Lithosphere Weakening During Arctic Ocean Opening: Evidence From Effective Elastic Thickness

Abstract

Evolution of the Arctic Ocean lithosphere has involved multiple stages of opening with crustal stretching and thinning prior to lithospheric breakup. Mapping lateral variations in lithospheric rheology can help unravel the detailed tectonic history of the Arctic. Here we perform a wavelet analysis of gravity and bathymetry data to map the effective elastic thickness () of the lithosphere in the Arctic. The low overall suggests that large shear stresses and serpentinization weakened the lithosphere at Arctic passive margins during a multistage opening process. Moderately low values along the Gakkel Ridge imply a relatively cold ultraslow-spreading center compared to typical mid-ocean ridges.

Key Points

• The Arctic effective elastic thickness (T_e) map is obtained by spectral analysis

• Sedimentary correction for vertical density variation improves the T_e estimation

• Pre-breakup lithospheric extension caused low T_e along passive margins

Plain Language Summary

Geological exploration and sampling in the Arctic Ocean are difficult due to sea ice cover and frigid weather. Gravity and bathymetry data obtained using remote sensing methods can be used to estimate the effective elastic thickness () of the lithosphere, which is a proxy for the strength of the tectonic plates. We draw a new map of the Arctic Ocean by combining the latest data with recent crustal structure models. is low on the edge of the oceanic basins, suggesting that the lithosphere was weakened by mechanical rifting and mantle metamorphism in the opening of the Arctic Ocean. The seafloor spreading centers are generally thought to be hot and weak, but the Gakkel Ridge shows moderately low , revealing unique mechanical and geothermal properties of this slowest spreading center in the world.

1 Introduction

The Arctic Ocean is characterized by two oceanic basins, the Eurasia and the Amerasia Basin (Figure 1a), which are separated by the Lomonosov Ridge, a remant of continental crust (Gaina et al., 2014). Spreading is currently only active in the Eurasian Basin at the ultraslow spreading (8–13 mm/yr) Gakkel Ridge (Dick et al., 2003). Debates on the tectonic reconstruction in the Amerasia Basin are centered on the nature of the crust and its relation to continental rifting (Nikishin et al., 2019), which began as early as 195 Ma (Grantz et al., 2011) and was followed by opening of the Canada Basin at 135–130 Ma (Hadlari et al., 2016). To the north, extensive volcanic overprinting by the Alpha-Mendeleev Ridge complex, part of the High Arctic Large Igneous Province (HALIP), hampers accurate tectonic reconstruction of the Amerasia Basin (Døssing et al., 2013). The Alpha-Mendeleev Ridge is interpreted as either (a) a thick oceanic plateau (e.g., Funck et al., 2011; Grantz et al., 2011; Jokat, 2003), (b) a volcanic province with possible fragments of extended continental crust (Bruvoll et al., 2012), or (c) substantially altered and under plated continental material (Lebedeva-Ivanova et al., 2006). Between the Alpha-Mendeleev Ridges and the Lomonosov Ridge, the oceanic Makarov Basin may represent the continuation of the Canada Basin formed during the late Cretaceous (Døssing et al., 2017), with anomalously thick oceanic crust (Sorokin et al., 1999).

Figure 1

(a) Bathymetry (data from Jakobsson et al., 2012) and (b) free-air gravity anomaly (data from Andersen et al., 2010) of the Arctic Ocean. AR = Alpha Ridge; BS = Barents Sea; CB = Canada Basin; ChB = Chukchi Borderland; CS = Chukchi Sea; ESS = East Siberian Sea; FJI = Franz Josef Land; GR = Gakkel Ridge; KS = Kara Sea; LR = Lomonosov Ridge; LS = Laptev Sea; MB = Makarov Basin; MR = Mendeleev Ridge; PB = Podvodnikov Basin; SV = Svalbard.

