training College, University of Yaoundé I, Cameroon, P.O. Box 47, Yaoundé I, Cameroon Faculty of Sciences, University of Maroua, P.O. Box 46, Maroua, Cameroon


Introduction
The Cameroon Volcanic Line (CVL) is characterized by alignment of oceanic and continental volcanic massifs and anorogenic plutonic complexes extending from Pagalu Island in the Atlantic Ocean to Chad with N30°E general orientation (Fig. 1). Volcanism along this line began in the Eocene by the formation of Bamoun plateau between 51.8 and 46.7 Ma (Moundi et al., 2007) and Mount Bangou between 44.7 and 43.1 Ma (Fosso et al., 2005) and continues until the present day on Mount Cameroon (Suh et al., 2003). The products of the CVL volcanism are mainly rhyolites, phonolite, trachyte and basalt. Welded and non-welded ignimbritic deposits are found only in the continental sector of the CVL, particularly in the Mts Bambouto and Bamenda (Fig. 2). Deposits of limited extension have been also described in Nkogam (Kamgang, 1986), Oku (Dunlop, 1983 andLissom, 1991) and Nganha volcanoes (Nono et al., 1994).
The Bamenda volcano covers an area of about 600 km 2 and culminates at 2621 m (Bambili Lake borders). It is characterized by two elliptically shaped calderas ( Fig. 2): Santa-Mbu caldera (6×4 km) and Lefo caldera (4×3 km). Their floor located respectively at 550 and 400 m is composed mainly of trachytic domes, which are also abundant on the external slopes of the massif. In addition to felsic, intermediate and mafic lavas, numerous outcrops of ignimbritic deposits are found in mount Bamenda. The massifs lay on a granitic Pan-African basement (Toteu et al., 2001;Nzolang et al., 2003). According to Kamgang et al. (2007;, intermediate and felsic lavas (27.40-18.98 Ma) are made of rhyolites, trachytes, benmoreites and mugearites. Mafic lavas (hawaiites, basalts and basanites) are dated from 17.4 Ma to the present. The presence of pyroclastic density currents deposits justify the dangerousness of this volcano because it is still considered as an active volcano in the CVL with the basalts recently dated to 0 Ma (Kamgang et al., 2007).

Pyroclastic Density Currents Deposits
Pyroclastic Density Currents (PDC) are volcanic emissions that lead to the formation of ignimbrite deposits. They constitute an inhomogeneous mixture of volcanic particles (lava fragments, ash and block) and gas that flow according to their density relative to the surrounding fluid and due to Earth's gravity (Branney and Kokelaar, 2002;Schminke, 2004). Speeds (up to 300 km/h) and temperatures (up to 600-700°C) of these flows are generally highon the earth surface (Wilson and Houghton, 2000). PDC are mainly controlled by gravity (Carey, 1991;Druitt, 1998a;Belousov et al., 2002;Burgisser and Bergantz, 2002) and commonly indicating a turbulent regime (Fisher, 1990;Carey, 1991;Dellino and La Volpe, 2000;Valentine and Fisher, 2000). A pyroclastic flow is diluted when he have a solid volume concentration equal or less than a few percent; a concentrated suspension or of high concentration will thus have a concentration of several percent or several tens of percent (Choux, 2001). PDC are subdivided on the basis of textural characteristics of the deposits, into two sub-groups which are either two particular phenomena, separated by a leap concentration (Walker, 1983;Wilson, 1997) or two extremes of a single continuous process (Fisher and Schmincke, 1984;Cas and Wright, 1987;Druitt, 1998b): Pyroclastic surges and flows.
Pyroclastic surges are diluted concentration flows whose stratified deposits drape topography. Pyroclastic flows are denser than surges and generally form more massive deposits, poorly sorted, filling the valleys. Surges can be associated with a pyroclastic flow, located at the periphery of this flow or generated by ash cloud surmounting it (case of Mt St Helens, Mellors et al. (1988) and the Soufriere Hills volcano, Calder et al. (1999).
On Mount Bamenda, deposits of these two types of flows have been identified. They can originate by fountain-like collapse of parts of an eruption column following explosive fragmentation of magma and rock in a volcanic conduit, or from laterally inclined blasts, or from hot avalanches derived from lava domes. They can transport large volumes of hot debris rapidly for many kilometers across the ground and they constitute a lethal and destructive volcanic hazard. Ground-hugging pyroclastic density currents produce a buoyant counterpart, known as a phoenix cloud or co-ignimbrite ash plume, which can carry ash and aerosols into the stratosphere and so cause significant climatic perturbation. Most processes within pyroclastic density currents are impossible to observe and so are commonly inferred from the associated deposits.

