Construction of the Caroline Ridge uppermost basement in the West Pacific: Implications from intrabasement seismic reflectors

The construction model of the Caroline Ridge uppermost basement is still unresolved, requiring more inference from limited geophysical observational data. Here, we systematically reveal intrabasement seismic reflectors of volcanic sequences within the rifted and subsidence domains of the Caroline Ridge. Extrusive centres and three types of intrabasement reflectors, that is, relatively horizontal, ridgeward‐dipping and folded reflectors, have been identified. Extrusive centres in the rifted domain are characterized by domal shapes and produce sub‐parallel stratified intrabasement reflectors within the conduits that connect with the relatively horizontal reflectors distributed on both sides of the basement highs. Intrabasement reflectors display increasing dip angles away from the extrusive centre and present ridgeward‐dipping reflectors but not troughward‐dipping reflectors in subsidence domain 1, suggesting a brittle deformation process. Layered intrabasement reflectors are developed within subsidence domain 2 but display folded and mounded morphologies, suggesting a ductile deformation process. We propose that the Caroline Ridge formation was supported by discrete extrusive centres, and the uppermost basement construction model has experienced stages of transition from brittle deformation to ductile deformation, which can provide new clues for the early‐stage crustal evolution of global oceanic plateaus.


Funding information
The construction model of the Caroline Ridge uppermost basement is still unresolved, requiring more inference from limited geophysical observational data.Here, we systematically reveal intrabasement seismic reflectors of volcanic sequences within the rifted and subsidence domains of the Caroline Ridge.Extrusive centres and three types of intrabasement reflectors, that is, relatively horizontal, ridgeward-dipping and folded reflectors, have been identified.Extrusive centres in the rifted domain are characterized by domal shapes and produce sub-parallel stratified intrabasement reflectors within the conduits that connect with the relatively horizontal reflectors distributed on both sides of the basement highs.Intrabasement reflectors display increasing dip angles away from the extrusive centre and present ridgeward-dipping reflectors but not troughward-dipping reflectors in subsidence domain 1, suggesting a brittle deformation process.Layered intrabasement reflectors are developed within subsidence domain 2 but display folded and mounded morphologies, suggesting a ductile deformation process.We propose that the Caroline Ridge formation was supported by discrete extrusive centres, and the uppermost basement construction model has experienced stages of transition from brittle deformation to ductile deformation, which can provide new clues for the early-stage crustal evolution of global oceanic plateaus.

K E Y W O R D S
Caroline Ridge, extrusive centres, intrabasement reflectors, uppermost basement, West Pacific
High-resolution intrabasement structures shown by seismic reflection data are generally induced by orderly magmatism and can provide important evidence to indicate crustal formation and tectonic deformation (e.g., Arai et al., 2017;Menzies et al., 2002;Phillips et al., 2016;Sauter et al., 2021;Taira et al., 2004).The analysis of intrabasement seismic reflection data is an important way to reveal the oceanic plateau uppermost basement (e.g., Sager et al., 2013Sager et al., , 2016Sager et al., , 2019)).Previous works have proposed a suite of different models for Caroline Ridge crustal evolution.These models are mainly inferred from interpretations of shallow topography, sediments and faulting characteristics or large-scale structural analyses of regional gravity anomaly characteristics and crustal thickness (e.g., Altis, 1999;Fujiwara et al., 2000;Lee, 2004;Dong et al., 2018;Zhang et al., 2017Zhang et al., , 2019;;Zhang, Dong, Sun, & Zhang, 2021;Zhang, Dong, Sun, Zhang, & Bai, 2021).What is unclear is the construction model of the Caroline Ridge uppermost basement due to the lack of direct systematic analysis for high-resolution intrabasement seismic reflectors, hindering the evaluation of geodynamic models of oceanic plateaus.To explore this issue, therefore, we first focus on intrabasement seismic reflectors in this study, which are identified as volcanic sequences within the
The rifting of the Caroline Ridge formed the Sorol Trough.Rifting likely occurred since the Early Miocene-Middle Miocene (Altis, 1999;Weissel & Anderson, 1978;Zhang, Dong, Sun, & Zhang, 2021).Rock samples within the western section of the Sorol Trough show medium-to high-grade metamorphic characteristics, indicating that the synkinematic crustal accretion process was accompanied by shear stress (Fornari et al., 1979).Altis (1999) inferred that rifting and crustal thinning might have been dominated by tensile stress at the Yap-Mariana subduction junction.Subduction rejuvenation at the Yap Trench occurred in the Miocene, indicated by island arc tholeiite and gabbro samples from southern Yap Island (Crawford et al., 1986).The rejuvenation of subduction was possibly related to shallow melting of the upper mantle due to hot Sorol Trough subduction (Ohara et al., 2002), suggesting that the Caroline Ridge front had already moved into the subduction domain (Dong et al., 2018;Zhang, Dong, Sun, Zhang, & Bai, 2021).In the Caroline Ridge front, N30 E-trending faults parallel to the trench have been developed (Fujiwara et al., 2000;Lee, 2004;Zhang, Dong, Sun, Zhang, & Bai, 2021) and rare sediment has deposited (e.g., Kobayashi, 2004;Nagihara et al., 1989;Xia et al., 2020), providing active tectonic processes at the trench.
