Preamble

Vol. 43, No. 2

June 2025

Progress in Astronomy Vol. 43, No. 2 June 2025 doi: 10.3969/j.issn.1000-8349.2025.02.02

Research Progress on Mantle Convection in Terrestrial Planets and Large Satellites

TANG Ruirui¹;², LI Ligang¹;²

(1. Shanghai Astronomical Observatory, Chinese Academy of Sciences, Shanghai 200030, China; 2. University of Chinese Academy of Sciences, Beijing 100049, China)

Abstract

Mantle convection is a crucial internal process in the evolution of terrestrial planets. This paper reviews the latest research progress on mantle convection in terrestrial planets and large satellites within the solar system, employing a comparative planetology approach to address three fundamental questions: (1) Does mantle convection exist within terrestrial planets? (2) If mantle convection exists, what are its patterns and characteristics? (3) What conditions must be satisfied for mantle convection to occur in terrestrial planets? We discuss the primary factors influencing mantle convection in terrestrial planets, including geometric scale, physical parameters, driving sources, fluid viscosity, thermal and kinematic boundary conditions, differentiation, and phase transitions, along with convection in rocky mantles and ice mantles of some large satellites. Clarifying the origins, mechanisms, and impacts of mantle convection across different celestial bodies holds significant importance for understanding planetary internal dynamics and evolution.

Keywords: mantle convection; terrestrial planet; large satellite; thermal evolution; convection pattern

1.1 Proposal of Mantle Convection Theory

Mantle convection refers to the movement of mantle material within a planet that transports heat and mass through natural convection, also known as subsolidus mantle convection [1]. This theory was first proposed by Hopkins in 1839 [2] and discussed by Fisher in 1889 [3]. In the first half of the 20th century, Holmes formally proposed the mantle convection hypothesis [4–6] to explain the driving force source for Wegener's continental drift hypothesis.

With continuous deep space exploration activities, years of research indicate that other terrestrial bodies (including terrestrial planets, large satellites, dwarf planets, asteroids, etc.) may similarly possess mantle convection analogous to Earth's, which we collectively refer to as mantle convection. Mantle convection is driven by internal heat sources within a planet, with energy primarily originating from gravitational potential energy release during planetary formation and evolution, radioactive isotope decay (such as ²³⁵U, ²³⁸U, ²³²Th, ⁴⁰K) [7], and impacts from small asteroids [8]. Mantle convection is closely related to a planet's internal structure (such as mantle and crust thickness, core size), plate tectonics, gravity field, distribution of radioactive isotope materials, surface heat flow, and rock rheology, representing one of the primary causes of observed surface anomalies including geoid anomalies, gravity anomalies, surface topography, and lithospheric stress field anomalies [9]. As the main mechanism for internal energy release to outer space in terrestrial planets (and large satellites), mantle convection determines a planet's cooling rate and thermal state, while also serving as a means of internal material migration and cycling. Mantle convection not only drives surface geological tectonic movements such as mountain-building processes and fault zones but also controls the formation, disappearance, and evolution of planetary magnetic fields [10–12], thus playing a vital dynamic role in the evolution of terrestrial planets.

1.2 Theoretical Basis of Mantle Convection

The theory of mantle convection is primarily established on the foundations of planetary internal thermodynamics and fluid mechanics, encompassing four main theoretical aspects [1, 13–15]. First, thermodynamic principles: mantle convection arises from temperature differences within a planet. Internal heat, mainly from the planetary core, conducts outward through the mantle, creating temperature gradients that drive material migration and heat transfer from high-temperature to low-temperature regions, forming convective motion. Second, fluid mechanics: although the mantle consists of relatively rigid and partially molten rock, it can be treated as a fluid over geological timescales. Fluid mechanics theory describes the fundamental principles of mantle convection, including fluid motion laws, momentum transfer, and energy conservation. Third, petrology and geochemistry: mantle convection is also influenced by petrological and geochemical factors. Petrology investigates the physical properties and chemical composition of mantle rocks, while geochemistry studies element composition and reaction characteristics, providing essential physical property foundations for understanding mantle convection. Fourth, internal structure and dynamics: the internal structure and dynamic processes directly affect the morphology and characteristics of mantle convection, including material stratification and plate tectonic movements.

1.3 Indirect Evidence of Mantle Convection

Currently, we cannot directly observe mantle convection due to two main reasons: (1) existing geological drilling depths remain far from reaching the mantle layer; and (2) mantle convection is an extremely slow process with timescales comparable to plate drift, reaching millions of years, while humans have not yet accumulated sufficiently long observational data. Present research focuses on demonstrating the existence of mantle convection, studying its patterns, establishing convection models, and exploring its effects on the crust, lithosphere, and core.

Although direct observation remains impossible, indirect methods can prove its existence. Among these, plate tectonic theory, particularly continental drift and seafloor spreading, is considered powerful evidence for mantle convection. In the asthenosphere, convection cells exert horizontal viscous forces on the lithosphere, dragging plate movements and providing driving forces [16–18]. Since plate motion represents primary evidence and a direct product of mantle convection, mantle convection can be indirectly observed through plate movement detection.

Beyond plate tectonics, mantle plumes represent another form of mantle convection—columnar thermal upwellings from the core-mantle boundary. Wilson [19] first proposed in 1963 that the Hawaiian island chain represents traces left by a fixed heat source region beneath a moving plate, termed "hotspots" [20, 21]. Later, he further clarified that hotspots are surface manifestations of mantle plumes rising from the mantle, marked by volcanism, high heat flow, and uplift. Morgan [22] formally proposed the mantle plume hypothesis in 1972, suggesting that fixed heat source regions originate from thermal columns in the lower mantle, formed by radioactive element heating and deep material buoyancy. Amit and Olson [12] found that formation of super mantle plumes in the lower mantle can cause geomagnetic reversals. Other terrestrial planets such as Venus and Mars similarly possess or may possess mantle plumes [23–25] with formation mechanisms similar to Earth's.

1.4 Convection Patterns and Characteristics

Mantle convection is strongly influenced by boundary conditions in the overlying lithosphere. Based on kinematic boundary conditions at the mantle top, mantle convection patterns are divided into two main categories (Figure 1 [FIGURE:1]): plate tectonic mode (or active plate mode) and stagnant lid mode.

Observations indicate that among solar system bodies with current or possible mantle convection, the stagnant lid mode is more common, while plate tectonic mode is rare [26]. Stern et al. [27] analyzed terrestrial planets and large satellites within the solar system, defining a tectonic activity index (TAI) that scores various bodies from 0 to 3 based on evidence of recent volcanic activity, deformation, and surface replacement (inferred from crater density). Nine terrestrial bodies have TAI > 2, indicating active tectonics and convection; 17 terrestrial bodies have TAI < 2, inferred as tectonically dead; Earth's TAI is 3, making it the only terrestrial planet in the solar system with plate tectonic convection mode.

Note: a) Plate tectonic mode, where the lithosphere consists of multiple relatively soft plates that can subduct into the mantle and melt in hot magma; b) Stagnant lid mode, where the lithosphere is relatively rigid.

Figure 1 Two patterns of mantle convection

1.4.1 Stagnant Lid Mode

The stagnant lid mode is characterized by a massive, rigid, and essentially immobile lid (mostly lithosphere) at the planetary surface, with convection occurring beneath this lid. This pattern is also called the single-plate mode, with corresponding planets termed single-plate planets. Typical representatives include present-day Mercury, Venus, and Mars [29]. Under stagnant lid mode, surface geological tectonic movements are weak, mantle convection intensity is relatively low, but multiple mantle plumes can form within the mantle, sustaining surface volcanism and plateaus. Due to low efficiency of heat transfer from the mantle bottom [30], temperatures remain high, which limits convection in the electrically conductive fluid layer of the planetary outer core, preventing formation of intrinsic magnetic fields or hindering their maintenance. This provides a reasonable explanation for Venus's lack of an intrinsic magnetic field. However, stagnant lid mode has exceptions: Driscoll and Bercovici [31] suggest that if melting in the upper mantle can provide substantial heat sinks, intrinsic magnetic fields may still form in stagnant lid planets.

