Extensional Tectonic Investigation based on Gravity and Magnetic Data Analysis in the Gulf of Tomini, Indonesia

The enigmatic genesis of Sulawesi Island remains a debate among earth scientists up to the present day. The Northern Arm of Sulawesi is designated as a volcanic province, while the Eastern Arm is identified as an ophiolite zone along with several micro continental fragments. However, the tectonic relationship in the marine area between these two arms is not clearly defined. The existence of isolated volcanic islands, namely the Una-Una and Togian islands increases the complexity of the area. A regional study utilizing gravity and magnetic methods is needed in the Gulf of Tomini, North Sulawesi, Indonesia. Therefore, a comprehensive analysis was conducted, involving 3D inverse modelling and Moho depth estimation based on the Parker-Oldenburg algorithm using 3D Euler deconvolution to obtain the average depth of causative bodies. As a result, the presence of extensional tectonics were revealed and identified as multiple parallel dextral strike-slip faults across the study area. All of these faults show northwest-southeast lineaments consistent with the existing seismic sections. The extensional tectonic mechanism triggers crustal thinning and rifting, leading to the build-up of graben structures such as the Gorontalo, Tomini, and Poso basins. Additionally, it influences volcanic activ - ity around the Togian and Una-Una islands. The major axes of the inferred opening area are oriented in the northeast-southwest direction, perpendicular to the lineaments of these faults. The opening area coincides with the orientation of Lalanga and Togian ridges. Finally, a regional geological structure is proposed across the Gulf of Tomini linking both the Northern and Eastern arms of Sulawesi.


Introduction
The Gulf of Tomini as a study area, located between the Northern Arm and the Eastern Arm of Sulawesi Island, Indonesia, is defined as a geologically complex region with abundant tectonic activities for both subduction-collision zones and an active extensional regime (Cottam et al., 2011;Pholbud et al., 2012;White et al., 2014;Sendjaja et al., 2020).These tectonic activities resulted in an island with a unique shape resembling the letter K (White et al., 2014).The island consists of several arms, namely the Northern Arm of Sulawesi defined as a volcanic province, Western Sulawesi consist-ing of a volcanic province and continental basement, Central Sulawesi recognized as a metamorphic-belt region, the Eastern Arm of Sulawesi defined as ophiolite, and other micro continental fragments (Eastern Arm and Buton), as shown in Figure 1.Several deep marine bays are located between the arms, such as Gorontalo Bay, Tomini Gulf, Tomori Bay, and Bone Bay (Pholbud et al., 2012;Watkinson et al., 2012;Nugraha and Hall, 2018).All of these arms and bays are sliced by several major strike-slip faults, such as the Palu-Koro Fault, the Lawanopo Fault, the Walanae Fault, the Kolaka Fault, the Gorontalo Fault, the Matano Fault, and the Balantak Fault (White et al., 2014).
The geological structures surrounding Tomini Bay are relatively well defined and have been the subject of the current tectonic settings across the Gulf of Tomini based on gravity-magnetic datasets and to establish correlations with the existing interpretations of geological structure on-land.As a result, a comprehensive synthesis of the geological structure interpretation over the study area and its surrounding will be provided.The study results will contribute to a deeper understanding of the tectonic settings of the region.Furthermore, it will serve as a valuable guide for both exploration of geological resources and the mitigation of geological disasters in the study area.

Geological Settings
Tectonically, Northern Sulawesi is formed due to the convergence of a triple junction involving several plates, including the Eurasian Continental Plate (ECP), the Celebes Sea Plate (CSP), the Australian Banda Arc Plate (ABP), and the Philippine Sea Plates (PSP).The interactions and collisions between these plates have played a significant role in the geological evolution of Sulawesi Island.The tectonic system of Sulawesi Island is characterized by the westward motion of the PSP, the southeastward movement of the ECP with the western region acting as a relatively steady reference point, the southward movement of the CSP, and the north-westward movement of the ABP.These plate interactions form a fascinating and enigmatic tectonic framework within Sulawesi Island (Walpersdorf et al., 1998a;1998b;Kopp et al., 1999;Kusnida, 2008;Martosuwito, 2012).The existence of isolated volcanic island in the study area, i.e.Una-Una and Togian islands increase the complexity of the tectonic system within the arc (Katili et al., 1963;Cottam et al., 2011;Sendjaja et al., 2020).Extensive research on the tectonic evolution of Sulawesi Island suggests that the region has passed through a complex tectonic deformation due to the interaction among plates.
Several research results show a relationship between the rapidly subsiding Gorontalo Basin, regional extensional tectonics, and rotation of the Northern Arm away from the Eastern Arm of Sulawesi in a clockwise direction (Cottam et al., 2011;Pholbud et al., 2012;Advokaat et al., 2017).All of these occurrences are happening at the present day and have been substantiated by time-lapse GPS measurements.Most earth scientists believe that the continuous northward slab rollback at the hinge of the North Sulawesi Trench resulted from the reversal of southward subduction of the Celebes Sea Plate (CSP), which has been occurring since the Pliocene (Advokaat et al., 2017).These conditions cause faults in both the Northern and Eastern Arms of Sulawesi, which predominantly have northwest-southeast lineaments, such as the Gorontalo Fault, the Teba Fault, the Harapan Fault, the Rompi Fault, the Sabo Fault, the Balantak Fault, etc. (Sidarto and Bachri, 2015), as shown in Figure 1.It is inferred that Una-Una Island (Colo Vol-cano) might be an expression of crustal thinning, while the Lalanga Ridge has been successfully identified based on bathymetric datasets and defined as a carbonate reef with several pinnacle reefs and limestones (Pholbud et al., 2012).
According to the plate tectonic reconstructions, during the early Cretaceous (120 Ma), Southwest Borneo (SWB) and East Java and West Sulawesi (EJWS) moved away from Australia and approached the margin of Sundaland or Eurasia (Hall, 2012; Baillie and Decker,

