The dominant tectonic mode operating on early Earth (before ca. 2.5 Ga) remains elusive, with an increasing body of evidence suggesting that non-plate tectonic modes were likely more prevalent at that time. Thus, how plate tectonics evolved after that remains contentious. We performed two-dimensional numerical modeling of mantle convection at temperatures appropriate for the Hadean–Archean eons and show that subduction and rift systems may have spontaneously emerged on Earth from an earlier drip-and-rift–dominated tectonic mode in response to the secular cooling of the mantle. This cooling of the mantle was mediated by repeated events of rifting and dripping that likely occurred over a few hundred million years. As the mantle cooled, its effective viscosity and the thickness and strength of the lithosphere increased, which helped establish rigid plates and initiate plate tectonics on Earth.At present, Earth’s interior cools by a mantle-convection mode known as plate tectonics, in which the lithosphere plays a critical role (cf. Lenardic, 2018). It is comprised of a globally linked network of rigid lithospheric plates that are separated by weak plate margins and that participate in rifting, subduction-collision, and transform faulting. However, non-plate tectonic modes, like stagnant or sluggish-lid tectonics, have been suggested to be more dominant on early Earth (e.g., Moore and Webb, 2013; Sizova et al., 2015; Fischer and Gerya, 2016; Rozel et al., 2017; Capitanio et al., 2020; Foley, 2020; Lourenço et al., 2020), albeit alternate views also exist (e.g., Miyazaki and Korenaga, 2022; Hastie et al., 2023). The hypothesis of early non-plate tectonic modes relies on two arguments: (1) mantle temperatures were higher in the Hadean–Archean than today (Herzberg et al., 2010; Condie et al., 2016), and (2) mantle viscosity is strongly dependent on temperature, which affects convective vigor and pattern (Karato and Wu, 1993; Tackley, 1993; Moresi and Solomatov, 1995). These factors would have reduced convective stresses during the Hadean–Archean, likely failing to overcome lithospheric strength (Moresi and Solomatov, 1995; O’Neill et al., 2007; Lenardic, 2018). Geological evidence increasingly favors a non-plate tectonic mode, at least before ca. 3.2–3.0 Ga (e.g., Johnson et al., 2017; Bédard, 2018; Stern, 2018; Brown et al., 2020; Chowdhury et al., 2021; Palin and Santosh, 2021; Cawood et al., 2022). This leads us to one of the highly contentious questions in solid Earth sciences: why and how did the Earth transition from a non-plate tectonic mode to plate tectonics?Subduction is the primary driver of plate motions on the modern Earth (Forsyth and Uyeda, 1975). Therefore, to understand how plate tectonics emerged, we need to resolve how episodes of subduction can initiate within a non-plate tectonic mode. While models suggest that meteoritic impacts or mantle plumes may cause isolated subduction events (Gerya et al., 2015; O’Neill et al., 2020), it remains unclear as to how subduction can initiate without such external forces. Although subduction may spontaneously result from lithospheric differentiation or gravitational collapse of continents (Rey et al., 2014; Sizova et al., 2015; Capitanio et al., 2020), a general mechanism showing how mantle cooling may initiate subduction and plate tectonics is missing.We carried out numerical modeling of mantle convection starting from high mantle potential temperatures (Tp) and investigated how lithosphere dynamics spontaneously change through time as Tp decreases. We identify various tectonic stages by observing surface heat flow (Qs) and horizontal lithospheric velocity (vx) that show a co-evolution of mantle Tp, mantle viscosity, and the style of lithosphere dynamics. By elucidating this dynamic feedback, we propose a self-consistent mechanism through which Earth progressed from a non-plate tectonic mode to plate tectonics.We modeled mantle convection in a 1200-km-deep × 4800-km-wide two-dimensional (2-D) Cartesian domain using Underworld2 (https://www.underworldcode.org/intro-to-underworld/; Fig. S1 in the Supplemental Material1). We varied the starting mantle Tp in the models between 1500 °C and 1600 °C (Table S1) following the Archean mantle Tp estimates (e.g., Herzberg et al., 2010; Ganne and Feng, 2017). We used suitable temperature-dependent rheology for each rock type and considered the effects of mantle melting and higher radiogenic heat production. Details of the modeling are given in the Supplemental Material.Figure 1 shows the evolution of our reference model with an initial mantle Tp of 1600 °C. With the establishment of mantle convection, the lithosphere enters into a stagnant-lid mode but undergoes weak internal deformation. The