Aim of Study
During the past years, many people have been killed on volcanoes by relatively small eruptions that otherwise represented no hazard to nearby populated and cultivated areas. Pyroclastic surges and flows have caused about 55,000 deaths since 1600 A.D. (about 48% of all volcano-related fatalities). The Bamenda volcano being still considered as active can potentially cause massive damage in case of an eruption occurred, because in the past, violent eruptions have set up significant ignimbrite deposits on the slopes of the volcano.The purpose of this paper is to map and assess pyroclastic density currents hazards at Mount Bamenda which are the most disastrous in terms of possible volcanic processes in this region. To achieve this goal, anisotropy of magnetic susceptibility has been used to determine flow directions in ancient ignimbritic deposits which are visually nearly isotropic in most outcrops. Magnetic fabrics of ignimbrites help us to determine palaeoflow directions of pyroclastic density currents, to then reassemble discontinuous deposits into their original ignimbrite sheets and to identify their source. Such reconstructions allow us to plan future potential ignimbritic eruption scenarios and draw a hazard map of pyroclastic density currents in the Bamenda volcano.

Petrography of Ignimbritic Deposits
Ignimbrites are predominantly observed in the central part of the CVL; they are concentrated principally in Mount Bamenda and its SW extension (Mount Bambouto) (Fig. 2). The ignimbrites of Bamenda volcano constitute roughly 7.5% of the massif representing about 45 km 2 with a volume estimated at 6.3 km 3 (Fig. 3). Ignimbrites of Mt Bamenda lie on a granito-gneissic basement and are covered by old lateritized basalt as in the case of Mounts Bambouto. The aspect ratios of these ignimbrites are generally high (2.77×10 −2 to 7.23×10 −2 ). According to Branney and Kokelaar (2002) classification, these ignimbrites consist of massif lithic breccias (mlBr) and massif lapilli tuff (mlT).

The Welded Ignimbrites
The welded ignimbrites outcrop as sheets, metric and decimetric blocks balls. The color of rock and the proportions of lithic fragments and minerals are variable depending of the type of the facies identified at Bamenda, Mbu, Mbengwi, Bambili, Sabga and Big Babanki (Gountié Dedzo et al., 2011).

The Ignimbrites of Big Babanki, Sabga and Bambili
The ignimbritic deposit of Big Babanki, Sabga Bambili and cover a surface of 3 km 2 , 5 km 2 and 6 km 2 respectively. These deposits form small outcrops with thickness varying from 65 from 120 m ( Fig. 4d-f). The remnant blocks (up to 7.5×12 m) of ignimbrites (surge deposits) (Fig. 5c) are also observed at Bambili (Gountié Dedzo et al., 2015). They lie on a basement rock made of migmatite, schist and granite. The welded mlT ignimbrite of Bambili possess trace of rubefaction in places and characterized by two cooling units (dark-grey and whitish) while the ignimbrite of Big Babanki is represented only by one whitish unit; the two outcrops are covered in places by a mlBr facies. The ignimbrite of Sabga ( Fig. 5d) is also a welded mlT constituted by one dark-gray unit covered by one light grey unit. The fiammes (25-40%) are generally distorted in the basal parts of the outcrops. Enclaves of rocks (8-15%) consist of granites vitrophyre, trachytes, ignimbrites and rhyolites; in mlBr facies of Bambili, they represent 60 to 70% where there are predominantly enclaves of black vitrophyre. The matrix is devitrified and consist of alkali feldspar (5-10%; sanidine: Clinopyroxene (1%), biotite (1%), oxides (1-3%), plagioclase (1-3%) and Ab 58-57 Or 57-41 ), quartz (2-3%).   The Non-Welded Ignimbrites The non-welded ignimbritic deposits (about 5 km 2 ) outcrop in the Santa-Mbu and Lefo caldera at Santa Coffee and Ndzah. They lie on a basement rock consisting of micaschist and covered by trachytic flow in the Lefo caldera. Their precise thickness (> 15 m) is difficult to evaluate because of the profuse vegetation and uneven topography. The rock is very powdery and consists of enclaves (15-25%) mainly trachytic with minor percentage of rhyolite, ignimbrite, obsidian and granite fragments. The average size of these enclaves is 3×2.3 cm; few enclaves reached 1.5×2.4 m; they have characteristics close to co-ignimbritic breccias as in Bambouto volcano. Probably due to the very advanced state of alteration in these deposits, the proportion of accretionary lapilli are low (<5%). There is presence of feldspar, quartz and remnants fragments of pumices in the ashy matrix of deposits.

