What Drives Plate Motion ?

Plate motion was widely thought to be a manifestation of mantle dynamics. However, an in-depth investigation shows this understanding incompetent. Here we propose, the daily tides yield varying pressures between oceans, the application of these pressures to the continent's sides forms enormously unequal horizontal forces, the net effect of these forces provides lateral push to the continent and may cause it to move horizontally, further, the moving continent by basal drag entrains its adjacent crust to move, these totally give birth to plate motion. A roughly estimation shows that the oceanic tide-generating force may independently give South American, African, Indian, and Australian continents a movement of respectively 2.8, 4.2, 5.6, and 6.3 cm/yr. Some torque effects may rotate North American and Eurasian continents, and a combination of two lateral pushes offers Pacific Plate unusual motion (nearly orthogonal to Australian plate’s motion).

The distribution of ridge push, basal friction, and collisional forces around a continental plate is outlined in Figure 1, in which 145 trench suction is neglected. These forces represent the primary driving forces. It is assumed that Plate A moves towards the left, Plate B exerts a collisional force (i.e., Fc) on its left side, the oceanic ridge exerts a push force (i.e., FRP) on its right side, and the asthenosphere exerts a friction force (i.e., Fb) on its base. This force distribution requires the stress produced by these forces to be distributed mostly in the lowermost part of the plate.
We developed a simplified model to verify this expectation (Figure 2 (top)). The model is made of rocks materials and is 150 straight, meaning that the Earth's curvature is not considered. The model is assumed to be homogeneous and isotropic, and its thickness and length are 100 km and 290 km, respectively. Along the horizontal direction, it receives a collisional force Fc, and this force is resistive and uniformly exerted on its left side; the oceanic ridge exerts a push force FRP on its right side, this force is driving and increases with depth; the mantle exerts a frictional force Fb on its base, and this force is resistive. These forces realize a horizontal force balance. Along the vertical direction, it is supported by the mantle, and its gravity is balanced 155 out by the supporting from the mantle.
Finite element analysis software (i.e., Abaqus) is used to resolve the stress caused by these forces. The model's bottom is given a remote boundary condition. As the upper part of the lithospheric plate is elastic and brittle, whereas the lower part is plastic and ductile, this requires to assume that the physical property of the model is vertically transited from elasticity to plasticity.
The inputs include the vertical pressure caused by the rock's weight and the lateral pressures caused by the loads (i.e.,FRP,Fb,160 and Fc). The outputs include the stress produced by the vertical pressure alone and the stress produced by a combination of the vertical and lateral pressures. The two-dimensional frame allows to know the horizontal stress (i.e., S11) and the vertical stress (i.e., S22).
However, in order to expedite the deduction, we will only discuss the horizontal stress (i.e., S11) in the following. The elastic modulus, Poisson ratio, and rock density of the model are set to 100,000 MPa, 0.3, and 2,690 kg/m 3 , respectively. The pressure 165 caused by the rock's weight yields Set I data; The FRP is given as 4.0×10 12 N m -1 , which is generally accepted by the scientific community (Turcotte and Schubert, 2004). It is assumed that Fb and Fc are 80% and 20% of FRP, respectively. The pressures caused by these loads yield Set II data; To test the stress variation when the resistive forces are moderately adjusted, we again assume Fb and Fc to be 50% and 50% of FRP. The pressures caused by these revised loads yield Set III data.
To gain a more accurate understanding of the resultant stress, we cut a rectangular area GHIJ to specifically discuss. The stress 170 clouds of this area are compared in Figure 2 (left). Please note that any of these loads (i.e., FRP, Fb, and Fc) is too small relative to the rock's weight. For example, when FRP =4.0×10 12 N m -1 is applied to the model's right side (which is 85 km length), its resultant mean pressure is 47.06 MPa, while the mean lithostatic pressure of the model (which is 100 km depth) is 1318.1 MPa.
This reality means that, if we use a stress cloud to compare the stress caused by a combination of the rock's weight and the loads with the stress caused by the rock's weight alone, both of them may be indistinguishable. To create a visual impression, 175 we magnify these loads 50 times, which yields Set II' data and Set III' data (see Figure 2(right)). We find that the horizontal stress caused by these loads is compressional and mainly concentrated on the lower part of section GHIJ.
We use three sections (i.e., M1N1, M2N2, and M3N3) from that rectangular area to quantify the comparison. Each section keeps a span of 50 km relative to one another. The stress diagrams for three sections are compared in Figure 3. When we subtract the stress caused by the rock's weight from the stress caused by the rock's weight and these loads, we obtain the stress caused by 180 the loads, which is exhibited in Set II (III) -Set I. These stress diagrams agree the result exhibited in the stress clouds. Kusznir and Bott (1977) argued that, due to the ductile nature of the lower part of the lithosphere, there would be a redistribution of any stress applied to the whole lithosphere that would result in stress amplification in the upper brittle part of the lithosphere.
This view is based on the assumption that force is uniformly exerted on the side of the lithospheric plate, but the reality is that the ridge push force would increase with depth; consequently, the redistribution of the stress and its amplification are not 185 applicable to the ridge push force. In contrast, we have considered this ductile nature in the modeling, but no evidence was 7 https://doi.org/10.5194/egusphere-2023-1458 Preprint. Discussion started: 3 August 2023 c Author(s) 2023. CC BY 4.0 License.
found for stress amplification in the upper part of section GHIJ. Our modeling analysis suggests that the stress caused by a combination of the ridge push, collisional, and basal friction forces, cannot be in accordance with the observed stress.

Geometry and kinematics of the continental plate
In examining the plate's shape around the globe, we see that the eastern coastline of the American continent is approximately 195 subparallel to the Atlantic ridge at the plate's boundary, and the coastline of Australia's continent is subparallel to the boundary of the Australian Plate. However, the coastline's length of the American continent is greater than that of the Atlantic ridge, whereas the coastline's length of the Australian continent is less than that of its boundary. This feature is also clear for the Indian Plate. Such fashion implies that the driving force of continental plate is likely related to the coastline.
All plates are steadily moving over the Earth's surface, this means that there would be separations and approaches between 200 plates. Separations would result in a gap between two plates. The gap would allow magma to erupt and form a mid-ocean ridge 9 https://doi.org/10.5194/egusphere-2023-1458 Preprint. Discussion started: 3 August 2023 c Author(s) 2023. CC BY 4.0 License.
(MOR). In this respect, the MOR may be the result of plate motion. Currently, the MOR is treated as the cause of the plate driving force. This treatment leads to a chicken-or-egg question: which came first? In physics, the object that exerts force must be clearly differentiated from the object that accepts this force. Some argue that once subduction and spreading are initiated, plates may drive themselves as part of the large circulation of the mantle and lithosphere, which resolves the chicken-or-egg 205 question. However, this view cannot be convincing. Ridge push force contributes to not only the oceanic plates but also the continental plates. The oceanic plates are subducting into the trenches, they take part in the large circulation. Instead, the continental plates never sink, they wouldn't take part in the large circulation, therefore, the chicken-or-egg question remains for the continental plates.
As mentioned in section 1, the latest understanding of plate dynamics is that the lithosphere, crust, and mantle compose the 210 large-scale circulation, and that the plate itself is an integral part of this circulation. Consequently, the dynamic source of plate motion is traced back to the Earth's interior. The terrestrial planets (Venus, Mercury, and Mars) share similar formation procession and interior structure (i.e., crust, mantle, and core) with the Earth, they also have the same spatial surroundings (i.e., asteroid impact) as the Earth does. Therefore, the question remains of why there is plate motion on Earth but not on the other terrestrial planets. This discrepancy of plate motion distribution, together with the above mentioned problems of slab pull and 215 ridge push, implies that some key factor of the Earth, which is still currently unknown, is more likely to cause plate motion.
3 An ocean-generating force driving mechanism for plate motion