stagnant lid is broken by thermal instabilities introduced at the lower boundary. Thereafter, the lithosphere undergoes extension to create rift-like settings, while undergoing compression at other places, creating thickened lithospheric segments (plateaus; Fig. 1A). Lithospheric thickening in these plateau regions leads to the formation of denser minerals like garnet within the deep basaltic crust, which in turn triggers Rayleigh-Taylor instabilities and dripping (Figs. 1B–1E). The hotter geotherm of the lithosphere lowers its strength, further facilitating the initiation and growth of drips. While individual dripping events last for <10 m.y., repeated events of lithospheric thickening followed by dripping (and concomitant extension) continue for a few hundred m.y. (Figs. 1A–1E). This tectonic mode of the model is referred to as the “drip-and-rift” mode, which evolves into a “subduction-and-rift” mode with time. This mode starts with the formation of a drip that grows until one side of the lithosphere begins to subduct (Figs. 1F–1I). The subducting lithosphere rolls back, mimicking the convective flow dynamics of a retreating subduction system (Figs. 1F–1I). The subduction terminates via slab break-off, which initiates through lithospheric necking at a shallow depth (<60–70 km; Fig. 1J). At the same time, the zone of rifting migrates toward the subduction zone. The overall duration of the subduction event from initiation to slab break-off is <30 m.y.Thus, our model shows a spontaneous transition from the drip-and-rift mode to subduction-and-rift mode after ~1100 m.y. of evolution, which is also evident in the Nusselt number versus time plot (shown in Fig. S3). We further quantified vx and Qs profiles from the model to better characterize the tectonic modes and measure the plateness of lithospheric segments (Fig. S2; Tackley, 2000). The drip-and-rift mode shows variable lithospheric mobility (Figs. 2A and 2B). vx is significantly high (~50–60 cm/a) around the dripping location and decreases as one moves away from the drip (Fig. 2B). The strong downward pull exerted by the drip causes this lithospheric mobility. However, the stresses are not uniformly transmitted laterally over large distances (<600 km), as is evident from the declining vx, reflecting the lithosphere’s non-rigid nature. In contrast, the subduction-and-rift mode displays the vx profile of a rigid lithosphere undergoing subduction. The subducting lithosphere shows a constant vx of ~4–6 cm/a over a length scale of ~1200 km (Figs. 2A and 2B), suggesting that it responds to large horizontal stresses via lateral displacement, and not by internal deformation, as expected by a coherent and rigid plate (Bercovici, 2003; van Hunen and van den Berg, 2008). The subduction trench shows a sharp velocity change in a narrow (<200-km-wide) zone. The overriding plate also shows a constant vx, suggesting that its mechanical response to stress is similar to that of the subducting plate (Fig. 2B). Such a tectonic setting—wide rigid lithospheres separated by narrow deforming margins—is consistent with plate tectonics.Furthermore, the lithosphere forms in a rift setting and becomes thicker as it migrates closer to the subduction trench during the subduction-and-rift mode (Fig. 2A). Thus, the corresponding Qs profile shows progressive decline in conductive heat-loss through the lithosphere as it thickens with increasing distance from the rift axis (Fig. 2C), analogous to modern oceanic lithospheres (cf. Turcotte and Schubert, 2014). The high heat-flow regions are restricted to rifts and overriding lithosphere inboard of the trench where the mantle rises due to slab roll-back, resembling the dynamics of a retreating subduction system. In contrast, the drip-and-rift mode shows high heat-flow regions corresponding to broad regions of thinned lithosphere, whereas the plateau and/or dripping regions are marked by low heat flow (Fig. 2C). Notably, the model evolution suggests that the drip-and-rift mode is more efficient in transmitting heat through the lithosphere over relatively shorter time scales as compared to the subduction-and-rift stage (Fig. 2C). This effect will be more pronounced in natural (three-dimensional) settings, where the radial flux of the lithosphere will feed the drip. Models with different initial mantle Tp values show tectonic evolution similar to the reference model (Figs. S4 and S5). They show the same development of symmetric drips followed by asymmetric subduction zones as a function of declining mantle temperature (Figs. S6 and S7); however, the mantle thermal condition at which the tectonic transitions occur, and thereby their timing, differ from those observed in the reference model (Fig. S8).Our models