The sedimentary sequences and basement properties of the Caroline Ridge have been calibrated by drilling cores from the Deep Sea Drilling Program (DSDP) sites (Heezen et al., 1971a(Heezen et al., , 1971b(Heezen et al., , 1971c)).
The results revealed that the strata section encountered was 130-335 m of Late Oligocene or Miocene ooze.The sediments at DSDP sites 55 and 56 overlie a smooth reflector (Heezen et al., 1971a(Heezen et al., , 1971b)), which was interpreted as igneous (Heezen et al., 1971c).
Basalt was found beneath the unaltered Upper Oligocene sediments at DSDP site 57.The lack of contact metamorphism of the sediments suggests emplacement as magma (Heezen et al., 1971c;Ridley et al., 1974).Since the Late Oligocene, with the relative motion direction of the Caroline Plate changing, the seamount chain with a descending age for Chuuk (12.7-4.7 Ma), Pohnpei (Ponape) (8.6-3 Ma) and Kusrae (2.6-1.2Ma) (Keating et al., 1984;Zhang, Wang, et al., 2020) that formed east of the Caroline Ridge.

| DATA AND METHODS
During 2016-2017, the Chinese Academy of Sciences conducted geophysical surveys in the Caroline Ridge.Approximately 360-km high-quality multichannel seismic lines were used in this study.Multichannel seismic data were obtained based on a 120-channel seismic acquisition system.The source was a four-gun array with a total volume of 1300 in 3 , and the length of the active section was 1500 m with a 12.5-m channel interval.Based on the seismic features of deep-water areas, data were processed by amplitude compensation, prestack amplitude-preserved integrated denoising, combined deconvolution, multiple attenuation, fine migration velocity field establishment, two-dimensional prestack time migration imaging, etc. Ghost wave effects were removed using broadband processing.During seismic acquisition, the rugged seafloor topography might cause reflection points of diffractive multiples located outside the cable ranges, which makes it difficult to accurately simulate and suppress the multiples at one time.In this study, we use complex diffraction multiple wave attenuation (DMA) to suppress multiple waves and improve the data fidelity by (1) strata signal modelling and amplitude statistics, (2) amplitude iteration and decomposition under consistent surface control, and (3) discrimination of residual multiple waves, calculating scale factors and implementing suppression processes.These methods accurately highlight the thin sedimentary layers and intrabasement seismic reflectors.Thus, we imaged the high-quality seismic reflections of the uppermost basement of the Caroline Ridge.We apply seismic data interpretation methods to track and analyse the intrabasement reflectors using the software 'Petrel'.Additionally, local structural dip attributes and root mean square (RMS) attributes have been used to analyse the intrabasement reflectors.The local structural dip attribute is an edge detection method estimating the variations in local dip from seismic data on the basis of event, gradient and principal components (Marfo et al., 2017;Omosanya et al., 2018).The instantaneous phase attribute is a method for improving reflector continuity and stratigraphic boundaries and enhancing lateral variations and seismic facies variations (e.g., Sarhan, 2017).The RMS amplitude attribute mathematically computes the root mean square values of amplitude by squaring individual traces to enhance high amplitudes, which is generally sensitive to amplitude variation (e.g., Brown, 2004;Omosanya & Alves, 2013;Kumar et al., 2021).In subsidence domain 2 of the Caroline Ridge, the layered, high relief and folded intrabasement reflectors are relatively clear and can be identified by seismic profiles.However, folded intrabasement reflectors have been bounded by some strong-amplitude, irregular and dipping intrabasement reflectors.Hence, the RMS amplitude attribute has been used to highlight the distributions of these strong-amplitude intrabasement reflectors, which can be interpreted as the decollement.

