Subduction zones are convergent plate boundary in which at least one lithospheric plate is oceanic and disappear beneath the other plate. The oceanic crust formed at mid-ocean ridges is progressively covered by sediments, as time goes, and the existing faults and cavities are sealed, stopping the hydrothermal circulation. This is confirmed by the decay of magnetization within the first 10 million years observed at all spreading centers followed by a relative stability (Dyment et al., 2015). Before subduction, the flexure of the oceanic plate generates new faults parallel to the trench and re-open existing faults of proper trend resulting in increasing the permeability of the crust to seawater.
The subducting oceanic plate slides under the overriding plate along a basal detachment fault called decollement, which structure is formed by high pressure on the boundary between the subducted sediment layer on the oceanic crust and the overriding plate: the friction creates the megathrust zones where the large earthquakes are triggered.
Subsequently, the subducting oceanic plate starts to emit plenty of seawater when the plate reaches high pressure and high temperature, resulting in partial melting in the overlying mantle. In this process, the mantle becomes hydrated and serpentinizes. Lighter partially molten rocks rise to the surface and form the volcanic arc (Marsh and Leitz, 1979). As a result, the converging tectonic processes outcome tremendous natural hazards such as earthquakes, tsunamis and volcanic eruptions.
In the subduction zone, two distinctive magnetic anomaly signals are observed: (1) seafloor spreading magnetic anomalies of the subducting slab with strong NRM intensities are erased when the increasing temperature pass the Curie temperature of its magnetic minerals, and (2) landward magnetic anomaly belt generated by induced magnetization of the serpentinized fore-arc mantle which contains high susceptibility magnetic minerals.
Figure 1-5. The world seafloor age map (Müller et al., 2008). The blue solid lines indicate the boundary of active subduction zones and red solid lines, inactive subduction zones.
This work focuses on the analysis of fading seafloor spreading magnetic anomalies in subduction zones and the causes of this decay. To address this problem, we selected subduction zone areas following several criteria: (1) dense marine magnetic data, (2) no perturbation from other tectonic structures, and (3) strike of the anomaly oblique to the trench. The three criteria aim to observe the amplitude variations of seafloor spreading magnetic anomalies before and after subduction alleviating possible biases related to paleomagnetic field intensity variation with time, by studying the same sets of anomalies. The Japan-Kuril Trench area fits all conditions for a detailed analysis. We later extend the analysis to four other subduction zones.
To build better high-resolution magnetic anomaly maps, we include shipboard three-component magnetometer (STCM) data to complement proton precession magnetometer (PPM) data in areas where the later are not available and designed a new crossover algorithm to combine these two types of marine magnetic data (See appendix A). PPM are preferred over STCM data because the PPM sensor measures absolute magnetic field, whereas the STCM sensor measures relative variations of vector magnetic field which, despite complicated correction of the ship induced and remanent magnetization effects, is still affected by viscous remanent magnetization (VRM) progressively acquired by the ship while following a constant heading.
We took advantage of a high-resolution magnetic anomaly map and designed an original analytic method to alleviate the effect of topography and isolate the variations of magnetization along individual magnetic anomalies. This method, which differs from that used by Okubo et al. (1991) in the same area, was applied on magnetic anomalies having entered subduction in the Japan Trench (Chapter 2) and approaching the Japan-Kuril Trench (Chapter 3). It is later applied to four other subduction zones for comparison and generalization (Chapter 4).
Early magnetic studies of the Japan Trench showed that seafloor spreading magnetic anomalies progressively fade away and disappear during subduction, reflecting the increasing distance to magnetized sources and the removal of their remanent magnetization with alteration and increasing temperature. An improved magnetic anomaly map derived from both scalar and vector magnetic anomaly data, coupled with a better knowledge of the slab geometry in one hand, of the magnetic structure of the oceanic crust on the other hand, allow us to constrain the thermal structure of the subducting slab. We, for the first time, identify two steps in the anomaly disappearance: first the magnetization of extrusive basalt is rapidly erased between 9-12 km, where titanomagnetite reaches its blocking temperature between 150-350°C, then the magnetization of deeper crustal layers slowly decreases down to ~20 km, reflecting the progressive slab heating toward the Curie temperature of magnetite, 580 °C. The resulting slab temperatures are higher than predicted by most thermal models. Recent observations and models suggest rejuvenated hydrothermal activity triggered by lithospheric flexure before subduction that may significantly heat up the subducting oceanic crust through thermal blanketing and possibly serpentinization, with consequences on the depth of the seismogenic zone.