2022).
The EJWS converged with SWB and joined with Sundaland in the late Cretaceous (85 Ma).A subduction with the northwest dipping direction occurred and was marked by volcanic activity in West Sulawesi and Sumba at around 65 Ma.Around 45 Ma, the Celebes Sea Plate (CSP) and Philippine Sea Plate (PSP) were spreading in a back-arc setting, while the Ceno-Tethys subducted northwards from Sumatra to Halmahera.In the late Oligocene period (around 25 Ma), the Sula spur was on the verge of making contact with the Northern Arm of the Sulawesi volcanic arc as the beginning of a collision between Australia and Southeast Sundaland.In the middle Miocene (around 15 Ma), the seafloor spreading of the Sulu Sea occurred in a back-arc setting due to the rollback of Celebes Sea Plate (CSP) subduction.In the late Miocene (around 10 Ma), the rollback mechanism also occurred in the Banda embayment and formed the North Banda Sea due to the extension of the Sula spur.
During the same period, Andaman seafloor spreading was also occuring.During the Pliocene period (around 5 Ma), the South Banda Sea formed as a result of the rollback into the Banda embayment, leading to extension in the region.The subduction of the Molucca Sea Plate was almost complete during this time, and the Halmahera and Sangihe arcs were prepared for collision.Finally, the regional tectonic settings that formed Sulawesi Island are comparable to the current tectonic conditions, as shown in Figure 1.Sulawesi Island underwent significant shaping and deformation due to the intensive movement (subduction, collision, and obduction) of the Sula platform (Banggai-Sula Microcontinent).These processes resulted in a clockwise rotation of the Northern Arm of Sulawesi and the development of the North Sulawesi trench since the early Miocene up to the present time (Silver et al., 1983).In the early Pliocene, a collision occurred between Sulawesi and Australia (New Guinea Micro Plate), leading to the significant transformation of Sulawesi into an island.As a result of this collision, Sulawesi experienced severe deformation, causing its convex side to face towards the continent.Additionally, obduction of ophiolite occurred in the Eastern Arm of Sulawesi Island (Katili, 1978).
In accordance with the geological consensus regarding the tectonic settings of Sulawesi Island in the past decade, it has been established that the island is primarily a result of contractional, strike-slip tectonics, accompanied by subsequent rotation (Baillie and Decker, 2022).The tectonic evolution regime of Sulawesi Island can be divided into two main periods.Firstly, during the Mesozoic-Paleogene period, there was significant tectonic activity that played a role in the main blocks or platform forming the island.Secondly, during the Neogene period, which started from the middle Miocene to present time, there was movement and interaction among these blocks or platforms, which contributed to the forming of Sulawesi Island (Hinschberger et al., 2005; Baillie and Decker, 2022).Massive time-lapse GPS meas-urements have been conducted for more than a decade with various purposes (Puntodewo et al., 1994;Walpersdorf et al., 1998a;1998b;Rangin et al., 1999;Socquet et al., 2006).In addition to monitor the local movement of the Palu-Koro Fault, GPS measurements have also confirmed the validity of the "Salami Slicer" model.This model suggests that the collision in Sulawesi Island is primarily driven by terranes from the east as the product of continent dispersion, which are drifted by the seafloor spreading mechanism (Pigram and Panggabean, 1984;Baillie and Decker, 2022).Furthermore, the GPS data measurements have also identified a small number of crustal blocks that show rapid rotation, which are confined within the Palu-Koro and Lawanopo fault systems (Socquet et al., 2006;Baillie and Decker, 2022).The rotation mechanism is caused by the coupled system of strike-slip and thrust faults (Silver et al., 1983;Hamilton, 1979;Baillie and Decker, 2022).Travel Time Tomography plays an important role in determining the characteristics of the subduction slab beneath the Earth's surface.The west-east double subductions beneath the Molucca Sea Plate have been successfully imaged using this method (Rachman et al., 2022).

