Abstract
Fluid flow pathway characterisation is critical to geothermal exploration and exploitation. In fractured geothermal reservoirs, it requires a good understanding of the structural evolution together with the fracture distribution and fluid flow properties. A fieldwork-based approach has been used to evaluate the potential fracture permeability characteristics of a typical high-temperature geothermal reservoir in the Southern Negros Geothermal Field, Philippines. This is a liquid-dominated resource hosted in the andesitic Quaternary Cuernos de Negros Volcano, Negros Island. Fieldwork reveals two main fracture groups based on fault rock characteristics, alteration type, relative age of deformation, and associated thermal manifestation, with the youngest fractures mainly related to the development of the current geothermal system. Fault kinematics, cross-cutting relationships, and palaeostress analysis suggest at least two distinct deformation events under changing stress fields since probably the Pliocene. We propose that this deformation history was influenced by the development of the Cuernos de Negros Volcano and the northward propagation of a major neotectonic structure located to the northwest, the Yupisan Fault. A combined slip and dilation tendency analysis of the mapped faults indicates that NW-SE structures should be particularly promising drilling targets under the inferred current stress regime, consistent with drilling results. However, existing boreholes also suggest that NE–SW structures can act as effective channels for geothermal fluids. Our observations suggest that these features were initiated as the dominant features in the older kinematic system and have then been reactivated at the present day.
1. Introduction
Permeability, heat source, fluid recharge, and capping mechanism are the vital elements to consider during the development of a geothermal reservoir [1, 2]. In a typical subduction-related geothermal system like those seen in the Philippines, the reservoir is mostly hosted in crystalline rocks in which permeability arises mainly from fractures, and less from the intrinsic permeability of the reservoir rocks [3]. It is therefore necessary to understand the type of fractures present, the fracture development history, and the nature of the transmissive present-day fracture networks developed at depth prior to tapping the reservoir through drilling. Ultimately, such understanding should allow geothermal production to be maximised through targeting optimal fractures in the subsurface.
However, the structural modelling of subsurface reservoirs in volcano-hosted geothermal systems has its own challenges. Subsurface analyses using seismic refraction and reflection data are difficult to use as thick volcanic rocks are usually opaque to seismic waves and geological structures are often difficult, or impossible, to image [4]. Resistivity data is commonly used to visualise the reservoir structure based on the implied fluid content and development of alteration zones [5]. Although this can highlight the presence of large-scale structures, it is still challenging to identify reservoir-scale fracture networks using this technique. Additionally, at the early stages of geothermal exploration, carrying out geophysical surveys (e.g., gravity, resistivity, and seismic) poses large financial risks to the developer. Thus, it is important to first maximise surface geological data before moving forward with the exploration stages. This work is aimed at illustrating how to better utilise surface geological data in understanding fractured geothermal systems using the Southern Negros Geothermal Field (SNGF) as a case study.
The SNGF lies in the municipality of Valencia in south Negros Oriental, Philippines. It is a volcano-hosted, high-temperature geothermal system, with temperatures ranging between 200 and 300°C, sitting on the northeastern flanks of the Cuernos de Negros (CDN) volcanic edifice (Figure 1). It is liquid-dominated with localised two-phase zones [6] containing fluids which are generally neutral in pH, moderately saline, and have low gas content [7]. The field was commissioned in 1983 with a total installed capacity of 192.5 MWe. However, despite its long history of geothermal exploration and development, its fracture systems are still poorly understood.
(a)
(b)
(c)
Here, we use field and thin-section observations to establish a deformation history for the SNGF particularly highlighting its role in influencing the development of the geothermal system. Given the limitations of outcrop quality and distribution that are typical in tropical countries (e.g., high rates of weathering and erosion and extensive vegetation cover) and the effects of volcanism (i.e., recent phreatic eruption of the volcanic centers may cover or erode exhumed structures), methodologies that optimise the field data are also examined. From the mapped structures, slip and dilation tendencies are evaluated and we show how these results could relate to and influence a drilling strategy for the SNGF.
2. Regional Geological Setting
The Philippine Archipelago is where four tectonic plates—SE Eurasia, Philippine Sea, Pacific, and Indo-Australia—meet [8]. The largely aseismic Palawan-Mindoro microcontinent lies to the west representing a fragment rifted from mainland Eurasia in the mid Cenozoic, and to the east lies the seismically active Philippine Mobile Belt on which the majority of the country is located (Figure 1(a)). The latter region is an actively deforming zone composed of terranes of various affinities (i.e., from the ancient Philippine Sea Plate and the Indo-Australian margin) [9] that are bordered by subduction zones of opposing polarities: the west-dipping Philippine Trench and East Luzon Trough to the east and the east-dipping Manila, Negros, Sulu, and Cotabato trenches to the west (Figure 1(a), [10–12]). Shallow earthquakes are dispersed across the Philippine Mobile Belt indicating its continued active deformation due to plate tectonic forces [8].
Traversing almost the entire length of the country, from northwest Luzon to southeast Mindanao, is the >1200 km long sinistral Philippine Fault (Figure 1(a)) which has formed due to the oblique convergence of the Philippine Sea Plate with the Philippine Mobile Belt [13]. It is suggested that the Philippine Fault formed 4 Mya after the plate convergence changed from north to NNW with respect to Eurasia, although its northern segment appears to have been initiated much earlier (10 Ma) [14]. GPS data show that the Philippine Fault has a slip rate of 2 to 3 cm/yr [14, 20] or 2.4 to 4 cm/yr [17] which represents a third of the oblique convergence of Philippine Sea Plate, whilst the two-thirds is accommodated along the Philippine Trench and other major structures across the country [14]. Maximum compression, , from recent studies is oriented between 90 and 110° in Luzon [17] and approximately east-northeast in Bicol region [16].
3. Geology of Negros
Negros Island is made up of three Cenozoic-Quaternary tectonostratigraphic terranes that represent part of the subduction arc system related to the Negros Trench (Figure 1(b)) and are underlain by oceanic volcaniclastic basement, which is thought to be Cretaceous in age [11, 21]. From west to east in Negros, two overlapping volcanic arcs of different ages, the Ancient and Recent Negros Arcs, and the sedimentary Visayan Sea Basin are present (Figure 1(b); [11, 22]).