Sampling Method and AMS Measurements
The response to an applied magnetic field on a material or rock refers to Magnetic Susceptibility (MS). The main element responsible for the magnetization is iron. Anisotropy of low field MS is a multipurpose petrofabric instrument. AMS predominantly defines grain-shape anisotropy for magnetite; it's also expresses crystallographic control on magnetic properties for other minerals. We may therefore, infer the orientationdistribution of a dominant mineral from the AMS of a rock. Flow directions from pyroclastic deposits, magma and current directions from sediment, can record by AMS main directions. In this study, AMS is applied on ignimbrites with a view to reconstitute palaeoflow directions of pyroclastic current. All measurements were made in Géosciences Environnement Toulouse (GET) laboratory at Paul Sabatier University (Toulouse, France). Sampling of the Bamenda ignimbritic deposits was realized on 15 sites (Fig. 3) using a portable, drillapparatus with a non-magnetic diamond-tipped drill bit. At each AMS station, a maximum of 9 oriented cores were collected in a surface covering about 6 to 12 m 2 . Evidence of rheomorphic flow or secondary flow (Wolff and Wright, 1981) that might have modified their primary emplacement magnetic fabric was not identified in the studied rocks. Samples were then oriented (azimuth/dip) using magnetic compass and carefully numbered. Each core sample was cut into 22×25 mm cylinder specimens in laboratory, using a diamond tipped, non-magnetic saw blade. Two to four specimens per core sample were obtained, hence a total of 297 specimens. AMS data were acquired using a Kappabridge susceptometer (KLY-3S, Agico, Czech Republic; operating at low alternating field (4×10 −4 T at 920 Hz) with a sensitivity of about 2×10 −7 SI, allowing anisotropy discrimination below 0.2% over a wide range of susceptibility. This method measures the orientation of the magnetic carriers (minerals) in a rock which help to regenerate ancient flow directions. The MS can also be measured in different directions and the results are usually expressed in terms of an ellipsoid characterized by the maximum (K 1 ), intermediate (K 2 ) and minimum (K 3 ) susceptibility directions. The anisotropy of the principal susceptibility axes is commonly considered to be inherited from the mechanism of emplacement and can be used to rebuild ancient flow directions in ignimbrites lavas and intrusions. The small axis K 3 (minimum susceptibility) is the foliation pole, i.e., the normal to the plane of magnetic foliation; the long axis K 1 (maximum susceptibility) defines the magnetic lineation, meanwhile K 2 is the mean susceptibility. The arithmetic mean of the lengths of the principal axes (Km = [K1 + K2 + K3]/3) is the mean MS (Km). The method also quantifies the anisotropy percentage achieved by the relation: P% = [(K1/K3)-1]×100. In addition, the linear (L%) and planar (F%) anisotropies are also provided by equations: L% = [(K1/K2)−1] ×100 and F% = [(K2/K3)−1]×100. The T parameter of Jelinek (1981) expressed by T = (2lnK 2 -lnK 1 -lnK 3 )/(lnK 1 -lnK 3 ) characterizing the shape of the MS ellipsoid; its ranging from +1 for oblate ellipsoid to -1 for prolate ellipsoid. T values between -0.5 and +0.5 are characterized by triaxial ellipsoids.
Since the early work of Ellwood (1982), several studies have used AMS in an attempt to locate source vents of large ignimbrites. Some authors (MacDonald and Palmer, 1990;Seaman et al., 1991) compared AMS data from ignimbrites with petrographic and field observations and recognized that AMS provides a precise and significantly faster means of approximating the transport direction of ignimbritic deposits than techniques based on macroscopic examination. It is established that the heterogeneous character of an ignimbritic deposits does not really affect the AMS fabric and that this method provides a realistic indication of the flow direction (Wolff et al., 1989). Results of previous MS studies that have inferred flow patterns for ignimbritic deposits propose that the alignment and distribution of magnetic grains produces the anisotropic magnetic fabric (Wolff et al., 1989;Palmer and MacDonald, 1999). Magnetic lineations and foliations in ignimbrites arise as a consequence of alignment of magnetic mineral grains parallel to the direction of flow and the rotation of somewhat elongate to sub-equant magnetic grains so that their long dimension approximately parallels the direction of flow (Ellwood, 1982;Knight et al., 1986). The mean magnetic foliation plane (characterized by K 2 -K 1 axes) is perpendicular to the K 3 axis and estimates the flow plane, for a normal magnetic fabric. Nevertheless, the plane of magnetic foliation frequently differs in orientation (imbrication angle) comparative to the flow plane (Ellwood, 1982;Knight et al., 1986) and dips in a direction opposite to the direction of flow. The imbrication dip direction is inferred to point in direction of the source area (Fig. 6b). The K 1 axis is usually interpreted to be parallel to the flow direction and thus plunges towards the source (Knight et al., 1986). The use of the K 1 axis alone as a proxy for flow direction is not always reliable (Seaman et al., 1991;Hillhouse and Wells, 1991;Ort, 1993;Le Pennec et al., 1998) and a different methodology must be used to interpret the magnetic data. The imbrication angle of the magnetic foliation plane, which equivalents the deviation of the K 3 axes from the normal to the macroscopic flow plane, offers another method to distinguish the flow direction of ignimbrite (Henry, 1980;Thompson and Oldfield, 1986) (Fig. 6a). Application of the imbrication angle methodology may result in other interpretation of the direction of flow (Alva-Valdivia et al., 2005). In this study, only the K 3 axes were used to determine flow directions because the K 1 axes are approximately perpendicular to the dip direction of the imbrication plane in most AMS stations.