| Extrusive centres
Basement highs have been identified within the oceanic plateau according to seismic lines.They are delineated by dome-shaped reflectors at their summits (Figures 2, 3 and 6) and present circular or elliptical features in plain view.Unlike the upwelling magmatic conduits with pull-up, low-amplitude reflectors, these structures present relatively clear-layered intrabasement reflectors within the conduits, connecting with the layered reflectors distributed on both sides of the basement highs.Reflectors within basement highs are relatively horizontal and not as continuous as those on the flanks, corresponding to the discontinuous reflectors in the instantaneous phase attribute map (Figure 2f) and discontinuous low values in the structural dip attribute map within the uppermost basement of extrusive centres (Figure 6a).On the basis of seismic profiles, the basement highs of the Caroline Ridge did not deform the overlying horizontal strata (Figure 2h).The extrusive centres (Figures 2, 3     Overall, multidirectional intrabasement reflectors are present below the already identified smooth basement top (Heezen et al., 1971c;Ridley et al., 1974;Zhang, Dong, Sun, Zhang, & Bai, 2021).
For volcanic margins, seaward-dipping reflectors (SDRs) show distinct synrift features accumulating on transitional crust during breakup (Menzies et al., 2002;Roberts et al., 1984;Sager et al., 2013).(e.g., Fischer et al., 2016;Omosanya et al., 2018;Sun et al., 2022;Zhao et al., 2014) because crystalline basement easily presents acoustically transparent features due to the typically low internal impedance contrasts (Phillips et al., 2016).The analysis of petrologic results of the Caroline Ridge basement revealed the unaltered nature of the overlying sediments, the lack of a chill zone at the top of the basalt and no sign of weathering and erosion near the top of the basement (Heezen et al., 1971c;Ridley et al., 1974).The relatively horizontal morphologies of the intrabasement reflectors are consistent in the northern and southern Caroline Ridge, which could be distributed within an 200-km region.These observations suggest that abundant magma can be capable of laterally flowing over distances and was produced at prodigious rates, indicating massive flows induced by highly effusive volcanism (e.g., Keszthelyi & Self, 1998;Self, Thordarson, & Keszthelyi, 1997).Thus, the layered intrabasement reflectors of the Caroline Ridge can be interpreted as lateral lava flows from the basement highs.
Additionally, seismic profiles reveal that no obvious deformation, such as folding and faulting, has been identified within the overlying sediments draping the basement highs in the rifted and subsidence domains, indicating that the basement highs have a relatively old age.
Because the DSDP results reveal that the sediments date to the Late Oligocene-Quaternary (Heezen et al., 1971a(Heezen et al., , 1971b(Heezen et al., , 1971c)), the youngest age of these basement highs might be the Late Oligocene, which formed before the overlying sediment formation.According to these observations, we prefer the interpretation that the basement highs with layered intrabasement reflectors were formed during the formation of the Caroline Ridge, which can be interpreted as extrusive centres distributed randomly across the oceanic plateau, in accordance with the extrusive centre interpretation of the seismic reflector analysis of the comparable Agulhas Plateau in the SW Indian Ocean, of which extrusive centre lateral extent is approximately 10 km (Gohl & Uenzelmann-Neben, 2001;Uenzelmann-Neben et al., 1999).
Especially, the potential temperature of the mantle beneath the Sorol Trough indicates that there is no thermal anomaly in the mantle source, and the Sorol Trough was the product of low-degree partial melting of a relatively enriched mantle source that passively upwelled (Yan et al., 2022).Thus, the generation of the Sorol Trough might not have developed basement highs related to intensive lateral magma flows.Additionally, analysis of petrology and geophysics indicates that the Sorol Trough did not progress into mature seafloor spreading (Altis, 1999;Fornari et al., 1979).Thus, the basement highs with layered intrabasement reflectors in the Sorol Trough (Figure 3) can be interpreted as the pre-existing extrusive centres (extrusive centres 2 and 3) during the Caroline Ridge formation.