The Ancient Negros Arc comprises Eocene to Oligocene andesitic to dacitic volcanic and clastic rocks intruded by a Miocene dacitic diatreme complex [22]. It is highly mineralised, hosting the Bulawan intermediate sulphidation gold deposit, the gold-poor Sipalay deposit, and the Hinobaan porphyry copper and molybdenum deposits [22], all of which are situated in the southwestern part of Negros. The Recent Negros Arc [11] or Negros Belt [23] is composed of Middle Miocene to Pliocene andesite flow breccias, volcaniclastics, and conglomerates that are overlain by Late Pliocene andesitic volcanics and Quaternary andesite and basalt stratovolcanoes [22]. Geomorphologically, the recent arc is represented by a 260 km chain of volcanoes, four of which are on Negros Island (from north to south): Mt. Silay, Mt. Mandalagan, Mt. Canlaon, and Cuernos de Negros (CDN) [23] (Figure 1(b)). Of these four, only Canlaon is active with the most recent volcanic activity (i.e., release of white plumes and volcanic earthquakes) occurring in January 2018 [24], whilst Mandalagan and Cuernos de Negros are considered to be in their fumarolic stage [23].
Towards the east, the Visayan Sea Basin, representing the back-arc region of the Negros arc system [25], underlies the eastern coast of Negros, the Tañon Strait, and the islands of Cebu and Bohol (Figure 1(b)). The basin is filled with up to 4 km thick carbonate and volcaniclastic sequences deposited from the Middle Oligocene to Middle Miocene [11]. These are generally folded, with fold axes oriented NNE-SSW on the average. Rangin et al. [25] proposed that the Visayan Basin comprises a series of NNE-SSW-trending horst and graben structures, with the Tañon Strait corresponding to a graben, which cuts the earlier folds. However, recent studies by Aurelio et al. [26] following the Mw 6.7 earthquake in February 2012 propose the existence of a northeast-striking and northwest-dipping reverse fault, the Negros Oriental Thrust, that runs from west of CDN towards eastern offshore Negros.
Southern Negros is formed mainly by Quaternary volcanic rocks that are part of the Recent Negros Arc. Miocene to Early Pleistocene clastics have been locally exposed in the northwest part of CDN as part of the Pamplona Anticline—a result of the regional fault-propagation folding associated with the Negros Oriental Thrust, which is mapped onshore as the Yupisan Fault (Figure 1(b); [26, 27]).
4. Local Geology
Deep drilling to 3300 m depth over the last three decades reveals that the CDN volcanic complex was created by several volcanic and intrusive events ([26]; Figure 2). The oldest rocks drilled are thick, 990 m on average, Miocene volcanic sequences of altered andesites intercalated with tuffs and calcarenites with occasional volcanic and sedimentary breccias, known as the Puhagan Volcaniclastic Formation. These rocks are cross cut by the Nasuji quartz monzodiorite to micromonzodiorite pluton which led to the formation of a metamorphic aureole known as the Contact Metamorphic Zone. Geochronological studies of the pluton have yielded contradicting age of Miocene (10.5 Mya using K-Ar in [28, 29]) and Pleistocene (0.7 to 0.3 Mya using Ar-Ar in Rae et al. [18]). By the Early Pliocene, the Okoy Sedimentary Formation and overlying undifferentiated andesitic volcanics and pyroclastics of the Southern Negros Formation were deposited. All the above mentioned formations have been intruded by at least two dyke events during the Pliocene. Lateral and vertical variations of lithologies and facies within the Okoy Sedimentary and Southern Negros Formations have been detected during drilling of wells and indicate the presence of a palaeotopography within the SNGF in which the western sectors were uplifted in the Early Pliocene [28]. This is confirmed by fossil assemblages within the two formations in the western region which are characteristic of a shallow to subaerial environment, whilst those preserved in the eastern region are typical of a deep marine environment [28]. These rocks are overlain by the Quaternary-aged andesitic Cuernos Volcanics, which can be subdivided into different members depending on which volcanic edifice of the CDN volcanic complex they are associated with (i.e., main CDN peak, Talines, Guinsayawan, Figure 1(c)). Radiocarbon dating of charred wood within the Cuernos Volcanics suggests a youngest eruption age of 14,450 years [29]. These young volcanics cover much of the surface of the present-day CDN volcanic complex, with exposures of the older Southern Negros Formation limited to the downstream river valley area of the E-W Okoy River (Figure 1(c)).
Topographic lineament analysis using combined high- and low-resolution digital elevation models carried out indicates a dominance of ENE-WSW and NW-SE to NNW-SSE features (Figure 1(c)). The most conspicuous lineaments are the ENE-WSW-trending set that coincide with the Okoy River, and appear to be discontinuous and arranged as right-stepping en echelon features, representing the traces of a known fault network in SNGF, designated here as the Puhagan Fault Zone (Figure 1(c)). Geomorphological kinematic indicators such as push-up ridges observed along the trace of this fault zone suggest it has a dextral sense of movement.
5. Mapped Structures
The SNGF rocks exposed at the surface are exclusively deformed by brittle structures, including different types of fractures, such as faults, joints, and veins. The fractures observed in 84 out of the 135 outcrops that were studied can be sorted into two main groups, here termed group 1 and group 2. This two-fold classification is mainly based on the types of associated fault rocks, key alteration minerals, and the host rock which the fractures cut. The field characteristics are summarised in Table 1 and discussed in detail in the sections that follow.
5.1. Group 1 Fractures
Group 1 structures are restricted to the river exposures along the ENE-draining Okoy River gorge and one of its SE-draining tributaries (Figures 1(c), i and 3). This restricted region corresponds to the exposures of the Southern Negros Formation adjacent to the river, but it is most likely that these structures may be widely developed in the older rock strata that underlie much, if not all, of the Quaternary CDN edifice. Textures in the Southern Negros Formation in general are almost completely obliterated due to the effects of hydrothermal and supergene alteration, appearing as dark grey with yellowish to reddish patches due to iron oxide and sulphur deposition. Ghost phenocrysts are observed suggesting that the protoliths were porphyritic, but the crystals have been almost entirely replaced by clays.