Results
Directional data of AMS study show a fairly reliable transport pattern in 6 sites inferred by magnetic foliation. These sites situated at Bambili, Mbu Mbengwi and Bamenda show well-defined magnetic fabrics characterized by mostly low plunges of the magnetic lineations (1 to 26°) and gently inclination of magnetic foliation planes (3 to 24°). The remaining six sites at Bamenda and Sabga cannot used to infer flow directions because they exhibit commonly highly plunging (20 to 80°) of magnetic lineations and sharply dipping foliations (52 to 81°).
Synthetic stereograms of AMS sites at Bamenda, Mbu and Mbengwi show mildly dipping and steeply dipping magnetic foliations (Fig. 6b). The two AMS stations of Mbengwi locality display normal fabrics (with K 1 axis dips at angle of less than 30°, whereas the magnetic foliation plane is subhorizontal i.e. K 3 axis is nearly vertical or sharply plunging). In the station GM31, the data shows a well disposition of magnetic foliation plane characterized by the SSW dipping of 24°; the same direction is observable for the magnetic lineation which is approximately parallel to flow. At the station GM32, the magnetic lineation appears transverse to the flow direction, while the K 2 -K 1 axes are also welldefined with magnetic foliation inclined somewhat (3°) to the SE. At Mbu, the imbrication (19°) of magnetic foliation plane to the WSW is observable in the site GM21. Stations GM28 and GM29 of Bamenda display inverse fabrics (in which K 1 axis dips at high angle, generally > 50°) whereas Station GM27 exhibits normal fabric with plane of magnetic foliation slightly plunge (11°) to NW. The magnetic lineation for Bamenda seems transverse to flow, i.e., sub-perpendicular to the direction of imbricated magnetic foliation plane.
At Sabga (Fig. 6b), the majority of magnetic foliation planes are steeply dipping (53-75°) in all stations. The magnetic foliation plane at Bambili, is well organized and dips moderately (6-28°) to the NE for the stations GM26 and GM24. Concerning the station GM26, magnetic lineations have roughly the same NE direction; the magnetic lineation is on the other hand, transverse to flow direction, at station GM24, i.e., subperpendicular to the imbricated direction of magnetic foliation. At Bambili, site GM33 shows an inverse fabric with subvertical plane of magnetic foliation and lineation. The magnetic foliation dips gently at 6° to the NE and the magnetic lineation has the same NE trend at station GM30.