illustrate the spontaneous initiation of subduction from a non-plate tectonic mode due to the cooling of mantle, which changes the thermomechanical properties of the lithosphere-mantle system. To illustrate this, we determined how average Qs and average (upper) mantle temperature and effective viscosity vary with model time (Fig. 3). During the stagnant-lid stage, the average Qs is at its minimum since the heat loss happens only by conduction (Fig. 3A). Hence, the average mantle temperature remains high, which keeps its effective viscosity low (Fig. 3B). As episodes of dripping and rifting begin, these are associated with rapid heat loss (~200–300 mW/m2) interspersed with longer intervals of reduced heat loss (~60–100 mW/m2; Fig. 3A). The high heat loss is due to rifting associated with large-scale mantle upwelling and lithospheric dripping. The cold drips further cool down the mantle. Notably, the magnitude of average Qs during periods of high heat loss shows a near-exponential decay with time, suggesting that the rate of mantle cooling is a function of the frequency of rifting and dripping. This cooling manifests in the steady decline in average mantle temperature over ~2.5 b.y. (Fig. 3B).During the subduction-and-rift stage, periods of low heat loss are longer, while periods of high heat loss show a near-constant magnitude of average Qs (~100–125 mW/m2) (Fig. 3A). The periods of high heat loss correspond to the episodes of subduction and rifting, where the mantle cools via convective heat loss at rifts, and via mixing with cold lithospheres at subduction zones. Notably, the magnitude of average Qs during periods of high heat loss is much lower than the corresponding average Qs during the drip-and-rift stage. The decline in mantle temperature leads to the increase of its effective viscosity with time (Fig. 3B). At this point, the pattern of mantle convection transitions from multiple small-wavelength cells to fewer large-wavelength cells (Fig. 2A). We explain this transition by re-balancing the main forces of plate tectonics: the buoyancy force and the viscous resistance to the mantle flow. Buoyancy declines with decreasing temperature, and viscous resistance increases with decreasing temperature (Fig. 3B). Mantle cooling also leads to cooling and consequent thickening of the lithosphere, which increases lithospheric strength. This is substantiated by the evolution of our models’ thermal and viscosity profiles (averaged over model width), which show a drop in lithospheric temperature and a concomitant increase in lithospheric thickness with time (Figs. 3C and 3D).Thus, our results suggest that efficient mantle cooling during dripping and rifting increases mantle viscosity and lithospheric strength, which helps establish rigid lithospheres (i.e., plates) and plate margin–like processes. The rate of heat loss during the drip-and-rift mode is higher than during the subduction-and-rift mode, implying that repeated occurrences of dripping and rifting may have significantly cooled Earth’s interior. However, our models predict that the mantle was still hotter by 130–190 °C than today, when the subduction-and-rifting mode may have begun (Fig. S8). This is why our modeled subduction style (e.g., shorter duration, shallow slab break-off) differs from modern, cold-style subduction, which agrees well with the results of previous modeling studies (e.g., Sizova et al., 2015, Fischer and Gerya, 2016; Gerya et al., 2021). The mantle thermal conditions of our model may have existed during the late Archean to early Proterozoic (Herzberg et al., 2010). We understand that the relation between Earth’s mantle Tp and age is debated (e.g., Herzberg et al., 2010; Ganne and Feng, 2017), and our 2-D modeling represents regional, rather than global, tectonics—both of which may affect the timing of tectonic transitions. Nevertheless, the concurrence of a wide range of evidence in the rock record toward a diachronous establishment of plate tectonics across Earth at ca. 3.2–2.2 Ga (cf. Brown et al., 2020; Cawood et al., 2022) supports our inferences and is consistent with how weak plate boundaries may have developed during that time (Bercovici and Ricard, 2014).This research is supported by Australian Research Council grant FL160100168, the National Institute of Science Education and Research (India) internal funding, and NASA grant 20-EW20_2-0026. We thank Tim Johnson, Jeroen van Hunen, Taras Gerya, Jun Korenaga, and an anonymous reviewer for their constructive comments and Urs Schaltegger for editorial handling. We also thank Fabio Capitanio and Oliver Nebel for insightful discussions about early Earth tectonics and Shayne McGregor for providing resources at National Computational Infrastructure, Australia.

