Relationship Between the High-Amplitude Magnetic Anomalies and Serpentinized Fore-Arc Mantle in the Cascadia Subduction Zone

A zone of significant high-amplitude magnetic anomalies is observed without a comparable gravity high along the Cascadia margin and is spatially correlated with the low-velocity fore-arc mantle wedge. This wedge is interpreted to be serpentinized fore-arc mantle and is further considered to be the main source of the high-amplitude magnetic anomalies. To test this hypothesis, the magnetization-density ratio (MDR) is estimated along the Cascadia margin to highlight the physical characteristics of serpentinization (reduced density and increased magnetization). Interestingly, high MDR values are found only in central Oregon, where slab dehydration and fore-arc mantle serpentinization (50–60% serpentinization) are inferred in conjunction with sparse seismicity. This result may indicate either poorly serpentinized fore-arc mantle (low degree of serpentinization) or that the fore-arc mantle is deeper than the Curie temperature isotherm for magnetite in northern and southern Cascadia. This finding means that magnetic anomaly highs and serpentinized fore-arc mantle may not always be correlated in subduction zones. On the other hand, the MDR pattern suggests segmentation of the Cascadia subduction zone, which is consistent with several previous geological and geophysical observations.


Introduction
Serpentinization of fore-arc mantle is a common feature that has been observed in many subduction zones worldwide and plays an important role in dynamic subduction processes (Bostock et al., 2002;Brocher et al., 2003;Chou et al., 2009;Doo et al., 2020;Hyndman & Peacock, 2003;Xia et al., 2008;Zhao et al., 2001).In subduction zones, the subducting oceanic crust generally contains free water (Fujie et al., 2018;Ranero et al., 2003;Shillington et al., 2015).As the temperature and pressure increase, large volumes of fluids are released, subsequently upwelling into and hydrating the overlying mantle (Kirby et al. 1996;Peacock et al., 2002;Hyndman & Peacock, 2003).Serpentinization reduces seismic velocities and densities as well as increases the intensity of magnetization (Christensen, 1966(Christensen, , 2004;;Horen et al., 1996;Hyndman & Peacock, 2003).The serpentinite in the fore-arc region is thus often characterized by very high magnetic susceptibility simultaneously with low density, which can result in relatively high magnetic and low gravity anomalies.
The Juan de Fuca plate is subducting obliquely northeastward beneath North America at a rate that ranges from 33 mm/year at the southern part of the subduction zone to 43 mm/year at the north (Fig. 1; Wilson, 1993).In central Oregon, Bostock et al. (2002) found a very low seismic velocity structure beneath the fore-arc region from a shear-wave velocity profile (black dashed line shown in Fig. 1) and interpreted the anomalies as serpentinized forearc mantle.In addition, according to seismic observations, Brocher et al. (2003) presented that weak or missing PmP North American plate reflections and low upper-mantle velocities mark a narrow zone along the Cascadia margin from southern British to northern California.They further proposed that serpentinization of the fore-arc mantle is the most likely explanation for their observations.Several seismological studies (Abers et al., 2009;Audet et al., 2009;Delorey & Vidale, 2011;Delph et al., 2018;Nicholson et al., 2005;Van Wagoner et al., 2002;Zhao et al., 2001) have similarly reported that serpentinization of the fore-arc mantle associated with slab dehydration is widespread along the Cascadia margin.
A global magnetic anomaly map (Meyer et al., 2017) reveals a roughly continuous zone of significant high-amplitude magnetic anomalies along the Cascadia subduction zone from 48.5°N to 43.5°N (Fig. 2).The cause of this prominent magnetic anomaly high has been interpreted as (1) Miocene granodiorite intrusions in western Cascadia (Finn, 1991), (2) the accreted basalt basement of the fore-arc (Wells et al., 1998), or (3) serpentinization of the fore-arc mantle (Blakely et al., 2005).To delineate the locations of the relevant magnetic sources, I apply the reduction to the pole (RTP) method to center magnetic anomalies over the causative bodies.The inclination and declination are set to 67.59°and 16.26°, respectively, in the study area [based on the International Geomagnetic Reference Field (IGRF) 11 model].The RTP magnetic anomaly is shown in Fig. 3a.Interestingly, this highly magnetic region has no comparable Bouguer gravity high (Fig. 3b; Bonvalot et al., 2012).Based on the results of Bostock et al. (2002), through 2D gravity and magnetic modeling processing, Blakely et al. (2005) proposed that these high-amplitude magnetic anomalies could be caused by serpentinized fore-arc mantle.Considering the characteristics of serpentinization, Blakely et al. (2005) further suggested that magnetic data are useful in mapping hydrated mantle in convergent margins, which seems feasible and reliable.To figure out the relationship between the high-amplitude magnetic anomalies and serpentinized fore-arc mantle in the whole Cascadia margin [the survey area of Bostock et al. (2002) only located covers central Oregon], a combined magnetic and gravity data analysis method leveraging the magnetization-density ratio (MDR; Doo et al., 2009) was used in this study.The advantage of using the MDR is that it can strongly highlight the characteristics of serpentinization (reduced density coincident with increased magnetization).

