New model for ultraslow-spreading ridges

New model for ultraslow-spreading ridges | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Physical Sciences - Article New model for ultraslow-spreading ridges Ståle Emil Johansen, Hans Amundsen, Børge Arntsen, Rune Mittet, and 4 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-3999138/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract The oceanic crust formed at mid-ocean ridges constitutes seventy percent of the earth's solid surface 1–3 . The crust in ocean basins is approximately seven km thick 4,5 , but when the spreading rate drops below 15-20 mm/a (ultraslow-spreading), crustal thickness decreases rapidly 6,7 . The paradigm view is that thickness depends on the spreading rate 6–8 . However, current models for the ultraslow-spreading ridges are not based on direct imaging 9 , and it is not well understood how tectonic processes, melt dynamics, 2,3,10,11 , lithospheric structure, and crustal formation 12–15 interact along ridges 12–14,16,17 . New electromagnetic (EM) data across the Mohns and Knipovich ridges show that, outside the volcanic centres, the lithosphere appears closed to melt migration, and instead of normal thinning, the lithosphere is unusually thick (35-45 km) beneath the ridges. Crustal thickness varies along the strike and is thinnest where the spreading rate is the highest, contrary to the prediction of conventional models 6–8 . In the new model, ambulatory volcanic centres, forming along weak zones, and fault-induced ultra-deep direct drainage of melt from the asthenosphere explain the EM data. Volcanic centres are point sources of melt supply and both centres, feeder channels and volcanism are episodic, relatively short-lived, and random in time and place. In this model with a thick and brittle lithosphere, the plate motions (rate and direction), local tectonics, lithology, weak zones, and deep faults associated with the deep drainage, control the development. Melt dynamics and crustal formation are passive buoyancy-driven responses to the tectonic development. The fact that the proposed ridge model is closely connected to fundamental tectonic processes support the idea that the model can also be applied to ultraslow-spreading ridges in general. Earth and environmental sciences/Solid Earth sciences/Geophysics Earth and environmental sciences/Solid Earth sciences/Geodynamics Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Away from fracture zones, hotspots, and marginal basins, the crust in ocean basins has a uniform thickness of approximately 7 kilometers 4,5 , and at fast and intermediate spreading rates, the melt supply may fully accommodate plate divergence. However, when the rate drops below 15–20 mm/a, the crustal thickness decreases rapidly 6,7 . The thin or absent crust is explained by the current models as reduced melt production or impeded melt migration due to conductive cooling 10,18–21 . Changes in ridge geometry, LAB topography, melt producing rock volumes, mantle composition, thermal structure and hydrothermal permeability are also believed to affect melt production and crustal thickness along the ultraslow-spreading ridges 10,14,18–20,22 . The ridges have been studied for decades, but present models are largely based on bathymetry, shallow seismic data, low resolution seismological models, geodynamic modeling, studies of ophiolites and not on direct deep imaging 9 . It is therefore still unclear how melt dynamics, 2,3,10,11 , lithospheric structure, lithology, and crustal formation 12–15 interact with the tectonic processes. Variations along ridge strikes, in particular, have received little attention and are poorly understood 12–14,16,17 . New electromagnetic data Technological advancements in electromagnetic (EM) methods have enabled the combination of high-quality controlled-source shallow electromagnetic imaging (CSEM) with low-frequency passive and deep magnetotelluric imaging (MT) 10,11 . This combined EM technique is particularly well suited for deep imaging below ridges, where melt and fluids are expected to cause considerable fluctuations in electrical resistivity 13,23 . The first deep EM data (ATLAB-1) 10 across an ultraslow-spreading ridge was collected by the Atlantic Lab. Consortium (ATLAB) and showed melt in a narrow and asymmetric feeder zone below the northern Mohns Ridge in the Norwegian Greenland Sea (NGS). This profile provided unique local insight, but to what degree this single transect is representative of the ridge as such is not clear. To address this problem ATLAB collected four new EM datasets along the Knipovich and Mohns Ridges. The ATLAB EM data base now consists of six unique data sets (three MT and three CSEM) (Fig. 1 ), collected during three distinct campaigns in the Norwegian-Greenland Sea (NGS). The ATLAB-1 data was reinverted for this study and the new CSEM model now images to 15 km depth compared to 8 km in the original version 10 . ATLAB-2 (this study) (Figs. 1ab) was acquired using 39 seafloor receivers along a 130 km long transect at the northernmost Knipovich Ridge, whereas ATLAB-3 (this study) was collected using 24 seafloor receivers along a 54 km long transect at the northernmost Mohns Ridge (Figs. 1 c and ED Fig. 1 0a). The ATLAB-2 transect crosses the Knipovich ridge before proceeding eastward across the continent-ocean boundary (COB) and onto continental crust west of Svalbard. ATLAB-3 was collected just south of the ATLAB-1 transect 10 to evaluate the potential changes over relatively short distances along the Mohns ridge. Using a nonlinear regularized inversion of the MT and CSEM data, smooth two-dimensional (2D) anisotropic electrical resistivity models were reconstructed from the datasets (see Methods). All surveys were optimized for local imaging across the ridges, but when combined they from a unique database that also opens for studying variation along the strike of the ridges. The crust The inverted ATLAB-2 CSEM-model (Fig. 2 a) shows a clear sub-horizontal three-layered subsurface at the northern Knipovich ridge, interpreted as a rigid oceanic lithospheric mantle overlain by oceanic crust, with a conductive sedimentary layer on the top. The rapid increase in conductivity (stippled in Fig. 2 a) is interpreted as a porosity threshold at the base of the lower crustal gabbro sequence. Drilling has never verified the Mohorovičić discontinuity (Moho), hence the coupling between geophysical and igneous ‘crust’ is not established for ultraslow-spreading ridges 8,24 . However, interpreting this level as the base of the oceanic crust yields a crustal thickness of ca. 3.5–5.5 km at the northern Knipovich Ridge, well below the global average of ca. 7 km 4,24 . The geometry of intra-crustal conductive anomalies is banana-shaped and are thought to represent the downward migration of seawater followed by lateral transport within extensively fractured sheeted dyke complexes. Both sediment and total crustal thickness increase towards the coast into the continental domain, reaching a total thickness of ca. 15–20 km at the outer Spitsbergen margin, which is approximately in line with earlier estimates of depth to Moho in this area 25 . The high conductivity below the shelf indicates that saline water penetrated deep into the continental crust in this area. Similar criteria were used to determine the base of the crust below at the northern Mohns Ridge as at ATLAB-2. The crust is significantly thinner here 10 , varying between 2.5 and 3 km (Fig. 2 b and ED Fig. 6). This is confirmed by the new CSEM data (ATLAB-3) collected 25 km south of ATLAB-1 (Fig. 2 c and ED Fig. 7). Other geophysical studies 21,26–28 have estimated average and approximate crustal thicknesses of 2–5 km along the Mohns and Knipovich ridges. Also at ATLAB-1 and ATLAB-3 the internal anomalous conductive zone in the crust is interpreted to be caused by saline fluids 10 . The western graben margin is uplifted, and the graben is characterized by substantial negative relief and faults extending deep into the upper mantle. The presence of active vent fields 29 and an inferred plumbing system suggest increased water circulation in this area 10 . Lithosphere, LAB, and upper asthenosphere Temperature, lithology, porosity, permeability, fluid content, and melt content is addressed when interpreting EM sections in terms of oceanic lithosphere. Of these factors, partial melting has the greatest effect on conductivity 10,30 . Using the same criteria as those used for the ATLAB-1 model 10 , the very low resistivities in the ATLAB-2 model (MT) is interpreted as concentration of melt in the mantle below LAB at approximately 50–80 kilometers depth. The electrical field shown in Fig. 3 a and ED Figs. 2 – 4 is perpendicular to the profile (transverse electric, TE mode), indicating a strong conductivity alignment with the N-S striking ridge and the area’s general N-S geological strike (Figs. 1ab) (See methods for details and interpretation strategies). Unlike the northern Knipovich ridge (ATLAB-2) the melt below northern Mohns Ridge (ATLAB-1) (Figs. 1 c, 2 b, 4 and ED Fig. 6) is concentrated along a narrow, oblique, and strongly asymmetric zone below the ridge, and the partly melted asthenosphere can be traced to the inferred depth of the Moho discontinuity. In both areas, the partially melted sections are enveloped by a resistivity contour of about 100 Ωm, establishing minimum melt content, interpreted as the base of the electrical lithosphere. This surface is also clearly visible at 30–50 km depth in the inverted model based on the ATLAB-3 data acquired just south of ATLAB-1 (Figs. 3 c, 4 and ED Fig. 8b). Above this surface there is a rapid transition to high resistivities typical of melt-free peridotite 10,31 . The surface appears as a permeability barrier preventing vertical melt movement and as a guide for lateral migration and is interpreted as the oceanic LAB within our study area. Because the 2D regularized inversion seeks the smoothest model, the real transition is probably significantly sharper than that seen in the inverted models (See also Methods for details). Geological models The basic assumption for oceanic crustal formation is that the thickness is proportional to the spreading rate 6,8,32 , and several models have been proposed to explain why the crust thins as the spreading rates decrease. More conventional models explain the thin crust by conductive cooling, which forms a thick thermal boundary layer 6 below the ridge that shuts off melt production (Model M1), or alternatively (Model M2), a comparable but much thinner lid that impedes vertical melt migration 18 . The third model (Model M3) predicts that available melt-producing rock volumes directly regulate crustal thickness 10 , whereas a fourth model (Model M4) proposes 19,33 that melt flows below the LAB topography to volcanic centers, explaining unequal crustal production. However, none of these single models account for all observations in the EM data along the Mohns and Knipovich Ridges. The northern Knipovich Ridge has thick lithosphere and Model M1 could explain the reduced crustal thickness at ATLAB-2. However, neither this model nor the M2 model, can then explain the much thinner crust in the melt-rich, slightly faster spreading ATLAB-1 area, with an open feeder system below the ridge (Figs. 2 , 3 , 4 ). Model M3 10 found a direct correlation between crustal thickness and the volume of melt available in the asthenosphere below the ridge. This model can explain the observations at ATLAB-1 but cannot easily explain the much thicker crust in the melt-poor, slower spreading area to the north (ATLAB-2). According to Model M4, melt can migrate below LAB along the ridges to volcanic centers, explaining the fluctuation in crustal thickness. This model does not rely on direct imaging of LAB or asthenosphere. It was extrapolated instead using shallower earthquake hypocenters 14 . The inverted EM models confirm that there is significant LAB topography below the ridges and that melt can migrate along strike. However, since melt is available at all EM locations, lateral migration is not necessary to explain the observations. Crust can form from local melt. Furthermore, the M4 model suggests that the crust is thickest where the asthenosphere is shallow. This contradicts the ATLAB models, which reveal that crustal thickness is not directly related to depth to LAB. New model Extension and large rotated fault blocks are the dominant tectonic styles (Figs. 1bc, 2 and ED 10ab) 34,35 spreading rates decreases northwards (15 − 12 mm/yr) and the crust is continuous along the ridges 21,26–28 , varying in thickness from ca. 2.5 to 5.5 km (Fig. 2 and ED Figs. 6,7). The crust is thickest in the north at ATLAB-2 (5.5 km) where the spreading rate is lowest. For an ultraslow-spreading ridge this is relatively thick crust 4,24 , but a feeder system connected to the deep melt is invisible and apparently currently inactive (Figs. 2 a, 3 a). Further south at ATLAB-3 where the spreading rates are higher the crust is much thinner, but a deep feeder system is not readily visible here either (Figs. 2 c, 3 c and ED Figs. 7, 8b, 9e). Just north of ATLAB-3, at ATLAB-1 the crust is also thin, but here a well-developed feeder system connects the melt at depth with the crust (Figs. 1 b, 2 b, 4 and ED Fig. 6). Along the Mohns Ridge, a series of AVRs appears along NE lineaments (weak zones) inside the graben, alternating with more quiet areas with individual volcanoes. This is also the case further north, but here the volcanic lineaments are more oblique to the graben owing to the N-S orientation of the Knipovich Ridge (Figs. 1 and ED Figs. 1 0 ab). Recent age dating shows that the lava flows at the rift valley floor is younger than predicted by conventional models, and typically not older than 180 Kyr 36 . Large areas with individual volcanoes