Methodology
The gravity dataset utilized in the study is obtained from the global satellite gravity mission TOPEX/PO-SEIDON.The data are collected in the form of Free Air Anomaly (FAA) and topography datasets, with a lateral resolution of approximately 1.5 km (Oceanography, 2023).Both the FAA and topography/bathymetry datasets are processed and combined to calculate the Complete Bouguer Anomaly (CBA) using the FA2BOUG software, which is implemented in a FORTRAN code (Fullea et al., 2008).To cover both sea and land gravity datasets, a density value of 1021.6 kg/m 3 is selected for the gravity data reduction.This density value is chosen based on the Parasnis method (Rao and Satyanarayana Murty, 1973).The Radially Averaged Power Spectrum (RAPS) plays a crucial role in the separation of regionalresidual gravity data.The calculation of the method can be performed using different techniques, such as the wavenumber filtering technique or the moving average technique with a specified window size.The Total Magnetic Intensity (TMI) data are converted to Reduction to Pole (RTP) using inclination and declination values based on the International Geomagnetic Reference Field (IGRF) provided by the National Oceanic and Atmospheric Administration (NOAA).In this study, the RTP transformation from the TMI data is performed using specific inclination and declination values.The inclination value used is -16.8352°, while the declination value used is 0.9001°.The magnetic data acquisition was conducted around 2010/2011 by the Marine Geological Institute (MGI), Geological Agency, Indonesia (Ministry of Energy and Mineral Resources-MEMR).In general, the research methodology (framework) involves optimizing the utilization of gravity and magnetic datasets through three main parts (see Figure 2).These parts include Mohorovičić discontinuity (Moho-depth) estimation based on the gravity dataset, conducting a single 3D inversion for both the gravity and magnetic data to visualize the distribution of density contrast and susceptibility below the surface, and performing lineament analysis of anomalies in the study area using gravity and magnetic datasets.The Moho depth estimation is calculated based on gravity anomalies, which are obtained by applying a low pass filter to the Complete Bouguer Anomaly (CBA) datasets.This filtering process is useful to obtain a long-wavelength gravity anomaly.The calculation of the Moho depth requires an initial depth interface, which is refined iteratively through the calculation using the Parker-Oldenburg algorithm (Gómez-Ortiz and Agarwal, 2005; Chen et al., 2021) or based on 3D gravity inversion using GROWTH 2.0 (Ashena et al., 2018).The initial depth interface use the average depth value of the causative body, which is derived from the spectral analysis or the 3D Euler deconvolution (Reid et al., 1990;Beiki, 2010;Golshadi et al., 2016;Parang et al., 2016) instead of using shear wave tomography (Ardestani and Mousavi, 2023).The spatial analysis of the Moho depth distribution is integrated with the findings from lineament analysis and 3D inversion to provide a comprehensive explanation of the tectonic settings within the study area.The lineament analysis is performed by comparing the residual gravity anomaly with the RTP magnetic anomaly.In this case, the residual gravity anomaly is obtained from the regional-residual data separation, while the RTP magnetic anomaly is a conversion result from the TMI using specified inclination and declination values.The identification of lineament anomaly patterns was conducted based on forward modelling using both gravity and magnetic datasets (Handyarso and Saleh, 2017;Handyarso, 2022Handyarso, , 2023)).The 3D gravity inversion is performed using low pass filtered CBA dataset instead of the result from regional-residual data separation, while the 3D magnetic inversion utilizes the RTP anomaly in the study area.These results are subsequently cross-validated and integrated to provide a comprehensive description of the subsurface conditions.Seismic data are used to confirm the results, ensuring the accuracy and reliability of the interpretation results.Lastly, the structural interpretation is superimposed onto the Digital Elevation Model (DEM) in a 3D perspective to visually verify the findings as an additional approach to confirm the results.

Frequency Filtering
Frequency filtering based on the Fourier Transform plays an important role in the processing of potential field data.The aims of these filtering stages are to remove unwanted noise and to enhance the desired signals.In this study, the filtering technique was performed based on the wavelength (λ) filtering approach.The long wavelength cut-off used for this purpose is within the range of 50-100 km, serving as a Low-Pass Filter (LPF) to represent the Mohorovičić discontinuity interface.The long wavelength trend contained in the data will correspond to any known large structures close to the surface, which are typically associated with the Mohorovičić discontinuity (Lefort and Agarwal, 2000).The Fourier Transform, which converts data from the spatial domain into frequency domain can be mathematically expressed using Equation 1.
(1) where: The spectral analysis formulation of the potential field for both frequency-based and wavelength-based filtering can be calculated using Equation 2. This equation known as the Radially Averaged Power Spectral (RAPS), is expressed in the form of the Fourier Transform of the potential field under the assumption of a horizontal plane (Blakely, 1995; Parlak and Elmas, 2023).
(2) where: F [g z ] -Fourier Transform of gridded gravity anomaly, G -Universal gravity constant, Δρ -Density contrast between two layers, z 0 -Height of the observation points, z -Depth of the causative body.Equation 2 can be simplified to Equation 3. Additionally, Equation 4 expresses Equation 3 in terms of the natural logarithm on both sides, resulting in a linear relationship between the wavenumber and spectrum amplitude.
(3) (4) where: The slope of the linear relationship between the wavenumber and spectrum amplitude in the Radially Averaged Power Spectral (RAPS) analysis provides an estimation of the average depth of the causative body.Hence, the average depth of the anomaly source can be estimated using RAPS approximation.

3D Euler Deconvolution
The 3D Euler deconvolution method is widely used in potential field data analysis, particularly in gravity and magnetic data interpretations.It provides an estimation of the location and depth of the causative bodies (Reid et al., 1990;Blakely, 1995;Reid, 1998;Golshadi et al., 2016).The 3D Euler deconvolution concisely can be expressed using Equation 5.
(5) where: x, y and z -Coordinates of the source location, x 0 , y 0 and z 0 -Coordinates of the observation points, B -Base level or Background value, N -Structural Index (SI) or degree of the homogeneity, g -Gridded potential field measured at observation points, which refers to a gravity or magnetic anomaly.The structural index measures the decay rate of the potential field from the source as the distance changes (Golshadi et al., 2016).This feature can be utilized to distinguish sources from different geometries or shapes.Every shape has a distinct level of "sharpness" relative to its depth, resulting in a unique attenuation rate (Thompson, 1982;Golshadi et al., 2016).The type of contact anomaly is assumed to have a structural index (SI) value between 0.0 up to 0.5 (Thompson, 1982;Yudistira and Grandis, 1998;Golshadi et al., 2016).The correct assignment of structural index (SI) parameter is significant to the successful implementation of the 3D Euler deconvolution method.As a result, the 3D Euler deconvolution solution provides both lateral locations and depth information.