The gradient slope map reveals that basement highs are widely developed within the Caroline Ridge and the interior of the Sorol Trough, of which the morphology and scales resemble these extrusive centres (Figure 2).We cannot confirm that all these basement highs within the Caroline Ridge correspond to the extrusive centres.However, it is considered that as an oceanic plateau with 100to 300-km crustal width, the formation of hundreds of kilometres lateral extending lava flows likely requires abundant magma supply.We model dominated by discrete extrusive centres has also been identified within other oceanic plateaus, such as the Agulhas Plateau (e.g., Barrett, 1977;Parsiegla et al., 2008;Uenzelmann-Neben et al., 1999), south-western Mozambique Ridge (Fischer et al., 2016), SW Indian Ocean, Manihiki Plateau, West Pacific (Pietsch & Uenzelmann-Neben, 2015) and Shatsky Rise, West Pacific (Sager et al., 2013).The view of discrete extrusive centres (multiple magmatism sources) in the Caroline Ridge relies on clear imaging of intrabasement reflectors from new and reprocessed seismic data.Generally, the origin models of oceanic plateaus can be divided into two types: (1) the spreading model in which the oceanic plateau crust originates from the decompression melting of a hot mantle plume (fertile mantle model) (Anderson & Natland, 2014;Foulger, 2007;Sager et al., 2016) and ( 2) the model in which the crust originates from massive lava flows due to underplating of a hot mantle plume (plume head model) (Clark et al., 2018;Coffin & Eldholm, 1994;Duncan & Richards, 1991;Huang et al., 2018;Richards et al., 1989;Sager et al., 2013).Intrabasement reflectors of the Caroline Ridge might indicate the prediction that a plume head arrived at the base of the oceanic lithosphere (e.g., Campbell, 2005Campbell, , 2007;;Coffin & Eldholm, 1994;Duncan & Richards, 1991;Hill, 1991;Li et al., 1999;Richards et al., 1989;Sager et al., 2016).Because the intrabasement reflector features of the Caroline Ridge suggest that the emplacement of a large volume of lateral lava flows was likely supplied by numerous discrete extrusive centres (hot sources), and these discrete extrusive centres might not appear in the fertile mantle model (Anderson & Natland, 2014;Foulger, 2007;Sager et al., 2016).
A hypothesis is that the northern Caroline Ridge was built by more extrusive centres than the southern Caroline Ridge on the basis of gradient slope maps (Figure 2).It is possible that the northern Caroline Ridge came from more magmatism during oceanic plateau formation.The reasons might include the following: (1) pre-existing fracture zones might exist in the oceanic lithosphere of the northern Caroline Ridge that provide more pathways for magma; and (2) seawater might be transported through pre-existing fracture zones that promote the hydration (e.g., Gou et al., 2018) and the melting process of the oceanic lithosphere.

| Deformation of the Caroline Ridge uppermost basement
Deformation of intrabasement structures occurred during later tectonic events, which fundamentally modified the uppermost basement (Phillips et al., 2016).On the one hand, the titling of originally relatively horizontal lava flows is evident in outcrops of the upper crust.teaus is closely connected to seafloor spreading processes, for example, the Shatsky Rise in the north-west Pacific (Sager et al., 2016(Sager et al., , 2019) ) and the Ontong-Java-Manihiki-Hikurangi oceanic plateau in the West Pacific (Taylor, 2006).Prograding hyaloclastite foresets and lava deltas (Eldholm et al., 1987;Franke, 2013;Keen et al., 2012;Menzies et al., 2002;Planke et al., 2015Planke et al., , 2017) ) dominated the formation of the oceanic plateau uppermost basement, similar to the volcanic continental margins, for example, Ceara Ride and conjugate Sierra Leone Rise in the Atlantic (Kumar, 1979;Basile et al., 2020) and Rio Grande Rise and conjugate Walvis Ridge (Hoyer et al., 2022).However, our observations imply that the original intense magmatism did not necessarily rift the oceanic plateau immediately, and the construction model of the oceanic plateau uppermost basement in the subsidence domain resembles the evolutionary model of the non-volcanic continental margins.