1.4.2 Plate Tectonic Mode

Plate tectonic mode refers to multiple active, deformable fragmented lids at the planetary surface, where the lithosphere consists of plates floating on a relatively plastic asthenosphere. These plates are softer than stagnant lids. Continuous dragging by mantle convection causes these plates to subduct into the mantle, melting and disappearing in hot magma. At mid-ocean ridges, upwelling asthenospheric material forms new plates. Earth's typical representation [32–34]. Unlike stagnant lid mode, plate tectonic mode typically occurs on geologically active planets with extensive material and energy exchange between surface and interior. This mode efficiently releases heat from the mantle and core [30], maintaining large temperature differences within and outside the conductive fluid layer of the planetary outer core, thereby facilitating formation and operation of planetary dynamos.

1.4.3 Factors Influencing Convection Patterns

Many factors influence these two convection patterns [35], including planetary mass and volume [26, 36], internal structure [37], material composition (such as phase and radioactive isotope abundance) [38], physical parameters (such as density, rheology, viscosity, initial temperature, plate hardness) [7, 28, 36, 39–41], and even the presence of liquid water. For example, if liquid water (or hydrous minerals) exists in the planetary crust, it can reduce frictional resistance during plate subduction, lowering yield stress on the top boundary layer and facilitating plate movement [42–44], thereby affecting mantle convection patterns. Noack et al. [45] found that planetary core size also influences mantle convection patterns because a large iron core occupying a significant volume fraction creates large pressure gradients in the silicate mantle, resulting in higher melting temperatures in shallow mantle regions. Additionally, these two convection patterns are not immutable and may transform during planetary evolution. Stevenson [35] and Solomatov and Moresi [46] speculate that Venus underwent a transition from plate tectonic to stagnant lid mode during its evolution; O'Neill et al. [37] suggest Io and Europa are currently transitioning from plate tectonic to stagnant lid mode.

This paper employs a comparative planetology approach, with Chapters 2–6 presenting research findings on mantle convection for Mercury, Venus, Mars, Earth, and large satellites (Moon, Galilean satellites, Titan), focusing on key scientific questions: (1) Does mantle convection exist within these bodies, and what evidence supports its presence or absence? (2) If mantle convection exists, what are its patterns and structures? Chapter 7 then discusses conditions for mantle convection in terrestrial planets and large satellites and major influencing factors, concluding with a summary in Chapter 8.

2 Mercury

As the planet closest to the Sun, Mercury is difficult to observe and challenging for probes to reach. Compared to other terrestrial planets, our understanding of Mercury remains limited. Beyond early ground-based radar and telescope observations, only Mariner 10 and MESSENGER have successfully reached Mercury. Close-range observations from these missions reveal that although large-scale volcanic activity has ceased, the existence of an intrinsic magnetic field and surface volatile activity [47] indicate that Mercury's interior remains active, far from being a "dead" planet.

The processes of Mercury's early evolution and internal structure formation remain unclear [48–51], so research on Mercury's mantle convection generally focuses on the period after gravitational differentiation when temperature began decreasing and the "crust-mantle-core" structure initially formed, primarily addressing whether Mercury currently possesses mantle convection, convection mechanisms, and maintenance duration.

If Mercury currently has mantle convection, it must be closely related to Mercury's unique characteristics. Mercury's moment of inertia can be calculated from its rotation parameters and gravity data returned by MESSENGER, revealing a five-layer internal structure: solid inner core, liquid outer core, FeS layer, mantle, and crust [52–55], with Fe as the main component of the inner and outer cores. Mercury's crust thickness is 26 ± 11 km, with total crust-mantle thickness of about 400 km [54, 56, 57], and its crust production rate ranks second only to the Moon among solar system terrestrial planets and large satellites. Compared to Earth, Mercury has a large, relatively thick crust proportion but a very thin mantle [52]: the Fe core radius ratio occupies 82%, mass ratio 73.9%, with FeS layer thickness less than 90 km [55]. This structure can be described as a massive iron sphere wrapped in a thin silicate layer.

Additionally, research suggests Mercury's surface has a regolith-like structure similar to the Moon, called megaregolith [59, 60]. Xie et al. [57] found through simulation studies that this megaregolith prolonged partial melting processes in Mercury's interior during early evolution, affecting Mercury's thermal evolution. The insulating effect of this megaregolith promoted crustal thickening and increased internal temperatures.

Mercury is the smallest terrestrial planet in the solar system by volume and mass. During formation and early evolution, heat loss was rapid, resulting in lower mantle temperatures than other terrestrial planets [61–65]; however, as the crust gradually thickened, heat loss became affected, making mantle temperature reduction a long, slow process [47].

Although Mercury's mantle is very thin, which is unfavorable for mantle convection, theoretical studies and observational results have not completely ruled out the possibility of present-day mantle convection. Numerical simulation studies by Michel et al. [61] indicate that mantle convection may still exist even with Mercury's mantle thickness of only about 300–400 km. Tosi et al. [66, 67] found through simulations that if mantle convection can occur in Mercury, its convection structure may be small-scale, with spatial scales comparable to mantle thickness. Observations reveal widespread volcanic activity features on Mercury's surface, indicating extensive and long-duration volcanic modification, primarily within the first 1 Ga after solar system formation. Subsequently, the metallic core gradually cooled, and large-scale volcanic eruptions ceased around 3.5 Ga ago; Mercury's crust then entered a global contraction state, hindering magma ascent [68–70]. Although we cannot yet definitively determine whether mantle convection motion stopped as Mercury gradually cooled, Mariner 10 detected a global weak magnetic field during 1974–1975, with average surface strength of about 450 nT [47, 71]. The origin of this magnetic field remains uncertain, with two main hypotheses [72]: (1) Due to Mercury's relatively large, Fe-rich core, the magnetic field may originate from a planetary dynamo in the liquid outer core [71, 73–75]; (2) Because Mercury's magnetic field data are incomplete, the observed field may represent remanent magnetization after Mercury's formation [76, 77]. These two possible origins have different implications for Mercury's mantle convection: (1) If Mercury's magnetic field originates solely from remanent magnetization, it indicates very weak current internal activity, most likely without mantle convection; (2) Unlike Earth, Mercury's mantle may not contain large amounts of radioactive isotopes [72, 78], making radioactive decay unlikely to provide sufficient heat for thermal convection. If the dynamo theory for Mercury's magnetic field is confirmed, efficient heat transport from Mercury's core would be required to maintain the liquid core dynamo operation, providing heat sources and driving forces for long-term mantle convection—strong evidence for present-day mantle convection in Mercury.

Due to limited data from Mariner 10 and MESSENGER, researchers cannot yet make definitive judgments about whether mantle convection exists in Mercury and its intensity [61, 62]. Furthermore, whether the FeS layer between Mercury's core and mantle affects mantle convection and dynamo processes remains under investigation [52, 53, 79]. To comprehensively understand Mercury, the next-generation Mercury probe BepiColombo was launched in October 2018 and is expected to arrive in 2025, hopefully bringing more information for Mercury research.

3 Venus

Venus's mass, volume, and density are 0.815M⊕, 0.866V⊕, and 5.2 g/cm³ respectively (M⊕ and V⊕ being Earth's mass and volume), with internal crust-mantle-core structure, making it highly similar to Earth and known as Earth's "sister planet."