outside the AVRs are younger than 25 Kyr and half of the AVRs are younger than 18 Kyr. The youngest volcanic events (8 Kyr) along the Mohns Ridge are found just south of ATLAB-3, between the AVR and the western boundary fault (Figs. 1 c and Fig. ED 10a). However, at the three ATLAB locations the volcanic activity at the sea floor and the crustal thickness seems to be independent of depth to conductive melt visible in the EM models (Figs. 2 , 3 , 4 ). How deep the extensional faults can penetrate into the lithosphere is not known, but the arctic ridges have very deep earthquakes (30–35 km) indicating the lower limit of brittle lithosphere 14 . Further, deep serpentinization (15 km) suggest that water circulation could extend deep exploiting weak zones, faults, and fractures 14 . Deep faulting, fracturing and water circulation in combination with slow spreading will promote effective cooling and deep and rapid solidification of magmatic events, explaining the much thicker than expected lithosphere along the ridges (Figs. 2 , 3 , 4 ). However, the general presence of melt in the asthenosphere (Figs. 3 , 4 ), the continuity of the crust along the ridges (Fig. 2 ) and the sustained volcanic activity along weak zones (WZ) at the graben floor (Figs. 1 bc and ED Figs. 1 0ab) suggest that, at least in a geological timeframe, melt is released steadily through the thick lithosphere. Very rapid drainage of melt from an ultra-deep reservoir was recently documented in an extensional system in the Madagascar Strait 37 . Here, the melt is believed to drain directly from the asthenosphere to the sea floor through dykes that intrude the ca. 55 km thick oceanic lithosphere via pre-existing deep faults, creating huge volcanic edifice at the sea floor. The conditions are favorable for this mechanism (direct deep drainage) to be active also along the Mohns and Knipovich Ridges with faults and weak zones potentially extending deep into the brittle lithosphere. Inversion of synthetic data show that the CSEM method is sensitive to thin conductive vertical features below the ridges (ED Fig. 5), but due to the inversion regulation, smoothing the inverted EM model, feeders will appear thicker than they are. Such a thin feeder associated with a deep fault system is visible below the spreading center in the reinverted ATLAB-1 CSEM model (Fig. 2 b and ED Fig. 6). Due to the effective cooling of the lithosphere similar and older feeders could already be solidified and thus invisible in the CSEM data. Direct deep drainage can explain young crust in areas with presently thick and apparently inactive lithosphere and adds an important element to the geological model. In the new model, the volcanic centers (Fig. 4 ) are formed at weak zones (Figs. 1 c,10) and are ambulatory point sources for melt supply to the ridges. The centers, feeder system and volcanic activity at the sea floor are episodic in geological time, relatively short-lived and random in both time and place. This explains the generally thinner crust, and why the crust can be abnormally thin even with an active feeder system, and why the crust can be both thick and thin in areas with no visible feeders. In this model the spreading rate cannot predict variation in crustal thickness along the ridges. Instead, it is the extensional forces, local tectonics, lithology, weak zones and deep faults, potentially also associated with direct deep drainage, that controls the development. However, in such a model with thick and brittle lithosphere, it is still the slow and asymmetric plate motions (rate and direction) that is the dominant factor. Melt dynamics is a passive buoyancy-driven response to the tectonic evolution, and as long as melt is available in the asthenosphere it finds its way to the surface when the tectonic process cracks and opens the lithospheric plate. That the melt- and geodynamic models, substantiated by the new EM data, are so closely coupled to the fundamental tectonic processes supports the idea that the model may also apply to ultraslow-spreading ridges in general. Declarations Data and code availability All data needed to evaluate the conclusions in the paper are present in the paper and/or the Methods section. NTNU holds the publishing rights to the data used in this study, and access to the data can be granted upon mutual agreement with the Centre for Geophysical Forecasting at the Norwegian University of Science and Technology in Trondheim. MARE2DEM is a parallel adaptive finite element code for two-dimensional forward and inverse modeling in electromagnetic geophysics. It was used for data modeling and inversion in this project. MARE2DEM is available for download at http://mare2dem.ucsd.edu/. Acknowledgements The data used in this research was obtained from the ATLAB consortium, a collaborative effort including a diverse group of stakeholders from commercial, regulatory, and academic domains, who jointly acquired multi-geophysics and environmental data 2017-2022. The consortium's funding was made possible through both monetary contributions and in-kind support from all participating partners. The ATLAB consortium includes the following entities: NTNU CGF (serving as the scientific operator and lead), NTNU IGP, NPD, EMGS, Equinor, AkerBP, NORCE, PGS, CGG, Shearwater, OFG, InApril, Allton, and TGS. Author contributions Conceptualization: SEJ, HEFA, BA Methodology: SEJ, HEFA, BA, RM, OMP, MP, ML, SW Investigation: SEJ, HEFA, BA, RM, OMP, MP, ML Visualization: SEJ, HEFA, BA, RM, OMP, MP, ML, KOL Funding acquisition: SEJ, BA, ML Project administration: SEJ, BA, ML, KOL Supervision: SEJ, BA Writing – original draft: SEJ, OMP, MP Writing – review & editing: SEJ, HEFA, BA, RM, OMP, MP, ML, KOL, SW Competing interest declaration The authors declare that they have no competing interests. References Solomon, S. C. & Toomey, D. R. The Structure of Mid-Ocean Ridges. Annu Rev Earth Planet Sci 20 , 329–364 (1992). Team, T. M. E. L. T. S. Imaging the Deep Seismic Structure Beneath a Mid-Ocean Ridge: The MELT Experiment. Science 280 , 1215–1218 (1998). Wanless, V. D. & Shaw, A. M. Lower crustal crystallization and melt evolution at mid-ocean ridges. Nat Geosci 5 , 651–655 (2012). Bown, J. W. & White, R. S. Variation with spreading rate of oceanic crustal thickness and geochemistry. Earth Planet Sci Lett 121 , 435–449 (1994). Raitt, M. The crustal rocks. The Sea 3 , 85–102 (1963). Reid, I. & Jackson, H. R. Oceanic spreading rate and crustal thickness. Mar Geophys Res 5 , 165–172 (1981). White, R. S., Minshull, T. A., Bickle, M. J. & Robinson, C. J. Melt Generation at Very Slow-Spreading Oceanic Ridges: Constraints from Geochemical and Geophysical Data. J Pet. 42 , 1171–1196 (2001). Zhou, H. & Dick, H. J. B. Thin crust as evidence for depleted mantle supporting the Marion Rise. Nature 494 , 195–200 (2013). Sauter, D. et al. Evidence for magma entrapment below oceanic crust from deep seismic reflections in the Western Somali Basin. Geology 44 , 407–410 (2016). Johansen, S. E. et al. Deep electrical imaging of the ultraslow-spreading Mohns Ridge. Nature 567 , 379–383 (2019). Key, K., Constable, S., Liu, L. & Pommier, A. Electrical image of passive mantle upwelling beneath the northern East Pacific Rise. Nature 495 , 499–502 (2013). Lay, T. Seismological Grand Challenges in Understanding Earth’s Dynamic Systems. in Report to the National Science Foundation (IRIS Consortium (2009). Rychert, C. A., Harmon, N., Constable, S. & Wang, S. The Nature of the Lithosphere-Asthenosphere Boundary. J Geophys Res Solid Earth 125 , 2018–016463 (2020). Schlindwein, V. & Schmid, F. Mid-ocean-ridge seismicity reveals extreme types of ocean lithosphere. Nature 535 , 276–279 (2016). Wang, S., Constable, S., Rychert, C. A. & Harmon, N. A Lithosphere-Asthenosphere Boundary and Partial Melt Estimated Using Marine Magnetotelluric Data at the Central Middle Atlantic Ridge. Geochem Geophys Geosyst 21 , 2020–009177 (2020). Michael, P. J. et al. Magmatic and amagmatic seafloor generation at the ultraslow-spreading Gakkel ridge, Arctic Ocean. Nature 423 , 956–961 (2003). Niu, X. et al. Along-axis variation in crustal thickness at the ultraslow spreading Southwest Indian Ridge (50°E) from a wide-angle seismic experiment. Geochem Geophys Geosystems 16 , 468–485 (2015). Cannat, M. How thick is the magmatic crust at slow spreading oceanic ridges? J Geophys Res Solid Earth 101 , 2847–2857 (1996). Cannat, M. et al. Spreading rate, spreading obliquity, and melt supply at the ultraslow spreading Southwest Indian Ridge. Geochem Geophys Geosystems 9 , 2007 001676 (2008). Dick, H. J. B., Lin, J. & Schouten, H. An ultraslow-spreading class of ocean ridge. Nature 426 , 405–412 (2003). Ritzmann, O., Jokat, W., Mjelde, R. & Shimamura, H. Crustal structure between the Knipovich Ridge and the Van Mijenfjorden (Svalbard. Mar Geophys Res 23 , 379–401 (2002). Standish, J. J., Dick, H. J. B., Michael, P. J., Melson, W. G. & O’Hearn, T. MORB generation beneath the ultraslow spreading Southwest Indian Ridge (9-25°E): Major element chemistry and the importance of process versus source: SWIR MORB-PROCESS VERSUS SOURCE. Geochem Geophys Geosystems 9 , (2008). Carbotte, S. M. & Scheirer, D. S. Variability of ocean crustal structure created along the global mid-ocean ridge. Hydrogeol Ocean Lithosphere 59–107 (2004). Jokat, W. et al. Geophysical evidence for reduced melt production on the Arctic ultraslow Gakkel mid-ocean ridge. Nature 423 , 962–965 (2003). Eiken, O. Seismic atlas of western Svalbard: a selection of regional seismic transects. (1994). Conley, M. M. & Dunn, R. A. Seismic shear wave structure of the uppermost mantle beneath the Mohns Ridge: UPPERMOST MANTLE BENEATH THE MOHNS RIDGE. Geochem Geophys Geosystems 12 , (2011). Hermann, T. & Jokat, W. Crustal structures of the Boreas Basin and the Knipovich Ridge, North Atlantic. Geophys J Int 193 , 1399–1414 (2013). Tectonostratigraphic atlas of the North-East Atlantic Region. in Geological Survey of Denmark and Greenland (GEUS (eds. Hopper, J. R. et al.) (Copenhagen, 2014). Pedersen, R. B. et al. Discovery of a black smoker vent field and vent fauna at the Arctic Mid-Ocean Ridge. Nat Commun 1 , 126 (2010). Ni, H., Keppler, H. & Behrens, H. Electrical conductivity of hydrous basaltic melts: implications for partial melting in the upper mantle. Contrib Miner. Pet. 162 , 637–650 (2011). Constable, S. SEO3: A new model of olivine electrical conductivity. Geophys J Int 166 , 435–437 (2006). Chen, Y. J. Oceanic crustal thickness versus spreading rate. Geophys Res Lett 19 , 753–756 (1992). Meier, M. et al. Segment‐Scale Seismicity of the Ultraslow Spreading Knipovich Ridge. Geochem Geophys Geosystems 22 , 2020 009375 (2021). Bruvoll, V., Breivik, A. J., Mjelde, R. & Pedersen, R. B. Burial of the Mohn‐Knipovich seafloor spreading ridge by the Bear Island Fan: Time constraints on tectonic evolution from seismic stratigraphy. Tectonics 28 , 2008TC002396 (2009). Lim, A., Brönner, M., Johansen, S. E. & Dumais, M. Hydrothermal Activity at the Ultraslow‐Spreading Mohns Ridge: New Insights From Near‐Seafloor Magnetics. Geochem. Geophys. Geosystems 20 , 5691–5709 (2019). Stubseid, H. H., Bjerga, A., Haflidason, H., Pedersen, L. E. R. & Pedersen, R. B. Volcanic evolution of an ultraslow-spreading ridge. Nat. Commun. 14 , 4134 (2023). Feuillet, N. et al. Birth of a large volcanic edifice offshore Mayotte via lithosphere-scale dyke intrusion. Nat. Geosci. 14 , 787–795 (2021). Anderson, D. L. Lithosphere, asthenosphere, and perisphere. Rev Geophys 33 , 125–149 (1995). Methods Six electromagnetic surveys (3 CSEM and 3 MT) have been collected at the northern Mohns- and the northern Knipovich ridges (Figs. 1). The data were collected by the ATLAB program (Atlantic Laboratory) and consists of ATLAB-1 10 (Figs. 2b,3b), ATLAB-2 (Figs. 2a,3a) and ATLAB-3 (Figs. 2c,3c). Data analysis and inversion of the ATLAB-1 data are also described in detail in online method in reference 10 . Data analysis and inversion ATLAB-2 The ATLAB-2 data set was collected across the northern Knipovich Ridge onto the Svalbard continental margin by 39 broadband ocean-bottom electromagnetic receivers deployed along a 120-km-long profile (Fig. 1). The receivers recorded the horizontal electric and magnetic fields in time series. Based on pre-survey modelling 39 , the receiver spacing was set to 2.5 km in the central part and 5 km for the last 6 receivers at both end of the profile. A 281-m-long horizontal electric dipole source was towed across the receivers, emitting a time-varying electric current of 1,200 A. The source was towed 30–50 meters above the seafloor, but in areas with very rough bathymetry the source was up to 250 meters above the seafloor. The source signal was optimized to ensure high current output over a certain low frequency range, and the output waveform of the source with a 4s fundamental period covers more than one decade with logarithmically spaced frequencies, ranging from 0.25 Hz to 5.25 Hz, with almost equally high current output. From the measured time series, the CSEM processing 40 extracts high-quality data for the frequencies of 0.25, 0.5, 1.0, 2.5 and 5.25 Hz. Due to variations in subsurface resistivity and water depth the electric field data vary significantly along the profile (ED Fig. 1a). For the extracted frequencies, we also obtained precise noise estimates, which were used to compute appropriate data misfits in the inversion. From up to 6-day-long time series, continuously registered by the seafloor electromagnetic receivers, it was also possible to process high-quality MT data for periods of 1.4 s to 1,000 s using robust multi-station processing 41 . MT data examples are shown in ED Fig. 1b. Dimensionality analysis of the MT data reveals 3D distortions for some receivers located in the deepest part of the Knipovich Ridge graben (Figs. 1b) and the