中文翻译：

---

将地幔冷却与早期地球构造转变联系起来

早期地球（约 2.5 Ga 之前）的主要构造模式仍然难以捉摸，越来越多的证据表明非板块构造模式在当时可能更为普遍。因此，板块构造在此之后如何演化仍然存在争议。我们在适合冥古宙-太古宙的温度下对地幔对流进行了二维数值模拟，结果表明，俯冲和裂谷系统可能是响应长期冷却而从早期的滴水和裂谷主导的构造模式中自发出现在地球上的。地幔的。地幔的冷却是由可能发生了几亿年的重复裂谷和滴水事件介导的。随着地幔冷却，其有效粘度以及岩石圈的厚度和强度增加，这有助于在地球上建立刚性板块并启动板块构造。目前，地球内部通过称为板块构造的地幔对流模式冷却，其中岩石圈起着至关重要的作用（参见 Lenardic，2018）。它由全球相连的刚性岩石圈板块网络组成，这些板块被薄弱的板块边缘分隔开，并参与裂谷、俯冲碰撞和转换断层。然而，非板块构造模式，如停滞或缓慢盖构造，被认为在早期地球上更为主导（例如，Moore 和 Webb，2013；Sizova 等，2015；Fischer 和 Gerya，2016；Rozel 等） al.，2017；Capitanio 等人，2020；Foley，2020；Lourenço 等人，2020），尽管也存在替代观点（例如，Miyazaki 和 Korenaga，2022；Hastie 等人，2023）。早期非板块构造模式的假说依赖于两个论据：（1）冥古宙-太古宙地幔温度比今天更高（Herzberg et al., 2010; Condie et al., 2016），（2）地幔粘度强烈依赖于温度，而温度影响对流强度和模式（Karato 和 Wu，1993；Tackley，1993；Moresi 和 Solomatov，1995）。这些因素会减少冥宙-太古代期间的对流应力，可能无法克服岩石圈强度（Moresi 和 Solomatov，1995；O'Neill 等，2007；Lenardic，2018）。地质证据越来越倾向于非板块构造模式，至少在大约之前。 3.2–3.0 Ga（例如，Johnson 等人，2017 年；Bédard，2018 年；Stern，2018 年；Brown 等人，2020 年；Chowdhury 等人，2021 年；Palin 和 Santosh，2021 年；Cawood 等人，2022 年）。这将我们引向固体地球科学中极具争议性的问题之一：地球为何以及如何从非板块构造模式转变为板块构造模式？俯冲作用是现代地球上板块运动的主要驱动力（Forsyth 和 Uyeda， 1975）。因此，为了了解板块构造是如何出现的，我们需要解决俯冲事件如何在非板块构造模式下启动的问题。虽然模型表明陨石撞击或地幔柱可能会导致孤立的俯冲事件（Gerya 等人，2015 年；O'Neill 等人，2020 年），目前尚不清楚如何在没有此类外力的情况下引发俯冲。尽管俯冲可能是由岩石圈分异或大陆重力塌陷自发引起的（Rey等，2014；Sizova等，2015；Capitanio等，2020），但显示地幔冷却如何引发俯冲和板块构造的一般机制是我们从地幔高位温 (Tp) 开始对地幔对流进行了数值模拟，并研究了随着 Tp 降低，岩石圈动力学如何随时间自发变化。我们通过观察地表热流 (Qs) 和水平岩石圈速度 (vx) 来识别各个构造阶段，这些速度显示了地幔 Tp、地幔粘度和岩石圈动力学类型的共同演化。通过阐明这种动态反馈，我们提出了一种自洽机制，地球通过该机制从非板块构造模式发展到板块构造模式。我们在 1200 公里深 × 4800 公里宽的二维（2 -D) 使用 Underworld2 的笛卡尔域（https://www.underworldcode.org/intro-to-underworld/；补充材料 1 中的图 S1）。我们根据太古代地幔 Tp 估计，将模型中的起始地幔 Tp 更改为 1500 °C 至 1600 °C（表 S1）（例如，Herzberg 等人，2010 年；Ganne 和 Feng，2017 年）。我们对每种岩石类型使用了合适的温度依赖性流变学，并考虑了地幔熔化和更高的放射热产生的影响。补充材料中给出了建模的详细信息。