Venus's surface is very young with uniform age distribution, averaging about 0.7 Ga [35]. Comparing crater counts with other terrestrial planets, some studies calculate Venus's surface age at about 0.5 Ga [80–82]. Research suggests Venus's young surface results from global catastrophic surface replacement events [83–86] occurring about 0.3–0.6 Ga [87]. Phillips et al. [80] calculated surface replacement rates of about 1 km²/a using their model, during which Venus lost substantial internal heat [88]. Wei et al. [89, 90] used gravity and topography data from the Magellan probe to invert for Venus's crust thickness using Venus mantle convection models, finding that Venus's gravity field and topography are strongly influenced by Venus's mantle dynamic processes, indicating mantle convection within Venus. Based on these facts, research and observational data confirm that Venus's mantle convection primarily manifests as mantle plumes, with about 10 mantle plumes whose depths reach scales of 10⁶ m [23]. Venus's surface is 65% volcanic plains, mainly basaltic [91], with very active magmatic activity and young volcanism [92], indicating considerable mantle convection intensity. Additionally, volcanic corona structures on Venus's surface result from mantle upwelling and diapirism [93, 94], further confirming mantle activity.

Unlike Earth, various features on Venus's lithosphere indicate no large-scale plate tectonic movement currently [95, 96], primarily because [97]: surface temperature is too high [98, 99], Venus lacks an asthenosphere [86, 100–102], and liquid water is absent [103–105]. However, Solomatov and Moresi [46] found that plate tectonic mode was the dominant convection pattern during Venus's early evolution; plate movement ceased around 0.5 Ga, with the lithosphere gradually thickening through thermal conduction, possibly reaching about 200–400 km thickness today [106]. Discoveries by Magellan and Venus Express indicate Venus is currently a single-plate planet without global plate movement, with stagnant lid mantle convection mode [27, 28, 35, 46, 87, 107–110]. This single-plate lithospheric structure has generated two viewpoints on Venus's mantle convection [35]: (1) A catastrophic event at 0.5–0.7 Ga caused global surface replacement, after which mantle convection evolved to the current stagnant lid mode [88, 110]; (2) Venus's mantle convection underwent a transition from active plate mode to stagnant lid mode (Figure 2 [FIGURE:2]), meaning mantle convection can alternate between plate tectonic and stagnant lid modes [37, 111]. Possible causes for this mode transition include decreased mantle convection velocity or loss of liquid water on Venus's surface, preventing stresses from breaking plates [85–87]. Solomatov and Moresi [46] hold similar views, suggesting Venus's plate movement cessation may stem from lithospheric brittle rheology: as Venus evolved, plate yield stress increased while maximum stress decreased, terminating plate movement when they became comparable. Strom et al. [112] believe Venus experienced one or more surface replacement events, a view widely accepted [1, 82, 87].

Although Earth resembles Venus in many aspects, its mantle convection follows plate tectonic mode, distinct from Venus and other terrestrial planets, demonstrating its uniqueness. Research on Venus's mantle convection inspires attention to Earth's special characteristics—could Earth's mantle convection undergo mode transitions like Venus? This question has attracted scholarly attention [105, 113] (discussed in Chapter 5), as transitions in Earth's mantle convection mode would represent major events in Earth's evolution and greatly affect human survival.

4 Mars

Mars is significantly smaller than Earth, with radius about half of Earth's, substantially affecting its evolution [114]. Mars is currently considered a single-plate planet; however, Mariner 9 observations in 1971 revealed global hemispheric dichotomy in its topography [115, 116], showing clear differences between northern and southern hemispheres. These differences manifest in three aspects [117]: (1) Southern hemisphere elevation is generally above the datum, while northern hemisphere is below; (2) Southern hemisphere has dense craters while northern hemisphere is sparse, though most northern plains are later-formed magmatic plains likely covering large impact craters; (3) Martian crustal thickness gradually thins from south to north, as shown in Figure 3 [FIGURE:3], based on a crustal thickness model by Neumann et al. [118] using Mars Global Surveyor data. Southern highland crust averages about 58 km thick, while northern plain crust averages about 32 km thick, showing significant differences.

Two mainstream hypotheses explain Martian topographic dichotomy: impact hypothesis (external process) [119–121] and large-scale Martian mantle convection hypothesis (internal process) [25]. In the mantle convection hypothesis, some studies [122, 123] suggest that if early Mars had a thin asthenosphere, degree-1 mantle convection (spherical harmonic function) could cause hemispheric dichotomy, with upwelling and downwelling in opposite hemispheres. However, due to limited comprehensive geophysical observation data and unclear Mars evolution history (such as impact events), the origin of Martian dichotomy requires more observational data for confirmation.

Based on long-term observations, we can roughly reconstruct Mars's thermal evolution history [124]: During the late Noachian period (4.1–3.7 Ga ago), Mars may have experienced plate tectonic movement; entering the Hesperian period (3.7–3.1 Ga ago), Martian mantle thermal activity was very active; after this active period, average mantle temperature decreased, viscosity increased, mantle convection slowed or stopped, reducing heat release efficiency and halting plate movement, forming a global lithospheric stagnant lid.

Different views exist regarding current mantle convection on Mars. Some scholars argue that due to Mars's small volume and rapid heat loss, large-scale mantle convection may not exist, making Mars more like a "dead" planet today. For example, although suspected crustal blocks were identified from Martian surface magnetic maps, no plate tectonic activity has been observed [117]. Mars had very active large-scale magmatic activity historically, such as Olympus Mons (the largest volcano in the solar system) and northern magmatic plains [125], but this cannot prove present mantle convection. Additionally, geochemical studies [117] found no large-scale mantle convection or crustal material recycling after Mars's formation; even if it occurred, it only lasted a very short time. Zhang and Shi [126] used a parameterized model to suggest Mars had early mantle convection that likely stopped before 1.6 Ga ago.

However, most scholars affirm that mantle convection continues to the present on Mars. Hauber et al. [127] analyzed new data from probes, such as crater size-frequency distributions, concluding that Mars not only had active volcanic activity historically but still has sufficient internal heat today to produce widespread plain-style volcanism. Kiefer [128] conducted numerical simulations of Martian mantle convection and magma formation, providing Rayleigh numbers for mantle convection with minimum values around 2 × 10⁶, suggesting currently active mantle convection on Mars. Phillips et al. [129] found through observational data that the lithospheric elastic thickness at Mars's north pole is much larger than previous estimates, reaching 300 km, with sediment ages less than 5 Ma; Kiefer and Li [38] explained this phenomenon using convection models, attributing lateral lithospheric thickness variations to mantle convection with stagnant lid pattern. Other scholars [30, 34, 35, 130–132] also believe Mars currently has mantle convection in stagnant lid mode, having undergone a transition from active plate mode similar to Venus. Solomon et al. [133] suggest this mode transition may be one reason Mars lost its global intrinsic magnetic field. Plesa et al. [134] used InSight lander heat flow data to model and correct present-day Martian average surface heat flow values between 0.023–0.027 W·m⁻² (Earth's average surface heat flow is about 0.087 W·m⁻² [135, 136]), reflecting that plate tectonic mode dissipates heat faster than stagnant lid mode.

Mars currently has neither plate movement nor active volcanism [137]. Data from Mars Global Surveyor show Mars lacks a global intrinsic magnetic field today, with its intrinsic magnetic field essentially disappearing about 4 Ga ago; observed Martian magnetic fields are mainly lithospheric remanent magnetism and magnetospheric fields [117, 132, 138–140]. These evidences indirectly indicate that even if mantle convection exists in Mars's interior today, its intensity is low.