On the other hand, the results from DSDP 57 reveal that unaltered Late Oligocene sediments rest on the surface of a doleritic basalt, indicating that the Caroline Ridge basement originated from lava flows (Heezen et al., 1971c).In subsidence domain 2, the thick and layered intrabasement reflectors present relatively high relief and folded morphology (Figures 7 and 8), indicating that the lava flows have experienced compressional tectonic deformation.Generally, layered intrabasement reflectors within oceanic plateaus, as shown in seismic profiles, originate from alternating massive flows and lava flows (Bangs et al., 2015;Klaus & Sager, 2002;Rotstein et al., 1992;Sager et al., 2013Sager et al., , 2016;;Schaming & Rotstein, 1990).During the oceanic plateau uppermost basement formation, the emplacement durations of each laterally massive flow are estimated to be years to decades (Inoue et al., 2008), and multiple emplacements formed alternating massive and pillow flows (Sager et al., 2013).Thus, there might exist differences in properties between the massive and pillow flows.and subduction in the Miocene (e.g., Altis, 1999;Yan et al., 2022;Zhang et al., 2019).
Overall, the construction of the Caroline Ridge uppermost basement is a complex process.The uppermost basement construction model of the Caroline Ridge can be divided into three stages ; thus, sedimentary infills above the basement top might preserve limited or indirect geological records about evolutionary history of oceanic plateau.At present, a fundamental question remains unresolved: what is the construction model of the uppermost basement of oceanic plateau.This necessitates further inference and testing of models based on the limited geophysical observational data available.
Regional tectonic map of the Caroline Plate and the Philippine Sea Plate.The white solid box represents the location of Figure 1a.The black dotted box represents the location of Figure 1b.The black dots show the locations of the Deep Sea Drilling Project (DSDP) sites.(b) Multichannel seismic lines in the Yap subduction zone.We use the seismic data sets that were collected in 2016-2017.The red lines represent the seismic data first used in this work.The blue line represents the seismic data reprocessed in 2020.The upper colour scale applies to Figure 1a, and the lower colour scale applies to Figure 1b.Caroline Ridge uppermost basement of the rifted domain, subsidence domain 1 (Sorol Trough) and subsidence domain 2 (Caroline Ridge front) (Figure 1b).We select the 240-km new high-quality multichannel seismic lines collected in 2017 and 120-km multichannel seismic lines reprocessed in 2020 (collected in 2016).These seismic lines have been used because they could display the clearest imaging for the uppermost basement of the Caroline Ridge compared with other seismic lines in our data sets.Multibeam bathymetric data were processed in 2022 to reveal the intrabasement seismic reflectors of the Caroline Ridge.Our main objective is to illustrate the Caroline Ridge uppermost basement construction model and provide new clues for a geodynamic model of oceanic plateaus globally.
and 6) display an average diameter of 3-6 km and occur at depths of 2-5 s of TWT.In particular, from the analysis of seismic line YP17-05, basement highs with radius of 10 km are distributed in the Sorol Trough.The top mounded reflectors of basement highs are dome-shaped reflectors (Figure 3).Clear-layered, discontinuous and horizontal reflectors have also been identified within basement highs.The basement properties are relatively homogeneous; thus, the wave impedance contrast between the adjacent layered reflectors is relatively small.F I G U R E 2 (a) Slope gradient map.The yellow circles represent the extrusive centres identified by seismic profiles.The blue circles represent the inferred extrusive centres based on the slope gradient map.In planar view, linear anomalies have been recognized based on abrupt changes in the slope gradient within the oceanic plateau, suggesting NW-trending rifting fault systems.(b) Original Caroline Ridge basement top morphology.It reveals that the oceanic plateau is a roughly semi-elliptical structure in profile, approximately 300 km in width and 2 km in height, with the highest summit in the northern region (Figure2b) before rifting.For this model, the thermal subsidence of the oceanic plateau basement has not been considered, which should be the focus of future research.(c-i) Extrusive centres and related intrabasement reflectors within the subsidence domain of the oceanic plateau revealed by seismic transects from reprocessed seismic line YP16-01.Figure2f, g is maps of the instantaneous phase attribute, showing continuous reflectors in the subsidence domain and discontinuous reflectors in the extrusive centre.