(a)
(b)
(c)
(d)
Group 1 structures form as conspicuously grey fractures filled with abundant <1 mm to 2 mm grain-size pyrite crystals (Figure 3). Along the central part of the Okoy River gorge, these are E-W to WNW-ESE-striking (096°/62°-S) sinistral faults with slickenlines typically pitching <15° and which run subparallel to the trace of the main Puhagan Fault Zone. A fault core is recognised and is interpreted to correspond to the region where most shear displacement has been localised. A series of cm- to m-spaced ENE-WSW-trending subvertical faults here are filled with cataclasite comprising fine angular (rare 10 mm, mostly <5 mm grain-size) fragments of altered protolith, weakly foliated in some cases, set in a cemented dark grey matrix (almost clay-sized, <1 μm) (Figure 3(a)). Petrographic work on the fault rock samples suggests a dominance of pyrite mineralization within a generally cryptocrystalline matrix carrying identifiable fine crystals of pyrite, quartz, opaque minerals, and amorphous silica (Figure 4(a)). Sinistral senses of motion are indicated by microscopic kinematic indicators (i.e., en echelon features) (Figure 4(a)). Few offset markers are seen along the fault plane, so it is difficult to estimate total fault displacements.
(a)
(b)
The sinistral faults along the downstream part of Okoy River are associated with smaller NE-SW- to E-W-striking sinistral fractures based on millimeter-scale offsets, interpreted to be synthetic Riedel shears that are completely cemented and mineralised. Likewise, dense arrays of NE-SW-striking steeply dipping quartz- and pyrite-filled tensile fractures are observed (in i and ii in Figure 1(c)). Fine (~1 to 2 μm) gypsum crystals also occur within some mineral fills. The tensile fractures lie obliquely anticlockwise to the main sinistral faults or form as en echelon features (Figure 3(c)). Where the two key alteration minerals, pyrite and quartz, are present, they usually occur as anhedral crystals, but when pyrite occurs on its own, it is typically sub- to euhedral, with crystals that can be as large as 5 mm and usually oriented perpendicular to the fracture walls. Cockade overgrowth textures [30] within some of the euhedral pyrite crystals are also observed in thin sections (Figure 4(b)).
5.2. Group 2 Fractures
Group 2 fractures are the dominant and ubiquitous features throughout the SNGF. Of the 84 outcrops mapped within the geothermal reservation, 90% expose faults classified under this group. The majority of large fractures are generally oriented WNW-ESE to NNW-SSE, usually steep to moderately dipping, exhibiting either normal, dextral, or normal-dextral oblique senses of movement, where kinematic indicators are preserved (Figure 5). Smaller, but poorly preserved and less frequent NE-SW and NNE-SSW-trending fractures have also been mapped. Where shear sense can be determined, it is often observed in normal faults and, more rarely, in sinistral structures.
Group 2 fractures generally contain incohesive fault rocks, typically fault gouges and fault breccias (Table 1). Occasionally, sharp slip planes or localised brittle shear zones occur. These structures cut both completely altered and fresh volcanic rocks. In some outcrops, the damage zones belonging to faults at least ten meters long have widths ranging a few centimeters to tens of meters. Most outcrops, however, preserve evidence that the brittle deformation is followed by intense clay alteration localised along the fault zone. Active and recently active thermal manifestations—such as hot springs, gas seepages, and hot ground—are often found along or adjacent to the traces of these group 2 fractures, implying the permeability present on these sets of structures that channel hydrothermal fluids. This further suggests that these faults play critical roles in the geothermal development and present-day fluid flow.
Two outcrops at localities 81 and 104 feature some of the best preserved group 2 structures and are discussed in detail below (more outcrop discussion can be found in Pastoriza [27]). These faults show the typical characteristics of the group, particularly the dominant NW-SE fractures. In locality 81, this fault is a large dextral-oblique structure, whilst at locality 104, the fault has a dip-slip normal sense of movement.
5.2.1. Locality 81
This WNW-ESE-oriented dextral-oblique fault (mean orientation of 118°/80°-S) cuts moderately to intensely altered porphyritic andesite (fault A in Figure 6; refer to box iv in Figure 1(c) for the location). Narrow 40 to 60 cm wide fault cores, with a wider 150 cm intensely damaged zone, are flanked by a still broader 600 cm wide moderately damaged zone. Background fracturing is present outside the damaged zones on both sides of the fault zone (Figure 6(b)). The fault preserves a crudely banded, variably altered pale gouge, formed by moderately to completely pulverised host rocks. In most parts, the fault is clay-altered and lined with fine-grained, hematite-filled and coarse-grained fibrous gypsum veinlets. Some less altered clasts of the porphyritic andesite protolith up to 15 cm in diameter are preserved within the fault core. Fault kinematics are gleaned from slickenlines and slickenfibres, varying between a purely dextral to less common dip-slip normal sense of motion (Figure 6(d)). Offset markers are poorly preserved, but are most likely minimal, suggesting generally small finite strains. One to two centimeter thick gypsum veins have crystallised along the walls of the fault planes and are oriented oblique to the wall (Figure 6(d)) suggesting that crystallization is contemporaneous with fault movements. Generally NW-SE-trending unfilled tensile fractures are observed to have formed adjacent to the fault plane, which are kinematically consistent with the main dextral shear sense inferred for the faults.
(a)
(b)
(c)
(d)
(e)
A left-stepping, generally E-W-striking dextral-normal fault occurs immediately to the south of the dextral fault (fault B in Figure 6). A 25 to 50 cm fault core here is filled with intensely oxidised fault rock (Figure 6(e)). Gypsum precipitation is preserved around some of the sheared margins of trapped protolith fragments. At the southern end of the outcrop, a network of associated smaller normal faults are generally oriented ENE-WSE to WNW-ESE with discrete slip planes (Figures 6(a) and 6(b)). Fault cores are not as well developed here compared to the two structures described above, but there is a strong indication that they channel hydrothermal fluids based on the formation of diffuse alteration haloes around most fractures. The network of smaller normal faults also appears to connect with the larger ones. Where they join, evidence of enhanced permeability is noted, such as the presence of empty vugs and suspected recent open fractures marking the sites of recently inactive gas seepage fissures, and preservation of increased deposition of secondary minerals (e.g., gypsum, travertine; Figure 6(e), ii).