Segmentation of the Cascadia Subduction Zone
The Cascadia margin is a convergent plate boundary where a relatively young (\ 15 Ma), warm, and thin plate is subducting beneath North America (Fig. 1).Notably, the entire subduction zone, including the subducting and overriding plates, is segmented north to south, revealing lateral variations in several geophysical and geological characteristics.The subducting plate consists of the main Juan de Fuca plate and smaller plates to the north and south: the Explorer and Gorda plates, respectively.In the overriding plate, the along-strike topography of the fore-arc region can be divided roughly into the Washington, Oregon, and Sierra Nevada segments from north to south, with flat and low-lying topography in the center and higher elevations to the north and south (Wells et al., 1998).In addition, the subducting slab and overriding North American crust appear roughly segmented into three parts according to the results of ambient noise tomography (Porritt et al., 2011).Variations in the distribution of seismicity also reflect this segmentation: seismicity occurs mainly in the northern and southern segments of the Cascadia subduction zone and sparsely in the central part (Chen et al., 2015).On the other hand, the non-volcanic tremor density reveals similar features (much higher in the northern and southern segments along the Cascadia margin) (Delph et al., 2018;Wells et al., 2017).Overall, northern and southern Cascadia exhibits roughly similar geophysical and geological features, but central Cascadia exhibits different characteristics (Fig. 4).Although no great earthquakes (M [ 8) have been detected instrumentally along the Cascadia margin (Ji et al., 2017) and even though the structural characteristics are segmented, the length of the subduction zone (approximately 1100 km) is sufficient to generate magnitude 9 earthquakes (Goldfinger et al., 2012), and the seismogenic zone is considered to be fully locked (Wang & Tre ´hu, 2016).According to an analysis of marine turbidites, the recurrence interval of large earthquakes in the Cascadia margin is approximately 500 years (Goldfinger et al., 2012), and the most recent great megathrust earthquake (M 9) occurred in 1700 (Satake et al., 1996).Considering the time span between this event and the present day, it is essential to better understand the background tectonic characteristics, which could be beneficial for hazard assessment and risk mitigation.