Subduction is a major geodynamic process that leads to the consumption of oceanic lithosphere, the creation of volcanic arcs and the largest volcanic edifices on Earth, and the generation of natural disasters such as earthquakes and tsunamis resulting in major damages and casualties. Many geophysical studies have been conducted over the Japan Trench to understand the structure of the subducting plate. Detailed slab geometry models Slab 1.0 and Slab 2 (Hayes et al., 2012; 2018) have been developed based on active seismic profiles and seismicity studies. Thermal structure models have been proposed based on heat flow surveys (Hyndman and Peacock, 2003; van Keken et al., 2012; Kawada et al., 2014; Wada et al., 2015). A pioneer study of magnetic anomalies on a subducting plate has shown that, from the Japan Trench landward, the magnetic anomalies decrease in amplitude and their short wavelength content attenuates (Okubo et al., 1991), as a result of increasing distance between the magnetized sources and the observation, thermal demagnetization and continuous oxidation of magnetic minerals within the extrusive oceanic crust (Okubo et al., 1991; Kido and Fujiwara, 2004). Today, the structure of the slab is well constrained (Hayes et al., 2012; 2018), moreover, the magnetic structure of the oceanic crust is better understood (e.g., Dyment and Arkani-Hamed, 1995; Gee and Kent, 2007). Based on this information, the magnetic anomalies over the Japan Trench offer an independent means to access to the slab thermal structure. To address this problem, we take advantage of a unique scalar and vector marine magnetic dataset to build an improved high-resolution magnetic anomaly map. We identify two steps of thermal demagnetization, corresponding to the two major magnetic minerals of the oceanic crust layers, which constrain the thermal structure at shallow depths within the subduction system.
Geological settings of North-West Pacific
The oceanic crust in the North-Western Pacific plate off Japan was formed about 125-140 Ma at the Pacific-Izanagi plate boundary (Nakanishi et al., 1989). The area is covered by pelagic sediments 1.6 km thick and subducts beneath the Japan islands at a speed of 70-90 km/Ma (Seno, 2017). A thicker and wider accretionary prism is observed in the northern Japan Trench (Kodaira et al., 2017). The free-air gravity anomaly locally delineates seamounts and fracture zones (Fig. 2-1A). In the whole area, it displays a negative anomaly associated to the trench and a positive anomaly on the flexural bulge ~150 km seaward. This bulge reflects the bending of the subducting plate and induces horsts and grabens as the plate approaches the trench (Tsuru et al., 2002; Kodaira et al., 2017).
Data and methods
Two different sets of marine magnetic data were gathered. Scalar magnetic data (i.e., total field vector intensity) acquired by proton precession magnetometer (PPM) were obtained from DARWIN of Japan Agency for Marine-Earth Science and Technology (2016), GEODAS of National Center for Environmental Information (2007) and Nautilus of Institut Français de Recherche pour l’Exploitation de la MER (2014). Vector magnetic data (i.e., total field vector components) acquired by shipboard three component magnetometer (STCM), mostly on Japanese research vessels, were obtained from JAMSTEC (DARWIN data base). The International Geomagnetic Reference Field (IGRF) model (Thébault et al., 2015) was subtracted from the PPM data. The STCM data were corrected for the ship magnetic effect and motion by the method of Isezaki (1986). To minimize the misfit at cross over points, a modified cross over error analysis technique was applied both to the PPM data and to the STCM data (see Supplementary information). We reduced the corrected magnetic anomaly grid to the pole (RTP) assuming a 53.6° inclination and -7.6° declination (IGRF averaged over 20 years), 33° paleoinclination and 11° paleoazimuth (average value for the study area from the global grids of Dyment and Arkani-Hamed, 1998).
High resolution bathymetric data were obtained from Global Multi-Resolution Topography (GMRT; Ryan et al., 2009). The top of the magnetic source, i.e. of the extrusive basalt layer, for the subducting plate was computed by merging and correcting different grids. The World sediment thickness grid (Divins, 2003) was subtracted from the bathymetry grid, and the resulting grid was merged with Slab 1.0 (Hayes. et al., 2012; for a discussion on the choice of Slab 1.0 see Supplementary Information) validated by published seismic profiles (e.g., Tsuru et al., 2002; Kodaira et al., 2017). The geometrical misfit between the two grids along the trench boundary was erased and re-interpolated using the Partial Differential Equation (PDE) surface method (D’Errico, 2005).
We carefully selected magnetic anomalies showing no tectonic or volcanic local complexities such as fracture zones, propagators, or seamount. Anomalies older than M10r and younger than M8 are therefore discarded (Fig. 2-1). We extracted profiles from the magnetic grid across the selected anomalies and considered separately the anomaly profiles before and after subduction, i.e. located East and West of the Japan Trench. The anomaly profiles before subduction were inverted to equivalent magnetization assuming a 500-m-thick magnetized source layer with no vertical variation of magnetization (Parker and Huestis, 1974). The equivalent magnetization shows little variation among profiles and has been averaged (Fig. 2-2A). We use this average equivalent magnetization and the inferred top of the magnetic source layer along the anomaly profiles after subduction (Fig. 2-2B) to compute synthetic magnetic anomalies along these profiles (Fig. 2-2C). These modeled anomalies represent the contribution of the subducting plate at the sea-surface if the magnetic structure of the plate remains unchanged. Comparison of these synthetic anomalies with the observed ones gives us the opportunity to estimate how the magnetic structure of the plate has been changed (Fig. 2-2C).