5.2.2. Locality 104
A NW-SE-trending moderately north-dipping fault (131°/58°-N on average) runs perpendicular to the Okoy River (Figure 7; located at box iii in Figure 1(c)). It cuts an intensely silicified andesite outcrop that can be traced through both sides of the ravine, forming cave-like features on either side (Figure 7(a)). Preserved slickenlines and steps in the hanging wall suggest dip-slip normal senses of shear. The magnitude of displacement is difficult to assess due to the limits of exposure and lack of discernible offset markers in the wall rocks. The hanging wall is clearly exposed whilst the footwall is less distinct (Figure 7(b)). The fault core for this structure is a 12 to 15 cm thick gouge in which several fractured lensoid slivers of the host rock are entrained. The gouge-filled fault core is only exposed by hammering and has a thin film of what appears to be amorphous silica on its surface. Thin-section studies indicate intense silica-replacement and fracture infill of both the host rock and the fault rock (Figure 7(e)). Smaller fractures within the fault zone are oriented mostly WNW-ESE and are interpreted to be conjugate fractures to the main normal faults (Figure 7(d)). Where these smaller fractures join, local pull-apart structures or dilational jogs are observed (Figure 7(c)). In many cases, uncharacterised fine white crystals have precipitated around these structures which strongly suggest a recent outflow of mineral-rich fluids and/or gas. NNW-SSE sealed quartz veins and NW-SE unfilled tensile fractures are observed throughout the exposure which is consistent with the inferred direction of extension for this normal fault (Figure 7(d)).
(a)
(b)
(c)
(d)
(e)
5.3. Cross-Cutting Relationships
Group 1 structures are limited to the older Late Pliocene to Pleistocene Southern Negros Formation and are not observed in the younger lava flows and pyroclastics belonging to the Quaternary Cuernos Volcanics. Group 2 structures, on the other hand, are observed in both lithological formations. Given the older age of the lithological formation where group 1 fractures are found, this suggests that they are most likely to have formed earlier than group 2 and prior to the deposition of the Quaternary and Recent rocks (i.e., Cuernos Volcanics). The distinct characteristics of the associated fault rocks from the group 1 and group 2 fractures (Table 1) also suggest that the two groups of fractures have formed under different deformational conditions at different times.
Cross-cutting relationships observed in the field support the relative age hypothesis. In the downstream regions of the Okoy River valley, sinistral NE-SW-trending fractures interpreted as group 2 faults based on the development of millimeter-scale gouges with sulphur and oxides formed along exposed surfaces, everywhere offset group 1 WNW-ESE sinistral pyrite-filled fractures by up to 4 cm (Figure 8(a)). Offsetting relationships are also observed for two sets of tensile fractures trending ENE-WSW and NNW-SSE, interpreted to belonging to group 1 and group 2, respectively. Here, the NNW-SSE fractures cut through the ENE-WSW ones (Figure 8(b)), suggesting that group 1 structures are older than group 2 structures.
Quartz- and occasionally sulphur-filled group 2 fractures offset the pyrite-dominated tensile fractures of group 1 (Figures 8(c) and 8(d)). In some cases, group 1 tensile fractures here have reopened, either as pure mode I or as hybrid fractures (Figure 8(c)). Based on this evidence, the fracturing event that formed the group 1 fractures is referred to as stage 1.
There is also some field evidence that the group 2 fractures may have accommodated more than one phase of deformation. This is based on several instances of group 2 fractures being offset by other group 2 faults with a different sense of slip which are inferred to have been reactivated. For example, in the northeast area of SNGF, a NW-SE-trending, gently dipping sinistral-reverse group 2 structure offsets a steeply dipping group 2 fault of similar trend by approximately one meter. An ENE-WSW- to E-W-trending group 2 fault in the northern area of SNGF preserves both near vertical (pitch 84°) and subhorizontal (pitch 20°) slickenlines (Figures 8(e) and 8(g)). These are related to normal and dextral senses of motion, respectively, based on slickenline stepping relationships. The steep lineations overprint the subhorizontal slickenlines suggesting that the fault initiated as a dextral structure and was then reactivated with a dip-slip sense of motion.
More generally, a distinct group of NW-SE-oriented dip-slip faults are seen to consistently offset all other fractures. In many cases, these structures show evidence of overprinting slickenlines, and in all cases, they reactivate as dip-slip structures rather than strike-slip features. It is therefore proposed that the group 2 fractures, identified on textural and alteration assemblage grounds, show evidence for at least two deformation events, herein termed as stage 2a and stage 2b.
Figure 5 summarises the key kinematic features of faulting during each of the stages. Stage 1 includes all the group 1-classified fractures, which are mostly E-W-trending sinistral faults and NE-SW-oriented (rare NW-SE) tensile fractures. Approximately 70% of the group 2 fractures are interpreted to have formed during stage 2a. This is dominated by NW-SE-oriented normal and dextral faults and NW-SE tensile fractures, and some N-S-striking normal and tensile fractures (Figure 5). Lastly, some NW-SE and E-W faults show evidence of later dip-slip normal movements and are considered to have formed during stage 2b.
6. Palaeostress Analysis
Following the classification of the mapped fractures into two stages based on field and microscopic characteristics, a stress inversion analysis was conducted to evaluate the possible stress conditions at the time of their formation. The inversion relies on the assumption that the faults and the blocks of rock they bound have not rotated significantly since their formation. This is a reasonably safe assumption given that the observed and inferred displacements of the key SNGF faults are minimal, suggesting that finite strains are overall low. The stress inversion was carried out in the Windows-based application, MyFault version 1.05, using the Minimised Shear Stress Variation inversion method which assumes that the magnitude of the shear stress on the fault is similar for all the fault planes at the time of rupture [31–33] (see the Supplementary Material for details of the methods (available here)). Amongst the various methods of stress inversions considered, this yielded the minimum misfit angles and thus is considered to be the most appropriate for the SNGF dataset. A weighting scheme was further applied in the inversion based on the thickness and length of the mapped faults to give more significance to larger structures over smaller ones.