Understanding the nature of the crust and its mechanical properties is key to deciphering the mechanisms of Arctic opening for the last 200 million years (Gaina et al., 2014). However, due to the thick sea-ice cover, mapping the geological structure and makeup of Arctic Ocean lithosphere is difficult. Geophysical data and models are therefore commonly used to constrain the tectonic structure and evolution in the Arctic (e.g., Døssing et al., 2013; Lebedeva-Ivanova et al., 2019). The effective elastic thickness () of the lithosphere is a proxy for long-term lithospheric strength and can be used to understand the thermo-mechanical structure of the lithosphere. corresponds to the thickness of an idealized elastic plate that would produce equivalent flexure under observed geological loading (Watts, 2001). In oceanic domains, has been shown to reflect the plate age at the time of loading, and to some degree crustal age and thermal structure (Lu et al., 2021; Kalnins & Watts, 2009; Watts, 1978; Watts & Burov, 2003).

can be modeled using lithospheric strength profiles calculated using empirical constitutive flow laws constrained by lithospheric temperature estimates (e.g., Burov & Diament, 1995; Brown & Phillips, 2000; Tesauro, Kaban, & Cloetingh, 2012). An Arctic map has recently been calculated using this approach (Struijk et al., 2018), where patterns mimic the predicted crustal age-temperature relation in the ocean. Alternatively, can be estimated by calculating cross-spectral properties between gravity and bathymetry data (e.g., Audet, 2014; Kirby & Swain, 2009) and inverting them using a loading model of the lithosphere. Comparing the results from these two approaches can yield important insight into the various factors controlling lithospheric strength (Tesauro, Audet, et al., 2012) and help determine the nature of the lithosphere in the Arctic. In this study, we obtain the first estimated map of the Arctic Ocean from the cross-spectral approach and use it to investigate the relationship between and tectonics of the Arctic Ocean.

2 Estimation

2.1 Method

We estimate over the Arctic Ocean by inverting the real-valued admittance between free-air gravity anomaly and bathymetry. The admittance is defined as:

(1)

where and are the spectra of 2D gravity anomaly and bathymetry data; the asterisk denotes complex conjugation; and indicate the one- and two-dimensional wavenumbers, respectively, and ; the triangular brackets denote an azimuthal averaging procedure. In this study, we use a continuous wavelet transform (CWT) with a Fan wavelet (Kirby, 2005) to calculate the admittance function. The CWT convolves a set of scaled wavelets with the data over the whole study area and allows determination of the admittance function at every point of the grid (Audet, 2014; Kirby & Swain, 2009). The Fan wavelet is calculated from a superposition of Morlet wavelets with different azimuths spanning 180°. The central wavenumber of the Morlet wavelet determines the spatial resolution of the resulting map; high- have poor spatial resolution but more reliable estimates and vice versa (Kaban et al., 2018; Kirby & Swain, 2009). The central wavenumber used in this study is 5.336.

The admittance function is modeled using the flexural equation of a thin elastic plate under initially uncorrelated surface and subsurface loading (Forsyth, 1985). Subsurface loading is emplaced at the expected largest density contrast boundary (i.e., the Moho). The subsurface-to-surface load ratio is described as . varies from 0 to 1; indicates that the surface loading is dominant, and implies the dominance of subsurface loading (McKenzie, 2003). and are estimated using a non-linear least squares minimization technique, where the misfit between observed and predicted admittance is calculated by a reduced chi-squared criterion (Audet, 2014):

(2)

where refers to the measured sample of real admittance with variance , indicates the predicted admittance, and denotes the number of wavenumber samples. The predicted admittance is calculated using analytical expressions (see Audet, 2014, for details), given our parameterization of lithospheric structure and loading. Error in is obtained from the diagonal elements of the error covariance matrix.