The ratio of peak to trough anomaly amplitudes of the observed and synthetic anomalies, hereafter named RAM (Remaining Amount of Magnetization), provides an estimate of the remaining fraction of magnetization in the subducting plate: when it is close to 1, demagnetization remains negligible, whereas when it tends to 0, demagnetization is almost complete. We expect demagnetization to progress as a function of depth, and display RAM as a function of the depth to the top of the magnetized layer under the maximum (resp. minimum) of the observed positive (resp. negative) anomalies (Fig. 2-3).
A new magnetic anomaly map on the japan trench area
We compiled both scalar and vector marine magnetic data available in the study area and merged them into a unique scalar magnetic anomaly map (Fig. 2-1B). Many gaps in the scalar anomaly coverage could be filled with the vector data. The resulting map shows two types of anomalies. Near the Japanese Islands, a strong positive anomaly (FIMAB in Fig. 2-1) is caused by the induced magnetization of serpentinite in the fore-arc mantle (Okubo and Matsunaga, 1994; Hyndman and Peacock, 2003; Blakely et al., 2005) and, possibly, of the volcanic arc. On the Pacific plate and subducting slab, alternating positive and negative lineated anomalies are caused by the remanent magnetization of the oceanic crust. Magnetic anomalies M5 to M17 (~124.6-139.7 Ma; Malinverno et al., 2012) are identified between two NNW-SSE-trending fracture zones depicted on the free-air gravity anomaly (Fig. 2-1A), confirming the interpretation of Nakanishi et al., 1989).
In our study area, only anomalies M8r-M10r (129.0-130.8 Ma) are suitable for a detailed analysis as they are linear and only disrupted by a few isolated seamounts. They display a clear seafloor spreading magnetic signal before and after subduction (Fig. 2-1A). Conversely, anomaly M5 is still recognizable but totally subducted, while anomalies M6-M7 are only partially subducted beneath the southern Kuril Trench, roughly parallel to the magnetic anomaly trend. The presence of propagating rifts (Nakanishi, 2011) and the Joban seamount chain affect anomalies M10n1-M15 south of 38°40’N. During this period the (half) spreading rate was 70-80 km/Ma and the studied oceanic crust was formed at a fast spreading center (Nakanishi et al., 1989).
Magnetic structure of the oceanic crust and progressive demagnetization
The amplitude of the slab magnetic anomalies continuously decreases landward and the anomalies disappear beyond 20 km depth below sea level (bsl) (Fig. 2-1C). The increasing depth of the magnetized source preferentially attenuates the shorter anomaly wavelengths of the anomaly, as does demagnetization of the oceanic crust induced by fluids (alteration) and increasing temperature (thermal demagnetization). To separate the effects of increasing depth and demagnetizationthese effects, we compute synthetic magnetic anomalies assuming unchanged magnetic structure of the subducting plate. The observed anomalies decrease faster than the synthetic ones (Fig. 2-2C). Indeed, the RAM decays rapidly, by 20% per km, between 9-12 km bsl, and more slowly, by 2% per km, beyond (Fig. 2-3A).
Table of contents :
CHAPTER 1. GENERAL INTRODUCTION
1.2 Subduction zones
1.3 Research framework
CHAPTER 2. FADING MAGNETIC ANOMALIES, THERMAL STRUCTURE AND EARTHQUAKES IN THE JAPAN TRENCH
2.3 Geological settings of North-West Pacific
2.4 Data and methods
2.5 A new magnetic anomaly map on the japan trench area
2.6 Magnetic structure of the oceanic crust and progressive demagnetization
2.7 Thermal structure and earthquakes in the shallow subducting oceanic crust
2.9 Supplementary materials
2.9.1 Crossover correction for scalar and vector magnetic anomalies
2.9.1 Uncertainties in the slab geometry
2.9.1 Effect of a dipping slab on the inclination and declination of magnetization, consequences on the amplitude of magnetic anomalies
CHAPTER 3. FADING MAGNETIC ANOMALIES, LITHOSPHERIC FLEXURE AND REJUVENATED HYDROTHERMALISM OFF THE JAPAN-KURIL SUBDUCTION ZONE
3.3 Data and methods
3.4 Magnetic structure of the subducting plate before entering the japan-kuril trench
3.5 Decay of magnetization before subduction: alteration as the main process
3.6 An integrated magnetization model of the subducting plate
CHAPTER 4. THE TWO CATEGORIES OF FADING MARINE MAGNETIC ANOMALY IN SUBDUCTION ZONES
4.3 Data and methods
4.3.1 Selection of study area
4.3.2 Magnetic data collection and processing
4.3.3 Basement geometry data and processing
4.3.4 Remaining amount of magnetization
4.4.1 Magnetization loss in old oceanic lithosphere
4.4.2 Magnetization loss in young oceanic lithosphere
4.5.1 Comparison betwee the two categories
18.104.22.168 Before subduction
22.214.171.124 After subduction
4.5.2 Discussion for the other subduction zones
4.7 Supplementary materials
4.7.2 West Aleutian
4.7.3 Central Aleutian
4.7.3 South Alaska
CHAPTER 5. GENERAL CONCLUSIONS
5.1 New high-resolution magnetic anomaly map
5.2 Decaying magnetic signals in subduction zones