For stage 1, is calculated to be steeply plunging (66°/138°), whilst both and are horizontal to shallowly plunging, trending ENE and NNW-SSE, respectively (Table 2 and Figure 5). The calculated extensional direction, , is consistent with the poles to the majority of associated tensile fracture planes (Figure 5). With a shape factor of 0.31, the stress configuration suggests a strike-slip to transpressional tectonic setting.
During stage 2a, the implied tectonic setting has changed to a strongly extensional or normal-faulting regime. The calculated is vertical whilst the principal extension direction, , is E-W, which coincides with the poles to the mapped N-S tensile fracture planes (Figure 5).
During stage 2b, remained steeply plunging (55°/269°, Table 2) during a generally extensional or transtensional tectonic regime with a shape ratio of 0.71. The is now oriented NW-SE.
The orientations of the principal stresses in stage 2b appear to be similar to the present-day configuration but with a swapped and . The World Stress Map [15] reports, based on limited data for Negros, a modern WNW-ESE and less common ENE-WSW-oriented in the southwest of Negros and Tañon Strait. The focal mechanism of the 2012 earthquake generator as discussed in Aurelio et al. [26] suggests that is oriented NW-SE. GPS studies by Rangin et al. [34] likewise suggest a NW-oriented (315°) convergence direction, similar in direction to that reported by Kreemer et al. [35], which is parallel to the overall convergence direction estimated for the Philippine Sea Plate [13]. Finally, restricted borehole breakout data from SNGF wells indicate that is WNW-ESE. Overall, these present-day data suggest that is generally oriented NW-SE to WNW-ESE, consistent with the principal stress orientations determined for stage 2b, except that . Thus, the youngest fractures in SNGF have probably formed in a stress field where the principal stress axis orientations were similar to the present day, but and may have switched due to the perturbing effect of the volcanic edifice in the region of the SNGF as discussed below.
7. Slip and Dilation Tendency Analysis
An attempt to identify which of the mapped structures might be the most favorable drilling targets is now presented, using a slip and dilation tendency analysis. The rationale here is that when a structure has a higher tendency to slip under the present-day stress conditions, there is a greater chance of increased fracture density and enhanced permeability [36]. Higher capacity to transport fluids is also probable when a structure is more prone to dilate, since fault aperture is most likely to readily enlarge. These concepts have been effectively applied, for example, in assessing fault reactivation potential in deep enhanced geothermal systems in Germany [37] and in understanding anisotropic transmissivity of the groundwater in the Yucca Mountain in Nevada [36].
All the analyses were carried out using 3DStress® version 5 software developed by the Southwest Research Institute. Present-day principal stress directions applied were based on the regional GPS studies of Negros Island (i.e., [34, 38]). Stress magnitudes were estimated from the lithological overburden pressure (for the vertical stress, ), borehole leak-off tests (for the minimum horizontal stress, ), and the derived shape ratios from the palaeostress analysis (which allows calculation of the maximum horizontal stress, ).
Overall, slip tendency values are quite low (<0.20) suggesting that in general, the structures in the SNGF are not prone to slip under the proposed present-day stress conditions. This contrasts with the strongly dilational tendency with widespread values approaching 1.0. The mapped steeply dipping to vertical NW-SE faults have the highest slip and dilation tendency followed by moderately dipping structures of similar orientations. The most stable faults are those striking NNE-SSW to NE-SW. These are the structures whose poles lie around the maximum principal stress (Figure 9).
8. Discussion
The fieldwork reveals the presence of two characteristically different groups of structures (groups 1 and 2) which accommodate three movement stages (stages 1, 2a, and 2b) in the SNGF. These, correspond to three brittle deformation events, with the most recent (stage 2b) mostly being limited to reactivation of pre-existing structures rather than involving new fracture formation. Palaeostress inversion analysis suggests a transition between a strike-slip to an extensional to transtensional tectonic regime. Stage 1 likely occurred under a strike-slip to transpressive tectonic regime where is oriented NE-SW. This formed mainly WNW-ESE-trending sinistral faults and NE-SW tensile fractures. Based on the reported age of the Southern Negros Formation in which these fractures were observed exclusively, the stage 1 fracturing event occurred no earlier than the Pliocene. The main sinistral faults run subparallel to the trace of the Puhagan Fault Zone which may suggest that it formed during stage 1, but was initiated as a sinistral structure, contrary to its observed present-day kinematics.
The stress conditions then changed to a strongly extensional regime when most of the group 2 fractures formed. New structures formed including WNW-ESE- to NNW-SSE-trending normal, dextral, and oblique (normal/reverse) faults together with less common ENE-WSW-oriented normal faults. Associated NNW-SSE tensile fractures and smaller faults offset earlier stage 1 fractures. The calculated principal extension direction is E-W for stage 2a whilst principal compression is steep to vertical. Horizontal compression is probably minor given the nature of the suggested tectonic regime, but could conceivably be oriented N-S. There is a perceived counterclockwise rotation of the compression direction around the vertical from stage 1 to stage 2a, which may have persisted to the present day, with a likely transitional phase captured by the stage 2b deformation reactivating mainly pre-existing group 2 fractures formed during stage 2a. The overall tectonic setting during stage 2b is also dominated by extension, with the principal extension direction, , now being NE-SW.
It is important to note here that the SNGF stress inversions have high misfit angles which imply that although the apparent directions of rotation are clear, the amount of rotation is rather less well-constrained. In the following subsections, we discuss geological processes that could potentially explain the directions of the interpreted local stress rotations within SNGF.