5 Earth

Among all terrestrial planets, Earth is the most extensively studied object, with the deepest understanding of its mantle convection. Research shows Earth is the only terrestrial planet in the solar system with global plate tectonic movement [27, 105]. Ernst [141] noted that this plate tectonic convection mode began as early as about 4.4 Ga ago during the Hadean eon, occupying most of Earth's evolution and continuing to the present. Nakagawa and Tackley [113] studied how tectonic modes affect modeled thermochemical evolution of mantle convection, concluding that plate tectonic convection mode formed early in Earth's evolution, better matching Earth's mantle structure and core evolution. Plate movement, as an efficient heat dissipation mechanism, is an important factor enhancing mantle convection. Earth is located in the stellar habitable zone, with suitable temperatures maintaining large amounts of liquid water on the surface, and crustal carbon participating in the carbon cycle—all favorable factors for plate tectonic formation and persistence [33, 37, 108, 142–144]. Additionally, Stern [105] suggests Earth underwent multiple transitions between stagnant lid and plate tectonic modes during certain historical periods, similar to Venus's mode transitions.

In the first half of the 20th century, the mantle convection hypothesis was proposed as a driving mechanism for plate movement. However, causal relationships between mantle convection and plate movement remain debated: whether plates drive mantle convection (active plate hypothesis) [145, 146] or mantle convection drives and maintains plate movement (passive plate hypothesis) [15, 22, 147], or whether they are coupled and mutually influential [148–151]. Different viewpoints and significant disagreements exist. Both active and passive plate hypotheses have limitations: active plate hypothesis can explain plate formation at subduction zones but cannot explain plate formation elsewhere, while passive plate hypothesis encounters difficulties due to the aspect ratio paradox [9, 152]. The aspect ratio paradox refers to opposite flow directions in adjacent convection cells creating opposite drag forces on overlying lithospheric plates, causing forces to cancel out [153–155]. Therefore, we cannot simply conclude that plate movement drives mantle convection or vice versa; even if mantle convection initiates plate movement, moving plates must反过来 constrain and influence mantle convection, meaning both have their own motion characteristics while interacting and influencing each other. To date, the dynamic mechanism of plate tectonics remains controversial [156].

Earth's magnetic field provides strong evidence for mantle convection. Strong mantle convection and rapid heat dissipation facilitate geomagnetic field formation. Research suggests geomagnetic reversals and secular variation are closely related to mantle heat flow distribution: mantle convection changes core-mantle boundary heat flow, affecting liquid core magnetohydrodynamic motion and thus altering the magnetic field [10, 157].

Seismic wave velocity and density studies reveal a mantle transition zone at 410–660 km depth, serving as a boundary between upper and lower mantle [158]. Regarding whether convection can penetrate this transition zone, two models exist: whole-mantle convection, where convection occurs throughout the mantle with material crossing the 660 km interface, where phase transitions at 410 km and 660 km depths have minor effects on temperature gradient and mantle convection; and layered mantle convection, where convection occurs separately in upper and lower mantle with heat exchange but no material exchange at the 660 km interface, causing radial velocity to be zero at this boundary [9, 159] (Figure 4 [FIGURE:4]).

a) Whole-mantle convection; b) Layered mantle convection. Left column shows viscosity (cid:22) distribution with depth; middle column shows mantle convection pattern schematic; right column shows density (cid:26), temperature T, and solidus temperature (cid:18) distribution with depth.

Figure 4 Earth's whole-mantle and layered mantle convection [160]

Whether Earth's mantle convection pattern is whole-mantle, layered, or a mixed pattern (where some regions have layered convection while others have whole-mantle convection) remains controversial [161]. Geochemical observations favor layered mantle convection [162], while geophysical and geodynamic observations support whole-mantle convection [163]. Mixed models have also been proposed, such as the blob model [164, 165] and lava lamp model [166, 167].

The blob model suggests that some large, high-viscosity, high-density blobs exist in the lower mantle that do not rise in whole-mantle convection and do not chemically mix with depleted mantle, but mantle plumes can bring them to the surface [161]. This research uses seismic observations and analysis to support a mixed mantle convection pattern, where convection between upper and lower mantle is blocked in some regions, causing regional layered convection that can transform between layered and whole-mantle convection over time [168]. The lava lamp model allows subducted plates to penetrate the 660 km transition zone to satisfy geophysical observations and calculations while preserving two chemically distinct mantle source regions (depleted and undepleted mantle) to explain geochemical characteristics, thus lowering the layered convection boundary from 660 km depth to below 1,000 km within the dynamic range between core and mantle [161].

Current models have limitations and cannot explain all observed data; each model can only explain some geological and geophysical phenomena and fit partial observational data [9].

Mantle plumes are secondary forms of mantle convection, with hotspots being surface eruption points of mantle plumes. Courtillot et al. [169] categorized Earth's hotspots into three types based on their origin: primary hotspots, secondary hotspots, and tertiary hotspots, originating from the core-mantle boundary, upper-lower mantle boundary, and upper mantle respectively. Hotspots from different origins may reflect a mixed convection pattern of whole-mantle and layered convection.

6 Large Satellites

In the solar system, some larger satellites and dwarf planets have internal structures similar to terrestrial planets and may also possess mantle convection. For small-mass and small-volume satellites, internal heat dissipates quickly during formation, making it difficult to form layered structures and generate mantle convection.

6.1 Moon

By mass or volume, the Moon is the fifth largest satellite in the solar system, with internal layered structure [170], formed before 4.1 Ga [171]. During historical periods before about 3.0 Ga, the Moon may have had lunar mantle convection [126, 172] in stagnant lid mode. Stegman et al. [173] suggested that lunar mantle convection drove early lunar liquid core dynamo. Zhi and Shi [174] simulated effects of latitude-dependent lunar surface temperature on early lunar mantle convection, finding this caused thicker lithosphere at lunar poles than at the equator. During lunar thermal evolution, due to the Moon's relatively small volume and early active volcanism [175, 176], internal heat dissipated quickly. Current research indicates the Moon is a "dead" terrestrial body—surface volcanism ceased 1.5 Ga ago [1]. Geochemical studies show lunar mantle is depleted in radioactive isotopes, with heat-producing elements mainly distributed near the surface. Tao et al. [177] simulated lunar thermal evolution using a parameterized model, finding that as the lithosphere gradually thickened, heat transfer transitioned from thermal convection to thermal conduction. Therefore, comprehensive research and observational results indicate that lunar mantle convection no longer exists in the Moon's interior today.

6.2 Io

Io is one of Jupiter's four Galilean satellites, with density and radius similar to the Moon, internally differentiated into a metallic core and silicate mantle. Its surface features extensive, active volcanism. These extremely high-temperature volcanic activities indicate a partially molten asthenosphere within Io [178]. Due to Io's proximity to Jupiter, its internal heat is mainly generated by tidal heating from Jupiter, resulting in surface heat flow of 2.4–4.8 W·m⁻² [179], several orders of magnitude greater than heat from radioactive isotope decay, sufficient to maintain internal melting and making Io one of the most volcanically active bodies in the solar system. Io's active volcanism indicates it is undergoing surface replacement events similar to Venus [180]. Most heat required for this surface replacement can be transported by convection: for silicate magma, convection transports 50% of surface replacement heat at 3 mm/a rate; for sulfate magma, the rate is 35 mm/a [181]. Active surface geological activity indicates very active mantle convection in its silicate mantle, driven by tidal dissipation [182], with stagnant lid convection mode [27]. Additionally, due to possible asthenosphere existence, small-scale layered convection may form internally, manifesting as dense short-wavelength heat flow distribution on the surface [183].

6.3 Icy Satellites

The other three Galilean satellites (Europa, Ganymede, Callisto) differ significantly from Io in appearance, belonging to icy satellites. Their surfaces are covered by ice layers mainly composed of water ice (phase Ih), within which ice layer convection similar to mantle convection may occur. Barr and Pappalardo [184] studied effects of ice grain size on convection, finding that overly large ice grain sizes are unfavorable for convection. Ganymede and Callisto's surface ice temperatures are about 100 K, far below Earth's surface temperature, making their surfaces extremely hard. Therefore, stagnant lid mode is considered the primary convection pattern for these ice layers, and these icy satellites do not experience plate tectonic activity [27]. When ice layers become thick enough to generate convection, convection can transport internal heat outward, causing rapid satellite cooling and thicker ice layers.