Parker-Oldenburg Algorithm
The Parker-Oldenburg algorithm is widely used to estimate the depth interface of causative bodies based on gravity datasets (Gómez-Ortiz and Agarwal, 2005; Kunnummal et al., 2018).The mathematical representation of the algorithm is expressed using Equation 6. (6) where: F [] -Fourier transform 2D discrete, G -Universal gravitational constant, ρ -Density contrast across the interface, Δg(x, y) -Gridded gravity anomaly, h(x, y) -Depth to interface (positive downwards), z -Average (mean) depth of horizontal interface, -Absolute value of the wave number.The equation above allows us to estimate the depth of the Mohorovičić discontinuity interface beneath the surface by means of an iterative calculation process.In the process, it is assumed that the average (mean) depth z as the initial horizontal interface is derived from the 3D Euler deconvolution result.As a comparison, the average depth in this study is also calculated based on the Radially Averaged Power Spectral (RAPS) of the gravity anomaly, which invoked the Fourier Transform (Blakely, 1995; Parlak and Elmas, 2023).

Gravity and Magnetic Inversion
A forward problem is a synthetic data calculation that arises from the distribution of a particular set of known physical properties, while the inversion is an attempt to recover the distribution of the physical properties based on the given datasets (Oldenburg and Li, 2005).Generally, a forward modelling operator produces the model response of the known model parameter distribution with some noises contained in the data expressed in Equation 7. (7)   In the case of gravity and magnetic methods, the forward modelling operators are based on the rectangular prism equation proposed by Plouff for the gravity method and Bhattacharyya for the magnetic method (Blakely, 1995).The model parameters or physical properties in our study are correlated with the density contrast (ρ) and susceptibility (χ) distribution beneath the surface.Inverse modelling is also known as the fitting of the model function F[m] from a vector of m as the physical property to a set n number of data points (Grandis, 2009;Ardestani and Mousavi, 2023).The solution is approximated by using weighted least square errors between the data points with the predicted dataset.The Gauss-Newton scheme is invoked for minimizing the objective function expressed in Equation 8, as described in previous studies (Grandis, 2009;Ghalehnoee et al., 2017).(8) where: -Kernel matrix, -Model weighting matrix, which consists of depth weighting and compactness weighting, -Noise weighting matrix, µ -Damping factor, d -Data points vector.

Lineament analysis
The forward modelling of both the gravity and magnetic anomalies are using a rectangular prism approximation to simulate the presence of a strike-slip fault anomaly.These synthetic data modelling will provide guidance in identifying any offset that is captured by both the gravity and magnetic datasets, as shown in Figure 3 (Handyarso and Saleh, 2017;Handyarso, 2022Handyarso, , 2023)).The subsurface model is designed in such a way that there is a two-thirds offset within the causative body.This offset corresponds to a net-slip occurring approximately 70% of the width of the causative body.A comprehensive summary of the physical properties of the model, including its dimensions, density contrast, and susceptibility value is provided in Table 1.The model responses of both gravity and magnetic methods are calculated using a gravity equation for a rectangular prism proposed by Plouff and a magnetic equation for a rectangular prism proposed by Bhattacharyya respectively (Blakely, 1995).The resulting potential field represents the measured value at the surface.In the presence of offset within the causative bodies, this offset will be reflect-ed as an anomaly pattern at the surface.Hence, the delineation of such patterns allows for the identification of strike-slip faults in the area.
The subsurface model consists of two blocky prisms that serve as sources of the anomaly bodies beneath the surface.Every single prism possesses identical physical properties such as density contrast or susceptibility.These two prisms are arranged in such a way to simulate the presence of a dextral strike-slip fault.The model is created using the real-world scale and is georeferenced under the UTM coordinates system.According to Table 1, the dextral strike-slip fault is located at Northing Y = 9197381.410m with lateral movement in the west-east direction.The width of the causative bodies can be calculated using prism #1 as follows X1end -X1start = 254679.200-241572.080= 13107.120m.Similarly, if we calculate using prim #2, it becomes X2end -X2start = 263753.360-250646.240= 13107.120m.The netslip is calculated using the start points of each prism as X2start -X1start = 250646.240-241572.080= 9074.160m.Similarly, if we calculate using the end points of each prism, it becomes X2end -X1end = 263753.360-254679.200= 9074.160m.In this case, the ratio of the net-slip to the width of the causative body is approximately 69.23%.In other words, the presence of net-slip amount in the synthetic modelling is approximately two-thirds of the width of the causative body.Figure 3 showcases the visualization of the synthetic modelling, revealing a dextral strike-slip fault characterized by west-east lateral movement and north-south major axes, as specified in Table 1.