Identification of Emission Centres
At Bamenda, Mbu and Mbengwi the overall flow direction appears to radiate from the same area, specifically to the Mbu caldera which is the probable emission center. Field indicators around station GM28 in this region also indicate that flow was towards azimuth 46° (NE).
Flow direction is from NE to SW at Big Babanki, Sabga and Bambili. Imbricated fiammes around stations GM22 and GM24 are the field indicators indicating that flow pattern was roughly toward azimuth 235 and 222° respectively. The probable emission center is therefore situated to the NE of these areas. The probable source of these ignimbrite deposits is then localized in Oku Lake. Finally, based on imbrication of magnetic foliation and field indicators, we conclude that the pyroclastic density current responsible for these deposits was probably generated from Mount Oku caldera concerning ignimbrites of Big Babanki, Sabga and Bambili and from Santa-Mbu caldera for ignimbritic deposits of Bamenda, Mbu and Mbengwi (Fig. 7).

Mapping
With the help of outcrop map of exposed ignimbrites at Mt Bamenda and their thickness, we have realized hazard map (Fig. 8) related to surge and pyroclastic flow, or ignimbritic eruptions responsible for these deposits. The region of high risk corresponds to: (i) zone where the deposits of ignimbrites are thick; and (ii) where highly sloping topography area would facilitate the displacement of pyroclastic currents. The regions of medium, moderate and low risk corresponds zones of less thick deposits or to areas where there are no ignimbritic deposits.   In the areas where ignimbrites are covered by younger rocks, the boundary of pyroclastic deposits was estimated from multiple quarries (for construction or engineering materials), wells and especially from water drillings (for domestic use), carried out on basaltic and trachytic covers. It should be noted that the areas where the basalt outcrops are densely populated zones because of fertility of soils. The drillings made in these areas are in significant numbers (over 185 listed) and deeper than 45 meters in most cases. The data collected (thickness of pyroclastic deposits) have permitted to estimate in more realistic manner, the surface, the volume of ignimbrites and reliable hazard map.

Assessment
The Bamenda volcano and its close surrounding are densely populated with more than 750,000 inhabitants. The information for location of the exposed elements and their inventories, followed by the investigation of the different values (economic, environmental, social, human, monetary and strategic) was obtained from the appropriate Cameroonian government departments and concerned administrations completed by field studies. The Cameroonian authorities consider the following elements as being the most important: strategic edifices for management of crisis (health centres, army and security, governance); main centers of financial interest and economic (tourist centres, factories and banks); Population (residential areas and places of sporadic concentrations of people, such as main centers of worship, stadiums, schools and markets); networks and buildings, infrastructures (roads and bridges, housing, power networks, telecommunications relays, water-supply systems,); farmland and natural environment (hydrography and vegetation, industrial plantations and food crops).
Measuring the value of the different element-at-risk was established on information obtained from the several services and organizations responsible for the social system (Table 2).
It is thus possible to assess by this method of calculation the total capital budget (updated to 2014) of the different element-at-risk within the study area at about 3,063 billion Fcfa or $US5.4 billion.

Conclusion
For the determination of flow directions in initially hot volcanic deposits and specifically in deposits of pyroclastic density currents, which are generally isotropic in most outcrops, anisotropy of magnetic susceptibility is a good tool. The high density of population in the Mount Bamenda (more than 750,000 inhabitants) increases the level of risk from future explosive eruptions. This type of study could be extended in this massif on other volcanic product such as basalt, trachyte and rhyolite which cover about 70% of Bamenda volcano in order to realize a reliable hazard map. In order to refine the hazard zoning map realized in Mt Cameroon (where the last eruptions occurred in 1999 and 2000) (Thierry et al., 2008), the AMS study will also very helpful in this volcano.

Acknowledgment
This work was stimulated by numerous discussions with the late Professor Nono Alexandre which has suddenly deceased on the morning of 24 July 2012; our posthumously thanks are addressed to him. The authors thank Professor Alain Fotso for accommodation facilities during the field work.

Funding Information
This work has been partly supported by the SCAC (Service de Coopération et d'Action Culturelle de la France au Cameroun), EGIDE (Centre français pour l'accueil et les échanges internationaux) and the French Government.

Author's Contributions
Merlin Gountié Dedzo: Designed the study, collected data in the field, made the interpretation and discussion and wrote the manuscript.
Pierre Kamgang: Collected data in the field and made the interpretation, wrote the manuscript and prepare some figures.
Emmanuel Njonfang: Made the interpretation, wrote the manuscript and prepare some figures.
Ghislain Zangmo Tefogoum: Collected data in the field and made the interpretation, wrote the manuscript and prepare some figures.

Ethics
This article is original. All authors have read and approved the manuscript and no ethical issues involved.