2.2 Input Data and Correction for Sedimentary Structure

The bathymetry data (Figure 1a) are obtained from the International Bathymetric Chart of the Arctic Ocean (IBCAO) v3.0 (Jakobsson et al., 2012), and the gravity data (Figure 1b) are obtained from the global gravity model DTU2010 (Andersen et al., 2010). We prefer the IBCAO bathymetry model because it is produced by interpolating shipboard measurements, rather than being derived from gravity data, which therefore guarantees the independence of the bathymetry and gravity datasets in the cross-spectral analysis. Crustal thickness (Figure 2a) and density (Figure 2b) data are extracted from the ArcCRUST (Lebedeva-Ivanova et al., 2019) model. We note that using a different sedimentary thickness model (e.g., Straume et al., 2019) did not change our final results. Crustal density ranges from 2.75 to 2.885 g/cm³ (Lebedeva-Ivanova et al., 2019), where the maximum and minimum values correspond to typical oceanic and continental crust, respectively (Figure 2b).

Figure 2

(a) Thickness of the crystalline crust, (b) crust density variation, and (c) sedimentary thickness from the ArcCRUST model (Lebedeva-Ivanova et al., 2019). The inset in (c) shows the relationship (Equation 3) between the vertical-integrated sedimentary density () and sedimentary thickness (). (d) The correction value for sediment effect, where is the corrected sedimentary thickness.

The seafloor of the Arctic Ocean is covered by a thick layer of low-density sediments (Petrov et al., 2016) (Figure 2c), which, if ignored, can affect the accuracy of estimates (Chen et al., 2015; Kaban et al., 2018; Ratheesh-Kumar & Xiao, 2018; Shi et al., 2017). To integrate the sedimentary layer into the loading model, we seek to estimate a correction to the bathymetry data that accounts for the added layer of sediments. We first calculate the vertical average of sediment density, , defined as:

(3)

where is the sedimentary layer thickness (Figure 2c) and is an exponential density-depth curve following Cowie and Karner (1990):

(4)

 = 2.65 g/cm³ is the sediment grain density, = 1.84 g/cm³ is the sedimentary density on the seafloor, = 0.4 km⁻¹ is the sedimentary compaction parameter (Lebedeva-Ivanova et al., 2019) and is depth below seafloor. The equivalent sedimentary load on the crystalline basement is given by:

(5)

where, is the crustal density (Figure 2b), = 1.03 g/cm³ is the density of sea water, and is the corrected sedimentary thickness. In our equivalent load correction, the reduction of sedimentary thickness is added to the water column, thus the corrected bathymetry () is given by:

(6)

where is original bathymetry (Figure 1a), and is the correction for both sedimentary thickness and bathymetry (Figure 2d).

3 Results

Results of and error estimated from both the original and the corrected bathymetry data are shown in Figure 3. Ignoring the effects of the sedimentary loading leads to anomalously high (>50 km) where sediment thickness is greatest, as well as areas with high error (Figures 3a and 3c). Applying the bathymetric correction reduces overall error and leads to sharper patterns (Figures 3b and 3d).

Figure 3

Spatial variations in of the lithosphere in the Arctic region: (a) without sediment correction; (b) with sediment correction; (c) and (d) are the error in estimates for (a) and (b), respectively. The color scale in (a) and (b) saturates at 50 km (black areas). The contours in (a) and (c) are sediment thickness (in km). Abbreviations as in Figure 1.

The corrected map (Figure 3b) further shows a better correlation with the tectonic features compared to the uncorrected map, especially on the continental shelves. values in the Barents-Kara sea are low in the west and high in the east, consistent with lithosphere-asthenosphere boundary and Moho temperature models that suggest weaker lithosphere in the west than in the east, with a sharp contrast at the location of the Franz Josef Land (Klitzke et al., 2015, 2016). In the North Chukchi Sea Basin, the thinned crust (Figure 2a) with low coincides with several rifting phases in the Aptian-Albian (125–100 Ma) and in the Cenozoic (45–37 Ma) (Nikishin et al., 2020). Interestingly, the western Gakkel Ridge is characterized by a thin crust (∼5 km) but a much higher (10–15 km) than other mid-ocean ridges in the world (e.g., Lu et al., 2021). Cannat (1996) proposed that the melts of slow-spreading ridges extracted from the asthenosphere crystallize in the mantle before they reach the crust. This would explain the thick axial lithosphere with sparse magmatism and cool mantle temperature (Cannat, 1996; Schlindwein & Schmid, 2016). Recent global Curie-point depth and crustal thickness models (Li et al., 2017; Zhou et al., 2020) also imply a relatively cold ultraslow spreading Gakkel Ridge.