图 1 显示了我们的参考模型的演化，初始地幔 Tp 为 1600 °C。随着地幔对流的建立，岩石圈进入停滞盖模式，但发生微弱的内部变形。停滞的盖子被下边界引入的热不稳定性打破。此后，岩石圈经历伸展，形成裂谷状环境，同时在其他地方经历压缩，形成增厚的岩石圈部分（高原；图1A）。这些高原地区的岩石圈增厚导致玄武岩地壳深处形成更致密的矿物，如石榴石，进而引发瑞利-泰勒不稳定性和滴落（图1B-1E）。岩石圈较热的地温降低了其强度，进一步促进了滴水的产生和增长。虽然单个滴落事件持续<10 my，但岩石圈增厚随后滴落（以及伴随的延伸）的重复事件持续数百 my（图1A-1E）。该模式的这种构造模式被称为“滴水裂谷”模式，随着时间的推移演化为“俯冲裂谷”模式。这种模式始于水滴的形成，水滴不断增长，直到岩石圈的一侧开始俯冲（图1F-1I）。俯冲岩石圈回滚，模仿后退俯冲系统的对流动力学（图1F-1I）。俯冲通过板片断裂终止，板片断裂是通过浅深度（<60-70 km；图 1J）的岩石圈颈缩开始的。与此同时，裂谷带向俯冲带迁移。俯冲事件从起始到板片断裂的总持续时间<30 my，因此，我们的模型显示在约 1100 my 的演化后，从滴水裂谷模式到俯冲裂谷模式的自发转变，这也是在努塞尔数与时间的关系图中很明显（如图 S3 所示）。我们进一步量化了模型中的 vx 和 Qs 剖面，以更好地表征构造模式并测量岩石圈段的平板度（图 S2；Tackley，2000）。滴水裂谷模式显示出不同的岩石圈活动性（图 2A 和 2B）。 vx 在滴水位置周围明显较高（~50-60 cm/a），并随着远离滴水而降低（图 2B）。水滴施加的强大向下拉力导致了这种岩石圈的流动性。然而，从 vx 的下降可以明显看出，应力在长距离（<600 km）上并没有均匀地横向传递，反映了岩石圈的非刚性性质。相比之下，俯冲和裂谷模式显示了正在俯冲的刚性岩石圈的 vx 剖面。俯冲岩石圈在约 1200 km 的长度范围内显示出约 4-6 cm/a 的恒定 vx（图 2A 和 2B），这表明它通过侧向位移而不是内部变形来响应大的水平应力，如连贯且刚性的板所期望的（Bercovici，2003；van Hunen 和 van den Berg，2008）。俯冲沟在狭窄（<200 公里宽）区域显示出急剧的速度变化。上覆板块也显示出恒定的 vx，表明其对应力的机械响应与俯冲板块相似（图 2B）。这种构造环境——宽的刚性岩石圈被狭窄的变形边缘分开——与板块构造一致。此外，岩石圈在裂谷环境中形成，并且在俯冲和裂谷模式期间随着其迁移到靠近俯冲海沟而变得更厚（图1）。 .2A）。因此，相应的 Qs 剖面显示，随着岩石圈随着距裂谷轴距离的增加而变厚，岩石圈的传导热损失逐渐下降（图 2C），类似于现代海洋岩石圈（参见 Turcotte 和 Schubert，2014）。高热流区域仅限于海沟内侧的裂谷和覆盖岩石圈，其中地幔因板块回滚而上升，类似于后退俯冲系统的动力学。相比之下，滴水和裂谷模式显示出与岩石圈变薄的广阔区域相对应的高热流区域，而高原和/或滴水区域则以低热流为标志（图2C）。值得注意的是，模型演化表明，与俯冲和裂谷阶段相比，滴水和裂谷模式在相对较短的时间尺度内通过岩石圈传输热量更有效（图 1）。2C）。这种效应在自然（三维）环境中会更加明显，其中岩石圈的径向通量将供给滴水。不同初始地幔Tp值的模型显示出与参考模型相似的构造演化（图S4和S5）。它们显示了对称滴水的相同发展，随后是不对称俯冲带，作为地幔温度下降的函数（图S6和S7）；然而，构造转变发生的地幔热条件及其时间与参考模型中观察到的不同（图S8）。我们的模型说明了由于地幔的冷却，改变了岩石圈-地幔系统的热机械特性。为了说明这一点，我们确定了平均 Qs 和平均（上）地幔温度以及有效粘度如何随模型时间变化（图 3）。在停滞盖阶段，平均 Qs 处于最小值，因为热量损失仅通过传导发生（图 3A）。因此，地幔平均温度仍然很高，这使其有效粘度保持在较低水平（图3B）。当滴水和裂痕开始出现时，这些现象与快速热损失（约 200–300 mW/m2）相关，并伴有较长时间间隔的热损失减少（约 60–100 mW/m2；图 3A）。高热量损失是由于与大规模地幔上涌和岩石圈滴水相关的裂谷造成的。