4. 2 |
Intrabasement reflectors related to extrusive centresThree seismic facies have been identified within the Caroline Ridge uppermost basement (Table1 and Figure4).In the rifted domain, intrabasement reflectors present sub-parallel stratified sequences.Typically, individual reflectors can be traced laterally, and neighbouring reflectors are connected to imply longer, continuous horizons, named HRs, which present continuous low values in the structural dip attribute map (Figures5d, e and 6a).The overall pattern of reflectors can be followed for dozens of kilometres down the shallow flank slopes to the neighbouring interaction region of the oceanic plateau and oceanic crust (Figures 6b).
The NW-trending seismic line YP17-02 (Figure5b) across the southern Caroline Ridge reveals that layered F I G U R E 3 (a-e) Extrusive centres 2-3 and (f-i) intrabasement reflectors of the Sorol Trough revealed by seismic transects from seismic line YP17-05.intrabasement reflectors extend horizontally 12 km, sub-parallel to the smooth basement.The vertical thickness of the intrabasement reflectors can reach 2 s of two-way travel time (TWT) (approximately 6-7 km).The amplitude and continuity are strongest near the basement top but progressively weaker with depth due to scattering and attenuation of seismic signals.The NE-trending seismic F I G U R E 4 Three seismic facies within the Caroline Ridge uppermost basement.(a) Seismic facies 1 represents the relatively horizontal reflectors.(b) Seismic facies 2 represents the ridgeward-dipping reflectors.(c) Seismic facies 3 represents the folded reflectors.

F
I G U R E 5 (a) Multibeam bathymetric data of the southern Caroline Ridge.The black dotted lines show the locations of Figure 5b, c. (c-g) Intrabasement reflectors of the southern Caroline Ridge revealed by the seismic transects from seismic lines YP17-02 and YP17-03.Figure 5d, e is maps of the local structural dip attribute, showing continuous low values in the rifted domain.T A B L E 1 Conclusions of three seismic facies within the Caroline Ridge uppermost crust.03 (Figure 5c), perpendicular to YP17-02, also reveals the development of relatively horizontal layered intrabasement reflectors.Intrabasement reflectors extend horizontally 20 km, and the total vertical thickness is 1.8 s of TWT (approximately 5-6 km).These reflectors show continuous low values in the local structural dip attribute map.

|
Figures 7 and 8, sub-parallel stratified intrabasement reflectors appear within the Caroline Ridge front uppermost basement, which resembles the reflective features of Caroline Ridge rifted domains.However, most of these intrabasement reflectors generally exhibit folded and mounded morphologies (Figures7b-d), distinct from the relatively horizontal intrabasement reflectors, named as FRs.The total vertical thickness of the FRs can reach 2.0 s of TWT.As shown in Figures7c and 8c,d, some intrabasement reflectors show prominent, high-amplitude and dipping features and present a larger acoustic impedance contrast and relatively high values in the RMS amplitude attribute map.

F
I G U R E 7 (a-d) Seismic transects and (e-g) root mean square (RMS) amplitude attribute map from seismic line YP17-03, (h-j) showing the layered and folded reflectors within the Caroline Ridge front uppermost basement.(k) Multibeam bathymetric data of the Caroline Ridge front crust.
believe that hotspot magmatism generated discrete extrusive centres in the oceanic basin, and lava flows overlapped on top of the Caroline Ridge, forming layered structures.The uppermost basement formation F I G U R E 8 (a-d) Seismic transects and (e-g) root mean square (RMS) amplitude attribute map from seismic line YP17-03, (h-j) showing the layered and folded reflectors and decollement within the Caroline Ridge front uppermost basement.(k) Multibeam bathymetric data of the Caroline Ridge front crust.