noisiest data were not included in the 2D inversion. The electric field CSEM data (Fig. 2a) were inverted using a regularized, nonlinear two-dimensional inversion with a finite element forward engine 42 . In the Cartesian coordinate system, used in the 2D inversion code, the x-axis is oriented normal to the observation profile, the y-axis is parallel to the profile and the z-axis is pointing downwards. The Occam-type inversion seeks a smooth resistivity model that explains the measured data within the data uncertainty with the consideration of triaxial electric anisotropy. However, we made the simple assumption of transverse isotropy, where the resistivities in the horizontal directions are assumed to be equal: where ρ x is the resistivity in the direction normal to the observation profile (parallel to the ridge), ρ y is the resistivity parallel to the observation profile, and ρ z is the resistivity in the vertical direction. To stabilize the inverse problem, both smoothness regularization and anisotropy regularization were applied, penalizing non-smooth resistivity variations and differences between the resistivity components in the horizontal and vertical directions. In the MT inversion we seek for a triaxially anisotropic resistivity model, allowing all three diagonal components of the resistivity tensor to vary individually: The transverse electric (TE) mode (electric current flowing parallel to the ridge) and the transverse magnetic (TM) mode (electric current flowing perpendicular to the ridge) of the MT data are then the off-diagonal elements of the impedance tensor, rotated into the observation profile. The diagonal elements are zero if the model is 2D: Due to strong bathymetry variations, especially in the eastern part of the profile, the MT source signal does no longer behave as a pure vertically propagating plane wave 43 . Consequently, the TE-mode data are primarily sensitive to the resistivity component normal to the observation profile ρ x , and the TM-mode data are sensitive to both the resistivity parallel to the profile ρ y and the vertical resistivity component ρ z . The MT inversion can be stabilized with anisotropy regularization. The seafloor topography in the starting model was composed of both the bathymetry data measured along the profile during the survey and global bathymetry data 44 for the rest of the model. In this way we could account for the Svalbard islands when simulating the long period MT data and get improved convergence in the inversions. Individual TE-mode inversion converged to an RMS of 1.15, assuming a 10% relative data uncertainty. The data from the receivers in the central Knipovich Ridge graben have slightly poorer fits than other receivers. The TE-mode inversion recovers a smooth geologically realistic resistivity model (ED Fig. 2a), suggesting strong conductivity alignment with the N-S striking Knipovich Ridge and the general N-S geological strike of the area. The individual TM-mode inversion result (ED Fig. 2b) has an RMS misfit of 1.5, assuming 10% relative data uncertainty. The data from receivers in the deep graben have a particularly poor fit and the resulting resistivity model has strong imaging artefacts beneath these receivers which were not present either in the TE-mode inversion or in the CSEM inversion. The CSEM data have good sensitivity to this depth interval and show a much simpler resistivity structure. Both the dominating conductive region ”x” and deep resistive region ”y” (ED Fig. 2b) appear as very unrealistic and suggest strong coastal effects in the TM mode data 43,45 . This effect, caused by the Svalbard Islands, is apparently much less pronounced or barely present in the TE-mode model. The combined TE- and TM-mode inversion model (ED Fig. 2c) has clear similarities to the single TE -mode inversion model, but coastal effects from the TM-mode are also preserved. A more detailed analysis of the MT data misfit for the combined inversion result show that the deep graben receivers have negative TE-mode phases for periods T < 250 s and negative TM-mode phases for periods T < 50 s. Further, the TM-mode amplitude data are significantly depressed below the TE amplitude responses (ED Fig. 1c). These observations are exactly the previous identified coast effects. The negative TM-mode phases and depressed TM-mode amplitude at relatively short periods are described as inductive and galvanic coastal effect distortions, respectively 45 , the negative TE-mode phases and resistivity cusp shape (Rx 17 and Rx18) at relatively long periods are identified as inductive coast effect distortion 43 . To find a common model in the joint CSEM and MT inversion we had to increase the relative uncertainty of the CSEM data to 5%, allowing for less constraint on the model from the CSEM data giving relatively higher weight to the MT data (ED Fig. 3). To test such effects and to optimize the interpretation we also performed inversion of synthetic data, including separate TE- and TM-mode inversions, joint TE- and TM-mode inversion and joint CSEM and MT inversion (Ed Fig. 4). The synthetic TE-mode inversion model (ED Fig. 4c) reconstructs the main features of the true model but underestimates the resistivities in the lithosphere (layer ”C”) due to the lower resolution to resistors than conductors and smears out the deep melt anomaly (Letter D). In contrast, the individual TM-mode inversion does not recover the deeper parts of the model (ED Fig. 4 d). The joint TE and TM-mode inversion also reconstructs important model elements (Letters C and D in ED Fig. 4b), but the modes are smeared together and complicates the interpretation with inversion artifacts (Letters f and g in ED Fig. 4b). The CSEM and MT joint inversion shows that including CSEM data in the inversion can improve resistivity imaging, also of the deeper parts of the model. The deep conductive anomaly (”D” in ED Fig. 4e ) has improved image focus compared to the TE-mode inversion, but is still a bit smeared out due to smoothness regularization. The high resistivities at depth (Letters g and g* in ED Fig. 4e) are probably end of line effects, but still difficult to remove due to limitations in extension of aperture. Similar effects are observed in the CSEM and MT joint inversion result for the measured data (ED Fig. 3). The Knipovich EM survey is characterized by partly symmetrical seafloor topography to the profile. In addition, data dimensionality analyses and small on-diagonal elements in the MT-impedance tensor all suggest 2D structures beneath the survey profile and limited data sensitivity to side structures. These reasons suggest that the 2D inversion models are representative for the geological structures below the survey line. However, reconstruction of the subsurface resistivity models from CSEM and MT data is a non-unique inverse problem and potentially many different models can fit the measured data. The subsurface interpretations are based on the various inversion models and tests. However, of the single inversion models it is the high frequency CSEM model that best represents the shallow subsurface and due to the apparently few inversion artifacts the TE-mode model is interpreted to best represents the deep subsurface in this case. Inversion of thin vertical features in the crust and the upper mantle To test how the CSEM models image vertical structures we generated synthetic data, added 3% random noise and inverted the data with the same inversion setup as for the real data. The inversion result are good representations of the general geological model (ED Fig. 5). For comparison inversions were done with and without a conductive melt feeder channel, thinning upwards (10 ohmm), and deep fault zones (1 ohmm). The synthetic inversion results show significant decrease in resistivity along the vertical channel not observed in the inversion without the channel or in the real data inversion (Fig. 2a). The conductive anomaly is smeared out due to resolution and inversion regularization (smoothing). The deep conductive fault zones reach several kilometers into the mantle and the inversion model images the fault zones, but also due to resolution and regularization they have a more complex response than the melt channel. Similar fault-like features are observed in the crust in the real data, but not in the mantle at the ATLAB-2 location. The inversions explain the synthetic data within the uncertainty (RMS = 1). New data analysis and inversion of ATLAB-1 Acquisition parameters and processing sequence of the ATLAB-1 data set is detailed in the methods section in the original publication 10 . A new EM data inversion of ATLAB-1 (Fig. 2b) confirms a vertical conductive feature beneath Loki’s Castle interpreted as thin feeder system. ED Fig. 6 (upper panel) shows the result of the electric data inversion where the model is imaged to 15 km compared to 8 km in the original version. This inversion reached a data misfit of RMS 1.0. A sensitivity modelling scenario designed to test the robustness of the vertical conductive feature is shown in the lower panel of ED Fig. 6. By removing the vertical conductor, the test shows that the ATLAB-1 CSEM data is clearly sensitive to changes in conductivity also between 5 and 15 km depth. It is mainly the long offsets of the CSEM data that record variations at these depths. Data analysis and inversion ATLAB-3 Acquisition parameters and processing sequence for ATLAB-3 are, with minor adjustments for local conditions, similar to ATLAB-1 10 and ATLAB-2. ATLAB-3 was collected at the northernmost Mohns ridge by 24 receivers along a 54 km long transect (Fig. 1c) just south of the ATLAB-1 transect 10 at Loki’s Castle to study subsurface changes over relatively short distances along the Mohns ridge. The ATLAB-3 area has very rough bathymetry (Fig. 1). Based on experience from the other ATLAB surveys, the EM receiver instruments at ATLAB-3 were carefully placed with an average spacing of 1700m. The 281-m-long electric dipole emitted a time-varying electric current of 1,200 A. Ideally the source should be towed 30–50 meters above the seafloor, but due to the very rough bathymetry, the distance from source to sea floor significantly. The source signal was optimized to ensure high current output over frequencies from 0.5 Hz to 9 Hz and the CSEM processing 40 extracted high-quality data for the selected frequencies and associated noise estimates were used to compute data misfits in the inversion. From up to 9-days registration, high-quality MT data was extracted with periods from 5.5 s to 1500 s using multi-station processing 41 . The phase tensor ellipses reveal 3D distortions for some of the receivers along the transect. Tests showed that omitting these receivers did not have major effects on the inversion result. The MT data was analyzed using the MTPy software package 46 . The EM data of ATLAB-3 was inverted using a regularized, nonlinear two-dimensional inversion code with a finite element forward engine 42 . The electric and magnetic fields CSEM data were first inverted separately (ED Fig. 7). The electric field CSEM data inversion reached data misfit of RMS 1.0 and the magnetic field CSEM data inversion also reached RMS of 1.0. The two independent CSEM inversion images show strong similarities which increases trust in the results. Subsequently, the MT data was inverted using both the TE and TM mode (ED Fig. 8a). The MT inversion reached data misfit RMS of 1.06. Finally, the MT and CSEM data of ATLAB-3 were inverted jointly (ED Fig. 8b), to find a resistivity model that explains both datasets. The joint CSEM and MT inversion reached an overall data misfit of RMS 1.06. It is expected that the shallow part of the joint inversion result is dominated by the CSEM data, while due to the frequency-dependent attenuation of the electromagnetic signal, the deeper part of the joint inversion model is dominated by the low-frequency MT data. It has also been shown that the CSEM data constrains the joint inversion and, in this way, improve imaging, also of the deeper parts of the model 10 . To test such effects, improve the interpretation and to illustrate the effect of joint CSEM and MT inversion we also performed inversion of synthetic data relevant for the ATLAB-3 area. We tested three different conceptual geological models, one without a feeder system (no feeder), a model where the feeder system is partially solidified into the lithosphere (not connected) and a completely open feeder system (connected) up to the oceanic crust (ED Fig. 9). After forward simulation of the three scenario models, the synthetic data was contaminated with uncertainty (noise) before inversions. The inverted MT data (ED Figs. 8b, c, d) provides a reasonable representation of the no-feeder model and of the open system (connected). Although there are changes also in the model where the supply system is terminated in the middle of the lithosphere (not connected), the imaging here is partly misleading. The results from the joint inversion of the synthetic data (Fig. 8e) show that combining all the data in this case significantly improves the imaging, also of the deeper parts of the models. The synthetic inversion results do not completely rule out that parts of the lithosphere may have increased conductivity, but if there had been a significant feeder system below ATLAB-3, this would have been imaged both in the MT models and in the joint models. Methods References 10. Johansen, S. E. et al. Deep electrical imaging of the ultraslow-spreading Mohns Ridge. Nature 567 , 379–383 (2019). 39. Tegnander, J. F. Electromagnetic Modelling and Inversion of Seafloor Massive Sulphide Deposits - Imaging of SMS Deposits and the Asthenosphere at the Atlantic Mid-Ocean Ridge (Master . (NTNU, Trondheim, 2017). 40. Myer, D., Constable, S. & Key, K. Broad-band waveforms and robust processing for marine CSEM surveys. Geophys J Int 184 , 689–698 (2011). 