Current探测results suggest that liquid oceans likely exist beneath ice layers of Europa, Ganymede, and Callisto [185–187], forming at ice layer bottoms: as ice layer depth increases, temperature gradually rises to melting point, forming liquid oceans. Beneath these oceans, one or more layers of water ice in different crystalline states may exist, such as ice VI and ice VII, possibly containing silicate material; beneath these warmer high-pressure ice layers are rocky mantles and metallic cores. Titan, the second-largest satellite in the solar system, is similar to these three Galilean satellites, also an icy satellite with internal structure sequentially comprising ice Ih layer, liquid ocean, high-pressure ice layer, and water-silicate core, possibly with a liquid metal inner core and silicate outer core [188, 189].

Due to structural and compositional similarities among icy satellites, their internal processes are also similar. This paper uses Titan as an example to illustrate internal convection processes. Titan's possible heat sources include radioactive decay in the core, tidal heating, cooling of different shells, and solidification heat release from the liquid ocean layer [188]. Cassini-Huygens probe data show Titan's outermost ice Ih layer is extremely hard, with thickness exceeding 40 km [190]. Although relatively thin, numerical simulation studies by Mitri and Showman [191] indicate that ice Ih layer may undergo thermal convection in stagnant lid mode. In their icy satellite model, a liquid ocean exists at the ice layer bottom. According to their Rayleigh number definition, it is proportional to D³/(cid:17)b (where D is ice layer thickness, (cid:17)b is viscosity at ice layer bottom). Ammonia in the liquid ocean can act as antifreeze, significantly lowering ice melting temperature, affecting changes in (cid:17)b. Therefore, for ice layers above ammonia-containing oceans, if convection can occur, ice layer thickness should be between two critical values: thin ice layers with low viscosity, warm bottoms and thick ice layers with high viscosity, cold bottoms. Only intermediate-thickness ice layers can have Rayleigh numbers exceeding critical Rayleigh number Rac. Experimental results show that only thermal conduction occurs in thin and thick ice layers, while thermal convection occurs in intermediate-thickness ice layers, with specific thickness depending on ammonia concentration in the liquid ocean. During 2.0–2.5 Ga after Titan's formation, ice Ih layer thickness decreased slightly because silicate core convection heated the ice layer during the same period [188, 189].

Due to limited observational data, current research on convection in the "ice-liquid-ice-rock(metal)" structure of icy satellites is scarce, and coupling effects of liquid layers between upper and lower ice layers remain unclear. However, with increasing attention to possible life in subsurface oceans of Europa and Enceladus [192, 193], related topics will gradually become research hotspots.

7.1 Conditions for Convection Occurrence

Heat transfer mechanisms from interiors of terrestrial planets and large satellites to their surfaces mainly include thermal conduction, thermal radiation, and thermal convection. Using Earth as an example, because lithospheric materials are relatively rigid, heat exchange in the lithosphere occurs primarily through thermal conduction; while in the mantle with higher temperatures and relatively soft materials, heat can be efficiently exchanged through thermal convection. If a planet's or large satellite's core (or partial core) is liquid metal, convection is also the main form of heat exchange, with convection speeds several orders of magnitude higher than mantle convection (e.g., Earth), though this is not the focus of this paper. The physical essence of thermal convection is density differences caused by medium thermal expansion leading to medium flow. Unlike thermal conduction, this heat transfer method causes material migration. In a fluid heated from below in a flat plate, if the lower layer temperature is higher than the upper layer, the fluid becomes unstable: lower material with lower density moves upward while upper material with higher density moves downward—this is the classic Rayleigh-Bénard convection, a natural convection caused by the fluid's own temperature field non-uniformity. Convective motion is constrained by fluid dynamics laws (conservation of mass, momentum, and energy). Scholars developed mantle convection theory starting from this simple Rayleigh-Bénard convection model.

Mantle convection research references classic Rayleigh-Bénard convection, treating mantle convection as a quasi-static process of incompressible fluid with Prandtl number Pr (= (cid:17)/(cid:20)) approximated as infinite. Under the Boussinesq approximation (where density varies with temperature only in buoyancy calculations but remains constant otherwise), the mass, momentum, and energy conservation equations governing mantle convection are [1, 194–196]:

(cid:17) ∇ · u = 0 ; ∇u + ∇Tu + (cid:26)ger = 0 ; + u · ∇T = k∇²T + (cid:26)H ; -∇p + ∇ · (cid:26)Cp

where u is velocity vector, p is pressure, (cid:17) is viscosity, (cid:26) is density, g is gravitational acceleration, er is radial unit vector, Cp is specific heat at constant pressure, T is temperature, k is thermal conductivity, and H is internal heat generation rate. Density affected by temperature is given by:

(cid:26) = (cid:26)₀ [1 - (cid:11) (T - T₀)]

where (cid:11) is thermal expansion coefficient, (cid:26)₀ is reference density, and T₀ is upper mantle boundary temperature. Thermal diffusivity (cid:20) relates to thermal conductivity k as (cid:20) = k/((cid:26)Cp). Normalized equations for each physical quantity are:

  (cid:26) = (cid:26)₀(cid:26)′ Cp = Cp₀C′ (cid:20)₀Cp₀ΔT D² H′ (cid:11) = (cid:11)₀(cid:11)′ g = g₀g′ (cid:20)₀ (cid:17)₀(cid:20)₀ D² p′ (cid:20) = (cid:20)₀(cid:20)′ xi = Dx′ (cid:20)₀ (cid:17) = (cid:17)₀(cid:17)′

where (cid:11)₀ is reference thermal expansion coefficient, (cid:20)₀ is reference thermal diffusivity, Cp₀ is reference specific heat, g₀ is reference gravitational acceleration, D is planetary mantle thickness, ΔT is temperature difference from mantle bottom to top (mantle bottom temperature is typically core-mantle boundary temperature), (cid:17)₀ is reference viscosity, and all primed symbols are dimensionless quantities. Substituting these normalized equations into equations (1)–(3) and removing primes yields dimensionless equations:

∇ · u = 0 ; -∇p + ∇ · (cid:17) ∇u + ∇Tu + Ra · T · er = 0 ; + u · ∇T = ∇²T + H ; (cid:26)₀g₀(cid:11)₀ · ΔT · D³ (cid:17)₀(cid:20)₀

where Ra is the Rayleigh number, expressed as:

Ra = (cid:26)₀g₀(cid:11)₀ΔTD³/(cid:17)₀(cid:20)₀

Equation (9) defines Rayleigh number based on temperature difference ΔT between upper and lower plates. Viscosity coefficient (cid:17) is an important physical parameter affected by temperature, pressure, and creep mechanisms. According to thermal convection theory, convection occurs when upward buoyancy exceeds resisting viscous forces. Quantitatively, the basic condition for thermal convection is that the Rayleigh number must exceed a critical value Rac; larger Ra means stronger convection and enhanced material exchange. Rayleigh number appears only in the body force term of the momentum equation, reflecting the degree of thermal instability and convection strength [159, 197]. The example formula above does not involve phase transitions; formulas become more complex when considering phase change factors.

Gurnis and Davies [198] noted that as Rayleigh number increases, the Péclet number Pe (= Re · Pr), representing the ratio of convective to conductive heat transfer (Re is Reynolds number), also increases. Numerical calculations also show that larger Rayleigh numbers produce narrower mantle plumes with faster ascent speeds [199]. Earth's mantle has large Rayleigh numbers; for example, in Jarvis and Peltier's [200] model, upper mantle Rayleigh number is 10⁶, whole mantle is 10⁷. When fluid Rayleigh numbers are very large, linear theory becomes inapplicable because nonlinear terms in the energy equation dominate. Boundary layer theory and numerical simulation methods can be used to analyze convection system behavior [195, 201].