Magnetization-Density Ratio (MDR) Estimation
Gravity and magnetic data can provide basic information on Earth's structures, but these data are usually processed and interpreted separately.The Poisson theorem (Poisson, 1826) provides a simple relationship between gravity potential (U g ) and magnetic potential (U m ) such that they can be interpreted together (Chandler & Malek, 1991;Doo et al., 2009;Garland, 1951;Hildebrand, 1985;Mendonca, 2004).Based on the Poisson theorem and characteristics of analytic signal (Hsu et al., 1996;Roest et al., 1992), Doo et al. (2009) provided a method to calculate the MDR of subsurface materials directly using magnetic and gravity anomalies.The formula is is the amplitude of the simple analytic signal of magnetic data (MAS 0 ), and is amplitude of the first order analytic signal of the gravity data (GAS 1 ).
Where G is the universal gravitational constant, Dq is the density contrast at the point within the source, and DM is the magnetization contrast at the same position.
Using this method, I do not need to obtain the magnetization and density of the subsurface materials individually.Thus, the MDR value is useful and convenient for establishing geophysical and geological interpretations (Chandler & Malek, 1991;Cordell & Taylor, 1971;Doo et al., 2009;Matos & Mendonca, 2020).The serpentinization of mantle peridotite increases the intensity of magnetization and reduces the density (Hyndman & Peacock, 2003;Saad, 1969).Therefore, in general, if the high-amplitude magnetic anomalies along the onshore part of the Cascadia subduction zone (Fig. 2) are the result of a serpentinized mantle wedge, the high MDR values should theoretically be pervasive along the Cascadia margin.
According to their definition (Doo et al., 2009), to obtain the MDR values, first, I estimate the zero-order analytic signal of the magnetic anomaly (MAS 0 ) and the first-order analytic signal of the Bouguer gravity anomaly (GAS 1 ), separately.Then, the MDR values of the study area are determined based on the amplitudes of MAS 0 and GAS 1 .The result is shown in Fig. 5.

Relationship Between the High-Amplitude
Magnetic Anomalies and Serpentinized Fore-Arc Mantle Bostock et al. (2002) identified a highly serpentinized fore-arc upper mantle beneath central Oregon from a dense array of broadband seismometers.In addition, on the basis of thermal modeling results, they pointed out that most of the serpentinized mantle wedge is at depths shallower than the Curie isotherm of magnetite in central Oregon.Along the same profile, two obvious peaks of the MAS 0 appeared in the boundaries (horizontal) of the serpentinized forearc mantle, which indicate that the cause of the major magnetization contrasts is serpentinized fore-arc mantle in this area.In addition, high MDR values show immediately above the serpentinized fore-arc mantle (Fig. 6).Overall, result of Fig. 6 is consistent with my initial idea and could be used as a typical example.The MDR method is useful in highlighting the feature (high degree of serpentinization and suitable temperature) observed by Bostock et al. (2002).High MDR values in the central Cascadia fore-arc region (Fig. 5) may indicate that the serpentinized fore-arc mantle is responsible for the highamplitude magnetic anomalies in this region.In contrast, the low MDR values in Washington seem inconsistent with the initial concept of serpentinization; in other words, the high-amplitude magnetic anomalies observed in this segment may not originate primarily from serpentinized fore-arc mantle.Highamplitude magnetic anomalies should not always be interpreted to correspond to a serpentinized fore-arc mantle in subduction zones.Brocher et al. (2003) proposed that the fore-arc mantle along the entire Cascadia margin is serpentinized according to seismic interpretation and the results of earthquake tomography.However, as shown in Fig. 5, high MDR values are distributed only in central Oregon.According to Fig. 6, these high MDR values may delineate the region with serpentinized fore-arc mantle, similar to the feature proposed by Bostock et al. (2002).The degree of serpentinization of the fore-arc mantle, which is closely associated with the slab dehydration, could be a major factor for this segmented MDR pattern.For example, Xia et al. (2014) proposed that the degree of serpentinization of the fore-arc mantle in the Kyushu subduction zone is strongly heterogeneous, varying from 0 to 12%.A similar feature has been reported along the strike of the Cascadia subduction zone, where the degree of serpentinization varies.Zhao et al. (2001) interpreted 15-20% serpentinization of the upper mantle beneath the northern Cascadia subduction zone, and an upper mantle P-wave velocity of 7.6 km/s was obtained in Washington (Miller et al., 1997), which may indicate * 25% serpentinization (Christensen, 1966), while Bostock et al. (2002) proposed that the degree of serpentinization in central Oregon may be as high as 50-60%.According to these results, the degrees of serpentinization are higher in the central Cascadia and lower in the north and south.Chen et al. (2015) similarly proposed that the degree of serpentinization is heterogenous along the whole Cascadia fore-arc region.As mantle peridotite is serpentinized from 0 to 95%, the magnetic susceptibility increases by several orders of magnitude, and the degree of variation in the susceptibility is larger than that in the density (Saad, 1969).Consequently, the degree of serpentinization may strongly affect the variations in the MDR value and the observed magnetic anomalies.Considering the formation mechanism of forearc mantle serpentinization, the volume of water released from slab dehydration could be the major reason to influence the degree of serpentinization (Hyndman & Peacock, 2003).
Another possibility is that the hydrated fore-arc mantle is deeper than the Curie temperature isotherm; this would cause ferromagnetic materials to lose their permanent magnetism in the northern and southern Cascadia margin.Oleskevich et al. (1999) proposed that the mantle wedges in many subduction zones are cooler than the Curie temperature of magnetite from 2D thermal modeling.However, the Cascadia margin is warmer than other subduction zones (due to young age of the subducting slab).In addition, McCrory et al. (2014) proposed that the fore-arc mantle corner is shallow in Oregon (35-40 km) and deep beneath Washington (41-43 km).These differences in depth may cause temperature differences reaching several tens of degrees due to the geothermal gradient.Ji et al. (2017) also proposed that the temperature gradient changes from high in the north to low in the central region and increases again in the southern Cascadia subduction zone.In summary, the thermal state may also be a cause of the low MDR values found in the northern and southern Cascadia margin.However, in this study, there is insufficient evidence to distinguish which of these two factors is the major control.