The possibility of block rotations is partially constrained by a palaeomagnetic study by McCabe et al. [39] that involved sampling and measurements at 86 sites across the Philippine Archipelago. Two key rotation events were suggested: first in the Early to Middle Miocene where it was suggested that the islands of Panay, Cebu, and Mindanao rotated clockwise, whilst Marinduque rotated counterclockwise. This, it was suggested, was related to the collision of the northern Palawan Block with the Philippine Mobile Belt. Since Negros Island is bordered by Panay, Cebu, and Mindanao, it may have also rotated clockwise at this time, but no Early to Middle Miocene samples were collected from Negros to determine this. In the Late Miocene to Pliocene, McCabe et al. [39] suggested that the central and northern parts of Luzon rotated clockwise potentially related to the collision of the Luzon Arc with Taiwan. No rotation was observed in other parts of Philippines during this period. Since then, it is suggested that the entire Philippine Arc has behaved as a single unit with no discernible rotations based on the available palaeomagnetic data [39]. It is possible, however, that the magnitude of any rotation that occurred close to the present-day may have been too small for the study to capture.
Thus, these studies suggest that since the Pliocene, there is no strong palaeomagnetic evidence to suggest that significant regional-scale block rotations have occurred in the island of Negros. This seems consistent with the generally low displacements inferred along the major fault structures in the SNGF, which suggests that the regional finite strain is low, meaning that significant fault-induced block rotations related to these faults are unlikely. Thus, it seems most likely that apparent changes in stress orientations due to block rotation are unlikely to have occurred in the last 5 Myr since the last plate reorganization happened [11, 40, 41]. Further, the direction of the observed rotations is also not consistent even if a “domino-style” rotation is considered (i.e., clockwise rotation of stress axes due to the movement of two large sinistral faults). Therefore, it is most likely that the observed stress rotations in the SNGF area are related to a smaller-scale heterogeneity in the regional stress field. We now go on to consider two possible geological processes which might account for such a locally controlled stress perturbation.
8.1. Possible Influences of the Philippine Fault and the Propagation of the Yupisan Fault
The observed stress rotation may reflect a smaller-scale disturbance that is related to the local lateral propagation of displacement along the Yupisan Fault. The Yupisan Fault is a NNE-SSW sinistral-reverse fault traversing the eastern coast of Negros in the north and passes through to the west of the CDN volcano in the southern part of the island (Figure 1(b); [26]). Stress rotation related to the Yupisan Fault is explored by looking at the displacement vectors of the Yupisan Fault as it continues to slip during the time of its formation (proposed to be during the Late Pliocene) and how the displacement could affect the surrounding blocks, including the area where the CDN volcano is located (details in Pastoriza [27]). Coulomb® 3.3 was used, which is a MatLab-based calculation and visualisation program designed for the determination of static displacements, strains, and stresses at any depth caused by a fault slip, magmatic intrusion, or dike expansion/contraction [42] following the concepts in Toda et al. [43] and Lin and Stein [44].
For this analysis, the southern inland trace of the Yupisan Fault was divided into five segments based on the curvature of the lineament observed on satellite imagery. The strike azimuth for each fault segment was extracted from the digital elevation models whilst the dip angle is taken from the earthquake focal mechanism data of the February 2012 Negros earthquake and was assumed to be the same for all five segments. Using the USGS-calculated 6.7 earthquake magnitude along Yupisan Fault in 2012, the amount of total co-seismic slip along the fault is estimated to be 0.524 m (calculation after Wells and Coppersmith [45]). This net slip was then broken down into dip and strike-slip components for each fault segment based on the estimated rake on that fault segment using geometrical rules. The geometrical rake was approximated with the aid of stereographical projections using the compression axis direction proposed by Rangin et al. [34].
A propagating Yupisan Fault was then modelled using Coulomb® 3.3 for two scenarios—one where the fault propagates northwards and another where it propagates southwards. The key assumption here is that every time a new segment slips, the older segments slip with it. This basically confers a cumulative displacement for each segment. Additionally, it is assumed that each time the Yupisan Fault slips, an earthquake with the same magnitude of 6.7 is generated.
Figure 10 illustrates that the total amount of slip for each segment gradually increases as the Yupisan Fault propagates. In a northward propagating model, at the first onset of the structure Time 1, the southernmost segment slips 0.43 and 0.30 m along the strike and the dip, respectively (Figure 10). By Time 5, it has slipped a total of 2.15 m sinistrally and 1.50 m along its dip. In a northward propagating Yupisan Fault, this configuration induced a progressive clockwise block rotation on both sides of the fault (Figure 10). The directions are the other way around for the stress rotations which would be counterclockwise. These observations are opposite if a southward growing Yupisan Fault is considered [27]. From stage 2a to the present day, an apparent counterclockwise rotation of the stresses around the horizontal is observed (Figure 5). This observation therefore fits a northward-propagating Yupisan Fault model, where the footwall, in which the CDN and SNGF are situated, appears to have rotated progressively clockwise, as the fault continued to move.
The model suggests that after five slip increments along the Yupisan Fault (Time 5), the southern part of the footwall has rotated roughly 11° (Figure 10). This is close to the 9° rotation along the horizontal from stage 2b (144°) to the present day (135° in [34]) (Figure 5) suggesting a potentially good fit with the northward-propagating Yupisan Fault as a potential trigger of the observed rotation of the stresses within the SNGF. Further, considering that this modelling utilised the present-day compression direction of 315°/135°, the consistency of the degree of rotation with the stress inversion results suggest that the present-day stress conditions may have actually remained relatively constant throughout stage 2a and that the growing Yupisan Fault has triggered local block rotations within its immediate vicinity resulting in an apparent rotation of the local stress fields. Potentially, the observed change in tectonic regime between stage 1 and stage 2a could be due to changes in the stress magnitudes, leading to a “flipping” of the stress axes, which does not necessarily require a drastic change in the far-field stress orientations. This illustrates how the growth of large-scale regional deformation structures may potentially affect smaller-scale stress fields.
Although we have initially eliminated the possibility of block rotation in Negros Island given that no palaeomagnetic data can support it, the smaller-scale rotation proposed herein is possible at low finite strains. A local palaeomagnetic study could help to test and refine this model, provided that a minimal rotation of 11° can be captured.
8.2. Changes in Tectonic Regime and the Development of the CDN Volcanic Activity
Given the presence of an active subduction zone (Negros Trench) located to the southwest of the SNGF and a large reverse fault (Yupisan Fault) on its western flank (Figure 1(a)), a dominant horizontal compression direction might be expected for the study area. Although this is true for stage 1, it does not seem to be the case during stage 2a and stage 2b which are both predominantly extensional based on the results of the palaeostress inversions and the observed dominance of normal faults.