41. Egbert, G. D. Robust multiple-station magnetotelluric data processing. Geophys J Int 130 , 475–496 (1997). 42. Key, K. MARE2DEM: a 2-D inversion code for controlled-source electromagnetic and magnetotelluric data. Geophys J Int 207 , 571–588 (2016). 43. Key, K. & Constable, S. Coast effect distortion of marine magnetotelluric data: Insights from a pilot study offshore northeastern Japan. Phys Earth Planet Inter 184 , 194–207 (2011). 44. G.E.B.C.O. The General Bathymetric Chart of the Oceans . (GEBCO, 2023). 45. Wang, S., Constable, S., Reyes-Ortega, V. & Rychert, C. A. A newly distinguished marine magnetotelluric coast effect sensitive to the lithosphere–asthenosphere boundary. Geophys J Int 218 , 978–987 (2019). 46. Kirkby, A., Zhang, F., Peacock, J., Hassan, R. & Duan, J. The MTPy software package for magnetotelluric data analysis and visualisation. J Open Source Softw 4 , (2019). Additional Declarations There is NO Competing Interest. Supplementary Files ExtendedDataFigs.docx Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-3999138","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Physical Sciences - Article","associatedPublications":[],"authors":[{"id":281080505,"identity":"8116199a-ef16-4349-83e2-b5ef5fc272be","order_by":0,"name":"Ståle Emil Johansen","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA7UlEQVRIiWNgGAWjYHACxgMMDGwMBhCODREa2BgYkLWkMfAQqYUBpuUwYS3y85sfHPjAwCdnzn728IePO87n2Uv3GDAXVODWYnCMzeDgDAY2Y8uevDTJmWduF/PInDFgnnEGjxagLw7zMLAlbjiQY8bM23Y7sUcixwDIwOOwNvYPh/+AtJx/Y/z5b9s5wloYjvEYHGYAabmRYyDN2HaAsBaDYzkFB3sM2IwNbrwxk+xtS07suZFWcJgHj1/km49vfPCj4picwfkc4w8/2+wS22ckb3zMgyfEYHah8g8Q0gAENUSoGQWjYBSMghELAKPxUA6MyxPuAAAAAElFTkSuQmCC","orcid":"","institution":"NTNU","correspondingAuthor":true,"prefix":"","firstName":"Ståle","middleName":"Emil","lastName":"Johansen","suffix":""},{"id":281080506,"identity":"20dbe576-bfdc-4394-b267-9e953617527a","order_by":1,"name":"Hans Amundsen","email":"","orcid":"","institution":"Vestfonna Geophysical AS","correspondingAuthor":false,"prefix":"","firstName":"Hans","middleName":"","lastName":"Amundsen","suffix":""},{"id":281080507,"identity":"03266ceb-e2af-4d5c-914b-ea6d275a3678","order_by":2,"name":"Børge Arntsen","email":"","orcid":"https://orcid.org/0000-0002-7265-4834","institution":"NTNU","correspondingAuthor":false,"prefix":"","firstName":"Børge","middleName":"","lastName":"Arntsen","suffix":""},{"id":281080508,"identity":"049ecad2-0db3-4f59-90fd-1eeab7b24ffb","order_by":3,"name":"Rune Mittet","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Rune","middleName":"","lastName":"Mittet","suffix":""},{"id":281080509,"identity":"a74f54d0-aaa0-47c7-a975-f7871a9c308c","order_by":4,"name":"Ole Pedersen","email":"","orcid":"","institution":"Allton","correspondingAuthor":false,"prefix":"","firstName":"Ole","middleName":"","lastName":"Pedersen","suffix":""},{"id":281080510,"identity":"2e7266e6-f071-4968-bb3c-7a8e82e393bb","order_by":5,"name":"Martin Panzner","email":"","orcid":"","institution":"Equinor Research","correspondingAuthor":false,"prefix":"","firstName":"Martin","middleName":"","lastName":"Panzner","suffix":""},{"id":281080511,"identity":"0592dfe7-d970-410a-955a-f5c266a25db8","order_by":6,"name":"Kamaldeen Omosanya","email":"","orcid":"","institution":"Oasisgeokonsult","correspondingAuthor":false,"prefix":"","firstName":"Kamaldeen","middleName":"","lastName":"Omosanya","suffix":""},{"id":281080512,"identity":"695da7a5-ede5-40da-98b5-cb40aae16033","order_by":7,"name":"Shunguo Wang","email":"","orcid":"","institution":"NTNU","correspondingAuthor":false,"prefix":"","firstName":"Shunguo","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2024-02-29 09:11:32","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-3999138/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-3999138/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":53185108,"identity":"ee241391-9104-4176-bd3c-8d0cb9678009","added_by":"auto","created_at":"2024-03-21 16:10:04","extension":"jpeg","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":637175,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eLocation of electromagnetic (EM) surveys (MT and CSEM) across the ultraslow-spreading northern Mohns Ridge and the Knipovich Ridge in the Norwegian- Greenland Sea. a, \u003c/strong\u003eLocation of the six ATLAB (Atlantic Laboratory Program) surveys along the ridges: ATLAB-1 \u003csup\u003e10\u003c/sup\u003e, ATLAB-2, and ATLAB-3. One CSEM and one MT survey were conducted at each location. The N-S-oriented western mountains of Svalbard are visible in red in the northeastern corner of the image. White stippled rectangles indicate the positions of the detailed bathymetric maps (ED Figs. 10 ab).\u003cstrong\u003e b, \u003c/strong\u003eSurface view of seafloor topography and seabed volcanoes at the northern Knipovich Ridge. The location of the EM receiver is indicated with white dot on the seafloor and the CSEM tow line is red. Northern Knipovich Ridge (1), Spitsbergen-Molloy fault zone (2). \u003cstrong\u003ec, \u003c/strong\u003eSurface view of the seafloor topography, axial volcanic ridges (AVR), and seabed volcanoes on the northern Mohns Ridge. The black smoker vent field, Loki’s Castle, is located at the crest of the northern AVR in the center of the ATLAB-1 survey line. Weak tectonic zones (WZ) are associated with AWRs and volcanoes on the graben floor (see ED Figs. 10 ab and text for details). Southern Knipovich Ridge (3), Northern Mohns Ridge (4).\u003c/p\u003e","description":"","filename":"image1.jpeg","url":"https://assets-eu.researchsquare.com/files/rs-3999138/v1/dee2ad0f7f879a79c172ad69.jpeg"},{"id":53186251,"identity":"d6451f99-4d93-45b1-9d7a-768e229904bc","added_by":"auto","created_at":"2024-03-21 16:18:04","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":647415,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eCSEM resistivity model across the ultraslow-spreading Knipovich and Mohns Ridges. \u003c/strong\u003eThe colors show resistivity ρh in the horizontal direction. The upper black line is the base of sediments and Moho is the interpreted Mohorovich discontinuity (see text and \u003csup\u003e10\u003c/sup\u003e for discussion). Location of CSEM profiles in Figs. 1, S1, and S2. \u003cstrong\u003ea, \u003c/strong\u003eResistivity model across the Northern Knipovich Ridge (ATLAB-2, upper panel) and Svalbard continental margin. The area between letters A and B is the interpreted transition zone from the oceanic to continental crust. The upper numbers (7-0-3) indicate the age of the oceanic crust in millions of years. The crust is ca. 3,5–5,5 km thick in the oceanic domain. This is notably below the global average of approximately 7 km \u003csup\u003e4,24\u003c/sup\u003e. The internal conductive pattern in the oceanic crust is interpreted as the convection of saline fluids along faults and cracks. Note that there is no visible active feeder system below the graben in this part of the Knipovich Ridge. \u003cstrong\u003eb,\u003c/strong\u003e Resistivity model across the northern Mohns Ridge \u003csup\u003e10\u003c/sup\u003e (ATLAB-1). The crust is ca. 2.0–3.5 km, which is much thinner than that at the northern Knipovich Ridge. Compared with the northern Knipovich ridge (Fig. 1a), high conductivity in the crust is more evident below the Mohn's ridge. Increased saline water infiltration and circulation are believed to be promoted by the conductive feeder system visible below Loki Castle. The upper part of the lithosphere is believed to be brittle to ca. 15–30 km deep \u003csup\u003e38\u003c/sup\u003e, and weak tectonic zones are expected to extend deep into the mantle. At the surface, weak zones are associated with faulting, AWRs, and volcanoes (ED Figs. ED 10 ab for details). The locations of the CSEM profiles are shown in Figs. 1. See the Methods and ED Figs. 5 and 9 for shallow feeder channel inversion tests. \u003cstrong\u003ec,\u003c/strong\u003e Resistivity model across the northern Mohns Ridge (ATLAB-3). The crustal thickness (2.0–3.0 km) is similar to ATLAB-1 (Loki’s Castle) but much thinner than below the northern Knipovich Ridge (ATLAB-2) and only ca. one-third of normal ocean crust. Note that all estimates of crustal thickness along the ridges are uncertain because the direct coupling between geophysical data and the igneous ‘crust’ has not been established for ultraslow-spreading ridges \u003csup\u003e8,24\u003c/sup\u003e. In addition, a feeder system comparable to that below Loki’s castle (ATLAB-1) is not visible below ATLAB-3. However, indications of a thin high conductivity feature is faintly visible in the magnetic component of the inverted data (ED Fig. 7).\u003c/p\u003e","description":"","filename":"image2.png","url":"https://assets-eu.researchsquare.com/files/rs-3999138/v1/504669b66e58673ec2e26e50.png"},{"id":53186249,"identity":"ad50d515-1efd-4379-9b2a-f192e80cd93c","added_by":"auto","created_at":"2024-03-21 16:18:04","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":639866,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eInverted resistivity images of the lithosphere and upper asthenosphere beneath the ultraslow-spreading Mohns and Knipovich Ridges. \u003c/strong\u003eThe ca. 100 Ω m contour, denoted electrical LAB in \u003csup\u003e10\u003c/sup\u003e is interpreted as the base of the lithosphere (LAB). This surface defines an impermeable surface, guiding melt migration to shallower levels and along the ridges. Outside and above this surface, there is a rapid transition to resistivities typical of melt-free peridotite (depleted mantle) \u003csup\u003e10,31\u003c/sup\u003e. Location of profiles in Fig. 1. \u003cstrong\u003ea,\u003c/strong\u003e Deep resistivity model across northern Knipovich Ridge (ATLAB-2). Colors indicate resistivity in the horizontal crossline direction from the inversion of the MT data (TE mode). The inversion model shows conductivity variations down to 100 km with a strong anomaly (red), indicating deep partial melt below the LAB. The letters A and B mark the transition zone from the oceanic to the continental crust. The Moho surface was transferred from the CSEM interpretation in Fig. 2a and showed better resolution in the higher-frequency data. Inversion tests of the synthetic data showed that the TE mode was the least affected by the coastal effect and provided the best geometrical imaging in this area. See Methods for a discussion on the processing, inversion, and interpretation strategies. \u003cstrong\u003eb,\u003c/strong\u003e Deep resistivity model across the northern Mohns Ridge (ATLAB-1) \u003csup\u003e10\u003c/sup\u003e. Colors show resistivity ρy in the horizontal inline direction from joint nonlinear inversion of the MT and CSEM data. Inversion model \u003csup\u003e10\u003c/sup\u003e shows conductivity variations down to 120 km with the ridge centered above an anomaly, indicating partial melt below the LAB and along an upward narrowing and asymmetric zone (melt feeder system). All three profiles showed deep melt: however, ATLAB-1 is the only deep EM profile along the ridges with a well-developed feeder system. \u003cstrong\u003ec,\u003c/strong\u003e Deep resistivity model across northern Mohns Ridge (ATLAB-3). Colors show resistivity ρy in the horizontal inline direction from joint nonlinear inversion of the MT and CSEM data. The inversion model showed conductivity variations to 120 km depth. The ATLAB-3 profile was acquired just 25 km south of Loki Castle (ATLAB-1; Fig. 1c), but the inverted model did not show a deep feeder below the ridge. However, indications of a thin high conductivity feature is faintly visible in the magnetic component of the inverted data (ED Fig. 7).\u003c/p\u003e","description":"","filename":"image3.png","url":"https://assets-eu.researchsquare.com/files/rs-3999138/v1/a901f8413ed1e095a6652152.png"},{"id":53186250,"identity":"25512e9c-687f-4897-bdb5-2e31c24b7f9c","added_by":"auto","created_at":"2024-03-21 16:18:04","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":855360,"visible":true,"origin":"","legend":"\u003cp\u003e\u003cstrong\u003eQuasi-3D resistivity model of a volcanic center including lithosphere, upper asthenosphere, and deep feeder system below the northern Mohns Ridge. \u003c/strong\u003eThe conceptual model is constructed from the ATLAB-1 and ATLAB-3 profiles in Fig. 3. E-W segments are copies of part of the profiles, while N-S segments are interpreted and interpolated. Top panel: Surface view of seafloor topography cut along EM profiles in the E-W direction, along center of the graben, and at the end of the ATLAB-1 profile. The model illustrates the discontinuity of the melt supply system below the ridges. The individual melt feeders are interpreted to be short lived, episodic, and random in time and place. Letter A, see Fig. 2c and ED Fig. 7 for details.