Relationship Between Intraslab Earthquake
and Serpentinized Fore-Arc Mantle Slab dehydration and eclogitization are believed to be mechanisms that produce intraslab earthquakes (Kirby et al. 1996;Peacock et al., 2002;Preston et al., 2003;Blakely et al., 2005;Chen et al., 2015).From this perspective, water released from a subducted slab could lead to brittle failure within the slab and hydration of the overlying mantle.Accordingly, the extent of serpentinized fore-arc mantle should spatially correlate with the distribution of intraslab earthquakes.Blakely et al. (2005) further suggested that magnetic anomaly highs in subduction zones could also be spatially correlated with intraslab earthquakes.As shown in Fig. 5, the seismicity in northern and southern Cascadia is clustered, while seismicity is lacking in the central segment, where high MDR values have been presented herein.The high MDR values calculated in this study may delineate the region with a high degree of serpentinization proposed by Bostock et al. (2002) and may further indicate that the volume of released water (the magnitude of slab dehydration) is larger in this segment than that in northern and southern Cascadia.In Fig. 5, the spatial correlation between the distribution of intraslab earthquakes and serpentinized fore-arc mantle (high MDR values region) seems to vary if I consider the information mentioned above [e.g., similar features are observed in the southern Alaska subduction zone by Blakely et al. (2005)].Consequently, the proposed slab dehydration mechanism seems insufficient to explain the occurrence of intraslab earthquakes in the Cascadia subduction zone.
Previous studies (Bilek, 2009;Doo et al., 2020;Wagner et al., 2005;Wang & Tre ´hu, 2016) have suggested that along-strike variations in the earthquake distributions in subduction zones may be attributable to several factors, including the thermal structure of the incoming plate, degree of slab dehydration, plate coupling state, curvature of the subducting slab, and subducting features (barriers or asperities remain controversial).To explain the distribution of seismicity in the Cascadia subduction zone, for instance, Brocher et al. (1994) proposed that changes in plate stress and/or plate geometry may be the cause of the paucity of earthquakes in Oregon.Alternatively, Blakely et al. (2005) postulated that the highly serpentinized fore-arc mantle beneath Oregon reduces friction on the downgoing slab, thereby explaining the scarcity of earthquakes.McCrory et al. (2012) suggested that the subducting Juan de Fuca b Figure 5 Magnetization-density ratio (MDR) value distribution map of the Cascadia margin.Black dashed line indicates the location of a shear-wave velocity model provided by Bostock et al. (2002).Purple shadow pattern shows location of magnetic anomalies of high amplitude.The triangles denote volcanoes.Earthquakes occurred between 1900 and 2019 (USGS database); the size of the spheres indicates the magnitude of the events slab appears to bend at a slightly steeper angle in central Cascadia than in the northern and southern parts.Ji et al. (2017) considered that the sparse seismicity in central Cascadia is attributed to an unusually warm segment of the subducting slab.Gao (2018) proposed that a highly hydrated oceanic lithosphere can significantly reduce its strength, which could reduce the seismicity in central Cascadia.Summarizing all of the above concepts, both the properties and the geometries of the subducting plate seem to play important roles in the different distribution densities of earthquakes and segmented features throughout the Cascadia subduction zone.
The MDR distribution pattern also reflects this segmentation and correlates well with the distribution of seismicity (Fig. 5).These likely factors may be related.However, the actual mechanism affecting the distribution of seismicity in Cascadia remains an open question and requires further investigation in the future.