Results and Interpretations
The data analysis started with the examination of the Complete Bouguer Anomaly (CBA) and the Total Magnetic Intensity (TMI) anomaly, as shown in Hence, it is necessary to perform a prior correction on the TMI dipole anomaly to convert it into a monopole anomaly known as RTP before any interpretation can be made.
Structural interpretation can be conducted using both the residual gravity and the RTP magnetic anomaly dataset, as shown in Figure 5.The high residual gravity anomaly is predominantly located around the land area, while the low residual gravity anomaly is more prevalent in the marine water area.The highest RTP magnetic anomaly is concentrated in the northeastern part of the study area, while the low RTP magnetic anomaly is relatively located around the marine water area.Multiple parallel dextral strike-slip faults with a northwest-southeast trend direction are identified in the study area, widespread across the Gulf of Tomini.The general behavior of a strike-slip fault is characterized by the presence of offset or net-slip, which can be identified by analyzing the synthetic model responses obtained from both gravity and magnetic methods.The similar anomaly characteristic is also successfully identified from the field dataset of these two methods.There are several offsets identified in both the residual gravity anomaly and RTP magnetic anomaly, which are aligned in a northwestsoutheast direction and spread across the study area, specifically the Gulf of Tomini, as shown in Figure 5.All of these offsets are interpreted as dextral strike-slip faults.
According to the residual gravity and the RTP anomaly maps, the identified multiple parallel dextral strikeslip faults from west to east (from left to right) consist of the Palu-Koro Fault zone, the Tambarana Fault zone, the Ampana Fault zone, the Buol Fault zone, and the Gorontalo Fault zone respectively.These fault zones show a relatively similar trend.The Palu-Koro Fault zone is characterized as a sinistral strike-slip fault, while the other fault zones are identified as dextral strike-slip faults.These multiple parallel dextral strike-slip faults with a northwest-southeast trend are driven by the extensional tectonics mechanism due to the clockwise rota- The multiple parallel dextral strike-slip faults have triggered the rifting or opening in this area.The lateral movement of the rifting is relatively in the same direction as the multiple parallel dextral strike-slip faults, which is a northwest-southeast direction.However, the major axes of the opening area of the graben align in a northeast-southwest trend direction, which is relatively perpendicular to the direction of dextral strike-slip faults.The extensional tectonics mechanism has caused rifting or opening in the Gulf of Tomini, which is associated with the existence of the Togian and Lalanga ridges.These ridges coincide with high residual gravity anomaly but low RTP anomaly, as shown in Figure 5.
Further analysis was performed to enhance the current analysis of the gravity and magnetic datasets.An estimation of the average depth of the causative bodies was determined by utilizing the low-pass filtered CBA gravity data.The low-pass filtered CBA gravity data provide information about long wavelength anomalies, which represent the deeper portions of the causative bodies.Thus, the average depth obtained from the lowpass filtered datasets will provide an estimation of the deeper anomaly sources such as the Mohorovičić discontinuity depth.The average depth is derived using both spectral analysis and 3D Euler deconvolution, which represents the depth solutions of the point masses of the causative bodies.Two different filtering techniques were employed in the study.Regional-residual filtering was used to separate gravity data for the purpose of identifying structural lineaments, while low-pass filtering was utilized to detect deeper anomaly sources.The low-pass filtering is used to generate a long-wavelength gravity dataset, which is assumed to represent the Mohorovičić discontinuity (Moho) undulation interface.In general, extensional tectonics cause the underlying mantle to uplift, decompress, and melt, resulting in shallow Moho depth.Additionally, the marine environment contributes to the rapid cooling of the rising magma and reduces its ability to penetrate further.The inferred rifting areas show major axes of the opening orientation in a northeast-southwest direction, coinciding with the presence of the Togian and Lalanga ridges.These two areas show high density anomalies but have a low magnetic susceptibility (see Figure 5), indicating that the rifting zones have been demagnetized due to the high temperatures associated with the uplift of the underlying mantle to the surface.The results show a good agreement with the fact that the Mohorovičić discontinuity becomes shallower around the rifting areas, as shown in Figure 6.
The subsurface modelling is obtained through a single 3D inversion process.This inversion is performed separately for the gravity and magnetic datasets.The gravity inverse modelling was conducted using low-pass filtered CBA gravity data, while the magnetic inverse modelling invoked the Reduction to Pole (RTP) magnetic data.The inversion results show an agreement with the current interpretations of the extensional tectonics in the study area, as shown in Figure 7.The red coloured zones highlight areas associated with a high-density body, while the purple-coloured zones indicate regions related to a high-susceptibility body.There are multiple parallel strike-slip faults spread across the study area.These faults can be identified by observing the lateral displacement or offset within the subsurface model, specifically in relation to both the density and susceptibility.The Mohorovičić discontinuity depth and the interpretation results are displayed together in a single frame or plot.The shallow Moho depth is superimposed on the high-density zones and Togian-Lalanga ridges, showing a northeast-southwest direction.Additionally, it is interpreted as the main axes of the rifting zones within the study area.Furthermore, this study incorporates the existing interpretations of geological structures on-land   The fault names identified in the marine area follow the existing fault names that have been defined on-land according to previous studies.The findings of this study are proposed to synthesize the overall tectonic settings in the northern part of Sulawesi, as shown in Figure 8.A previous study reported that in the early Miocene (around 23 Ma), the Sula spur began to collide with the Northern Arm of the Sulawesi volcanic arc and pushed north-westward.This collision eventually resulted in the welding of the Northern Arm and the Eastern part of Sulawesi into the West Sulawesi in the late Pliocene (2 Ma).As a result, the tectonic situation closely resembles the recent tectonic settings (see Figure 1).The Banggai-Sula moving northwestward collide, deformed, and lead to the following outcomes, such as the obduction of ophiolites in the Eastern Arm of Sulawesi and the clockwise rotation of the Northern Arm of Sulawesi with a pivot point at its eastern end (marked with a red circle in Figure 8).This clockwise rotation mechanism triggers the multiple parallel dextral strike-slip faults with a northwest-southeast direction and controls the extensional tectonics regime in the study area.The trend of these multiple parallel dextral strike-slip faults is relatively northwestsoutheast direction, which is consistent with the kinematic plate movement vector with respect to the Sundaland (Rangin et al., 1999;Baillie and Decker, 2022).
Extensional tectonics causes rifting or opening in several areas and results in the upwelling, decompressing, and melting of the underlying mantle.The seawater environment potentially influences the rapid cooling of the upwelling mantle.Thus, the rising mantle manifests as the shallow Moho depth interface and causes uplift in several locations.The uplift occurs in the Togian and Lalanga ridges, which are inferred as the main axes of the rifting zones with a northeast-southwest trend direction.Additionally, the seabed mapping reveals a bathymetric ridge for both Lalanga and Togian islands (Pholbud et al., 2012).
A DEM is visualized in 3D perspective to confirm any topographic evidence of faulting system identified in the study area.Fault can be revealed by using topographic relief analysis (Florinsky, 1996).According to Figure 8, this study proposed the presence of multiple dextral strike-slip faults extending across the Gulf of Tomini.These faults are identified as connecting or linking the Northern Arm and the Eastern Arm of Sulawesi.The multiple parallel dextral strike-slip faults deduced from the analysis of gravity and magnetic datasets in the Gulf of Tomini are interpreted as extension faults (represented by red lines).These faults are thought to have originated from the existing interpretation of geological structures on-land (represented by blue lines) and the known regional structures (represented by white lines).A superimposed visualization of the DEM with the geological structures demonstrates notable consistency with the topographic relief observed in the area, as shown in Figure 9.The superimposed visualization includes the existing fault structures (the white and blue lines) and the interpretation results of this study (the red lines).