\u003c/p\u003e","description":"","filename":"image4.png","url":"https://assets-eu.researchsquare.com/files/rs-3999138/v1/66858de3fa64460a07242179.png"},{"id":53186691,"identity":"7e9f832e-1865-453b-9809-5c6294486cca","added_by":"auto","created_at":"2024-03-21 16:26:05","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2410382,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-3999138/v1/445004c1-5114-4dbd-9488-2bdcde733d95.pdf"},{"id":53185111,"identity":"d86cc435-7e92-49d6-afc3-bd299bc0a744","added_by":"auto","created_at":"2024-03-21 16:10:04","extension":"docx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":9024578,"visible":true,"origin":"","legend":"","description":"","filename":"ExtendedDataFigs.docx","url":"https://assets-eu.researchsquare.com/files/rs-3999138/v1/6a9e1b36c9b62d41d21c3522.docx"}],"financialInterests":"There is \u003cb\u003eNO\u003c/b\u003e Competing Interest.","formattedTitle":"New model for ultraslow-spreading ridges","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAway from fracture zones, hotspots, and marginal basins, the crust in ocean basins has a uniform thickness of approximately 7 kilometers \u003csup\u003e4,5\u003c/sup\u003e, and at fast and intermediate spreading rates, the melt supply may fully accommodate plate divergence. However, when the rate drops below 15\u0026ndash;20 mm/a, the crustal thickness decreases rapidly \u003csup\u003e6,7\u003c/sup\u003e. The thin or absent crust is explained by the current models as reduced melt production or impeded melt migration due to conductive cooling \u003csup\u003e10,18\u0026ndash;21\u003c/sup\u003e. Changes in ridge geometry, LAB topography, melt producing rock volumes, mantle composition, thermal structure and hydrothermal permeability are also believed to affect melt production and crustal thickness along the ultraslow-spreading ridges \u003csup\u003e10,14,18\u0026ndash;20,22\u003c/sup\u003e. The ridges have been studied for decades, but present models are largely based on bathymetry, shallow seismic data, low resolution seismological models, geodynamic modeling, studies of ophiolites and not on direct deep imaging \u003csup\u003e9\u003c/sup\u003e. It is therefore still unclear how melt dynamics, \u003csup\u003e2,3,10,11\u003c/sup\u003e, lithospheric structure, lithology, and crustal formation \u003csup\u003e12\u0026ndash;15\u003c/sup\u003e interact with the tectonic processes. Variations along ridge strikes, in particular, have received little attention and are poorly understood \u003csup\u003e12\u0026ndash;14,16,17\u003c/sup\u003e.\u003c/p\u003e"},{"header":"New electromagnetic data","content":"\u003cp\u003eTechnological advancements in electromagnetic (EM) methods have enabled the combination of high-quality controlled-source shallow electromagnetic imaging (CSEM) with low-frequency passive and deep magnetotelluric imaging (MT) \u003csup\u003e10,11\u003c/sup\u003e. This combined EM technique is particularly well suited for deep imaging below ridges, where melt and fluids are expected to cause considerable fluctuations in electrical resistivity \u003csup\u003e13,23\u003c/sup\u003e. The first deep EM data (ATLAB-1) \u003csup\u003e10\u003c/sup\u003e across an ultraslow-spreading ridge was collected by the Atlantic Lab. Consortium (ATLAB) and showed melt in a narrow and asymmetric feeder zone below the northern Mohns Ridge in the Norwegian Greenland Sea (NGS). This profile provided unique local insight, but to what degree this single transect is representative of the ridge as such is not clear. To address this problem ATLAB collected four new EM datasets along the Knipovich and Mohns Ridges.\u003c/p\u003e \u003cp\u003eThe ATLAB EM data base now consists of six unique data sets (three MT and three CSEM) (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e), collected during three distinct campaigns in the Norwegian-Greenland Sea (NGS). The ATLAB-1 data was reinverted for this study and the new CSEM model now images to 15 km depth compared to 8 km in the original version \u003csup\u003e10\u003c/sup\u003e. ATLAB-2 (this study) (Figs.\u0026nbsp;1ab) was acquired using 39 seafloor receivers along a 130 km long transect at the northernmost Knipovich Ridge, whereas ATLAB-3 (this study) was collected using 24 seafloor receivers along a 54 km long transect at the northernmost Mohns Ridge (Figs.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003ec and ED Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e0a). The ATLAB-2 transect crosses the Knipovich ridge before proceeding eastward across the continent-ocean boundary (COB) and onto continental crust west of Svalbard. ATLAB-3 was collected just south of the ATLAB-1 transect \u003csup\u003e10\u003c/sup\u003e to evaluate the potential changes over relatively short distances along the Mohns ridge. Using a nonlinear regularized inversion of the MT and CSEM data, smooth two-dimensional (2D) anisotropic electrical resistivity models were reconstructed from the datasets (see Methods). All surveys were optimized for local imaging across the ridges, but when combined they from a unique database that also opens for studying variation along the strike of the ridges.\u003c/p\u003e "},{"header":"The crust","content":"\u003cp\u003eThe inverted ATLAB-2 CSEM-model (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea) shows a clear sub-horizontal three-layered subsurface at the northern Knipovich ridge, interpreted as a rigid oceanic lithospheric mantle overlain by oceanic crust, with a conductive sedimentary layer on the top. The rapid increase in conductivity (stippled in Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ea) is interpreted as a porosity threshold at the base of the lower crustal gabbro sequence. Drilling has never verified the Mohorovičić discontinuity (Moho), hence the coupling between geophysical and igneous ‘crust’ is not established for ultraslow-spreading ridges \u003csup\u003e8,24\u003c/sup\u003e. However, interpreting this level as the base of the oceanic crust yields a crustal thickness of ca. 3.5–5.5 km at the northern Knipovich Ridge, well below the global average of ca. 7 km \u003csup\u003e4,24\u003c/sup\u003e. The geometry of intra-crustal conductive anomalies is banana-shaped and are thought to represent the downward migration of seawater followed by lateral transport within extensively fractured sheeted dyke complexes.\u003c/p\u003e\u003cp\u003eBoth sediment and total crustal thickness increase towards the coast into the continental domain, reaching a total thickness of ca. 15–20 km at the outer Spitsbergen margin, which is approximately in line with earlier estimates of depth to Moho in this area \u003csup\u003e25\u003c/sup\u003e. The high conductivity below the shelf indicates that saline water penetrated deep into the continental crust in this area.\u003c/p\u003e\u003cp\u003eSimilar criteria were used to determine the base of the crust below at the northern Mohns Ridge as at ATLAB-2. The crust is significantly thinner here \u003csup\u003e10\u003c/sup\u003e, varying between 2.5 and 3 km (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003eb and ED Fig.\u0026nbsp;6). This is confirmed by the new CSEM data (ATLAB-3) collected 25 km south of ATLAB-1 (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003ec and ED Fig.\u0026nbsp;7). Other geophysical studies \u003csup\u003e21,26–28\u003c/sup\u003e have estimated average and approximate crustal thicknesses of 2–5 km along the Mohns and Knipovich ridges. Also at ATLAB-1 and ATLAB-3 the internal anomalous conductive zone in the crust is interpreted to be caused by saline fluids \u003csup\u003e10\u003c/sup\u003e. The western graben margin is uplifted, and the graben is characterized by substantial negative relief and faults extending deep into the upper mantle. The presence of active vent fields \u003csup\u003e29\u003c/sup\u003e and an inferred plumbing system suggest increased water circulation in this area \u003csup\u003e10\u003c/sup\u003e.\u003c/p\u003e"},{"header":"Lithosphere, LAB, and upper asthenosphere","content":"\u003cp\u003eTemperature, lithology, porosity, permeability, fluid content, and melt content is addressed when interpreting EM sections in terms of oceanic lithosphere. Of these factors, partial melting has the greatest effect on conductivity \u003csup\u003e10,30\u003c/sup\u003e. Using the same criteria as those used for the ATLAB-1 model \u003csup\u003e10\u003c/sup\u003e, the very low resistivities in the ATLAB-2 model (MT) is interpreted as concentration of melt in the mantle below LAB at approximately 50\u0026ndash;80 kilometers depth. The electrical field shown in Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003ea and ED Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e\u0026ndash;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e is perpendicular to the profile (transverse electric, TE mode), indicating a strong conductivity alignment with the N-S striking ridge and the area\u0026rsquo;s general N-S geological strike (Figs.\u0026nbsp;1ab) (See methods for details and interpretation strategies).\u003c/p\u003e\n\u003cp\u003eUnlike the northern Knipovich ridge (ATLAB-2) the melt below northern Mohns Ridge (ATLAB-1) (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003ec, \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eb, \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e and ED Fig.\u0026nbsp;6) is concentrated along a narrow, oblique, and strongly asymmetric zone below the ridge, and the partly melted asthenosphere can be traced to the inferred depth of the Moho discontinuity. In both areas, the partially melted sections are enveloped by a resistivity contour of about 100 Ωm, establishing minimum melt content, interpreted as the base of the electrical lithosphere. This surface is also clearly visible at 30\u0026ndash;50 km depth in the inverted model based on the ATLAB-3 data acquired just south of ATLAB-1 (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003ec, \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e and ED Fig.\u0026nbsp;8b). Above this surface there is a rapid transition to high resistivities typical of melt-free peridotite \u003csup\u003e10,31\u003c/sup\u003e. The surface appears as a permeability barrier preventing vertical melt movement and as a guide for lateral migration and is interpreted as the oceanic LAB within our study area. Because the 2D regularized inversion seeks the smoothest model, the real transition is probably significantly sharper than that seen in the inverted models (See also Methods for details).\u003c/p\u003e"},{"header":"Geological models","content":"\u003cp\u003eThe basic assumption for oceanic crustal formation is that the thickness is proportional to the spreading rate \u003csup\u003e6,8,32\u003c/sup\u003e, and several models have been proposed to explain why the crust thins as the spreading rates decrease. More conventional models explain the thin crust by conductive cooling, which forms a thick thermal boundary layer \u003csup\u003e6\u003c/sup\u003e below the ridge that shuts off melt production (Model M1), or alternatively (Model M2), a comparable but much thinner lid that impedes vertical melt migration \u003csup\u003e18\u003c/sup\u003e. The third model (Model M3) predicts that available melt-producing rock volumes directly regulate crustal thickness \u003csup\u003e10\u003c/sup\u003e, whereas a fourth model (Model M4) proposes \u003csup\u003e19,33\u003c/sup\u003e that melt flows below the LAB topography to volcanic centers, explaining unequal crustal production. However, none of these single models account for all observations in the EM data along the Mohns and Knipovich Ridges.\u003c/p\u003e\n\u003cp\u003eThe northern Knipovich Ridge has thick lithosphere and Model M1 could explain the reduced crustal thickness at ATLAB-2. However, neither this model nor the M2 model, can then explain the much thinner crust in the melt-rich, slightly faster spreading ATLAB-1 area, with an open feeder system below the ridge (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e,\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e,\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e). Model M3 \u003csup\u003e10\u003c/sup\u003e found a direct correlation between crustal thickness and the volume of melt available in the asthenosphere below the ridge. This model can explain the observations at ATLAB-1 but cannot easily explain the much thicker crust in the melt-poor, slower spreading area to the north (ATLAB-2). According to Model M4, melt can migrate below LAB along the ridges to volcanic centers, explaining the fluctuation in crustal thickness. This model does not rely on direct imaging of LAB or asthenosphere. It was extrapolated instead using shallower earthquake hypocenters \u003csup\u003e14\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe inverted EM models confirm that there is significant LAB topography below the ridges and that melt can migrate along strike. However, since melt is available at all EM locations, lateral migration is not necessary to explain the observations. Crust can form from local melt. Furthermore, the M4 model suggests that the crust is thickest where the asthenosphere is shallow. This contradicts the ATLAB models, which reveal that crustal thickness is not directly related to depth to LAB.\u003c/p\u003e"},{"header":"New model ","content":"\u003cp\u003eExtension and large rotated fault blocks are the dominant tectonic styles (Figs.\u0026nbsp;1bc, 2 and ED 10ab) \u003csup\u003e34,35\u003c/sup\u003e spreading rates decreases northwards (15\u0026thinsp;\u0026minus;\u0026thinsp;12 mm/yr) and the crust is continuous along the ridges \u003csup\u003e21,26\u0026ndash;28\u003c/sup\u003e, varying in thickness from ca. 2.5 to 5.5 km (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e and ED Figs.\u0026nbsp;6,7). The crust is thickest in the north at ATLAB-2 (5.5 km) where the spreading rate is lowest. For an ultraslow-spreading ridge this is relatively thick crust \u003csup\u003e4,24\u003c/sup\u003e, but a feeder system connected to the deep melt is invisible and apparently currently inactive (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003ea, \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003ea). Further south at ATLAB-3 where the spreading rates are higher the crust is much thinner, but a deep feeder system is not readily visible here either (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003ec, \u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003ec and ED Figs.\u0026nbsp;7, 8b, 9e). Just north of ATLAB-3, at ATLAB-1 the crust is also thin, but here a well-developed feeder system connects the melt at depth with the crust (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003eb, \u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eb, \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e and ED Fig.