Ocean-generating force
Ocean water covers approximately 71% of the Earth's surface, and its total volume is almost 1.35 billion km 3 , with an average depth of nearly 3,700 meters. Geochemical study of zircons suggests that liquid water has existed for more than 4 Gy ago 220 (Mojzsis et al., 2001;Bercovici et al., 2015;Valley et al., 2002). Ocean water is supported by the upper part of the lithosphere, this loading allows the weight of the ocean water to be vertically transferred to the lithosphere. The impact of ocean water on the isostatic balance and heat process of the lithosphere has been widely discussed (Bercovici et al., 2015;Fleitout and Froidevaux, 1983;Osei Tutu et al., 2018;Ricard et al., 1984;Steinberger et al., 2001;Lithgow-Bertelloni, 2014;Ghosh and Holt, 2012;Naliboff et al., 2012). The absence of plate tectonics on the other terrestrial planets (e.g., Mars and Venus) is 225 already recognized; this absence is typically ascribed to the lack of liquid water because the water on Mars possibly appeared in liquid form in the planet's early history and water on Venus has disappeared through a runaway greenhouse (Squyres et al., 2004). Nevertheless, the mechanism by which liquid water contributes to plate tectonics remains enigmatic (Bercovici et al., 2015). A view is that the Earth's surface is cooled by liquid water since the Earth's temperature needs to be stabilized by a negative feedback in the formation of plate tectonics (Bercovici et al., 2015;Walker et al., 1981;Berner, 2004). 230 Liquid exerts pressure on the wall of a container that holds it. The pressure generated on the wall of a cubic container may be written as P=ρgy/2, and the application of this pressure across the wall yields a force. This force may be expressed as 10 https://doi.org/10.5194/egusphere-2023-1458 Preprint. Discussion started: 3 August 2023 c Author(s) 2023. CC BY 4.0 License.
F=PS=ρgy 2 x/2, where S is the wall area, g and ρ are the gravitational acceleration and liquid density, respectively, and x and y are the liquid width and depth, respectively, in the container. Referring to the real world, ocean basins are naturally gigantic containers, and their depths are often more than a few kilometers and vary from one place to another. All of these factors imply 235 that oceans can generate enormous pressure everywhere and that this pressure is unequal among oceans. Furthermore, the application of pressure against oceanic basin walls, which consist of the continents, can yield enormous unequal forces on the continents. Geometrically, ocean pressure is always exerted vertically to the continental slope, by which a normal force is formed. This normal force is called the ocean-generated force, which may be further decomposed into a horizontal force and a vertical force. The horizontal force may be further written as F =0.5ρgLh 2 , where ρ, g, L, and h are the density of water, 240 gravitational acceleration, ocean width that fits the continent's width, and ocean depth, respectively.
In practice, the continent's side is not flat, and the continent's base is generally wider than its top, making the continent appear more like a circular truncated cone standing in the ocean. As the horizontal force is related to the ocean's width (i.e., the continent side's width), we need to horizontally project the continent onto a polygonal column, dissect the whole side of this column into a series of smaller rectangular sides connecting one to another and subsequently calculate the horizontal force 245 generated at each of these rectangular sides. Figure 4 exhibits the horizontal forces generated on the continents. The horizontal forces and the parameters for calculating them are listed in Table 1. Figure 4 exhibits the horizontal forces generated on the continents. The horizontal forces and the parameters for calculating them are listed in Table 1.  Notes: all geographic sites refer to Figure 4.

Resultant movements of plates
The continents are fixed on the top of the lithosphere, and the lithospheric plates connect to each other, this relationship allows the ocean-generated force to be laterally transferred to the lithospheric plates. Subsequently, we list the plausible forces that act on a sample continental plate ( Figure 5) and discuss the physical nature of these forces. 305 These forces can be classified into two categories: the forces acting on the parts of the continent that connect to the oceans and those acting at both the bottom surface of the continental plate and the parts of the continental plate that connect to adjacent plates. The forces acting on the parts of the continent that connect to the ocean derive from ocean pressure and are treated as ocean-generated forces, denoted as FR on the right and FL on the left. The horizontal forces decomposed from these forces are denoted as FR' on the right and FL' on the left. The force acting on the bottom surface of the continental plate arises from a 310 coupling between the plate and underlying viscous asthenosphere. This force is called the basal friction force and is denoted as fbase. According to Forsyth & Uyeda (1975), if there is thermal convection in the asthenosphere, fbase would be a driving force (Runcorn, 1962a, b;Turcotte & Oxburgh, 1972;Morgan, 1972). If, instead, the asthenosphere is passive relative to plate motion, fbase would be a resistive force. Here, we assume fbase to be a resistive force. Given that the continental plate moves toward the right, the forces acting on the parts of the continental plate that connect to adjacent plates include the collisional 315 force from the plate on the right side and the push force from the ridge on the left side, they are denoted as FC and Fridge, respectively. respectively. fbase denotes the basal friction force exerted by the underlying asthenosphere, while FC and Fridge denote the collisional force from Plate C on the right side and the push force from the ridge on the left side, respectively. hL and hR are the ocean depths on the left and right, respectively. r1, r2, r3, r4, and r5 denote the distances of these forces to the Earth's center.
Note that the ocean depth and lithospheric plate thickness are highly exaggerated.

325
Plate motion is conventionally understood as a rigid plate rotating about an axis that penetrates the Earth's center, and this rotation must be a consequence of the integrated effect of all torques acting on the plate (e.g., Richardson, 1992;Forsyth & Uyeda, 1975). Following this understanding, we use torque balance to discuss the movement caused by these forces. According to Figure 5, a combined torque for Plate A may be written as where the first term (r1FL' -r2FR') denotes the torque yielded by the final horizontal force, the second term r3Fridge denotes the torque yielded by the ridge push force, both of them represent the driving torque for the continental plate, and the third term (r4FC + r5fbase) denotes the resisting torque, which hinders the movement of the continental plate. Taking into consideration the reality that the plate is too thin (e.g., less than few hundred kilometers) relative to the Earth's radius (e.g., more than six thousand kilometers), we approximate r1= r2= r3= r4= r5. FL' and FR' may be further written as FL'=0.5ρgLhL 2 and 335 FR'=0.5ρgLhR 2 , where ρ, g, L, hL, and hR are the density of water, gravitational acceleration, ocean width that fits the continent's width, ocean depth at the left, and ocean depth at the right, respectively. Equation (1) provides three possibilities for the continental plate. If the driving torque is greater than the resisting torque, the combined torque is greater than zero, and the continental plate is subjected to an accelerating motion. Practically, it is impossible for the continental plate to undergo such a movement. If the driving torque is equal to the resisting torque, the 340 combined torque is zero, and the continental plate would be subjected to a steady motion. If the driving torque is less than the resisting torque, the combined torque is less than zero, and the continental plate remains motionless.
Plate A's movement exhibited in Figure 5 is parallel to that of Plate B and Plate C, this situation is rather idealized. Practically, the movements of most plates intersect with each other. For instance, the South American Plate moves northwest, the Nazca Plate moves eastward, the African Plate moves northeast, the Eurasian Plate moves eastward. These nonparallel movements 345 would yield additional collisional forces and shearing forces between plates. If two plates are not moving in the opposite direction, the collisional and shearing forces between them may be driving; and if the two plates are moving in the opposite direction, the collisional and shearing forces between them may be resisting. Below, we develop two semi-analytic methods (I and II) to independently resolve plate motion.