Discussion
The research has yielded several significant findings that need to be emphasized.One of these findings is the identification of extensional tectonics through the analysis of gravity and magnetic datasets.Furthermore, the interpretation is validated or confirmed by the existing seismic sections.The seismic sections referred to as "LineDeep" (Rudyawan et al., 2014) and "Line_30" (Kusnida, 2008;Kusnida & et al., 2012;Subarsyah & Sahudin, 2016) are shown in Figure 10 respectively.The positions and orientations of the line locations are displayed on the map in Figure 8.These two figures serve different purposes but are generally used as constraints in the interpretation of the results of our current work.The first seismic section, "LineDeep" is a long and deep section with a west-east direction, revealing several normal faults that correspond to multiple parallel dextral strike-slip faults in the northwest-southeast trend direction.The red lines in Figure 10 (top) indicate these dextral strike-slip faults.The dot sign denotes the side moving forward to (or approaching) the reader of the paper, while the cross sign indicates movement backward to leave away from the reader.This seismic line is interpreted into various rock units, but they are not discussed in this paper.Detailed information is suggested to be referred to the original work directly (Rudyawan et al., 2014).Another seismic section, "Line_30" is located at the Tomini Bay, crossing the Lalanga Ridge and has a northwest-southeast line direction.This seismic section exposes the extensional tectonic settings, which generate a graben structure with indications of the mud-volcanoes / mud-diapirs located in the middle of the rifting zones around the study area.The mud-volcanoes or mud-diapirs are identified as the chaotic seismic section, characterized by a taper structural pattern observed in the seismic section, as shown in Figure 10 (bottom).Previously, it was interpreted as pinnacle reefs (Pholbud et al., 2012).
The multiple parallel dextral strike-slip faults identified in the study area are caused by the clockwise rotation of the Northern Arm of Sulawesi, where its eastern end behaves as the pivot-point and the rotation is confined by the Palu-Koro and Lawanopo fault systems.These types of tectonic movements trigger extensional tectonics and cause rifting or opening in the area.As a result, the underlying mantle rises and manifests as the Togian and Lalanga ridges.The existence of a normal fault observed in the seismic section "Line_30", which is located crossing the Lalanga Ridge, as shown in Figure 10 (bottom), serves as evidence supporting the occurrence of extensional tectonics.The graben structure is found in the southern portion of the line, while the Lalanga Ridge is situated in the northern portion.The weak zone of the rifting area is located relatively in the northern side of the mentioned normal fault.This weak zone has experienced intrusion by the activity of mud-volcanoes (mud-diapirs), resulting in the development of the Lalanga Ridge.By using these two seismic sections, it was demonstrated that the interpretation results of the current work presented in the paper are well matched with the existing seismic section as our constraints.
The study area is inferred as a transitional marine rift based on the inference that the Moho depth interface is approximately 20 km, which is thicker than oceanic crust but thinner than typical continental crust.Typically, the Moho depth in the oceanic crust is less than or equal to 10 km, whereas in the continental crust area, it commonly ranges between 20-70 km.The Moho depth of about 20-30 km is often associated with the transitional zone.In a transitional marine rift, there are several factors that can trigger the occurrence of mud-diapirs or mud-volcanoes.The process of crustal extension or crustal thinning in a rift zone results in the stretching and thinning of the Earth's crust.This leads to the development of grabens or half-grabens and may trigger the upwelling of the underlying asthenosphere.The graben or half-graben behaves as a depositional area, facilitating the accumulation of fine-grained mud.During a long period of sedimentation, a thick and water-saturated sedimentary sequence can be formed, providing a potential source for mud mobilization and upward migration.The composition and properties of the subsurface geology, including the presence of clay-rich or fine-grained sediments can influence the formation of mud-volcanoes and mud-diapirs.Certain lithological and stratigraphic conditions can promote the accumulation and mobilization of mud.The extensional tectonics can yield faults that serve as pathways for the ascent of fluids, including mud from deeper sedimentary layers.Fault zones may act as conduits for the ascent of mud, allowing mud to migrate towards the surface and leading to the formation as mudvolcanoes or mud-diapirs.The upwelling of the underlying mantle results in a higher gravity anomaly compared to its surroundings.In some rift areas, hydrothermal activity may be present due to the interaction between hot fluids and the subsurface geology.Hydrothermal fluids can mobilize and transport mud, contributing to the formation of mud-volcanoes or mud-diapirs.The existence of submarine hydrothermal activity is also confirmed by a previous study on geochemistry, which successfully identified a manganese-rich sediment layer in the Gulf of Tomini (Yuliyanti et al., 2022).
The extensional tectonic is still ongoing up to the present time, where the Palu-Koro and Lawanopo fault zone serve as the boundary for the rotational tectonic movements (Socquet et al., 2006;Baillie and Decker, 2022).The monitoring of the Palu-Koro Fault using GPS and earthquake slip vector data shows that the southern (or western) area of the Palu-Koro Fault shows small relative motion (< 2 mm/year) with respect to Sundaland.However, in the northern (or eastern) area of the fault system, the relative motion varies from 11 to 42 mm/year (Socquet et al., 2006;Baillie and Decker, 2022).The Banggai-Sula Microcontinent moves northwestward and collides with Sulawesi Island, resulting in the obduction of ophiolites over the Eastern Arm of Sulawesi.However, these processes initiated with a short period of subduction, leading to the belief that remnants of the Banggai-Sula Microcontinent still exist beneath the Eastern Arm of Sulawesi (Walpersdorf et al., 1998a;Darman, 2011;Pholbud et al., 2012).Hypothetically, the North Banda Sea Basin also moves in a north-westward direction and exerts pressure on the Palu-Koro-Lawanopo faulting system.Hence, the extensional tectonic settings in the Gulf of Tomini are primarily controlled by the Palu-Koro-Lawanopo faulting system, which is influenced by the north-westward movement of the North Banda Sea Basin Plate.In contrast, the Banggai-Sula Microcontinent is postulated to show a slight contrasting behaviour, potentially causing shortening tectonics instead of extensional tectonics in the area.The Banggai-Sula Microcontinent has the potential to exert pressure on the Matano Fault zone, indirectly influencing the Palu-Koro faulting system, thereby contributing to the extensional tectonics observed in the Gulf of Tomini.The occurrence of shortening tectonics can suppress volcanic activity in the area, whereas extensional tectonics can trigger new volcanic activity.Both processes are presence in the Gulf of Tomini.The presence of multiple parallel dextral strike-slip faults serves as evidence of the segmentation of the Northern Arm of Sulawesi, resulting from its clockwise rotation.A further study is needed to compare the kinematic movement based on time-lapse GPS measurements of both the Palu-Koro-Lawanopo faulting system and the Banggai-Sula Microcontinent with reference to the southern (or western) area of the Palu-Koro Fault, which represents the Eurasian Continental Plate (ECP).