寒冷的水滴进一步冷却了地幔。值得注意的是，在高热量损失期间，平均 Qs 的大小显示出随时间的近指数衰减，这表明地幔冷却速率是裂谷和滴水频率的函数。这种冷却表现为平均地幔温度稳定下降约 2.5%（图 3B）。在俯冲和裂谷阶段，低热量损失周期较长，而高热量损失周期则表现出近乎恒定的幅度。平均 Qs (~100–125 mW/m2)（图 3A）。高热损失时期对应于俯冲和裂谷时期，其中地幔通过裂谷处的对流热损失以及通过在俯冲带与冷岩石圈混合而冷却。值得注意的是，高热损失期间的平均 Q 值远低于滴水和裂谷阶段相应的平均 Q 值。地幔温度的下降导致其有效粘度随时间增加（图3B）。此时，地幔对流模式从多个小波长单元转变为较少的大波长单元（图2A）。我们通过重新平衡板块构造的主要力量来解释这种转变：浮力和地幔流的粘性阻力。浮力随温度降低而降低，粘性阻力随温度降低而增加（图3B）。地幔冷却还会导致岩石圈冷却并随之增厚，从而增加岩石圈强度。我们模型的热分布和粘度分布（在模型宽度上取平均值）的演变证实了这一点，这些分布显示岩石圈温度随时间下降，岩石圈厚度随之增加（图 3C 和 3D）。因此，我们的结果表明滴水和裂谷过程中有效的地幔冷却增加了地幔粘度和岩石圈强度，这有助于建立刚性岩石圈（即板块）和类板块边缘过程。滴水和裂谷模式期间的热损失率高于俯冲和裂谷模式，这意味着滴水和裂谷的重复发生可能显着冷却了地球内部。然而，我们的模型预测，当俯冲和裂谷模式可能已经开始时，地幔的温度仍然比今天高 130-190 °C（图 S8）。这就是为什么我们模拟的俯冲类型（例如，较短的持续时间、浅板片断裂）不同于现代的冷型俯冲，这与之前的建模研究的结果非常吻合（例如，Sizova 等人，2015 年，Fischer 和Gerya，2016；Gerya 等人，2021）。我们模型中的地幔热条件可能存在于晚太古代到早元古代(Herzberg et al., 2010)。我们知道地幔 Tp 和年龄之间的关系存在争议（例如，Herzberg 等人，2010 年；Ganne 和 Feng，2017 年），并且我们的二维模型代表了区域构造，而不是全球构造，这两者都可能影响构造转变的时间。尽管如此，岩石记录中的大量证据一致表明，大约在大约 1970 年，地球各地的板块构造已经建立了历时性。 3.2–2.2 Ga（参见 Brown 等人，2020 年；Cawood 等人，2022 年）支持我们的推论，并且与当时板块边界薄弱的发展情况一致（Bercovici 和 Ricard，2014 年）。这项研究得到支持由澳大利亚研究委员会拨款 FL160100168、国家科学教育与研究学院（印度）内部资助以及 NASA 拨款 20-EW20_2-0026。我们感谢 Tim Johnson、Jeroen van Hunen、Taras Gerya、Jun Korenaga 和一位匿名审稿人提出的建设性意见，并感谢 Urs Schaltegger 的编辑处理。我们还感谢 Fabio Capitanio 和 Oliver Nebel 对早期地球构造的富有洞察力的讨论，感谢 Shayne McGregor 在澳大利亚国家计算基础设施提供的资源。这意味着重复发生的滴水和裂谷可能显着降低了地球内部的温度。然而，我们的模型预测，当俯冲和裂谷模式可能已经开始时，地幔的温度仍然比今天高 130-190 °C（图 S8）。这就是为什么我们模拟的俯冲类型（例如，较短的持续时间、浅板片断裂）不同于现代的冷型俯冲，这与之前的建模研究的结果非常吻合（例如，Sizova 等人，2015 年，Fischer 和Gerya，2016；Gerya 等人，2021）。我们模型中的地幔热条件可能存在于晚太古代到早元古代(Herzberg et al., 2010)。我们知道地幔 Tp 和年龄之间的关系存在争议（例如，Herzberg 等人，2010 年；Ganne 和 Feng，2017 年），并且我们的二维模型代表了区域构造，而不是全球构造，这两者都可能影响构造转变的时间。尽管如此，岩石记录中的大量证据一致表明，大约在大约 1970 年，地球各地的板块构造已经建立了历时性。 