\u0026nbsp;6).\u003c/p\u003e\n\u003cp\u003eAlong the Mohns Ridge, a series of AVRs appears along NE lineaments (weak zones) inside the graben, alternating with more quiet areas with individual volcanoes. This is also the case further north, but here the volcanic lineaments are more oblique to the graben owing to the N-S orientation of the Knipovich Ridge (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e and ED Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e0 ab). Recent age dating shows that the lava flows at the rift valley floor is younger than predicted by conventional models, and typically not older than 180 Kyr \u003csup\u003e36\u003c/sup\u003e. Large areas with individual volcanoes outside the AVRs are younger than 25 Kyr and half of the AVRs are younger than 18 Kyr. The youngest volcanic events (8 Kyr) along the Mohns Ridge are found just south of ATLAB-3, between the AVR and the western boundary fault (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003ec and Fig. ED 10a). However, at the three ATLAB locations the volcanic activity at the sea floor and the crustal thickness seems to be independent of depth to conductive melt visible in the EM models (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e,\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e,\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e).\u003c/p\u003e\n\u003cp\u003eHow deep the extensional faults can penetrate into the lithosphere is not known, but the arctic ridges have very deep earthquakes (30\u0026ndash;35 km) indicating the lower limit of brittle lithosphere \u003csup\u003e14\u003c/sup\u003e. Further, deep serpentinization (15 km) suggest that water circulation could extend deep exploiting weak zones, faults, and fractures \u003csup\u003e14\u003c/sup\u003e. Deep faulting, fracturing and water circulation in combination with slow spreading will promote effective cooling and deep and rapid solidification of magmatic events, explaining the much thicker than expected lithosphere along the ridges (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e,\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e,\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e). However, the general presence of melt in the asthenosphere (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e3\u003c/span\u003e, \u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e), the continuity of the crust along the ridges (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003e) and the sustained volcanic activity along weak zones (WZ) at the graben floor (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e bc and ED Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003e0ab) suggest that, at least in a geological timeframe, melt is released steadily through the thick lithosphere.\u003c/p\u003e\n\u003cp\u003eVery rapid drainage of melt from an ultra-deep reservoir was recently documented in an extensional system in the Madagascar Strait \u003csup\u003e37\u003c/sup\u003e. Here, the melt is believed to drain directly from the asthenosphere to the sea floor through dykes that intrude the ca. 55 km thick oceanic lithosphere via pre-existing deep faults, creating huge volcanic edifice at the sea floor. The conditions are favorable for this mechanism (direct deep drainage) to be active also along the Mohns and Knipovich Ridges with faults and weak zones potentially extending deep into the brittle lithosphere. Inversion of synthetic data show that the CSEM method is sensitive to thin conductive vertical features below the ridges (ED Fig.\u0026nbsp;5), but due to the inversion regulation, smoothing the inverted EM model, feeders will appear thicker than they are. Such a thin feeder associated with a deep fault system is visible below the spreading center in the reinverted ATLAB-1 CSEM model (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e2\u003c/span\u003eb and ED Fig.\u0026nbsp;6). Due to the effective cooling of the lithosphere similar and older feeders could already be solidified and thus invisible in the CSEM data.\u003c/p\u003e\n\u003cp\u003eDirect deep drainage can explain young crust in areas with presently thick and apparently inactive lithosphere and adds an important element to the geological model. In the new model, the volcanic centers (Fig.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e4\u003c/span\u003e) are formed at weak zones (Figs.\u0026nbsp;\u003cspan class=\"InternalRef\"\u003e1\u003c/span\u003ec,10) and are ambulatory point sources for melt supply to the ridges. The centers, feeder system and volcanic activity at the sea floor are episodic in geological time, relatively short-lived and random in both time and place. This explains the generally thinner crust, and why the crust can be abnormally thin even with an active feeder system, and why the crust can be both thick and thin in areas with no visible feeders.\u003c/p\u003e\n\u003cp\u003eIn this model the spreading rate cannot predict variation in crustal thickness along the ridges. Instead, it is the extensional forces, local tectonics, lithology, weak zones and deep faults, potentially also associated with direct deep drainage, that controls the development. However, in such a model with thick and brittle lithosphere, it is still the slow and asymmetric plate motions (rate and direction) that is the dominant factor. Melt dynamics is a passive buoyancy-driven response to the tectonic evolution, and as long as melt is available in the asthenosphere it finds its way to the surface when the tectonic process cracks and opens the lithospheric plate. That the melt- and geodynamic models, substantiated by the new EM data, are so closely coupled to the fundamental tectonic processes supports the idea that the model may also apply to ultraslow-spreading ridges in general.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eData and code availability\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAll data needed to evaluate the conclusions in the paper are present in the paper and/or the Methods section. NTNU holds the publishing rights to the data used in this study, and access to the data can be granted upon mutual agreement with the Centre for Geophysical Forecasting at the Norwegian University of Science and Technology in Trondheim. MARE2DEM is a parallel adaptive finite element code for two-dimensional forward and inverse modeling in electromagnetic geophysics. It was used for data modeling and inversion in this project. MARE2DEM is available for download at http://mare2dem.ucsd.edu/.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe data used in this research was obtained from the ATLAB consortium, a collaborative effort including a diverse group of stakeholders from commercial, regulatory, and academic domains, who jointly acquired multi-geophysics and environmental data 2017-2022. The consortium\u0026apos;s funding was made possible through both monetary contributions and in-kind support from all participating partners. The ATLAB consortium includes the following entities: NTNU CGF (serving as the scientific operator and lead), NTNU IGP, NPD, EMGS, Equinor, AkerBP, NORCE, PGS, CGG, Shearwater, OFG, InApril, Allton, and TGS.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contributions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eConceptualization: SEJ, HEFA, BA\u003c/p\u003e\n\u003cp\u003eMethodology: SEJ, HEFA, BA, RM, OMP, MP, ML, SW\u003c/p\u003e\n\u003cp\u003eInvestigation: SEJ, HEFA, BA, RM, OMP, MP, ML\u003c/p\u003e\n\u003cp\u003eVisualization: SEJ, HEFA, BA, RM, OMP, MP, ML, KOL\u003c/p\u003e\n\u003cp\u003eFunding acquisition: SEJ, BA, ML\u003c/p\u003e\n\u003cp\u003eProject administration: SEJ, BA, ML, KOL\u003c/p\u003e\n\u003cp\u003eSupervision: SEJ, BA\u003c/p\u003e\n\u003cp\u003eWriting \u0026ndash; original draft: SEJ, OMP, MP\u003c/p\u003e\n\u003cp\u003eWriting \u0026ndash; review \u0026amp; editing: SEJ, HEFA, BA, RM, OMP, MP, ML, KOL, SW\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eCompeting interest declaration\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eSolomon, S. C. \u0026amp; Toomey, D. R. The Structure of Mid-Ocean Ridges. \u003cem\u003eAnnu Rev Earth Planet Sci\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 329\u0026ndash;364 (1992).\u003c/li\u003e\n\u003cli\u003eTeam, T. M. E. L. T. S. Imaging the Deep Seismic Structure Beneath a Mid-Ocean Ridge: The MELT Experiment. \u003cem\u003eScience\u003c/em\u003e \u003cstrong\u003e280\u003c/strong\u003e, 1215\u0026ndash;1218 (1998).\u003c/li\u003e\n\u003cli\u003eWanless, V. D. \u0026amp; Shaw, A. M. Lower crustal crystallization and melt evolution at mid-ocean ridges. \u003cem\u003eNat Geosci\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 651\u0026ndash;655 (2012).\u003c/li\u003e\n\u003cli\u003eBown, J. W. \u0026amp; White, R. S. Variation with spreading rate of oceanic crustal thickness and geochemistry. \u003cem\u003eEarth Planet Sci Lett\u003c/em\u003e \u003cstrong\u003e121\u003c/strong\u003e, 435\u0026ndash;449 (1994).\u003c/li\u003e\n\u003cli\u003eRaitt, M. The crustal rocks. \u003cem\u003eThe Sea\u003c/em\u003e \u003cstrong\u003e3\u003c/strong\u003e, 85\u0026ndash;102 (1963).\u003c/li\u003e\n\u003cli\u003eReid, I. \u0026amp; Jackson, H. R. Oceanic spreading rate and crustal thickness. \u003cem\u003eMar Geophys Res\u003c/em\u003e \u003cstrong\u003e5\u003c/strong\u003e, 165\u0026ndash;172 (1981).\u003c/li\u003e\n\u003cli\u003eWhite, R. S., Minshull, T. A., Bickle, M. J. \u0026amp; Robinson, C. J. Melt Generation at Very Slow-Spreading Oceanic Ridges: Constraints from Geochemical and Geophysical Data. \u003cem\u003eJ Pet.\u003c/em\u003e \u003cstrong\u003e42\u003c/strong\u003e, 1171\u0026ndash;1196 (2001).\u003c/li\u003e\n\u003cli\u003eZhou, H. \u0026amp; Dick, H. J. B. Thin crust as evidence for depleted mantle supporting the Marion Rise. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e494\u003c/strong\u003e, 195\u0026ndash;200 (2013).\u003c/li\u003e\n\u003cli\u003eSauter, D. \u003cem\u003eet al.\u003c/em\u003e Evidence for magma entrapment below oceanic crust from deep seismic reflections in the Western Somali Basin. \u003cem\u003eGeology\u003c/em\u003e \u003cstrong\u003e44\u003c/strong\u003e, 407\u0026ndash;410 (2016).\u003c/li\u003e\n\u003cli\u003eJohansen, S. E. \u003cem\u003eet al.\u003c/em\u003e Deep electrical imaging of the ultraslow-spreading Mohns Ridge. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e567\u003c/strong\u003e, 379\u0026ndash;383 (2019).\u003c/li\u003e\n\u003cli\u003eKey, K., Constable, S., Liu, L. \u0026amp; Pommier, A. Electrical image of passive mantle upwelling beneath the northern East Pacific Rise. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e495\u003c/strong\u003e, 499\u0026ndash;502 (2013).\u003c/li\u003e\n\u003cli\u003eLay, T. Seismological Grand Challenges in Understanding Earth\u0026rsquo;s Dynamic Systems. in \u003cem\u003eReport to the National Science Foundation (IRIS Consortium\u003c/em\u003e (2009).\u003c/li\u003e\n\u003cli\u003eRychert, C. A., Harmon, N., Constable, S. \u0026amp; Wang, S. The Nature of the Lithosphere-Asthenosphere Boundary. \u003cem\u003eJ Geophys Res Solid Earth\u003c/em\u003e \u003cstrong\u003e125\u003c/strong\u003e, 2018\u0026ndash;016463 (2020).\u003c/li\u003e\n\u003cli\u003eSchlindwein, V. \u0026amp; Schmid, F. Mid-ocean-ridge seismicity reveals extreme types of ocean lithosphere. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e535\u003c/strong\u003e, 276\u0026ndash;279 (2016).\u003c/li\u003e\n\u003cli\u003eWang, S., Constable, S., Rychert, C. A. \u0026amp; Harmon, N. A Lithosphere-Asthenosphere Boundary and Partial Melt Estimated Using Marine Magnetotelluric Data at the Central Middle Atlantic Ridge. \u003cem\u003eGeochem Geophys Geosyst\u003c/em\u003e \u003cstrong\u003e21\u003c/strong\u003e, 2020\u0026ndash;009177 (2020).\u003c/li\u003e\n\u003cli\u003eMichael, P. J. \u003cem\u003eet al.\u003c/em\u003e Magmatic and amagmatic seafloor generation at the ultraslow-spreading Gakkel ridge, Arctic Ocean. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e423\u003c/strong\u003e, 956\u0026ndash;961 (2003).\u003c/li\u003e\n\u003cli\u003eNiu, X. \u003cem\u003eet al.\u003c/em\u003e Along-axis variation in crustal thickness at the ultraslow spreading Southwest Indian Ridge (50\u0026deg;E) from a wide-angle seismic experiment. \u003cem\u003eGeochem Geophys Geosystems\u003c/em\u003e \u003cstrong\u003e16\u003c/strong\u003e, 468\u0026ndash;485 (2015).\u003c/li\u003e\n\u003cli\u003eCannat, M. How thick is the magmatic crust at slow spreading oceanic ridges? \u003cem\u003eJ Geophys Res Solid Earth\u003c/em\u003e \u003cstrong\u003e101\u003c/strong\u003e, 2847\u0026ndash;2857 (1996).\u003c/li\u003e\n\u003cli\u003eCannat, M. \u003cem\u003eet al.\u003c/em\u003e Spreading rate, spreading obliquity, and melt supply at the ultraslow spreading Southwest Indian Ridge. \u003cem\u003eGeochem Geophys Geosystems\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, 2007 001676 (2008).\u003c/li\u003e\n\u003cli\u003eDick, H. J. B., Lin, J. \u0026amp; Schouten, H. An ultraslow-spreading class of ocean ridge. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e426\u003c/strong\u003e, 405\u0026ndash;412 (2003).\u003c/li\u003e\n\u003cli\u003eRitzmann, O., Jokat, W., Mjelde, R. \u0026amp; Shimamura, H. Crustal structure between the Knipovich Ridge and the Van Mijenfjorden (Svalbard. \u003cem\u003eMar Geophys Res\u003c/em\u003e \u003cstrong\u003e23\u003c/strong\u003e, 379\u0026ndash;401 (2002).\u003c/li\u003e\n\u003cli\u003eStandish, J. J., Dick, H. J. B., Michael, P. J., Melson, W. G. \u0026amp; O\u0026rsquo;Hearn, T. MORB generation beneath the ultraslow spreading Southwest Indian Ridge (9-25\u0026deg;E): Major element chemistry and the importance of process versus source: SWIR MORB-PROCESS VERSUS SOURCE. \u003cem\u003eGeochem Geophys Geosystems\u003c/em\u003e \u003cstrong\u003e9\u003c/strong\u003e, (2008).