Conclusions
The study area, encompassing the Gulf of Tomini is inferred as a transitional marine rift.This inference is supported by the observed variation in the Moho depth interface, which ranges approximately 12-28 km beneath the surface.Additionally, the presence of extensional tectonics has been successfully identified through the occurrence of multiple parallel dextral strike-slip faults that show a northwest-southeast trend direction spread across the area.These fault lineaments are recognized based on the synthetic data modelling of both gravity and magnetic anomaly responses.The interpretation results obtained from the analysis of gravity and magnetic show a good agreement with the existing seismic sections used as constraints in the study area.The multiple parallel dextral strike-slip faults are controlled by the clockwise rotational tectonics of the Northern Arm of Sulawesi, with its Eastern end acting as the pivot point.These tectonic movements trigger the extensional tectonic mechanism, resulting in crustal thinning around the study area.This process leads to several outcomes, including the development of graben and fault structures known as the Gorontalo, Tomini, and Poso basins.Additionally, it influences the presence of mud-volcanoes or mud-diapirs (and possibly volcanic activity) around the Lalanga and Togian ridges, including Una-Una Island (Colo Volcano).The rising, decompressing, and melting of the underlying mantle is reflected in shallow Moho depths, which also coincide with the Lalanga and Togian ridges.The graben structures serve as depositional areas and facilitate the accumulation of finegrained mud as water-saturated sedimentary sequences.Meanwhile, faults act as pathways for the upward migration of the ascent of fluids (mud) to the surface, forming mud-diapirs or mud-volcanoes.The Lalanga and Togian ridges are inferred to represent the opening area with the major axes showing a northeast-southwest direction that is relatively perpendicular to the trend of the multiple dextral strike-slip faults.This research has the capability to accurately determine the regional tectonic settings and makes a significant contribution in guiding exploration activities effectively.One particular benefit of the research is the identification and exposure of existing mud-diapirs within the study area, which is considered highly advantageous.This knowledge has the potential to reduce exploration risks, especially in oil and gas exploration because distinguishing between reef-building and mud-diapirs can be challenging.Additionally, a comprehensive understanding of the volcanic environment would significantly enhance mineral ore exploration in the future, particularly considering that the majority of the Indonesian region is comprised of volcanic areas.

Figure 1 :
Figure 1: Main tectonic features of the Gulf of Tomini are marked with the black and orange lines (from Pholbud et al., 2012), overlaid with the existing onshore geological structures of the Northern Arm and the Eastern Arm of Sulawesi-MEMR, which are marked with the red lines (from Sidarto and Bachri, 2015).Northern Sulawesi's subdivisions are color-coded on the map with explanations in the legend.Inset map shows the regional geological structures with several major plates and trenches (from Hinschberger et al., 2005).Several major plates are marked with CAR (Caroline Sea Plate), PSP (Philippine Sea Plate), EUR (Eurasian Continental Plate), AUS (Australian Continental Plate), and NBSB-SBSB (North/South Banda Sea Basin).The study area is outlined by a blue rectangle and covered the marine area known as the Gulf of Tomini.The Lalanga and Togian ridges inferred from bathymetric data are shown as orange double lines.
The data Copyright held(s) by author(s), publishing rights belongs to publisher, pp.1-18, DOI: 10.17794/rgn.2023.4.1 was subsequently compiled and incorporated into the Magnetic Anomaly Map of East and Southeast Asia Revised Version (3 rd Edition) in the year of 2021 by Coordinating Committee for Geoscience Programmes (CCOP) in East and Southeast Asia and Geological Survey of Japan (GSJ), National Institute of Advanced Industrial Science and Technology (CCOP and GSJ, 2021).