The spatial and temporal effects of volcanism on the stress fields in the summit region of a volcano have been explored by a number of authors, e.g., [46–48]. Being centrally located in a volcanic complex, the potential influence of the CDN volcanic activity on the stress fields over time and the style of fracturing within the SNGF should not be disregarded. The CDN volcano is characterised by several episodes of volcanism and intrusion marked by thick sequences of volcanic deposits seen in the subsurface, with three distinct volcanic centers of varying ages exposed at the surface. Terakawa et al. [47] have shown that volcanic activity can induce temporal stress changes in the summit regions of erupting volcanoes. Thus, during the 2014 eruption of Mount Ontake in Japan, focal mechanisms indicate that normal-faulting dominated pre-eruption whilst reverse faulting prevailed thereafter. Further, it was demonstrated that the average misfit angle of the focal mechanisms around and in the periphery of the edifice significantly increased prior to the eruption. An inflation under the volcano, which was driven by magmatic or hydrothermal fluids, was identified as the cause of the stress perturbation, particularly resulting in the rotation of the maximum and minimum principal stresses [47]. The results of this study and that of Vargas-Bracamontes and Neuberg [49], amongst others, clearly illustrate how magma pressures can locally perturb and even overpower the regional stress field in an area of active volcanism. This potentially results in local stress fields being different from the prevailing large-scale conditions.
The dominance of mostly normal faults in the group 2 fractures during both stages 2a and 2b may potentially be related to the inflation and gravity spreading of the CDN volcano and associated intrusive emplacements. Formation of grabens along the flanks and en echelon strike-slip faults, folds, and reverse structures at the base are typically observed on gravitational spreading volcanoes based on analogue modelling [50] and field observations [51]. Similar structures are also observed when the spreading is associated with magmatic intrusion. The abundance of extensional structures in the SNGF may be associated with this type of spreading behaviour. With continued dyke emplacements underneath the CDN, the volcano would continue to grow, requiring the surface to expand. Such expansion is most easily accommodated by fracturing. This is consistent with the observation that the southwest area of the geothermal field is most dissected by normal faults, which is the closest to the main edifice (highest elevation mapped) and where the Nasuji Pluton is laterally situated in the subsurface. Rae et al. [52] have illustrated that the intrusive events underneath the CDN have influenced the alteration type and the propagation of heat below the SNGF. It is perhaps not surprising then that during the several intrusion events, hydrofractures have formed and are later reactivated as shear or as tensile structures.
Thus, volcanism, spreading, intrusion, and the geothermal processes, are likely to have independently or altogether influenced the observed dominant normal-faulting regime within the SNGF since stage 2a.
9. Synthesis and Conclusion
Remote sensing studies and surface-based geological fieldwork conducted within the SNGF reveal that the region is dominated by brittle deformation. Age indicators and cross-cutting relationships suggest that at least two fracture-forming events occurred and have been locally reactivated under evolving stress fields. The changes are most likely driven by regional and local tectonic and volcanic processes within Southern Negros. Although limited by the age of the oldest rocks exposed at the surface, the deformation history proposed is constrained from the Middle Pliocene to the present day. This assumes that the reported age of Late Pliocene to Early Pleistocene of the Southern Negros Formation, which is purely based on field stratigraphic position, is correct. A regional structural evolution with time is illustrated in Figure 11(a) whilst a more detailed and localised view is shown in Figure 11(b).
(a)
(b)
By the Pliocene, the Philippine Fault had started to propagate in the eastern part of the Philippines and the present-day plate vectors were already in place. Southern Negros at this time was still partly submerged underwater and the Southern Negros Formation, which formed as a result of the volcanic activity of the palaeo-CDN, was deposited underwater in Puhagan (central part of the field) but subaerially in the west [28]. An emergent volcanic edifice in the western part of the present-day CDN may have existed at this time (Figure 11(a), i). An early phase of brittle deformation, stage 1, occurred, affecting the SNF and, presumably, older lithologies. Stress inversion analyses of the rather limited field data suggest a NE-SW-oriented horizontal maximum compression under a strike-slip or transpressional regime, which is also consistent with the equivalent-age structures observed in the country rocks to the northwest of SNGF in the Pamplona-Sta. Catalina area [27]. In southern Negros, a sinistral ENE-WSW-trending en echelon palaeo-Puhagan Fault is thought to have been present, running across the northern flanks of the palaeo-CDN edifice (Figure 11(b)). Associated with this major structure were a series of E-W sinistral faults and NE-SW tensile fractures cutting the early SNF and older rocks (Figure 11(b)). During this time, a smaller hydrothermal system may have existed which may be related to early intrusions (e.g., Puhagan dykes in [28]). The SNF rocks experienced intense preliminary alteration whilst fractures were channeling sulphide-rich fluids. A dominantly reducing environment, maybe because of less interaction with surface waters, prevailed, meaning that these fluids deposited widespread pyrite along the fractures. Cockade structures within the pyrite veins suggest episodic influxes of sulphide-rich fluids into open-fracture systems over significant timescales [30].
By the end of the Pliocene, it is suggested that the Yupisan Fault had started to propagate northwards from the SSW coast of Southern Negros towards the eastern coast of the island, under a stress regime similar to the present day, i.e., NW-SE horizontal compression ([34]; Figure 11(a), ii). The palaeo-CDN continued its activity, and deposition of SNF was maintained. In the Early to Middle Pleistocene, emplacement of the main Nasuji Pluton [18] and associated intrusions occurred, particularly in the western part of the SNGF. This intrusion triggered significant steam-heated argillic alteration in the western area of the geothermal field which is one of the earliest dated hydrothermal alteration episodes (0.7 Mya in Rae et al. [52] and 0.81 Mya in Takashima and Reyes [53]) in the SNGF. Leach and Bogie [54] concluded that this process may have introduced a barren porphyry-copper-type deposit, which they considered to be the relict alteration suite observed in the SNGF boreholes today. An overpressured magma chamber resulting from the intrusion may have induced hydrofracturing in the immediate vicinity of the pluton. This drastically increased the degree of fracturing in the western sector, which is exhumed today.