Method I
It is assumed that the Earth's surface is covered with Plate A, Plate B, Plate C, Plate D, and others, and that Euler pole of each plate has been established ( Figure 6). For Plate A, the horizontal force Fi (i=1, 2, 3, 4, and 5) acts on the side of the continent 20 https://doi.org/10.5194/egusphere-2023-1458 Preprint. Discussion started: 3 August 2023 c Author(s) 2023. CC BY 4.0 License. that is fixed on top of Plate A. The horizontal force (F1, for instance) yields a component (F1 ' , for instance) that is orthogonal to the rotation axis of the plate; this component then yields a torque (τ1 ' , for instance). The torques yielded by all the components 355 decomposed from the horizontal forces are summed into first driving torque. The ridge push force Fr-i (i=1, 2, 3, and 4) acts on the edge of the plate, this force also yields a component that is orthogonal to the rotation axis; this component also yields a torque. The torques yielded by all the components decomposed from the ridge push forces are summed into second driving torque. Given that Plate A, Plate B, Plate C, and Plate D move eastward, southward, westward, and eastward, respectively, and that Plate D moves faster than Plate A, these make Plate A undergo a collisional driving force FB-c from Plate B, a shearing 360 driving force FD-s from Plate D, a collisional resistive force FC-c from Plate C, a shearing resistive force FB-s from Plate B, and a basal friction force Fbasal from the underlying viscous asthenosphere. The collisional driving force FB-c and the shearing driving force FD-s also yield two components that are orthogonal to the rotation axis of the plate; these components also yield torques. The torques yielded by these two components are summed into third driving torque. The collisional resistive force FCc and the shearing resistive force FB-s also yield two components that are orthogonal to the rotation axis; these components also 365 yield torques. The torques yielded by these two components are summed into first resistive torque. The basal friction force Fbasal yields second resistive torque.
Then, we divide these five sets of torques into two exerting parts: one, which includes the second driving torque and a little portion of the first and third driving torque, balances out the first resistive torque, and the other, which including the remaining portion of the first and third driving torque, balances the second resistive torque. Consequently, all these torque balances allow 370 Plate A to be steadily rotated under the assumption that the acceleration and inertia of the plate are neglected. The remaining portion of the first and third driving torque is called the net driving torque, and the second resistive torque is called the net resistive torque. The balance between the net driving torque and the net resistive torque may be written as where τdriving is the net driving torque, τdriving =ετ, and ε is the ratio of the net driving torque to the first and third driving torque. 375 As shown in Figure 6(A), the component decomposed from a force (the horizontal force, for instance) may be written as Fi'= Ficosηi, and ηi=γi-λi, where γi is the inclination of this force to latitude. λi is the azimuth of arc PiE with respect to latitude. This component yields a torque τi with respect to the rotation axis, i.e., τi=ri Fi', where ri denotes the lever arm distance of the component Fi', ri=RearthsinφPi, Rearth is the Earth's radius and Rearth=6,371 km, and φPi is the angle of site Pi and the Euler pole.
A sum of the torques yielded by the components, which are decomposed from the horizontal forces, collisional driving forces, 380 and shearing driving forces, forms the first and third driving torque τ. τbasal denotes the net resistive torque yielded by the basal friction force, it can be written as τbasal =rKFbasal, where rK denotes the lever arm distance of the basal friction force. According to the principle of viscous fluid mechanics, the basal friction force may be expressed as Fbasal = μAu/y, μ, A, u, and y are the viscosity of the asthenosphere, the plate's area, the plate's speed, and the thickness of the asthenosphere, respectively. Therefore, u=yτdriving/μA, this speed represents average level of the plate's movement. In general, the largest speed of a plate occurs at the plate's equator, while the smallest speed occurs at the location whose angle distance to the Euler pole is minimal or maximal. As shown in Figure 6(B), we assume that the geometric center (i.e., location K) of Plate A moves at the average speed, namely, u=uk. And then, the speed of any location S within this plate may be expressed with us=uksinφs/sinφk, where φs(φk) is the angle distance of location S (K) to the Euler pole (i.e., location E) relative to the Earth's center. The speed us can be further decomposed into the longitudinal speed us-lo and latitudinal speed us-la, and 390 us-lo=ussin(λs-90 o ), us-la=uscos(λs-90 o ), where λs is the azimuth of arc SE with respect to latitude. The azimuth of the movement is then calculated through the longitudinal and latitudinal speeds. All these angles and distances (i.e., ηi, γi, λi, λs, φPi, φk, φs, ri, rK) may be further calculated through the latitudes and longitudes of related locations. Here, we use six plates (South American, African, Eurasian, North American, Australian, and Pacific plates) to demonstrate their movements. In order to simplify the following deduction, we plot globally tectonic plates into a grid of 10°×10° and use these grid nodes, which are within plate, to obtain the geometric center of each plate. The geometric center is approximately 405 calculated through the average of the latitudes and longitudes of these nodes. The Euler pole location of each of these plates is cited from the GSRM v.2.1 (Kreemer et al., 2014). Both the geometric center of each plate and its Euler pole location are exhibited in Figure 6(C).
Besides the horizontal forces, the other possible forces (i.e., collisional and shearing) for these plates must be considered. For example, the African, Indian, and Australian plates provide the collisional driving forces FAF-EU-C, FIN-EU-C, and FAU-EU-C for 410 the Eurasian Plate, respectively. The Nazca plate provides a collisional driving force FNaz-SA-C for the South American Plate.
The Eurasian Plate provides a shearing resistive force FEU-NA-S for the North American Plate; vice versa, the North American Plate provides a shearing driving force FNA-EU-S for the Eurasian Plate. The Australian, North American, and Eurasian Plates provide the collisional driving forces FAU-PA-C, FNA-PA-C, and FEU-PA-C for the Pacific Plate, respectively. Taking into consideration the long argument of slab pull that is listed in section 2.1, we presently neglect slab pull. And if this force can 415 be confirmed in the future, it can be added into this model. The details of these forces are exhibited in Figure 4 and listed Table   1. The resultant torques from all related forces are listed in Table 2.  Note: the negative symbol "-" beneath torque denotes counterclockwise with respect to the axis of rotation.  Note: the negative symbol "-" beneath torque denotes counterclockwise with respect to the axis of rotation. The asthenosphere viscosity is not yet exactly determined. Many numerical studies using glacial isostatic adjustment and geoid modeling have shown that asthenospheric viscosity ranges from 10 17 to 10 20 Pas (e.g., Steinberger, 2016;Hager and Richards, 1989;Mitrovica, 1996;King, 1995;Kido et al., 1998;James et al., 2009;Pollitz et al., 1998;Berker, 2017;Kaufmann and Lambeck, 2000;Hu et al., 2016). Laboratory experiments, however, suggested that the magnitude of the asthenospheric 450 viscosity could be substantially different from those constrained by numerical studies. The viscosity is variable and likely related to the thermodynamic state, grain size, composition of the medium, and state of stress (Bercovici et al., 2015). Both the melt contents of the asthenosphere and the water in the asthenosphere may greatly affect the viscosity (Mei et al., 2002;Hirth and Kohlstedt, 1996). Hirth and Kohlstedt (1996) reported a variable viscosity profile for a melt-free oceanic lithosphere with a mean value of ~10 18 Pas. These authors (e.g., Doglioni et al., 2011;Scoppola et al., 2006) concluded that, in 455 consideration of the water-and melt-rich layers characterized by much lower viscosities, a strong vertical variability of viscosity may be more realistic. The asthenosphere's effective viscosity can be greatly lowered to 10 15 Pas if the water content in the case of both diffusion and dislocation creep is included (Korenaga and Karato, 2008). Scoppola et al. (2006) conducted a more detailed review of asthenospheric viscosity and concluded that the presently accepted values of viscosity might be reduced through a combined experiment including these parameters (i.e., melt content, water content, mechanical anisotropy, 460 and shear localization). A "superweak", low-viscosity asthenosphere supported by recent observations is being accepted by the geophysical community (Kawakatsu et al., 2009;Hawley et al., 2016;Holtzman, 2016;Naif et al., 2013;Freed et al., 2017;Hu et al., 2016;Stern et al., 2015;Bercker, 2017). Jordan (1974) treated the asthenospheric thickness as 300 km.
Taking into account the present status of the viscosity and thickness of the asthenosphere above, we adopt y=300 km for each of the six selected plates, μ=10 18 Pas for the South American, African, North American, and Eurasian plates, μ=0.6×10 18 Pas 465 for the Australian Plate, and μ=0.12×10 18 Pas for the Pacific Plate.
The other parameters (i.e., plate area, the ratio of the net driving torque and the first and third driving torque) and the resultant average movements of these six plates are listed in Table 3.
There have been many plate motion models (i.e., GSRM, NUVEL-1, and MORVEL) that include global navigation satellite systems (GNSS) and paleomagnetic data. For instance, GSRM v.2.1 includes more than 6,739 continuous GPS velocity 470 measurements (Kreemer et al., 2014). The movements reproduced by these models may approximately represent observations.
Here, the movements of 450 locations (41 for