Figure 2 :
Figure 2: Workflow implemented within this research

Figure 3 :
Figure 3: Subsurface model of a dextral strike-slip fault and corresponding model responses calculated based on the rectangular prisms equation proposed by Plouff for gravity forward modelling and Bhattacharyya for magnetic forward modelling (Blakely, 1995).The top left image shows the synthetic model designed to simulate the existence of a dextral strike-slip fault.The top right image shows the gravity data response to the synthetic model.The bottom figures show the magnetic data responses of the synthetic model in terms of Total Magnetic Intensity or TMI (bottom left) and Reduction to Pole or RTP (bottom right).
Figure 4.The CBA contains a wide range of values, spanning from -152.7 mGal to 291.2 mGal, reflecting its coverage of both land and marine areas.The CBA is characterized by a combination of low and high frequencies.The study area includes the Gulf of Tomini, which is surrounded by the Northern Arm and the Eastern Arm of Sulawesi.These arms form the northern and southern boundaries of the study area.The low CBA values are located around Gorontalo, Tomini, and Poso basins, while the high CBA values are found in the central part of the marine area.The high CBA values coincide with the existence of Togian and Una-Una islands, as well as the Lalanga Ridge, which is situated between Poso Basin and the Tomini Basin, as shown in Figure 4 (left).The magnetic anomaly (TMI) is not associated with the basinal structures.The TMI anomaly ranges from -382.3 nT to 196.8 nT, with the high TMI anomaly concentrated in the center part of the study area.The strongest magnetic dipole pairs of the TMI anomaly are located at the Northern Arm of Sulawesi, as shown in Figure 4 (right).The Copyright held(s) by author(s), publishing rights belongs to publisher, pp.1-18, DOI: 10.17794/rgn.2023.4.1 TMI anomaly characterizes a dipole pattern, while the CBA anomaly has a monopole pattern.As we can observe from the forward modelling case presented in Figure 3, despite our knowledge that the synthetic subsurface model represents a dextral strike-slip fault (see Figure 3 top-left), there are no evident indications of this fault in the TMI anomaly (see Figure 3 bottom-left).

Figure 4 :Figure 5 :
Figure 4: Gravity anomaly map (Complete Bouguer Anomaly-CBA) on the left, and magnetic anomaly map (Total Magnetic Intensity-TMI) on the right, superimposed with the coastline of the Gulf of Tomini, Northern Sulawesi, Indonesia

Figure 6 :
Figure 6: Power spectrum of the average depth curves plotted with a 3-points moving average (represented by a blue line) and 5-points moving average (represented by a red line) respectively.The power spectrum result is compared with the average depth of the 3D Euler deconvolution solutions under the UTM 51S zone projection in the table (top figure).The 3D Euler deconvolution depth solutions are overlaid with the low-pass filtered gravity anomaly (bottom left), while the Moho depth estimation result based on the Parker-Oldenburg algorithm is superimposed with the interpretation results (bottom right).

Figure 7 :
Figure 7: A 3D Perspective view of high-density contrast (ρ) bodies (coloured in red) and high susceptibility (χ) bodies (coloured in purple), superimposed on the undulation of the Mohorovičić discontinuity beneath the surface including structural interpretations in the area.The black lines represent the coastline in the study area, while the white lines represent interpretation of the geological structures as the result of the study.

Figure 8 :
Figure 8: Redrawing of previous research on tectonic settings in the northern part of Sulawesi with consideration of extensional tectonics occurring in the Gulf of Tomini.The white lines represent the existing interpretation of the regional geological structures (Walpersdorf et al., 1998b; Hinschberger et al., 2005; Pholbud et al., 2012), the blue lines represent the existing interpretation of on-land geological structures (Sidarto & Bachri, 2015), and the red lines represent the multiple parallel dextral strike-slip faults identified in this study.The gray transparent area denotes the proposed rifting zone (opening area) based on gravity and magnetic datasets.The blue square represents the boundaries of the study area.The red circle at the eastern end of the Northern Arm of Sulawesi marks the pivot point of the clockwise rotation mechanism (Song et al., 2022).

Figure 9 :
Figure 9: Perspective visualization in 3D of the Northern Arm (top) and the Eastern Arm (bottom) using DEM with a vertical exaggeration of about 15 in QGIS software.The visualization includes the overlaying of the existing interpretation of geological structures (represented by white and blue lines) and the proposed geological structures (represented by red lines).The gray arrows indicate the north direction.

Figure 10 :
Figure 10: Two seismic sections show key features in the study area."LineDeep" (top) is a 250 km long, 11 s two-waytraveltime (TWT) deep section and revealing normal faults perpendicular to the parallel dextral strike-slip faults."Line_30" (bottom) spans 10 km, 4.5 s TWT, and showing volcanic mud-diapirs on the Lalanga ridge.Seismic line locations, namely "LineDeep" and "Line_30", are marked in the inset map and Figure 8 (orange lines).

Table 1 :
Physical parameters of the Causative Bodies used in the Synthetic Modelling (Referenced to Figure 3)