The Pleistocene marked the peak of the volcanic activity of the CDN complex which led to the deposition of lava flows and pyroclastics. It is suggested that a regional NW-SE compressional stress regime dominated in the SNGF and reactivated the palaeo-Puhagan Fault with dextral-strike slip kinematics (Figures 11(a) and 11(b)). Its kinematics is clearly observed today based on the modern geomorphology, offsetting the CDN complex and the Yupisan Fault right-laterally. This formed NW-SE dextral Riedel structures related to the Puhagan Fault in the northern part of SNGF. Locally, with a more active volcano formation, a localised extensional setting prevailed under stage 2a, with the dominant extension somewhat consistent with the regional extension direction (Figure 11(a), iii), where WNW-ESE to NNW-SSE normal, dextral, and oblique (normal/reverse) and rare ENE-WSW normal faults propagated. NNW-SSE-trending tensile fractures clearly offset the earlier tensile and sinistral structures. In the northern area of the field, the earlier hydrofractures were reactivated as shear fractures. The heat in the subsurface particularly shifted towards the central part of the SNGF (central Okoy) following the conduction-convection model of Takashima and Reyes [53]. With the now crystallised dykes and/or pluton, Rae et al. [52] suggested that the circulation became more diluted, being affected by meteoric water, and less influenced by hydrothermal sources. This led to a propylitic and illite-rich alteration, affecting the immediate country rock. Consequently, stage 2a fractures are in most cases host to thick clay alteration and iron-oxide formation.
During the Holocene, the volcanic activity in the CDN waned, thus the regional stress fields became more prevalent within the SNGF, but still under a local extensional/transtensional setting (Figure 11(a), iv and Figure 11(b)). The final fracturing transpired during stage 2b formed mostly NW-SE-oriented normal faults. Reactivation of the stage 2a fractures, mostly as dip-slip structures was also common. By this time, the heat was now centered in Puhagan as a result of either shifting due to convection-conduction [53] or because a younger intrusion event occurred underneath the area as proposed by Rae et al. [18]. Many of these young structures served as channels in the present-day hydrothermal system towards the surface, appearing as a result of the concentration of active thermal manifestations around Puhagan today. This suggests that a majority have remained open and active in the present-day stress configuration.
The results of the slip and dilation tendency analysis suggesting that NW-SE-oriented fractures should in theory be the best and most permeable drilling targets. This agrees with actual drilling results in SNGF [55]. However, the slip and dilation tendencies indicate that NE-SW and NNE-SSW-oriented structures should be the least permeable as they are likely presently stable; they are, however, also known to be important in the fluid flow regime of the SNGF based on actual boreholes [55]. These large-scale NE-SW faults could have been original tensile fractures formed during stage 1 and could have been reactivated during succeeding fracturing events. Such reactivation may have re-opened these structures at depth as tensile fractures in the predominantly extensional regime of stage 2a +/− a component of shearing. As the geothermal system was already developing during stages 2a and 2b, temperatures at the fracturing depths may well have been much higher, making rapid cementation/sealing of the fracture planes less likely. This reactivation model may explain why some larger NE-SW faults have remained opened until the present day and consequently are proven to be permeable channels of geothermal fluids in the SNGF [55].
This work is by far the most comprehensive geological analysis and the first microtectonic work on the SNGF surface geological data. Several structural field campaigns and studies have been done in the field over the last 30 years [56–62] as part of its exploration and development works. However, geological interpretations were limitedly based on orientation analysis (i.e., the orientation of the fractures with the highest frequency correlated to the orientation of the regional structure) with very minimal attempts to rank or prioritize structures with varying characteristics. There was a significant lack of the appreciation of the kinematics of mapped faults and their implication to potential subsurface permeability, and consequently the drilling strategy.
The work presented here addresses that gap by illustrating a more comprehensive approach to surface geological data analysis. From a thorough characterisation of the mapped faults (e.g., fracture fills and kinematics), their cross-cutting relationships, and a stress inversion, one can build the deformation history of the field. With a refined geological history and the information on the associated fluid alteration for each phase, a better understanding on which structures related to the geothermal development can be provided. Further, the utilization of the slip and dilation tendency analysis is a useful tool in the identification of the potential interaction of the mapped faults with the present-day stresses and how this may influence fluid flow. This could help to delineate structures which are currently stressed and thus may be potential drilling targets given the likely increased fracture intensity and widening of the fracture aperture.
This work illustrates how a strongly field-based geological approach can inform exploration and eventual development strategies for drilling, even at the early stages of the field exploration (i.e., even prior to drilling) and can be further enhanced by incorporating a multiscale fracture attribute and topology analysis as presented in Pastoriza [27]. This geological workflow should supplement existing workflows for building conceptual models at the exploration stage (e.g., Cumming [63]) and, when integrated with geochemical, geophysical, and hydrological data, be usefully employed for field-scale development of geothermal resources.
Data Availability
The data that support the findings of this study are available from the Durham University E-Theses Repository, but restrictions apply to the availability of these data, under a confidentiality agreement between the Energy Development Corporation and Durham University. Data are however available from the authors upon reasonable request and with permission to the Energy Development Corporation.
Disclosure
This work is part of the PhD thesis of the main author at Durham University.
Conflicts of Interest
The authors declare that there is no conflict of interest regarding the publication of this paper. Energy Development Corporation has reviewed and granted permission to publish this work.
Acknowledgments
Due acknowledgment is to the Energy Development Corporation for granting access to internal reports and well data; to Jonathan Kit Reyes and Ferdie Castillejo for the assistance in the field campaigns in Negros; to Mario Aurelio, Jon Gluyas, and Richard Walker for their contributions in the entire project, and to Nick Primaleon and the two anonymous reviewers for their constructive criticisms to improve this paper. This work is funded by the Energy Development Corporation.
Supplementary Materials
The supplementary material details the stress inversion methods applied in the paper. It highlights the assumptions considered and how the minimum shear stress variation method was selected amongst other methods available in the literature. (Supplementary Materials)