\u003c/li\u003e\n\u003cli\u003eCarbotte, S. M. \u0026amp; Scheirer, D. S. Variability of ocean crustal structure created along the global mid-ocean ridge. \u003cem\u003eHydrogeol Ocean Lithosphere\u003c/em\u003e 59\u0026ndash;107 (2004).\u003c/li\u003e\n\u003cli\u003eJokat, W. \u003cem\u003eet al.\u003c/em\u003e Geophysical evidence for reduced melt production on the Arctic ultraslow Gakkel mid-ocean ridge. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e423\u003c/strong\u003e, 962\u0026ndash;965 (2003).\u003c/li\u003e\n\u003cli\u003eEiken, O. Seismic atlas of western Svalbard: a selection of regional seismic transects. (1994).\u003c/li\u003e\n\u003cli\u003eConley, M. M. \u0026amp; Dunn, R. A. Seismic shear wave structure of the uppermost mantle beneath the Mohns Ridge: UPPERMOST MANTLE BENEATH THE MOHNS RIDGE. \u003cem\u003eGeochem Geophys Geosystems\u003c/em\u003e \u003cstrong\u003e12\u003c/strong\u003e, (2011).\u003c/li\u003e\n\u003cli\u003eHermann, T. \u0026amp; Jokat, W. Crustal structures of the Boreas Basin and the Knipovich Ridge, North Atlantic. \u003cem\u003eGeophys J Int\u003c/em\u003e \u003cstrong\u003e193\u003c/strong\u003e, 1399\u0026ndash;1414 (2013).\u003c/li\u003e\n\u003cli\u003eTectonostratigraphic atlas of the North-East Atlantic Region. in \u003cem\u003eGeological Survey of Denmark and Greenland (GEUS\u003c/em\u003e (eds. Hopper, J. R. et al.) (Copenhagen, 2014).\u003c/li\u003e\n\u003cli\u003ePedersen, R. B. \u003cem\u003eet al.\u003c/em\u003e Discovery of a black smoker vent field and vent fauna at the Arctic Mid-Ocean Ridge. \u003cem\u003eNat Commun\u003c/em\u003e \u003cstrong\u003e1\u003c/strong\u003e, 126 (2010).\u003c/li\u003e\n\u003cli\u003eNi, H., Keppler, H. \u0026amp; Behrens, H. Electrical conductivity of hydrous basaltic melts: implications for partial melting in the upper mantle. \u003cem\u003eContrib Miner. Pet.\u003c/em\u003e \u003cstrong\u003e162\u003c/strong\u003e, 637\u0026ndash;650 (2011).\u003c/li\u003e\n\u003cli\u003eConstable, S. SEO3: A new model of olivine electrical conductivity. \u003cem\u003eGeophys J Int\u003c/em\u003e \u003cstrong\u003e166\u003c/strong\u003e, 435\u0026ndash;437 (2006).\u003c/li\u003e\n\u003cli\u003eChen, Y. J. Oceanic crustal thickness versus spreading rate. \u003cem\u003eGeophys Res Lett\u003c/em\u003e \u003cstrong\u003e19\u003c/strong\u003e, 753\u0026ndash;756 (1992).\u003c/li\u003e\n\u003cli\u003eMeier, M. \u003cem\u003eet al.\u003c/em\u003e Segment‐Scale Seismicity of the Ultraslow Spreading Knipovich Ridge. \u003cem\u003eGeochem Geophys Geosystems\u003c/em\u003e \u003cstrong\u003e22\u003c/strong\u003e, 2020 009375 (2021).\u003c/li\u003e\n\u003cli\u003eBruvoll, V., Breivik, A. J., Mjelde, R. \u0026amp; Pedersen, R. B. Burial of the Mohn‐Knipovich seafloor spreading ridge by the Bear Island Fan: Time constraints on tectonic evolution from seismic stratigraphy. \u003cem\u003eTectonics\u003c/em\u003e \u003cstrong\u003e28\u003c/strong\u003e, 2008TC002396 (2009).\u003c/li\u003e\n\u003cli\u003eLim, A., Br\u0026ouml;nner, M., Johansen, S. E. \u0026amp; Dumais, M. Hydrothermal Activity at the Ultraslow‐Spreading Mohns Ridge: New Insights From Near‐Seafloor Magnetics. \u003cem\u003eGeochem. Geophys. Geosystems\u003c/em\u003e \u003cstrong\u003e20\u003c/strong\u003e, 5691\u0026ndash;5709 (2019).\u003c/li\u003e\n\u003cli\u003eStubseid, H. H., Bjerga, A., Haflidason, H., Pedersen, L. E. R. \u0026amp; Pedersen, R. B. Volcanic evolution of an ultraslow-spreading ridge. \u003cem\u003eNat. Commun.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 4134 (2023).\u003c/li\u003e\n\u003cli\u003eFeuillet, N. \u003cem\u003eet al.\u003c/em\u003e Birth of a large volcanic edifice offshore Mayotte via lithosphere-scale dyke intrusion. \u003cem\u003eNat. Geosci.\u003c/em\u003e \u003cstrong\u003e14\u003c/strong\u003e, 787\u0026ndash;795 (2021).\u003c/li\u003e\n\u003cli\u003eAnderson, D. L. Lithosphere, asthenosphere, and perisphere. \u003cem\u003eRev Geophys\u003c/em\u003e \u003cstrong\u003e33\u003c/strong\u003e, 125\u0026ndash;149 (1995).\u003c/li\u003e\n\u003c/ol\u003e"},{"header":"Methods","content":"\u003cp\u003eSix electromagnetic surveys (3 CSEM and 3 MT) have been collected at the northern Mohns- and the northern Knipovich ridges (Figs. 1). The data were collected by the ATLAB program (Atlantic Laboratory) and consists of ATLAB-1 \u003csup\u003e10\u003c/sup\u003e (Figs. 2b,3b), ATLAB-2 (Figs. 2a,3a) and ATLAB-3 (Figs. 2c,3c). Data analysis and inversion of the ATLAB-1 data are also described in detail in online method in reference \u003csup\u003e10\u003c/sup\u003e.\u003c/p\u003e\n\u003ch2\u003eData analysis and inversion ATLAB-2\u003c/h2\u003e\n\u003cp\u003eThe ATLAB-2 data set was collected across the northern Knipovich Ridge onto the Svalbard continental margin by 39 broadband ocean-bottom electromagnetic receivers deployed along a 120-km-long profile (Fig.\u0026nbsp;1). The receivers recorded the horizontal electric and magnetic fields in time series. Based on pre-survey modelling \u003csup\u003e39\u003c/sup\u003e, the receiver spacing was set to 2.5 km in the central part and 5 km for the last 6 receivers at both end of the profile. A 281-m-long horizontal electric dipole source was towed across the receivers, emitting a time-varying electric current of 1,200 A. The source was towed 30\u0026ndash;50 meters above the seafloor, but in areas with very rough bathymetry the source was up to 250 meters above the seafloor. The source signal was optimized to ensure high current output over a certain low frequency range, and the output waveform of the source with a 4s fundamental period covers more than one decade with logarithmically spaced frequencies, ranging from 0.25 Hz to 5.25 Hz, with almost equally high current output.\u003c/p\u003e\n\u003cp\u003eFrom the measured time series, the CSEM processing \u003csup\u003e40\u003c/sup\u003e extracts high-quality data for the frequencies of 0.25, 0.5, 1.0, 2.5 and 5.25 Hz. Due to variations in subsurface resistivity and water depth the electric field data vary significantly along the profile (ED Fig.\u0026nbsp;1a). For the extracted frequencies, we also obtained precise noise estimates, which were used to compute appropriate data misfits in the inversion. From up to 6-day-long time series, continuously registered by the seafloor electromagnetic receivers, it was also possible to process high-quality MT data for periods of 1.4 s to 1,000 s using robust multi-station processing \u003csup\u003e41\u003c/sup\u003e. MT data examples are shown in ED Fig.\u0026nbsp;1b. Dimensionality analysis of the MT data reveals 3D distortions for some receivers located in the deepest part of the Knipovich Ridge graben (Figs.\u0026nbsp;1b) and the noisiest data were not included in the 2D inversion.\u003c/p\u003e\n\u003cp\u003eThe electric field CSEM data (Fig. 2a) were inverted using a regularized, nonlinear two-dimensional inversion with a finite element forward engine \u003csup\u003e42\u003c/sup\u003e. In the Cartesian coordinate system, used in the 2D inversion code, the x-axis is oriented normal to the observation profile, the y-axis is parallel to the profile and the z-axis is pointing downwards. The Occam-type inversion seeks a smooth resistivity model that explains the measured data within the data uncertainty with the consideration of triaxial electric anisotropy. However, we made the simple assumption of transverse isotropy, where the resistivities in the horizontal directions are assumed to be equal:\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" height=\"93\" width=\"301\"\u003e\u003c/p\u003e\n\u003cp\u003ewhere \u0026rho;\u003csub\u003ex\u003c/sub\u003e is the resistivity in the direction normal to the observation profile (parallel to the ridge), \u0026rho;\u003csub\u003ey\u003c/sub\u003e is the resistivity parallel to the observation profile, and \u0026rho;\u003csub\u003ez\u003c/sub\u003e is the resistivity in the vertical direction. To stabilize the inverse problem, both smoothness regularization and anisotropy regularization were applied, penalizing non-smooth resistivity variations and differences between the resistivity components in the horizontal and vertical directions.\u003c/p\u003e\n\u003cp\u003eIn the MT inversion we seek for a triaxially anisotropic resistivity model, allowing all three diagonal components of the resistivity tensor to vary individually:\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" height=\"99\" width=\"173\"\u003e\u003c/p\u003e\n\u003cp\u003eThe transverse electric (TE) mode (electric current flowing parallel to the ridge) and the transverse magnetic (TM) mode (electric current flowing perpendicular to the ridge) of the MT data are then the off-diagonal elements of the impedance tensor, rotated into the observation profile. The diagonal elements are zero if the model is 2D:\u003c/p\u003e\n\u003cp\u003e\u003cimg src=\"data:image/png;base64,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\" height=\"86\" width=\"252\"\u003e\u003c/p\u003e\n\u003cp\u003eDue to strong bathymetry variations, especially in the eastern part of the profile, the MT source signal does no longer behave as a pure vertically propagating plane wave \u003csup\u003e43\u003c/sup\u003e. Consequently, the TE-mode data are primarily sensitive to the resistivity component normal to the observation profile \u0026rho;\u003csub\u003ex\u003c/sub\u003e, and the TM-mode data are sensitive to both the resistivity parallel to the profile \u0026rho;\u003csub\u003ey\u003c/sub\u003e and the vertical resistivity component \u0026rho;\u003csub\u003ez\u003c/sub\u003e. The MT inversion can be stabilized with anisotropy regularization. The seafloor topography in the starting model was composed of both the bathymetry data measured along the profile during the survey and global bathymetry data \u003csup\u003e44\u003c/sup\u003e for the rest of the model. In this way we could account for the Svalbard islands when simulating the long period MT data and get improved convergence in the inversions.\u003c/p\u003e\n\u003cp\u003eIndividual TE-mode inversion converged to an RMS of 1.15, assuming a 10% relative data uncertainty. The data from the receivers in the central Knipovich Ridge graben have slightly poorer fits than other receivers. The TE-mode inversion recovers a smooth geologically realistic resistivity model (ED Fig. 2a), suggesting strong conductivity alignment with the N-S striking Knipovich Ridge and the general N-S geological strike of the area.\u003c/p\u003e\n\u003cp\u003eThe individual TM-mode inversion result (ED Fig. 2b) has an RMS misfit of 1.5, assuming 10% relative data uncertainty. The data from receivers in the deep graben have a particularly poor fit and the resulting resistivity model has strong imaging artefacts beneath these receivers which were not present either in the TE-mode inversion or in the CSEM inversion. The CSEM data have good sensitivity to this depth interval and show a much simpler resistivity structure. Both the dominating conductive region \u0026rdquo;x\u0026rdquo; and deep resistive region \u0026rdquo;y\u0026rdquo; (ED Fig. 2b) appear as very unrealistic and suggest strong coastal effects in the TM mode data \u003csup\u003e43,45\u003c/sup\u003e. This effect, caused by the Svalbard Islands, is apparently much less pronounced or barely present in the TE-mode model. The combined TE- and TM-mode inversion model (ED Fig. 2c) has clear similarities to the single TE -mode inversion model, but coastal effects from the TM-mode are also preserved. A more detailed analysis of the MT data misfit for the combined inversion result show that the deep graben receivers have negative TE-mode phases for periods T \u0026lt; 250 s and negative TM-mode phases for periods T \u0026lt; 50 s. Further, the TM-mode amplitude data are significantly depressed below the TE amplitude responses (ED Fig. 1c). These observations are exactly the previous identified coast effects. The negative TM-mode phases and depressed TM-mode amplitude at relatively short periods are described as inductive and galvanic coastal effect distortions, respectively \u003csup\u003e45\u003c/sup\u003e, the negative TE-mode phases and resistivity cusp shape (Rx 17 and Rx18) at relatively long periods are identified as inductive coast effect distortion \u003csup\u003e43\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eTo find a common model in the joint CSEM and MT inversion we had to increase the relative uncertainty of the CSEM data to 5%, allowing for less constraint on the model from the CSEM data giving relatively higher weight to the MT data (ED Fig. 3). To test such effects and to optimize the interpretation we also performed inversion of synthetic data, including separate TE- and TM-mode inversions, joint TE- and TM-mode inversion and joint CSEM and MT inversion (Ed Fig. 4). The synthetic TE-mode inversion model (ED Fig. 4c) reconstructs the main features of the true model but underestimates the resistivities in the lithosphere (layer \u0026rdquo;C\u0026rdquo;) due to the lower resolution to resistors than conductors and smears out the deep melt anomaly (Letter D). In contrast, the individual TM-mode inversion does not recover the deeper parts of the model (ED Fig. 4 d). The joint TE and TM-mode inversion also reconstructs important model elements (Letters C and D in ED Fig. 4b), but the modes are smeared together and complicates the interpretation with inversion artifacts (Letters f and g in ED Fig. 4b).