From a convection pattern perspective, mantle convection in Earth and other terrestrial planets differs significantly from classic Rayleigh-Bénard convection. Rayleigh-Bénard convection describes convection between two flat plates, while mantle convection is bounded by spherical surfaces, which is important for studying large-scale whole-mantle convection problems [195, 202] because actual spherical shell geometry limits the number and geometric distribution of convection cells and mantle plumes.

7.2 Factors Influencing Convection

Mantle convection varies among terrestrial planets and large satellites because it is influenced by multiple factors, such as mantle geometric scale, material composition, physical parameters (viscosity, density, thermal diffusivity, thermal expansion coefficient), convection driving sources, kinematic boundary conditions (free or rigid boundary conditions), and thermal boundary conditions (heat flow, temperature). These factors are briefly discussed below.

First, geometric scale: mantle fluid Rayleigh number determines whether mantle convection can occur and its intensity. Typically, larger planetary mantle volumes have greater thickness, making convection easier to occur than in smaller planets (with thinner mantles) according to Rayleigh number expression (equation (9)). Conversely, small-mass terrestrial bodies have larger surface-area-to-volume ratios, losing heat faster, and are often considered unlikely to have mantle convection after substantial internal heat loss. Due to lower stress levels in small-mass bodies, thicker planetary crusts form more easily, limiting convection formation [35]. Most small-mass terrestrial bodies dissipate heat mainly through thermal conduction or melt migration.

Second, physical parameters: mantle rock physical properties such as density, thermal expansion coefficient, thermal diffusivity, and viscosity affect convection patterns and intensity. These effects are reflected in the dimensionless Rayleigh number, where they combine with mantle temperature differences to determine convection onset and intensity. These parameters affect not only overall convection patterns but also play important roles in secondary dynamic processes. For example, in Earth's interior, mantle thermal expansion coefficient determines buoyancy magnitude and density differences between upper and lower mantle at the 660 km transition zone, a key factor determining whether subducted plates can penetrate the 660 km transition zone, thus affecting mantle convection material cycling. For outer thick ice shells (or mixed rock/metal) of large satellites, although ice layer linear convection can occur, their physical parameters differ significantly from silicate mantle, with lower density and larger thermal expansion coefficient making ice layer convection quite active. This ice convection accelerates heat loss, causing ice shell thickening and affecting ocean existence beneath ice layers [203, 204].

Third, driving sources: early mantle convection in terrestrial planets and satellites primarily occurred through compositional convection accompanying mantle differentiation, driven mainly by light material upwelling and heavy material sinking. After layered structure formation, thermal convection dominated, driven by hot material upwelling and cold material sinking. Mantle thermal convection energy mainly comes from radioactive isotope decay, with mantle heating through bottom heating (core-mantle boundary), internal heating, or mixed heating, even including external heating such as asteroid impacts [8, 205] and tidal forces from primary bodies. Earth's mantle uses mixed heating, with heat from core to mantle accounting for only 10%–20% of total mantle heat [1]. Different mantle heating methods relate to planetary thermal evolution history and radioactive isotope distribution, creating significantly different convection patterns [7]. Studies show that with the same temperature difference across mantle boundaries, mixed heating creates more chaotic mantle convection and turbulence compared to pure bottom heating [206]; increasing internal heating proportion causes more localized downwelling and more dispersed upwelling distribution [7, 35]. Additionally, radioactive isotopes in planetary interiors are consumed over time, affecting mantle convection evolution.

Fourth, viscosity: mantle viscosity is an important physical parameter affecting temperature distribution and convection patterns. From equation (9), viscosity appears in the denominator, meaning larger viscosity yields smaller Rayleigh number, hindering convection. Mantle viscosity is affected by temperature and pressure [207, 208]; variable viscosity affects mantle convection [209]. For example, Travnikov et al. [210] found through experiments and numerical simulations that temperature-dependent viscosity significantly affects critical Rayleigh number changes, influencing convection onset, representing another connection between motion and temperature in mantle convection equations. Temperature-dependent viscosity affects convection patterns, making them more likely to be stagnant lid mode [7, 211] because the coldest upper boundary layer has highest viscosity, easily forming a rigid lid. If mantle viscosity is constant, temperature changes are limited to near upper and lower boundaries with small internal temperature variations; if temperature-dependent viscosity is used, temperature changes are limited to near the upper boundary, with lower mantle temperature remaining essentially constant and average mantle temperature higher than constant-viscosity cases. Under pseudo-plastic rheology assumptions, different yield stresses also affect convection patterns. With large yield stress, convection is stagnant lid mode; as yield stress decreases, convection transitions to active plate mode (plate tectonics), with episodic overturning of the upper boundary layer. This mechanism has been used to explain Venus's young surface age. Additionally, when viscosity decreases with increasing temperature, convection intensity increases and mantle plumes appear at bottom boundaries [212].

Fifth, boundary conditions: mantle convection boundary conditions include kinematic and thermal boundary conditions. Earth is the only terrestrial planet with plate tectonics; other terrestrial planets lack this tectonic movement, with stagnant lid convection mode. Such nearly rigid boundaries can reduce internal heat loss to some extent, maintaining internal temperature. Therefore, even small-mass terrestrial planets with stable outer shells like Mercury may have weak mantle convection. Under moving (free) boundary conditions, heat loss speeds increase significantly, enhancing mantle convection (given sufficient internal energy supply) [213]. Studies show plate movement significantly affects mantle convection velocity fields, particularly shallow mantle convection [214, 215].

Regarding thermal boundary conditions, temperature differences across mantle upper and lower interfaces determine mantle fluid Rayleigh number, a key convection parameter. Additionally, if the overlying lithosphere temperature is low, its strength is high and not prone to fracture, leading to stagnant lid convection. Lower mantle thermal boundary conditions relate to heat transfer to the mantle, which together with radioactive heat generation within the mantle provides energy for convection and affects mantle temperature. Heterogeneity in lower mantle thermal boundaries also affects mantle plume generation and planetary liquid core convection. For example, Earth D" heterogeneity may cause geomagnetic reversals [157]; for Venus, Moresi and Solomatov [33] found that if Venus's lithosphere is highly brittle, elevated temperatures make lithosphere deformation easier, enabling mantle convection mode transitions between stagnant lid and plate tectonic modes during its evolutionary history.

Sixth, differentiation: differentiation also affects planetary evolution, particularly during early formation stages, mainly manifested as [35]: (1) heavy material sinking and light material floating redistributes heat-producing elements, causing enrichment of radioactive elements in the crust; (2) density differences from differentiation are more significant than thermal effects, causing layered convection; (3) if differentiation-produced density distribution (light on top, heavy on bottom) offsets temperature-induced instability (heavy on top, light on bottom), differentiation may terminate mantle convection.

Seventh, phase transitions: due to temperature and pressure changes inside planets, mantle materials undergo phase transitions, including multiple ice phase transitions in icy satellites. These phase transitions change material properties, affecting convection onset. For example, in Earth's mantle, material phase transitions create transition zones at 410 km and 660 km depths. Based on geochemical evidence, these transition zones cause layered convection: phase transitions create larger mantle material density, hindering subducted slabs [161]. Venus's mantle also has material phase transitions; related research suggests that due to phase transition effects, its convection pattern has transformed from layered to whole-mantle convection, possibly causing Venus's surface replacement events [216, 217]. Yang et al.'s [86, 218, 219] three-dimensional models show that in Venus's mantle, endothermic phase transition hindering effects cause material accumulation at phase transition boundaries, with little temporal change in material exchange between upper and lower layers, resulting in relatively stable mantle convection structure.