Method II
We assume that the Earth's surface is covered with Plate A, Plate B, Plate C, Plate D, and others ( Figure 8). For Plate A, it undergoes the horizontal force Fi (i=1, 2, 3, 4, and 5), the ridge push force Fr-i (i=1, 2, 3, and 4), the collisional driving force FB-c, the shearing driving force FD-s, the collisional resistive force FC-c, the shearing resistive force FB-s, and the basal friction force Fbasal. One horizontal force (F1, for instance) yields a torque (τ1, for instance), another horizontal force (F2, for instance) 500 yields another torque (τ2, for instance), a combination of these two torques results in a new torque (τ1-2, for instance), this new torque then combines the torque yielded by third horizontal force to form another new torque. Subsequently, the torques yielded by all the horizontal forces are combined into a final torque. The collisional driving force FB-c yields a torque, the shearing driving force FD-s yields a torque, the final torque combines these two torques to firm first driving torque. The collisional resistive force FC-c and the shearing resistive force FB-s also yield two torques, they combine to firm first resistive torque. The 505 basal friction force Fbasal yields second resistive torque. The ridge push force Fr-i (i=1, 2, 3, and 4) also yields a torque, the torques yielded by all the ridge push forces are combined into second driving torque.
Then, we divide these four sets of torques into two exerting parts: one, which includes the second driving torque and a portion of the first driving torque, balances out first the resistive torque, and the other, which including the remaining portion of the first driving torque, balances the second resistive torque. Consequently, all these torque balances allow Plate A to be steadily 510 rotated under the assumption that the acceleration and inertia of the plate are neglected. The remaining portion of the first driving torque is called the net driving torque, and second resistive torque is called the net resistive torque. We assume that the net driving torque exerts on the geometric center (i.e., location K) of Plate A, this makes the plate move along a big circle that represents the equator of this Plate. And then, the balance between the net driving torque and the net resistive torque may be expressed with Equation (2). According to Figure 8(A), a force Fi yields a torque τi with respect to the Earth's center, i.e., τi= 515 Rearth Fi, where Rearth is the Earth's radius and Rearth=6,371 km. The combination of two torques follows the trigonometric principle and may be written as τj 2 =τi 2 + τi+1 2 + 2τiτi+1cos(γi-γi+1) where τj is the combined torque, τi and τi+1 are the torque yielded by the force Fi and Fi+1, respectively. γi and γi+1 denote the inclination of the forces Fi and Fi+1 to latitude, respectively. τbasal denotes the net resistive torque yielded by the basal friction 520 force, it can be written as τbasal = Rearth Fbasal. According to the principle of viscous fluid mechanics, the basal friction force may be expressed as Fbasal = μAu/y, μ, A, u, and y are the viscosity of the asthenosphere, the plate's area, the plate's speed, and the thickness of the asthenosphere, respectively. Therefore, u=yτdriving/μA, this speed represents average level of the plate's movement.
On the whole, the largest speed of a plate occurs at the plate's equator, while the smallest speed occurs at the location whose 525 angle distance to the Euler pole is minimal or maximal. According to Figure 8 Table 4. The viscosity and thickness of the asthenosphere for these three plates are the same as that listed in the method I. The other parameters (i.e., plate area, the ratio of the net driving torque to the first driving torque, and the amplification coefficient) and the resultant average movements are listed in Table 5. 555 The movements of 215 locations (41 for the South American Plate, 70 for the African Plate, and 104 for the Pacific Plate) are reproduced by GSRM v.2.1. The calculated and reproduced movements are compared in Figure 9. We find that the calculated movements for these locations are basically consistent with the reproduced movements in both speed and azimuth, the RMS of the calculated speed against the reproduced speed for the South American, African, and Pacific plates is 0.98, 3.18, and 6.51 mm/yr, respectively. This result is not as good as that demonstrated in the method I. One major cause for this is that, in the 560 method II the collisional and/or shearing forces considered are not enough. For example, the Australian, North American, and Eurasian plates collide the Pacific Plate extensively, the three collisional forces FAU-PA-C, FNA-PA-C, and FEU-PA-C are too spare relative to the long collisional zone; In addition, there may be shearing force between the North American Plate and the Pacific Plate, but we omit this force. As a result, the first driving torque that we calculate is not too accurate in both magnitude and orientation; Another cause for this is that the geometric center of a plate is strictly not calculated through the average of the 565 latitudes and longitudes of those nodes. The less accurate first driving torque adds to the less accurate geometric center, naturally, the calculated Euler pole location and the resultant movement of the plate cannot be accurate. Even so, our goal is realized that a combination of the ocean-generated force, the ridge push force, the collisional force, and the shearing force indeed may be responsible for plate motion.

Resultant stress
As mentioned in section 2.2, the observed stresses are mostly concentrated on the uppermost part of the lithosphere (Zoback, 1992;Zoback et al., 1989;Zoback & Magee, 1991), whereas our modeling analysis suggests that the stress caused by the existing forces (i.e., the ridge push, basal friction, and collisional) are mainly concentrated on the lower part of the lithosphere. 615 This discrepancy indicates that other force may be responsible for the observed stresses. Ocean water is loaded on the top of the lithosphere, this allows to create a stress field associated with the upper part of the lithosphere.
To examine this expectation, we add the ocean-generated force onto the model that is exhibited in Figure 2 (top). The final model is shown in Figure 10 (top left). The inputs include the vertical pressure caused by the rock's weight and the lateral pressures caused by these loads (i.e., FRW, FLW, FRP, Fb, and Fc). FRW and FLW are the ocean-generated forces, they correspond 620 to a result of 5 km water depth at the right and 3 km water depth at the left, respectively, and FRW=0.12×10 12 N m -1 , FLW=0.04×10 12 N m -1 . The outputs include the stress produced by the vertical pressure alone and the stress produced by a combination of the vertical and lateral pressures. Similarly, we will only discuss the horizontal stress (i.e., S11) in the following section. At this time, we first use these loads to yield Set A data and Set B data. The stress clouds of the area GHIJ are compared in Figure 10 (middle left). To realize a visual impression, we magnify these loads 50 times, which yields Set A' data and Set 625 B' data. We find that the horizontal stress caused by these loads is compressional and tends to distribute across the middle part of section GHIJ. We then minify FRP and Fb 100 times, remain FRW and FLW stable, and adjust Fc properly so as to sustain the horizontal force balance, this yields Set C data and Set D data. To realize a visual impression, we again magnify these revised loads 50 times, which yields Set C' data and Set D' data. A more detailed description of these loads for different sets is exhibited in Figure 10 (top right). It can be found that, after FRP and Fb are reduced, the horizontal stress caused by these loads 630 are mainly concentrated on the upper part of section GHIJ.
The stress diagrams for three sections (i.e., M1N1, M2N2, and M3N3) are also collected and compared in Figure 11. When we subtract the stress caused by the rock's weight from the stress caused by the rock's weight and these loads (i.e., FRW, FLW, FRP, Fb, and Fc), we obtain the stress caused by these loads, which are expressed with Set A/B/C/D -Set I. We find that the result of the stress diagrams agrees that of the stress clouds. The continental plates are not only rigid but also curved, this allows the 635 ocean-generated forces (i.e., the horizontal forces) to laterally penetrate across the plate. Our modeling analysis suggests that the stress caused by a combination of the ocean-generated force, ridge push force, collisional force, and basal friction force may be in accordance with the observed stresses. The lithospheric plates are curved, the rocks within them are not homogeneous and isotropic, and their thickness and density also vary spatially; in addition, as seen in Figure 4, the directions of the ocean-generated force are various. We expect that the 660 stresses caused by a combination of the ocean-generated force and all of these factors may realize a better match with the observed stresses in the WSM, and this will be included in the following research.