\u003c/p\u003e\n\u003cp\u003eThe CSEM and MT joint inversion shows that including CSEM data in the inversion can improve resistivity imaging, also of the deeper parts of the model. The deep conductive anomaly (\u0026rdquo;D\u0026rdquo; in ED Fig. 4e ) has improved image focus compared to the TE-mode inversion, but is still a bit smeared out due to smoothness regularization. The high resistivities at depth (Letters g and g* in ED Fig. 4e) are probably end of line effects, but still difficult to remove due to limitations in extension of aperture. Similar effects are observed in the CSEM and MT joint inversion result for the measured data (ED Fig. 3).\u003c/p\u003e\n\u003cp\u003eThe Knipovich EM survey is characterized by partly symmetrical seafloor topography to the profile. In addition, data dimensionality analyses and small on-diagonal elements in the MT-impedance tensor all suggest 2D structures beneath the survey profile and limited data sensitivity to side structures. These reasons suggest that the 2D inversion models are representative for the geological structures below the survey line. However, reconstruction of the subsurface resistivity models from CSEM and MT data is a non-unique inverse problem and potentially many different models can fit the measured data. The subsurface interpretations are based on the various inversion models and tests. However, of the single inversion models it is the high frequency CSEM model that best represents the shallow subsurface and due to the apparently few inversion artifacts the TE-mode model is interpreted to best represents the deep subsurface in this case.\u003c/p\u003e\n\u003ch2\u003eInversion of thin vertical features in the crust and the upper mantle\u003c/h2\u003e\n\u003cp\u003eTo test how the CSEM models image vertical structures we generated synthetic data, added 3% random noise and inverted the data with the same inversion setup as for the real data. The inversion result are good representations of the general geological model (ED Fig.\u0026nbsp;5). For comparison inversions were done with and without a conductive melt feeder channel, thinning upwards (10 ohmm), and deep fault zones (1 ohmm). The synthetic inversion results show significant decrease in resistivity along the vertical channel not observed in the inversion without the channel or in the real data inversion (Fig.\u0026nbsp;2a). The conductive anomaly is smeared out due to resolution and inversion regularization (smoothing). The deep conductive fault zones reach several kilometers into the mantle and the inversion model images the fault zones, but also due to resolution and regularization they have a more complex response than the melt channel. Similar fault-like features are observed in the crust in the real data, but not in the mantle at the ATLAB-2 location. The inversions explain the synthetic data within the uncertainty (RMS\u0026thinsp;=\u0026thinsp;1).\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eNew data analysis and inversion of ATLAB-1\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAcquisition parameters and processing sequence of the ATLAB-1 data set is detailed in the methods section in the original publication \u003csup\u003e10\u003c/sup\u003e. A new EM data inversion of ATLAB-1 (Fig.\u0026nbsp;2b) confirms a vertical conductive feature beneath Loki\u0026rsquo;s Castle interpreted as thin feeder system. ED Fig.\u0026nbsp;6 (upper panel) shows the result of the electric data inversion where the model is imaged to 15 km compared to 8 km in the original version. This inversion reached a data misfit of RMS 1.0. A sensitivity modelling scenario designed to test the robustness of the vertical conductive feature is shown in the lower panel of ED Fig.\u0026nbsp;6. By removing the vertical conductor, the test shows that the ATLAB-1 CSEM data is clearly sensitive to changes in conductivity also between 5 and 15 km depth. It is mainly the long offsets of the CSEM data that record variations at these depths.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData analysis and inversion ATLAB-3\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eAcquisition parameters and processing sequence for ATLAB-3 are, with minor adjustments for local conditions, similar to ATLAB-1 \u003csup\u003e10\u003c/sup\u003e and ATLAB-2. ATLAB-3 was collected at the northernmost Mohns ridge by 24 receivers along a 54 km long transect (Fig.\u0026nbsp;1c) just south of the ATLAB-1 transect \u003csup\u003e10\u003c/sup\u003e at Loki\u0026rsquo;s Castle to study subsurface changes over relatively short distances along the Mohns ridge. The ATLAB-3 area has very rough bathymetry (Fig.\u0026nbsp;1). Based on experience from the other ATLAB surveys, the EM receiver instruments at ATLAB-3 were carefully placed with an average spacing of 1700m. The 281-m-long electric dipole emitted a time-varying electric current of 1,200 A. Ideally the source should be towed 30\u0026ndash;50 meters above the seafloor, but due to the very rough bathymetry, the distance from source to sea floor significantly. The source signal was optimized to ensure high current output over frequencies from 0.5 Hz to 9 Hz and the CSEM processing \u003csup\u003e40\u003c/sup\u003e extracted high-quality data for the selected frequencies and associated noise estimates were used to compute data misfits in the inversion. From up to 9-days registration, high-quality MT data was extracted with periods from 5.5 s to 1500 s using multi-station processing \u003csup\u003e41\u003c/sup\u003e. The phase tensor ellipses reveal 3D distortions for some of the receivers along the transect. Tests showed that omitting these receivers did not have major effects on the inversion result. The MT data was analyzed using the MTPy software package \u003csup\u003e46\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eThe EM data of ATLAB-3 was inverted using a regularized, nonlinear two-dimensional inversion code with a finite element forward engine \u003csup\u003e42\u003c/sup\u003e. The electric and magnetic fields CSEM data were first inverted separately (ED Fig.\u0026nbsp;7). The electric field CSEM data inversion reached data misfit of RMS 1.0 and the magnetic field CSEM data inversion also reached RMS of 1.0. The two independent CSEM inversion images show strong similarities which increases trust in the results. Subsequently, the MT data was inverted using both the TE and TM mode (ED Fig.\u0026nbsp;8a). The MT inversion reached data misfit RMS of 1.06. Finally, the MT and CSEM data of ATLAB-3 were inverted jointly (ED Fig.\u0026nbsp;8b), to find a resistivity model that explains both datasets. The joint CSEM and MT inversion reached an overall data misfit of RMS 1.06.\u003c/p\u003e\n\u003cp\u003eIt is expected that the shallow part of the joint inversion result is dominated by the CSEM data, while due to the frequency-dependent attenuation of the electromagnetic signal, the deeper part of the joint inversion model is dominated by the low-frequency MT data. It has also been shown that the CSEM data constrains the joint inversion and, in this way, improve imaging, also of the deeper parts of the model \u003csup\u003e10\u003c/sup\u003e. To test such effects, improve the interpretation and to illustrate the effect of joint CSEM and MT inversion we also performed inversion of synthetic data relevant for the ATLAB-3 area. We tested three different conceptual geological models, one without a feeder system (no feeder), a model where the feeder system is partially solidified into the lithosphere (not connected) and a completely open feeder system (connected) up to the oceanic crust (ED Fig.\u0026nbsp;9).\u003c/p\u003e\n\u003cp\u003eAfter forward simulation of the three scenario models, the synthetic data was contaminated with uncertainty (noise) before inversions. The inverted MT data (ED Figs.\u0026nbsp;8b, c, d) provides a reasonable representation of the no-feeder model and of the open system (connected). Although there are changes also in the model where the supply system is terminated in the middle of the lithosphere (not connected), the imaging here is partly misleading. The results from the joint inversion of the synthetic data (Fig.\u0026nbsp;8e) show that combining all the data in this case significantly improves the imaging, also of the deeper parts of the models. The synthetic inversion results do not completely rule out that parts of the lithosphere may have increased conductivity, but if there had been a significant feeder system below ATLAB-3, this would have been imaged both in the MT models and in the joint models.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods References\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003e10. Johansen, S. E. \u003cem\u003eet al.\u003c/em\u003e Deep electrical imaging of the ultraslow-spreading Mohns Ridge. \u003cem\u003eNature\u003c/em\u003e \u003cstrong\u003e567\u003c/strong\u003e, 379\u0026ndash;383 (2019).\u003c/p\u003e\n\u003cp\u003e39. Tegnander, J. F. \u003cem\u003eElectromagnetic Modelling and Inversion of Seafloor Massive Sulphide Deposits - Imaging of SMS Deposits and the Asthenosphere at the Atlantic Mid-Ocean Ridge (Master\u003c/em\u003e. (NTNU, Trondheim, 2017).\u003c/p\u003e\n\u003cp\u003e40. Myer, D., Constable, S. \u0026amp; Key, K. Broad-band waveforms and robust processing for marine CSEM surveys. \u003cem\u003eGeophys J Int\u003c/em\u003e \u003cstrong\u003e184\u003c/strong\u003e, 689\u0026ndash;698 (2011).\u003c/p\u003e\n\u003cp\u003e41. Egbert, G. D. Robust multiple-station magnetotelluric data processing. \u003cem\u003eGeophys J Int\u003c/em\u003e \u003cstrong\u003e130\u003c/strong\u003e, 475\u0026ndash;496 (1997).\u003c/p\u003e\n\u003cp\u003e42. Key, K. MARE2DEM: a 2-D inversion code for controlled-source electromagnetic and magnetotelluric data. \u003cem\u003eGeophys J Int\u003c/em\u003e \u003cstrong\u003e207\u003c/strong\u003e, 571\u0026ndash;588 (2016).\u003c/p\u003e\n\u003cp\u003e43. Key, K. \u0026amp; Constable, S. Coast effect distortion of marine magnetotelluric data: Insights from a pilot study offshore northeastern Japan. \u003cem\u003ePhys Earth Planet Inter\u003c/em\u003e \u003cstrong\u003e184\u003c/strong\u003e, 194\u0026ndash;207 (2011).\u003c/p\u003e\n\u003cp\u003e44. G.E.B.C.O. \u003cem\u003eThe General Bathymetric Chart of the Oceans\u003c/em\u003e. (GEBCO, 2023).\u003c/p\u003e\n\u003cp\u003e45. Wang, S., Constable, S., Reyes-Ortega, V. \u0026amp; Rychert, C. A. A newly distinguished marine magnetotelluric coast effect sensitive to the lithosphere\u0026ndash;asthenosphere boundary. \u003cem\u003eGeophys J Int\u003c/em\u003e \u003cstrong\u003e218\u003c/strong\u003e, 978\u0026ndash;987 (2019).\u003c/p\u003e\n\u003cp\u003e46. Kirkby, A., Zhang, F., Peacock, J., Hassan, R. \u0026amp; Duan, J. The MTPy software package for magnetotelluric data analysis and visualisation. \u003cem\u003eJ Open Source Softw\u003c/em\u003e \u003cstrong\u003e4\u003c/strong\u003e, (2019).\u003c/p\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"","lastPublishedDoi":"10.21203/rs.3.rs-3999138/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-3999138/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eThe oceanic crust formed at mid-ocean ridges constitutes seventy percent of the earth's solid surface \u003csup\u003e1–3\u003c/sup\u003e. The crust in ocean basins is approximately seven km thick \u003csup\u003e4,5\u003c/sup\u003e, but when the spreading rate drops below 15-20 mm/a (ultraslow-spreading), crustal thickness decreases rapidly \u003csup\u003e6,7\u003c/sup\u003e. The paradigm view is that thickness depends on the spreading rate \u003csup\u003e6–8\u003c/sup\u003e. However, current models for the ultraslow-spreading ridges are not based on direct imaging \u003csup\u003e9\u003c/sup\u003e, and it is not well understood how tectonic processes, melt dynamics, \u003csup\u003e2,3,10,11\u003c/sup\u003e, lithospheric structure, and crustal formation \u003csup\u003e12–15\u003c/sup\u003e interact along ridges \u003csup\u003e12–14,16,17\u003c/sup\u003e. New electromagnetic (EM) data across the Mohns and Knipovich ridges show that, outside the volcanic centres, the lithosphere appears closed to melt migration, and instead of normal thinning, the lithosphere is unusually thick (35-45 km) beneath the ridges. Crustal thickness varies along the strike and is thinnest where the spreading rate is the highest, contrary to the prediction of conventional models\u003csup\u003e6–8\u003c/sup\u003e. In the new model, ambulatory volcanic centres, forming along weak zones, and fault-induced ultra-deep direct drainage of melt from the asthenosphere explain the EM data. Volcanic centres are point sources of melt supply and both centres, feeder channels and volcanism are episodic, relatively short-lived, and random in time and place. In this model with a thick and brittle lithosphere, the plate motions (rate and direction), local tectonics, lithology, weak zones, and deep faults associated with the deep drainage, control the development. Melt dynamics and crustal formation are passive buoyancy-driven responses to the tectonic development. The fact that the proposed ridge model is closely connected to fundamental tectonic processes support the idea that the model can also be applied to ultraslow-spreading ridges in general.\u003c/p\u003e","manuscriptTitle":"New model for ultraslow-spreading ridges","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-03-21 16:09:59","doi":"10.21203/rs.3.rs-3999138/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"aee0260d-c396-470e-ad8c-9b0bbf7d1295","owner":[],"postedDate":"March 21st, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":29598120,"name":"Earth and environmental sciences/Solid Earth sciences/Geophysics"},{"id":29598121,"name":"Earth and environmental sciences/Solid Earth sciences/Geodynamics"}],"tags":[],"updatedAt":"2024-03-21T16:09:59+00:00","versionOfRecord":[],"versionCreatedAt":"2024-03-21 16:09:59","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-3999138","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-3999138","identity":"rs-3999138","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}