In summary, although many factors influence mantle convection, terrestrial planet mantle convection must satisfy three basic conditions [195, 220]: (1) actual mantle temperature gradient exceeds adiabatic self-compression temperature gradient; (2) mantle material must be soft enough to allow convection, meaning viscosity should not be too large; (3) sufficient energy must exist to maintain mantle convection, such as large amounts of radioactive isotopes or heat from planetary core crystallization. The first two conditions together mean the mantle Rayleigh number must exceed the critical value for thermal instability, while the third condition ensures convection does not stop due to dissipation.

8 Conclusion

Using comparative planetology, this paper discusses research progress on mantle convection for Mercury, Venus, Mars, Earth, and large satellites, with a brief comparative summary shown in Table 1 [TABLE:1].

Surface morphological features (including topography, geological structures, rock and mineral composition) and surface physical quantities (such as temperature, heat flow, deformation) provide important clues for understanding planetary internal activity and mantle convection status, being the most easily obtainable information from ground-based telescopes and planetary probes. Among all morphological features, volcanic landforms directly reflect internal activity during different evolutionary stages. For example, Mars's extensive magmatic activity continued until a few million years ago [127], while lunar magmatism essentially ended about 3.5 Ga ago, allowing accurate judgment that Mars's mantle convection (or activity) lasted much longer than the Moon's. Additionally, observing impact crater morphologies modified by later magmatic activity can infer the chronological order of astronomical and geological events, outlining a brief planetary evolution history. Furthermore, planetary magnetic fields, surface thermal radiation, and temperature data also provide evidence for planetary internal activity.

To determine whether mantle convection exists in a planet (or large satellite), we must first understand its internal structure, including layered structure, geometric dimensions, material composition, and temperature-pressure state. This information usually comes from data obtained by planetary approach or in-situ探测. Seismological methods are considered the most effective and precise means to understand terrestrial planet internal structure; however, deploying large seismic networks and conducting high-energy artificial earthquakes is impractical for planetary exploration. Therefore, the most commonly used method is measuring planetary gravity fields, moments of inertia, solid tides, and other parameters using orbiters or landers, combined with physical properties of rock minerals under different temperature-pressure conditions, to invert for internal structure [221]. This inversion method can obtain not only planetary density and temperature distributions but also mechanical parameters such as elastic moduli and viscosity coefficients, which can be used in mantle convection models for mutual verification.

Plate movement is considered important evidence for mantle convection. If plate movement is not observed, mantle convection can only be inferred through parameter models or numerical simulations, with volcanic activity, faults, and other surface features serving as auxiliary evidence. When planetary differentiation is largely complete and layered structures have formed, physical and mathematical models can be established to study planetary mantle convection status. Fluid material components and thermodynamic parameters used in these models can be obtained through various approaches, including comparative planetology studies, internal structure inversion, high-temperature high-pressure experiments, and first-principles calculations. In mantle convection models, bottom heat flow values and radioactive isotope distributions are relatively difficult parameters to determine because they are closely related to planetary evolution history, about which our understanding remains limited. However, despite many uncertainties in mantle convection simulations, they must satisfy constraints from gravity fields, surface heat flow, and morphological features.

Therefore, extensive and comprehensive planetary research, combined with planetary formation and evolution history and considering interactions between different layers, will be future research directions for constructing integrated planetary (and large satellite) mantle convection models. Meanwhile, studying terrestrial planets and large satellites within the solar system can also provide valuable references for exoplanet research.