Why may ocean-generated force drive plate motion?
Although we have demonstrated that the movements estimated from ocean-generated force are consistent with observations, 665 many people still refuse this force to be a plate driving force for the following reasons: 1) The ocean constitutes just another deviation from the true radial density distribution of the Earth. Any "lateral" density heterogeneity creates stresses that in turn lead to deformation, and their extent is controlled by the rheological properties of the involved materials. 2) Plate motion determines the shape of the ocean basin; as a result, ocean water cannot contribute to plate motion. 3) Ocean loading on top of the lithosphere does not allow ocean-generated force to drive the lithospheric plates to move along the asthenosphere, this is 670 similar to that the water held in a container standing on the ground cannot drive the container to move along the ground. 4) Ocean-generated force is too small to drive plate motion. These issues need to be clarified here. First, the view that any "lateral" density heterogeneity would lead denser materials (i.e., rocks) to flow toward lighter materials (i.e., air or water) is rather idealized. The Himalayas are denser than the surrounding air and water, but this density heterogeneity does not allow the mountains to reduce their height; instead, they increasingly rise up. The continents are also denser than the oceans, and if they 675 flow toward the oceans, the ocean basin will be filled by the continent's rock substances. Then, the sea level would rise, causing water to submerge the coast. This would result in a decrease in the landmass area. Given the continent's volume is constant, then, the continent's height would reduce. This is evidently contrary to the continental accretion that is widely accepted by the geophysical community. The continental accretion indicates that the continents have been growing since the Archean. A further review of this topic can be found in this recent research (Zhu, et al., 2021). The examples of the Himalayas and continents 680 suggest that a system composed of ocean water and the crust is permanently disturbed by the hydrostatic pressure force and tides; As a result, it is difficult for the continents to follow the principle of "lateral" density heterogeneity to flow toward the oceans. It is an opinion of this author that ocean water compresses the crust, the elasticity of the crust's rock allows the crust to deform as a response to the hydrostatic pressure force. With the passage of time, the ocean basin expands gradually, and then the water in the seas flows toward the ocean basin, and the sea level decreases to cause part of the seafloor to expose and 685 become landmass. In addition, the continent's height increases relative to the sea level. This process simply accords with the continental accretion. Indeed, plate motion may reshape ocean basin, but ocean water is not passive, it may provide feedback through energy dissipation on the plate, and as a result, affect plate motion. Ocean loading on the lithosphere is far different from water loading in a container. Since the lithosphere has been broken into individual plates and these plates are attached to 45 https://doi.org/10.5194/egusphere-2023-1458 Preprint. Discussion started: 3 August 2023 c Author(s) 2023. CC BY 4.0 License. the asthenosphere on the plate. In contrast, a container is perfect, and the force produced by water pressure within the container is balanced out by the container itself and cannot interact with the basal friction exerted by the ground on the container. In physics, the interaction of a driving force and a resistive force is a precondition for an object to move. Figure 12 outlines how force balances may be created for the plates. Three plates are totally designed in the model; along the vertical direction, the weight of each plate is balanced out by the support from the asthenosphere; thus, we will only consider 695 the force along the horizontal direction. FAR, FAL, FCL, and FCR are the horizontal forces, FRL and FRR are the ridge push forces, FBA, FAB, FCB, and FBC are the collisional forces, fA, fB, and fC are the basal friction forces. Slab pull and trench suction are neglected. Each of these forces can yield a torque relative to the Earth's center, because torque is a product of force and lever arm, and here the lever arm may be represented with the Earth's radius since plate is too thin relative to the Earth's radius, the lever arm length of one plate is approximately equal to that of another plate. This situation allows to simplify torque balance 700 into force balance for the following discussion. We assume that Plate A rotates counterclockwise and Plate C rotates clockwise.
For Plate A, we set FRL= 4.0×10 12 N m -1 , this magnitude is presently accepted by geophysical community (Turcotte and Schubert, 2004). We assume the ocean depth to be 5.00 km at the right and 3.00 km at the left, respectively. These two depths correspond to FAR= 0.245×10 12 N m -1 and FAL= 0.0882×10 12 N m -1 , the final horizontal force of these two forces would be FAR-FAL = 0.1568×10 12 N m -1 . We set FBA= 4.05×10 12 N m -1 , and use FRL = 4.0×10 12 N m -1 and a little portion of the final 705 horizontal force, which is represented by FARL = 0.05×10 12 N m -1 , to balance out FBA, and use the remaining final horizontal force FRARL= 0.1068×10 12 N m -1 to balance out the basal friction force fA. The force balances for this plate would be FBA-FRL-FARL= 0 and FRARL-fA=0.
For Plate B, we set FCB=4.0×10 12 N m -1 , due to FBA= FAB= 4.05×10 12 N m -1 , thus, FBA-FCB =0.05×10 12 N m -1 . We use this net force to balance out the basal friction force fB. The force balance for this plate would be FBA-FCB -fB=0. 710 For Plate C, we set FRR= 3.95×10 12 N m -1 , and assume the ocean depth to be 4 km at the left and 6 km at the right, respectively.
These force balances allow three plates to be steadily rotated. We find, even if the ridge push force FRL (FRR) is given a smaller amplitude (~ 10 10 N m -1 , for example), so long as the collisional force FBA (FAB, FCB, and FBC) is properly valued, these force balances can always be created. Nevertheless, as demonstrated in section 3.3, a ridge push force of 4.0×10 12 N m -1 would result in a horizontal stress that is mostly concentrated on the lower part of the lithosphere, which is not in accordance with 720 observation. Hence, we prefer to accept the ridge push force to be smaller than ocean-generated force.

How does plate motion realize mechanically?
Thus far, we have concluded that ocean-generated force is able to combine the ridge push force, the collisional force, and the shearing force to satisfy the kinematics and geometry of plate motion. Now, let us discuss how plate motion can be 730 mechanically realized. As shown in Figure 12, it is assumed that the depth of Ocean 1 is greater than that of Ocean 2. If we use a part of Ocean 2 that connects to Plate A, which is equal in length to Ocean 1, to do comparison, the depth difference between this part of Ocean 2 and Ocean 1 creates a net gravitational potential energy relative to the asthenosphere reference level. As Plate A and Plate B move away from each other, this separation would require the Ocean 1 depth to decrease as the basin elongates horizontally, and require the Ocean 2 depth to increase as the basin shortens horizontally. Consequently, the 735 net gravitational potential energy decreases. Therefore, if there were no external energy inputs to compensate, the net gravitational potential energy would eventually disappear, terminating plate motion. Tides may be supplying this energy. Tides represent the regular alternations of high and low water on Earth; when high water falls, the gravitational potential energy converts into kinetic energy, then, ocean water obtains movement. As all oceans are physically connected, part of the water in Ocean 2 may travel via passages to compensate the decreasing ocean depth of Ocean 1, thus sustaining the net gravitational 740 potential energy. Given the basal friction force fbasal = 1.62×10 18 N and the movement distance u=3 cm/yr for the lithosphere, an energy of Q1=fbasal×u=4.86×10 16 J/yr is required to satisfy this movement distance. This energy also represents the net gravitational potential energy. The ocean water level often increases twice a day due to tides, and the resultant height is assumed to be h=0.3 m. Given the gravitational acceleration g=9.8 m/s, the volume v=1.35×10 9 km 3 and density ρ=1000 kg/m 3 for the whole ocean, and consequently, the gravitational potential energy obtained by ocean water due to tides during a year 745 would be Q2=2*365*ρvgh=2.9×10 21 J/yr. The transformation from gravitational potential energy to kinetic energy within ocean water and the energy transition between oceans must be complicated, and we believe that a small part of this tidal energy is enough to supply the net gravitational potential energy. In fact, the impact of tidal energy on plate motion has long been discussed. Wegener (1924) proposed that tides cause a slight progressive displacement of the crust. Rochester (1973) showed that the total energy released due to tidal friction exceeds 5*10 19 ergs/s. Several authors (e.g., Miller, 1966;Munk, 1968) 750 concluded that the dissipation in both shallow seas and on the solid Earth is approximately 2*10 19 ergs/s, and this amount of energy exceeds the lower bound set by seismic energy release by 2 orders of magnitude (Gutenberg, 1956) and might be driving the plate motion. Other authors (e.g., Riguzzi et al., 2010;Egbert and Ray, 2000) reevaluated the energy budget and found that the total energy released by tidal friction may reach up to 1.2*10 20 J/yr, and approximately 0.8*10 20 J/yr is dissipated in the oceans, shallow seas, and mantle, and the remaining energy is enough to maintain the lithosphere's rotation, estimated 755 at approximately 1.27*10 19 J/yr. In contrast to these studies, we provide another insight: the tidal energy obtained by ocean water may feed plate motion.