The ridgeward-dipping reflectors suggest the brittle deformation process of the Caroline Ridge uppermost basement.In the subsidence domain of the Caroline Ridge, as shown in seismic profiles, the ridgeward-dipping reflectors display large dip angles, and these reflectors are oblique to relatively horizontal to the basement top reflector.The angular unconformity R20 between the ridgeward-dipping reflectors and the basement top reflector marks the transition from the pre-rifting to rifting of the Caroline Ridge.Especially, these intrabasement reflectors did not present the features of prograding or retrograding volcanic sequences.It is suggested that the ridgeward-dipping reflectors originated from brittle deformation induced by normal faulting during basement subsidence and tilting.The lack of trough-dipping reflectors in the subsidence domain indicates that the typical lava flow gravitational effect might not have occurred during oceanic plateau uppermost basement formation.The identification of ridgewarddipping reflectors indicates that the original layered intrabasement structures of the Caroline Ridge were preserved during crustal rifting and subsidence.Generally, upwelling mantle plumes tend to move towards active spreading centres, and the formation of oceanic pla- During deformation, the original alternating lava flows can more easily to deform, corresponding to the layered, relatively high relief and folded intrabasement reflectors shown in seismic profiles(Figures 7f, g and 8f, g).Additionally, the folded intrabasement reflectors have been bounded by some irregular and dipping intrabasement reflectors with high RMS amplitudes (Figures8e, f), which can be interpreted as the decollement.Hence, we consider that the original layered intrabasement reflectors of subsidence domain 2 experienced a ductile deformation process.Sauter et al. (2021) proposed that crustal ductile deformation is always related to short timescale changes in magma supply and subsequent thermal structure.The folded intrabasement reflectors of subsidence domain 2 might be linked to the Caroline Ridge crustal thinning, mantle upwelling and related thermal structure changes during the Sorol Trough formation

(
Figure9): (1) The oceanic plateau was formed by discrete extrusive centres.The contribution of this view is that the Caroline Ridge formation was likely supported by multiple magmatic sources rather than a single magmatic source.The tholeiitic basalt samples of main-stage volcanism from the Caroline Ridge show various ages in the western and central parts of the CarolineRidge, 15.62-19.26Ma (Zhang,   Zhang, et al., 2020), which could correspond to our view that the activity initiation of magmatic sources might be different.(2) Ridgeward-dipping reflectors formed in subsidence domain 1 of the Caroline Ridge induced by the brittle deformation process.The extrusive centres might have been preserved.(3) The folded reflectors formed in subsidence domain 2 of the Caroline Ridge were induced by the ductile deformation process.The uppermost basement construction model has transferred from the rifted to subsidence domains of the Caroline Ridge.It is demonstrated that the extrusive centres and deformation intrabasement structures jointly affect the structural style of the uppermost basement of the Caroline Ridge.6 | CONCLUSIONS We illustrate the Caroline Ridge uppermost basement construction model based on the analysis of intrabasement seismic reflectors.Basement highs have been identified within the rifted and subsidence domains of the Caroline Ridge, and sub-parallel reflector sequences originating from the centre of basement highs can be connected with the reflectors in the surrounding crust and extend for a long distance in the rifted and subsidence domains, indicating that hotspot magmatism has generated discrete extrusive centres and that basalt flows progressively overlap on top of the Caroline Ridge uppermost basement.In the subsidence domain of the Caroline Ridge, the ridgewarddipping reflectors display large dip angles; these reflectors are oblique to relatively horizontal to the basement top reflector, and these intrabasement reflectors do not present the features of prograding or retrograding volcanic sequences, suggesting the brittle deformation process of the uppermost basement.The thick and layered intrabasement reflectors show relatively high relief and folded morphology within subsidence domain 2, suggesting that the original intrabasement reflectors of the Caroline Ridge have been affected by the F I G U R E 9 The construction sequences of the uppermost basement of the Caroline Ridge.The generation of discrete extrusive centres at seafloor and subsequent intrabasement structures brittle and ductile deformation.EC, extrusive centre; HRs, relatively horizontal reflectors; RDRs, Ridgeward-dipping reflectors; FRs, folded reflectors.ductile deformation process.Hence, the construction of the Caroline Ridge uppermost basement is a complex process.The generation of extrusive centres and related intrabasement structural deformation played dominant roles in the construction of the Caroline Ridge uppermost basement.