4 Weak Lithosphere of Arctic Passive Margins

4.1 Transitional Lithospheres

Along the Arctic passive margins, low values are prominent (Figure 3b). In particular, is ∼10 km in areas of early pre-breakup extension around the Chukchi Borderland (two stages, ∼195–160 Ma and ∼145.5–140 Ma, Grantz et al., 2011), and no more than 5 km in regions of late extension of the Makarov Basin and the Eurasia Basin margin (∼69–57 Ma and ∼56 Ma, Døssing et al., 2017; Minakov et al., 2012). We speculate that the low at Arctic passive margins reflects pre-breakup extension during the Arctic multi-stage opening process, where rifting-related lithospheric extension and serpentinization might have played a leading role in reducing the strength of the lithosphere in these regions. The crustal gravity model across the Northwind Ridge (western part of the Chukchi Borderland) to the Canada Basin reveals a thin serpentinized peridotite layer (∼3 km) in the ocean-continent transition (Grantz et al., 2011). Seismic profiles and gravity models indicate that continuous spreading thinned the serpentinized mantle and subsequent normal faulting produced basement blocks of the Eurasia Basin margins (Lutz et al., 2018). The presence of only ∼10% of serpentine dramatically reduces the strength of the oceanic lithosphere and is conducive to the formation of low-angle faults (Escartín et al., 2001). However, the Makarov Basin and the Podvodnikov Basin are not well studied, and the nature of the crust (oceanic or continental) is ambiguous (Døssing et al., 2013; Sorokin et al., 1999). Nevertheless, these basins are the result of the Late Cretaceous-Cenozoic extension between North America and Eurasia (Gaina et al., 2014), and the weak crust and uppermost mantle inferred from the low values might be caused by large-scale extension that led to the breakup from the Lomonosov Ridge.

Similarly low values have also been observed at other rifted margins worldwide. At the India-Madagascar conjugate passive margins, is lower than 5 km (Ratheesh-Kumar et al., 2015). However, low at the India-Madagascar margins may be due to a combination of lithospheric stretching during the early stage of continental breakup dating back to ca. 90 Ma, and subsequent hotspot-related thermal weakening during the drift stage at ca. 65 Ma. The Red Sea and the Gulf of Aden rifting zones also display values lower than 5 km (Chen et al., 2015), but these low values may be the result of pre-existing high lithospheric temperature rather than weakening due to mechanical stretching during breakup. In the Arctic, we can probably rule out these weakening effects, because seafloor spreading in the Canada Basin stopped about 127.5 Ma ago (Grantz et al., 2011) and the High Arctic Large Igneous Province magmatic event (∼122–125 Ma) caused by a mantle plume (Døssing et al., 2017) is located away from the Chukchi Borderland passive margins and is older than the breakup of the Makarov Basin.