Tidal force versus plate motion
The impact of tidal drag on plate motion has been debated for many years. Wegener (1915) attributed the continent's drift to 760 tidal drag and centrifugal forces, but these forces were shortly found to be too weak to work. Jeffreys (1929) claimed that the mean tidal friction corresponds to a westward stress of the order of only 10-4 dyn/cm 2 over the earth's surface, this stress is too small to maintain the drift. The notion of tidal drag revived after the discovery of a net rotation or westward drift of the lithosphere relative to the mantle (Le Pichon, 1968;Knopoff and Leeds, 1972). Another argument in favor of this notion stems from the assessment of energy budget, as discussed in section 4.2, it shows that tides are energetically enough for feeding plate 765 motion. However, a satisfaction in energy cannot shield the notion of tidal drag anymore. Jordan (1974) and Jeffreys (1975) attached the theoretical basis of tidal drag, they claimed that the viscosity both related to tidal drag and necessary to allow decoupling between lithosphere and mantle (~10 11 Pas) is far less than the present-day asthenosphere viscosity. Ranalli (2000) also showed that any non-zero torque due to difference in angular velocity between the mantle shell and lithosphere shell would be extremely transient, and cannot be a factor in the origin of the westward drift of the lithosphere. Despite these fierce 770 objections, the advocators of tidal drag didn't give up. Scoppola et al. (2006) proposed the westward rotation of the lithosphere as a consequence of the combined effect of tidal torque, downwelling of the denser material into the mantle, and thin layers of very low viscosity hydrate channels in the asthenosphere. Several authors (e.g., Riguzzi et al., 2012;Doglioni and Panza, 2015) had suggested that, if an ultra-low viscosity layer exists in the upper asthenosphere, the horizontal component of the tidal oscillation and torque may be able to move the lithosphere. As demonstrated in section 3.2, laboratory experiments tend to 775 support this possibility (Bercovici et al., 2015;Mei et al., 2002;Hirth and Kohlstedt, 1996;Scoppola et al., 2006;Doglioni et al., 2011). The asthenosphere's effective viscosity can be lowered to 10 15 Pas if the water content in the case of both diffusion and dislocation creep is included (Korenaga and Karato, 2008). A "superweak", low-viscosity asthenosphere is being accepted by the geophysical community (Kawakatsu et al., 2009;Hawley et al., 2016;Holtzman, 2016;Naif et al., 2013;Freed et al., 2017;Hu et al., 2016;Stern et al., 2015;Bercker, 2017). Zaccagino et al. (2020) recently investigated a 20-year series of plate 780 motion to conclude that plate motion relates to tidal drag in some way. For example, the lithospheric plates retain a non-zero horizontal component of the solid Earth tidal waves, and they move faster with frequencies of 8.8 and 18.6 years that correlate to lunar apsides migration and nodal precession.
We provide a few points to respond to the notion of tidal drag. First of all, the lithosphere's net rotation or westward drift is different from plate motion. The former indicates that the lithosphere is moving in a single direction, while the latter indicates 785 that the lithospheric plates are moving in different directions. Second, it is already established in the astronomical field that tidal drag is operated through the tractive force. This force is geometrically decomposed from a tide-generating force, and its direction always follows the Earth's surface. Apparently, this force may be divided into two symmetric fields that are aligned with the Earth-Moon system. In each field the force vector is uniformly directed to the sub-lunar points, which are projections of the Moon on the Earth's surface. Two patterns are expected for the lithospheric plates under the tractive force. One is that, 790 as the Earth rotates around its axis, all the plates are continuously swept from east to west. Another is that parts of the lithosphere, which are located at middle and high latitudes, would be dragged toward lower latitudes. Nevertheless, upon comparison with the plate motion vector (Figure 13), it becomes evident that the tractive force is not in accordance with the plate motion. The Pacific Plate, for instance, moves northwest, the Eurasian Plate rotates clockwise, the North American Plate rotates counterclockwise, the African and Indian-Australian Plates move northeast. Moreover, the movements of the Pacific, 795 Eurasian, and Indian-Australian plates intersect with each other. The diversity of plate motion implies that each plate is being operated by a set of independent forces, with a leading force controlling the direction of plate motion. Third, it has been found that the plates performed a cycle of dispersal and aggregation during a geological timescale, and that three supercontinents (i.e., Pangaea, Rodinia and Columbia) occurred over the past 2 billion years (Mitchell et al., 2021). This cycle of dispersal and aggregation is a manifestation of plate motion. When projected on the Earth's surface, the Moon's position is mostly between 800   (Robert, 2008). ITRF2014 horizontal velocity field of plate motion is from Altamimin et al. (2016).

Tidal force versus seismicity
Whether the tidal forcing relates to earthquake occurrence is a considerably hot topic. Most studies with global earthquake 810 catalogs tend to show no correlation between the two (e.g., Schuster, 1897;Morgan et al., 1961;Hartzell and Heaton, 1989). Some of regional earthquake catalogs reveal a significant correlation (e,g., Young and Zurn, 1979;Ulbrich et al., 1987;Shirley, 1988), but others also revealed no correlation (Knopoff, 1964;Shlien, 1972;Shudde and Barr, 1977;Vidale et al., 1998). Nevertheless, these studies explored only the effect that is caused by solid Earth tide, the effect caused by ocean tide (i.e., the loading) is commonly ignored. By adding ocean tide to solid Earth tide, Tsuruoka et al. (1995) reached a point that a 815 decrease in the confining pressure due to the tidal forcing is responsible for triggering earthquake occurrence. Tanaka et al. 51 https://doi.org/10.5194/egusphere-2023Tanaka et al. 51 https://doi.org/10.5194/egusphere- -1458 Preprint. Discussion started: 3 August 2023 c Author(s) 2023. CC BY 4.0 License.
(2002) expanded the method taken by Tsuruoka et al. (1995), they investigated 9350 globally distributed earthquakes with magnitude 5.5 or larger to conclude that a small stress change due to the tidal forcing encourages earthquake occurrence. The results of these studies (e.g., Tsuruoka et al., 1995;Tanaka et al., 2002) suggest that ocean tide plays a crucial role in determining the correlation. 820 If ocean tide really relates to earthquake occurrence, then it must depend on ocean water exerting its impact. For example, when a tide is added to the ocean, the ocean water depth will vary, the ocean water pressure will also vary. Ocean water pressure variation is not only vertically applied to the oceanic crust that is below the ocean, but also horizontally applied to the continental crust that connects to the ocean. Guillas et al. (2010) presented a link between the El Niño-Southern Oscillation (ENSO) and earthquake occurrences on the East Pacific Rise (EPR), and proposed that a reduction in ocean-825 bottom pressure over the EPR may encourage seismicity. A recent study found that sea level changes affect seismicity rates in a hydrothermal system near Istanbul (Martínez-Garzón et al., 2023). As pointed out by Tanaka et al. (2002), the most likely component to control the earthquake occurrence is the stress. Since tide represents a periodic oscillation, it should be expected that the stress produced by ocean water will behave periodically. In these studies (e.g., Tsuruoka et al., 1995;Tanaka et al., 2002;Martínez-Garzón et al., 2023), the tidal stress is theoretically estimated, but there is no modelling or 830 experimental evidence for the tidal stress. In section 3.3, we have modelled the crust's stress through a combination of various forces (i.e., the ocean-generated force, the ridge push force, the collisional force, and the basal drag force), in which the ocean-generated force is constant. Below, we explore the stress when the ocean-generated force varies due to tide, in order to provide support for past and future studies.
The model (Figure 14) is based on that of Figure 10 (top left). The inputs include the hourly vertical pressure caused by the 835 rock's weight and the hourly lateral pressures caused by these loads (i.e., FRW, FLW, FRP, Fb, and Fc). Note, the solid body tide is neglected. FRW and FLW are the ocean-generated forces that assimilate the effect of tides. We here design a water level variation of totally 12 hours, which corresponds to a semidiurnal tide. The information of tidal height and loads is listed in Table 6. The outputs include the hourly stress produced by the vertical pressure alone and the hourly stress produced by a combination of the vertical and lateral pressures. Using the latter to subtract the former, we obtain the hourly stress produced 840 by the lateral pressures. Similarly, we only discuss the horizontal stress (i.e., S11). At this time, we collect the results of 6 locations (i.e., ①, ②, ③, ④, ⑤, and ⑥) to do comparison. These locations belong to the 30 km depth and 60 km depth of three sections (i.e., M1N1, M2N2, and M3N3).
The stress diagrams for these locations are compared in Figure 15. We find that the stress oscillation due to ocean tide has laterally penetrated the crust's rock. This study has not yet considered the stress in the oceanic crust, but the result 845 is expected. For example, ocean water exerts pressure on the oceanic crust, which produces stress for the oceanic crust; when tide is added to the ocean, the stress is mechanically entrained to oscillate. As depicted in Set C(D) -Set I of Figure   11, the stress generated over a depth of 50 km is approximately 2.0 to 6.0 Mpa. This magnitude has fallen within the range of earthquake stress drops (1~30 Mpa) (Kanamori, 1994), indicating that the ocean-generated force may closely relate to 52 https://doi.org/10.5194/egusphere-2023-1458 Preprint. Discussion started: 3 August 2023 c Author(s) 2023. CC BY 4.0 License. earthquake occurrence. Please be aware of that our model is straight, the rock's materials within it are assumed to be 850 homogeneous and isotropic. In practice, the Earth's surface is curved, the rock's materials are not only inhomogeneous but also anisotropic. The oceans circle the continents, which leads to the continents being laterally compressed inward. All of these factors allow the ocean water pressure to be amplified in the crust, resulting in a higher stress level.