Conclusions
I estimate the MDR along the Cascadia margin to investigate the relationship between the high-amplitude magnetic anomalies and low gravity anomalies potentially associated with serpentinized fore-arc mantle.The high MDR values observed in the Cascadia margin are interpreted here to highlight the significant characteristics of the serpentinized mantle wedge (Hyndman & Peacock, 2003).Figure 6 could represent the typical example (high degree of serpentinization and suitable temperature).Contrary to expectations, high MDR values are segmented in the fore-arc region rather than along the entire Cascadia margin: the MDR values are higher in the central region and low in the north and south.According to the MDR results (Fig. 5), I propose that the serpentinized mantle wedge contributes to the highamplitude magnetic anomalies only in central Cascadia.Serpentinized fore-arc mantle has been thought to be ubiquitous in subduction zones worldwide.However, this feature may not necessarily result in high-amplitude magnetic anomalies due to the degree of serpentinization and/or the background thermal status.In other words, high magnetic anomalies in subduction zones may not inevitably result from a serpentinized mantle wedge (or vice versa).In addition, several previous studies have proposed that slab dehydration and embrittlement reactions are mechanisms that produce intraslab earthquakes.However, several significant features, including high MDR values, high degrees of serpentinization in the forearc mantle (Bostock et al., 2002), and corresponding rare intraslab earthquakes, have been observed in central Cascadia.Integrating all these observations and estimation results, I propose that the high-amplitude magnetic anomalies, the serpentinized forearc mantle, and the occurrence of intraslab earthquakes in subduction zones do not necessarily have a positive correlation.

Figure 1 A
Figure1A regional map(Amante & Eakins, 2009) of the Cascadia subduction zone showing boundaries of tectonic plates.Black arrows illustrate the motion of the Juan de Fuca and Gorda plates regarding the North American plate.Pink triangles denote active volcanoes of the Cascadia magmatic arc.Black dashed line indicates the location of a shear-wave velocity model provided byBostock et al. (2002)

Figure 2
Figure 2Magnetic anomaly map of the Cascadia margin(Meyer et al., 2017).Pink triangles denote active volcanoes of the Cascadia magmatic arc.White dashed line indicates the location of a shear-wave velocity model provided byBostock et al. (2002)

Figure 3 a
Figure 3 a Reduction to the pole magnetic anomaly map of the Cascadia margin.b Bouguer gravity anomaly map (Bonvalot et al., 2012) of the Cascadia margin.Yellow shadow pattern shows location of magnetic anomalies of high amplitude.Pink triangles denote volcanoes.White dashed line indicates the location of a shear-wave velocity model provided by Bostock et al. (2002)

Figure 4
Figure 4 Epicentral distribution of 1900-2019 earthquakes (USGS database) in the Cascadia subduction zone.The pink triangles show the Cascadia volcanoes.Orange dashed lines indicate segmentation boundaries of the Cascadia subduction zone [identified from Delph et al. (2018)]