4.2 Submerged Microcontinents

The Chukchi Borderland and the Lomonosov Ridge are microcontinents that experienced multiple stages of extension and crust thinning and are characterized by lower (∼15–20 km) than the crustal thickness (>25 km). This situation can arise due to the mechanical decoupling between the crust and lithospheric mantle, characterized by a weak lower crustal layer that cannot propagate the loading-induced elastic stress across the plate, which significantly reduces (Burov & Diament, 1995; Steffen et al., 2018). Weakening at lower crustal depth is primarily driven by high lithospheric temperatures, but also by large tectonic stresses (Brown & Phillips, 2000; Burov & Diament, 1995). Unfortunately, there are no surface heat flow data in the Chukchi Borderland (Ruppel et al., 2019); however, relatively high Curie-point depth values (>30 km; Li et al., 2017) do not point to high temperatures at lower crustal depth. Furthermore, surface heat flow measurements from the Lomonosov Ridge are not notably higher than predictions for moderately stretched continental crust (Shephard et al., 2018). Instead of high lithospheric temperatures, we propose that mechanical decoupling is produced by large shear stresses associated with lithospheric stretching that occurred in these micro continents. The Chukchi Borderland rifted from the Canadian Arctic shelf (∼195–160 Ma) and was later rotated clockwise away from the East Siberian Shelf into the Canada Basin prior to ∼ 145.5–140 Ma (Grantz et al., 2011). Similarly, the Lomonosov Ridge separated from the Alpha-Mendeleev Ridges in the Late Cretaceous and then rifted away from the northern Barents Sea during the Paleocene (Brozena et al., 2003). Large tectonic shear stresses produce slip at the crust-mantle boundary, resulting in distinct bending stress distributions in each layer (Brown & Phillips, 2000), and can lead to mechanical decoupling. In other rifted-related weak continental lithosphere, such as the Ethiopian and East African rifts, reduction is proportional to the amount of extension (Pérez-Gussinyé et al., 2009).

Unlike the Chukchi and Lomonosov micro continents, the Alpha-Mendeleev Ridge is interpreted as a submerged block of continental crust affected by intraplate volcanism and under plating (Bruvoll et al., 2012). Buchan et al. (2006) attributed the related magmatic activities to a mantle plume (135–75 Ma). Thermal weakening due to a mantle plume has been suggested in central Greenland (Steffen et al., 2018), where the is slightly lower than the Moho depth, similar to the Alpha-Mendeleev Ridge. The maps suggests that the mechanical strength of the Mendeleev Ridge lithosphere is higher than at the Alpha Ridge. A thick (7–8 km) complex of underplated magmatic material is inferred below the middle crust of the Alpha Ridge (i.e., below 12–13 km), where the normal lower crustal layer is elusive, but the Mendeleev Ridge has a strong lower crust (20 km) (Funck et al., 2011; Lebedeva-Ivanova et al., 2006; Petrov et al., 2016).

5 Conclusion

In this study, we estimate the spatial variations in the effective elastic thickness () of the lithosphere over the Arctic region from the inversion of the real free-air admittance between bathymetry and free-air gravity anomaly data using a continuous wavelet transform. We implement a bathymetric correction to account for the loading effect of sediments and show how it improves estimation in the Arctic. The estimated values range from 0 to 50 km and patterns correspond well to known tectonic features. We find relatively high (∼10–15 km) at the western Gakkel Ridge, which suggests a cool and thick lithosphere at this ultraslow and sparsely magmatic spreading center compared to typical mid-ocean ridges. We interpret widespread low values at Arctic passive margins as the result of lithospheric stretching-induced weakening during the multistage opening process. Large shear stresses and serpentinization at the time of loading might play prominent roles in lowering the mechanical strength of the lithosphere at these passive margins.

Acknowledgments

The authors would like to thank the Editor and two anonymous reviewers for their constructive comments. This research is funded by National Natural Science Foundation of China (Grant Nos. 41761124051, 91858213, and 41776057) and Natural Sciences and Engineering Research Council of Canada (RGPIN-2018-03752).

Open Research

Data Availability Statement

All figures are drawn by GMT (Wessel et al., 2013). The International Bathymetric Chart of the Arctic Ocean (IBCAO) v3.0 (Jakobsson et al., 2012) is available at https://www.gebco.net/about_us/committees_and_groups/scrum/ibcao/ibcao_v3.html. The gravity model DTU2010 (Andersen et al., 2010) is available at http://www.space.dtu.dk. The ArcCRUST (Lebedeva-Ivanova et al., 2019) model is available at https://doi.org/10.1594/PANGAEA.899841. The wavelet analysis and elastic thickness estimation were done using the open-source software PlateFlex (Audet, 2019).