How may plate motion initiate and proceed?
The dispersal and aggregation of plates represent that the ocean basin had been periodically adjusted, and this change is often 865 called the Wilson Cycle (Wilson, 1963). Figure 16 outlines how such a cycle may be realized. It is assumed that the left end of the model is connected to its right end and that the depth of Ocean 1 depth is greater than that of Ocean 2. The greater ocean depth corresponds to greater ocean-generated force. Slab pull and trench suction are neglected. The ocean-generated force, the collisional force, and the basal friction force combine to form force balances. For instance, the force balance for Plate A is FAL-FAR-FCA-fA=0, the force balance for Plate B is FBR-FBL-FCB-fB=0, and the force balance for Plate C is FAC-FBC-fC=0. At time 870 t1 and t2, these force balances allow Plate A and Plate B to move toward each other, Plate C is pushed to move. These movements make Ocean 2 basin shorten while make Ocean 1 basin elongate. Tides make ocean water move periodically, the passages between ocean basins allow water to travel and compensate. At time t3, Plate A and Plate B meet, forming an aggregation. Meanwhile, Plate C sinks and becomes disappeared, Ocean 2 basin closes. Since plate motion stops, the forming oceanic crusts cannot spread away from the ridge, they gradually accumulate and plug up magma eruptions, and the ridge 875 tends to die. Once the fractures of the lithosphere are repaired, the ocean-generated force cannot interact with the basal friction force, and then, these force balances terminate. After some time, a large asteroid violently collides with the aggregated plate, forming extensive fractures on the plate, and one of the fractures penetrates down to the lower part of the plate. At time t4, the large fracture induces water entry, forming a large body of water that is deeper than Ocean 1. The deeper water body corresponds to greater ocean-generated force, this force may further expand the fracture. The large fracture also represents a 880 mass loss of the upper part of the lithosphere; the isostasy would require the upper mantle to melt, the lower part of the aggregated plate is apparently broken. At time t5, both the ocean-generated force and the molten material finally cut the plate into Plate D and Plate E. As the left end of the model is connected to its right end, the greater ocean-generated force would require the left part of Plate D to compress the right part of Plate E. Together with the basal friction exerted by the asthenosphere, the left part of Plate D is eventually detached from the plate, forming one subduction. Similarly, the right part of Plate E is 885 detached from the plate, forming another subduction. These detachments and subductions allow to form a new oceanic plate--Plate F. At this moment, the ocean-generated forces may interact with the basal friction force, some new force balances are created. For instance, the force balance for Plate D is FDR-FDL-FFD-fD=0, the force balance for Plate E is FEL-FER-FFE-fE=0, the force balance for Plate F is FDF-FEF-fF=0. These force balances allow Plate D and Plate E to move away from each other. A new oceanic ridge gradually forms, and the increasing separation between the two plates results in a new Ocean 3 basin. 890 Ocean depth cannot be stable during a long geological timescale, it may change with the deepening/shallowing of basins. The ocean depth change in turn leads ocean-generated force to vary. We assume that, at time t1, the Ocean 1 depth and the Ocean 2 depth are h1=5,000 km and h2=3,000 km, respectively, the length and width of Plate A are D=6,000 km and L=2,000 km, the water density is ρ=1000 kg/m 3 , the gravitational acceleration is g=9.8 m/s. Then, the total ocean-generated force for Plate A would be Ftotal= FAL-FAR=0.5ρgL(h1 2 -h2 2 )=1.5680×10 17 N. We divide this total force into two exerting parts: one, as an 895 opposing force, balances out the collisional force from Plate C, and the other, as a driving force, balances the basal friction force. We also assume that half of the total force is used to act as the driving force, and then, according to Equation (2) exhibited in section 3, there would be 50%*Ftotal=Fdriving=fbasal=μSu/y (where μ, S, u, and y are the viscosity of the asthenosphere, the area of Plate A, the speed of Plate A, and the thickness of the asthenosphere, respectively). Given μ=10 18 Pas, S=DL=1.2×10 13 m 2 , and y=300 km, we get u=6.18 cm/yr. And now we assume that, at time t2, which has passed 30,000,000 years since time 900 t1, the Ocean 1 depth reduces from 5,000 km to 4,500 km, the Ocean 2 depth increases from 3,000 km to 3,800 km, and that other parameters remain constant, and then, the speed of Plate A would be turned into u=2.24cm/yr. During a period of 30,000,000 years, the rate of Plate A is (6.18-2.24)/30000000=1.31×10 -7 cm/yr. We again assume that, at time t3, which has passed 50,000,000 years since time t2, plate motion stops, and then, the speed of Plate A should be u=0.00 cm/yr. During a period of 50,000,000 years, the rate of Plate A is (2.24-0.00)/50000000=4.48×10 -8 cm/yr. Figure 4 and Table 1 show that the 905 ocean-generated forces around a continental plate are various in both magnitude and orientation, the ocean depth change would require these forces to vary. As a result, the final horizontal force varies, this leads plate motion to vary with time. Our calculation of the speed of Plate A suggests that the change of plate motion is considerably slow, thus, plate motion may be treated as near-steady.
Asteroid impacts are frequent events in the solar system, and it is widely believed that the initiation of plate motion relates to 910 large asteroid impacts (Alvarez, et al., 1980;Rampino and Stothers, 1984;Prinn and Fegley, 1987;Marzoli, et al., 1999;Hames, et al., 2000;Condie, 2001;Wan, 2018), but the details of this coupling remain elusive. Our demonstration here provides the first insight into this issue: asteroid impact fractures the lithosphere, initiating plate motion; ocean water yields force to maintain plate motion; and tides provide energy for plate motion. Ocean depth cannot be stable during a long geological timescale, it may change with the deepening/shallowing of basins. The ocean depth change in turn leads ocean-generated force 915 to vary. We assume that, at time t1, the Ocean 1 depth and the Ocean 2 depth are h1=5,000 km and h2=3,000 km, respectively, the length and width of Plate A are D=6,000 km and L=2,000 km, the water density is ρ=1000 kg/m 3 , the gravitational acceleration is g=9.8 m/s. Then, the total ocean-generated force for Plate A would be Ftotal= FAL-FAR=0.5ρgL(h1 2 -h2 2 )=1.5680×10 17 N. We divide this total force into two exerting parts: one, as an opposing force, balances out the collisional force from Plate C, and the other, as a driving force, balances the basal friction force. We also assume that half of the total 920 force is used to act as the driving force, and then, according to Equation (2) exhibited in section 3, there would be 50%*Ftotal=Fdriving=fbasal=μSu/y (where μ, S, u, and y are the viscosity of the asthenosphere, the area of Plate A, the speed of Plate A, and the thickness of the asthenosphere, respectively). Given μ=10 18 Pas, S=DL=1.2×10 13 m 2 , and y=300 km, we get u=6.18 cm/yr. And now we assume that, at time t2, which has passed 30,000,000 years since time t1, the Ocean 1 depth reduces from 5,000 km to 4,500 km, the Ocean 2 depth increases from 3,000 km to 3,800 km, and that other parameters remain constant, 925 and then, the speed of Plate A would be turned into u=2.24cm/yr. During a period of 30,000,000 years, the rate of Plate A is (6.18-2.24)/30000000=1.31×10 -7 cm/yr. We again assume that, at time t3, which has passed 50,000,000 years since time t2, plate motion stops, and then, the speed of Plate A should be u=0.00 cm/yr. During a period of 50,000,000 years, the rate of Plate A is (2.24-0.00)/50000000=4.48×10 -8 cm/yr. Figure 4 and Table 1 show that the ocean-generated forces around a 56 https://doi.org/10.5194/egusphere-2023-1458 Preprint. Discussion started: 3 August 2023 c Author(s) 2023. CC BY 4.0 License.
continental plate are various in both magnitude and orientation, the ocean depth change would require these forces to vary. As 930 a result, the final horizontal force varies, this leads plate motion to vary with time. Our calculation of the speed of Plate A suggests that the change of plate motion is considerably slow, thus, plate motion may be treated as near-steady.
Asteroid impacts are frequent events in the solar system, and it is widely believed that the initiation of plate motion relates to large asteroid impacts (Alvarez, et al., 1980;Rampino and Stothers, 1984;Prinn and Fegley, 1987;Marzoli, et al., 1999;Hames, et al., 2000;Condie, 2001;Wan, 2018), but the details of this coupling remain elusive. Our demonstration here provides the 935 first insight into this issue: asteroid impact fractures the lithosphere, initiating plate motion; ocean water yields force to maintain plate motion; and tides provide energy for plate motion.