3.2–2.2 Ga（参见 Brown 等人，2020 年；Cawood 等人，2022 年）支持我们的推论，并且与当时板块边界薄弱的发展情况一致（Bercovici 和 Ricard，2014 年）。这项研究得到支持由澳大利亚研究委员会拨款 FL160100168、国家科学教育与研究学院（印度）内部资助以及 NASA 拨款 20-EW20_2-0026。我们感谢 Tim Johnson、Jeroen van Hunen、Taras Gerya、Jun Korenaga 和一位匿名审稿人提出的建设性意见，并感谢 Urs Schaltegger 的编辑处理。我们还感谢 Fabio Capitanio 和 Oliver Nebel 对早期地球构造的富有洞察力的讨论，感谢 Shayne McGregor 在澳大利亚国家计算基础设施提供的资源。这意味着重复发生的滴水和裂谷可能显着降低了地球内部的温度。然而，我们的模型预测，当俯冲和裂谷模式可能已经开始时，地幔的温度仍然比今天高 130-190 °C（图 S8）。这就是为什么我们模拟的俯冲类型（例如，较短的持续时间、浅板片断裂）不同于现代的冷型俯冲，这与之前的建模研究的结果非常吻合（例如，Sizova 等人，2015 年，Fischer 和Gerya，2016；Gerya 等人，2021）。我们模型中的地幔热条件可能存在于晚太古代到早元古代(Herzberg et al., 2010)。我们知道地幔 Tp 和年龄之间的关系存在争议（例如，Herzberg 等人，2010 年；Ganne 和 Feng，2017 年），并且我们的二维模型代表了区域构造，而不是全球构造，这两者都可能影响构造转变的时间。尽管如此，岩石记录中的大量证据一致表明，大约在大约 1970 年，地球各地的板块构造已经建立了历时性。 3.2–2.2 Ga（参见 Brown 等人，2020 年；Cawood 等人，2022 年）支持我们的推论，并且与当时板块边界薄弱的发展情况一致（Bercovici 和 Ricard，2014 年）。这项研究得到支持由澳大利亚研究委员会拨款 FL160100168、国家科学教育与研究学院（印度）内部资助以及 NASA 拨款 20-EW20_2-0026。我们感谢 Tim Johnson、Jeroen van Hunen、Taras Gerya、Jun Korenaga 和一位匿名审稿人提出的建设性意见，并感谢 Urs Schaltegger 的编辑处理。我们还感谢 Fabio Capitanio 和 Oliver Nebel 对早期地球构造的富有洞察力的讨论，感谢 Shayne McGregor 在澳大利亚国家计算基础设施提供的资源。2022）支持我们的推论，并且与在那段时间可能发展的薄弱板块边界一致（Bercovici 和 Ricard，2014）。这项研究得到了澳大利亚研究委员会拨款 FL160100168、国家科学教育与研究学院（印度）内部的支持资金，以及 NASA 拨款 20-EW20_2-0026。我们感谢 Tim Johnson、Jeroen van Hunen、Taras Gerya、Jun Korenaga 和一位匿名审稿人提出的建设性意见，并感谢 Urs Schaltegger 的编辑处理。我们还感谢 Fabio Capitanio 和 Oliver Nebel 对早期地球构造的富有洞察力的讨论，感谢 Shayne McGregor 在澳大利亚国家计算基础设施提供的资源。2022）支持我们的推论，并且与在那段时间可能发展的薄弱板块边界一致（Bercovici 和 Ricard，2014）。这项研究得到了澳大利亚研究委员会拨款 FL160100168、国家科学教育与研究学院（印度）内部的支持资金，以及 NASA 拨款 20-EW20_2-0026。我们感谢 Tim Johnson、Jeroen van Hunen、Taras Gerya、Jun Korenaga 和一位匿名审稿人提出的建设性意见，并感谢 Urs Schaltegger 的编辑处理。我们还感谢 Fabio Capitanio 和 Oliver Nebel 对早期地球构造的富有洞察力的讨论，感谢 Shayne McGregor 在澳大利亚国家计算基础设施提供的资源。