Why does ocean water contribute to tectonics?
All continents are surrounded by oceans, ocean-generated forces are extensively exerted on the sides of the continents that are fixed on top of the lithospheric plates, and all plates connect to each other; consequently, all the plates may interact with each other (Figure 17). Under the effect of ocean-generated force, a moving continental plate would ride on an oceanic plate, and 945 the front part of the oceanic plate is forced to subduct, forming a sinking slab. Additionally, a moving plate would move away from another plate, and a gap would form between them. The gap allows magma to erupt, forming an MOR. From this point, the ridge push force may be treated as an auxiliary force for ocean-generated force. Since ocean-generated force is exerted on the continental wall (represented by coastline), and the oceanic crust is extensively connected to the continental crust, this allows ocean-generated force to be laterally transferred to the oceanic crust, and then, the continental crust drags the oceanic 950 crust to move, causing the plate's boundary to follow the shape of coastline.
Many people are extraordinarily perplexed as to why the Earth owns plate tectonics whereas the Venus does not. Many studies have shown that water provides the right conditions (maintaining a cool surface, for example) for plate tectonics, while the absence of water on Venus prohibits plate formation (Driscoll and Bercovici, 2013;Hilairet et al., 2007;Lenardic et al., 2008;Korenaga, 2007;Tozer, 1985;Lenardic and Kaula, 1994;Hirth and Kohlstedt, 1996;Landuyt and Bercovici, 2009). Our 955 understanding of ocean water provides a new perspective on this issue: no water on Venus means that there is no contribution of ocean-generated force and no further development of plate tectonics on that planet. 59 https://doi.org/10.5194/egusphere-2023-1458 Preprint. Discussion started: 3 August 2023 c Author(s) 2023. CC BY 4.0 License. Figure 17: Global view of the distribution of ocean-generated forces (yellow arrows) and ridge push forces (green arrows). The supporting tidal data are mainly from the Global Sea Level Observing System (GLOSS) database (Caldwell et 960 al., 2015).

What's the trouble with mantle convection?
Although mantle convection had been given up by most of geophysicists since the late 1990s, this cannot prevent it from becoming popular. We here restate how mantle convection cannot be realistic. The main problems with this paradigm include: 965 1) the convection cells proposed to exist in the asthenosphere require strong fitting to plate size. Seismic tomography shows that rising mantle material beneath ridges only extends down 200 to 400 km (Foulger, et al., 2001). This depth gives an upper limitation on the scale of the proposed cells. Most of plates (North American, Eurasian, and Pacific, for instance), however, are very wide, generally more than thousands of kilometers. 2) the movement of a large plate would yield a net mass flux in the asthenosphere to compensate the mass transport in the 970 moving plate, this requires the plate to be treated as an integral part of the circulation (Forsyth & Uyeda, 1975). Richter (1973) employed a model to show that the asthenosphere exerts viscous resistive forces rather than driving forces on the plates, which actually opposes the mantle convection currents to act as the drivers.
3) in the scenario of mantle convection, poloidal motion involves vertical upwellings and downwellings, while toroidal motion undertakes horizontal rotation (Bercovic, et al., 2015). The generation of torioidal motion requires variable viscosity, but 975 numerous studies of basic 3-D convection with temperature-dependent viscosity had failed to yield the requisite toroidal flow (Bercovic, 1993(Bercovic, , 1995bCadek et al., 1993;Christensen and Harder, 1991;Stein et al., 2004;Tackley, 1998;Trompert and Hansen, 1998;Weinstein, 1998). 4) mantle convection models are unsuccessful in yielding plate motions, although some of them had yielded plate-like behavior and mathematically got solution for plate motion velocity by means of a non-Newtonian way, i.e., a balance relationship of 980 buoyancy force and drag force (Bercovici, et al., 2015). Doglioni and Panza (2015) concluded that none of mantle convection models is really able to satisfy the constraints posed by the plate kinematics, the temperature, and the asymmetry of plate boundaries. 5) when mantle convection is treated as the plate driving force, it requires not only the mantle to couple the plates but also the movement of mantle currents to keep consistent with that of the plates. But reality is not so. The Pacific Plate is the fastest W-985 ward moving one relative to the mantle and is slipping over an Low Velocity Zone (LVZ) (Doglioni et al., 2005) with low viscosity (Pollitz et al., 1998). Evidently, the Pacific is the most decoupled plate, while mantle convection requires the faster moving plates to be more coupled (higher viscosity) with the mantle. The Hawaii hot spot volcanic chain represents that the underlying mantle is moving E-SE-ward. These authors (Hammond and Toomey, 2003;Doglioni et al., 2003) modeled, beneath the East Pacific Rise (EPR), an eastward-migrating mantle. The hot spot reference frame remains consistent with the 990 existence of an eastward relative mantle flow beneath the South America plate (Van Hunen, van den Berg, & Vlaar, 2002). A relatively moving eastward mantle flow has been proposed also beneath North America (Silver & Holt, 2002) and beneath the Caribbean plate (Gonzalez, Alvarez, Moreno, & Panza, 2011;Negredo, Jiménez-Munt, & Villasenor, 2004). All these results indicate that the movements of the mantle currents are reverse to that of the plates, the opposed moving mantle currents provide resistive force rather than driving force for these plates. 995 6) our modeling analysis suggests that, if mantle convection were considered as a driving force, what it performs is a drag on the plate's base, more like the basal friction force Fb exhibited in Figure 1, and then, its resultant stress must be mainly concentrated on the lowermost part of the plate, which cannot be in accordance with observed stress.