References

[1] Schubert G, Turcotte D L, Olson P. Mantle Convection in the Earth and Planets. Cambridge: Cambridge University Press, 2001: 1
[2] Hopkins W. Philosophical Transactions of the Royal Society of London, 1839, 129: 381
[3] Fisher O. Physics of the Earth's Crust. London: Macmillan, 1889: 1
[4] Holmes A. Mining Magazine, 1929, 40: 205
[5] Holmes A. Transactions of the Geological Society of Glasgow, 1931, 18(3): 559
[6] Holmes A. Geologiska Föreningens I Stockholm Förhandlingar, 1945, 67(1): 115
[7] Korenaga J. Journal of Geophysical Research: Solid Earth, 2017, 122(5): 4064
[8] 黄川, 傅容珊. 地球物理学报, 2014, 57(5): 1534
[9] 朱涛. 地球物理学进展, 2003, 18(1): 65
[10] Biggin A J, Steinberger B, Aubert J, et al. Nature Geoscience, 2012, 5(8): 526
[11] Coltice N, Gerault M, Ulvrova M. Earth-Science Reviews, 2017, 165: 120
[12] Amit H, Olson P. Earth and Planetary Science Letters, 2015, 414: 68
[13] Tackley P J. Science, 2000, 288(5473): 2002
[14] Turcotte D L, Schubert G. Geodynamics. 2nd Edition. New York: Cambridge University Press, 2002: 1
[15] Hager B H, O'Connell R J. Journal of Geophysical Research: Solid Earth, 1981, 86(B6): 4843
[16] 奥菲塞 C B. 理论地球物理学引论. 王子昌, 吴庆鹏, 朱仁益, 等译. 北京: 科学出版社, 1980: 402
[17] 吴泰然, 何国琦. 普通地质学. 北京: 北京大学出版社, 2003: 227
[18] Elsasser W M. Journal of Geophysical Research, 1971, 76(5): 1101
[19] Wilson J T. Canadian Journal of Earth Sciences, 1963, 41(6): 863
[20] 加林 G D. 地球物理学引论: 地幔、地核和地壳. 陈颙, 唐晓明, 赵晓敏, 译. 北京: 地震出版社, 1987: 513
[21] 王登红. 地幔柱及其成矿作用. 北京: 地震出版社, 1998: 1
[22] Morgan W J. Am Assoc Petroleum Geologists Bull, 1972, 56(2): 203
[23] Stofan E R, Smrekar S, Bindschandler D L, et al. Journal of Geophysical Research: Planets, 1995, 100(E11): 23317
[24] Spohn T, Sohl F, Breuer D. Astronomy and Astrophysics Review, 1998, 8: 181
[25] Harder H, Christensen U R. Nature, 1996, 380(6574): 507
[26] O'Rourke J G, Korenaga J. Icarus, 2012, 221(2): 1043
[27] Stern R J, Gerya T, Tackley P J. Geoscience Frontiers, 2018, 9(1): 103
[28] Karato S I, Barbot S. Scientific Reports, 2018, 8: 11884
[29] Solomon S C. Geophysical Research Letters, 1978, 5(6): 461
[30] Reese C C, Solomatov V S, Moresi L N. Journal of Geophysical Research: Planets, 1998, 103(E6): 13643
[31] Driscoll P, Bercovici D. Physics of the Earth and Planetary Interiors, 2014, 236: 36
[32] Solomatov V S, Moresi L N. Geophysical Research Letters, 1997, 24(15): 1907
[33] Moresi L, Solomatov V. Geophysical Journal International, 1998, 133(3): 669
[34] Loddoch A, Stein C, Hansen U. Earth and Planetary Science Letters, 2006, 251(1-2): 79
[35] Stevenson D J. Comptes Rendus Geoscience, 2003, 335(1): 99
[36] Noack L, Breuer D. Planetary and Space Science, 2014, 98: 41
[37] O'Neill C, Jellinek A M, Lenardic A. Earth and Planetary Science Letters, 2007, 261(1-2): 20
[38] Kiefer W S, Li Q S. Geophysical Research Letters, 2009, 36(18): 1
[39] Thiriet M, Breuer D, Michaut C, et al. Physics of the Earth and Planetary Interiors, 2019, 286: 138
[40] Schulz F, Tosi N, Plesa A C, et al. Geophysical Journal International, 2020, 220(1): 18
[41] 李琪桦. 硕士论文. 合肥: 中国科学技术大学, 2024: 1
[42] Hirth G, Kohlstedt D L. Earth and Planetary Science Letters, 1996, 144(1): 93
[43] Korenaga J. Journal of Geophysical Research: Solid Earth, 2010, 115(B11): 1
[44] Hirauchi K, Fukushima K, Kido M, et al. Nature Communications, 2016, 7: 12550
[45] Noack L, Godolt M, von Paris P, et al. Planetary and Space Science, 2014, 98: 14
[46] Solomatov V S, Moresi L N. Journal of Geophysical Research: Planets, 1996, 101(E2): 4737
[47] Johnson C L, Hauck II S A. Journal of Geophysical Research: Planets, 2016, 121(11): 2349
[48] Malavergne V, Toplis M J, Berthet S, et al. Icarus, 2010, 206(1): 199
[49] Asphaug E, Reufer A. Nature Geoscience, 2014, 7(8): 564
[50] Malavergne V, Cordier P, Righter K, et al. Earth and Planetary Science Letters, 2014, 394: 186
[51] Knibbe J S, van Westrenen W. Earth and Planetary Science Letters, 2018, 482: 147
[52] Smith D E, Zuber M T, Phillips R J, et al. Science, 2012, 336(6078): 214
[53] Stevenson D J. Nature, 2012, 485(7396): 52
[54] Sori M M. Earth and Planetary Science Letters, 2018, 489: 92
[55] Margot J L, Hauck II S A, Mazarico E, et al. Mercury's Internal Structure. Cambridge Planetary Science. New York: Cambridge University Press, 2018: 85
[56] Dumberry M, Rivoldini A, Van Hoolst T, et al. Icarus, 2013, 225(1): 62
[57] 谢景椿, 黄乘利, 张冕. 天文学进展, 2019, 37(04): 367
[58] Wieczorek M A, Neumann G A, Nimmo F, et al. Science, 2013, 339(6120): 671
[59] Spray J G. Annual Review of Earth and Planetary Sciences, 2016, 44: 139
[60] Egea-González I, Ruiz J. Icarus, 2014, 232: 220
[61] Michel N C, Hauck II S A, Solomon S C, et al. Journal of Geophysical Research: Planets, 2013, 118(5):
[62] Breuer D, Hauck II S A, Buske M, et al. Space Science Reviews, 2007, 132(2-4): 229
[63] Ke Y, Solomatov V S. Journal of Geophysical Research: Planets, 2006, 111(E10): 12
[64] Ogawa M, Yanagisawa T. Journal of Geophysical Research: Planets, 2014, 119(4): 867
[65] Boehler R. Annual Review of Earth and Planetary Sciences, 1996, 24: 15
[66] Tosi N, Grott M, Plesa A C, et al. Journal of Geophysical Research: Planets, 2013, 118(12): 2474
[67] Breuer D, Spohn T, Van Hoolst T, et al. Surveys in Geophysics, 2022, 43(1): 177
[68] 王一尘, 肖智勇. 南京大学学报(自然科学), 2021, 57(6): 981
[69] 李泽明, 李元. 地球化学, 2022, 51(2): 133
[70] Thomas R J, Rothery D A. Elements, 2019, 15(1): 27
[71] Christensen U R. Nature, 2006, 444(7122): 1056
[72] Redmond H L, King S D. Physics of the Earth and Planetary Interiors, 2007, 164(3-4): 221
[73] Ness N F, Behannon K W, Lepping R P, et al. Journal of Geophysical Research: Space Physics, 1975, 80(19): 2708
[74] Stevenson D J, Spohn T, Schubert G. Icarus, 1983, 54(3): 466
[75] Wicht J, Mandea M, Takahashi F, et al. Space Science Reviews, 2007, 132(2-4): 261
[76] Stephenson A. Earth and Planetary Science Letters, 1976, 28(3): 454
[77] Sharpe H N, Strangway D W. Geophysical Research Letters, 1976, 3(5): 285
[78] Cameron A G W, Benz W, Fegley J B, et al. Mercury. Tucson: University of Arizona Press, 1988: 692
[79] Namur O, Charlier B, Holtz F, et al. Earth and Planetary Science Letters, 2016, 448: 102
[80] Phillips R J, Raubertas R F, Arvidson R E, et al. Journal of Geophysical Research: Planets, 1992, 97(E10):
[81] Schaber G G, Strom R G, Moore H J, et al. Journal of Geophysical Research: Planets, 1992, 97(E08):
[82] Bullock M A, Grinspoon D H, Head J W. Geophysical Research Letters, 1993, 20(19): 2147
[83] Kaula W M. Philosophical Transactions of the Royal Society of London. Series A: Physical and Engineering Sciences, 1994, 349(1690): 345
[84] Price M, Suppe J. Earth Moon and Planets, 1995, 71(1-2): 99
[85] Smrekar S E, Stofan E R, Mueller N, et al. Science, 2010, 328(5978): 605
[86] Huang J S, Yang A, Zhong S J. Earth and Planetary Science Letters, 2013, 362: 207
[87] Nimmo F, McKenzie D. Annual Review of Earth and Planetary Sciences, 1998, 26: 23
[88] Turcotte D L, Morein G, Roberts D, et al. Icarus, 1999, 139(1): 49
[89] Wei D Y, Yang A, Huang J S. Science China-Earth Sciences, 2014, 57(9): 2025
[90] 魏代云, 杨安, 黄金水. 中国科学：地球科学, 2014, 44(5): 934
[91] McKenzie D, Ford P G, Johnson C, et al. Journal of Geophysical Research: Planets, 1992, 97(E08): 13533
[92] Herrick R R. Geology, 1994, 22(8): 703
[93] Koch D M, Manga M. Geophysical Research Letters, 1996, 23(3): 225
[94] Guseva E N, Ivanov M A. Solar System Research, 2020, 54(6): 497
[95] Phillips R J, Kaula W M, McGill G E, et al. Science, 1981, 212: 879
[96] Solomon S C, Head J W, Kaula W M, et al. Science, 1991, 252: 297
[97] 陈乐, 张少兵, 余金霏. 地球与行星物理论评, 2022, 53(01): 66
[98] Bercovici D, Ricard Y. Nature, 2014, 508(7497): 513
[99] Davaille A, Smrekar S E, Tomlinson S. Nature Geoscience, 2017, 10(5): 349
[100] Kiefer W S, Hager B H. Journal of Geophysical Research: Planets, 1991, 96(E04): 20947
[101] Richards M A, Yang W S, Baumgardner J R, et al. Geochemistry Geophysics Geosystems, 2001, 2: 16
[102] Höink T, Lenardic A, Richards M. Geophysical Journal International, 2012, 191(1): 30
[103] Smrekar S E, Elkins-Tanton L, Leitner J J, et al. Exploring Venus as a terrestrial planet, 2007, 176: 45
[104] Korenaga J. Annual Review of Earth and Planetary Sciences, 2013, 41: 117
[105] Stern R J. Philosophical Transactions of the Royal Society, 2018, 376(2132): 20170406
[106] Smrekar S E, Phillips R. Earth and Planetary Science Letters, 1991, 107: 582
[107] Solomon S C, Smrekar S E, Bindschadler D L, et al. Journal of Geophysical Research: Planets, 1992, 97(E08): 13199
[108] Reese C C, Solomatov V S, Moresi L N. Icarus, 1999, 139(1): 67
[109] Weeraratne D, Manga M. Earth and Planetary Science Letters, 1998, 160(3): 563
[110] Turcotte D L. Journal of Geophysical Research: Planets, 1993, 98(E09): 17061
[111] Sleep N H. Journal of Geophysical Research: Planets, 2000, 105(E07): 17563
[112] Strom R G, Schaber G G, Dawson D D. Journal of Geophysical Research: Planets, 1994, 99(E05): 10899
[113] Nakagawa T, Tackley P J. Geochemistry Geophysics Geosystems, 2015, 16(10): 3400
[114] Zuber M T. Nature, 2001, 412(6843): 220
[115] Hartmann W K. Journal of Geophysical Research, 1973, 78(20): 4096
[116] Moore W B, Simon J I, Webb A A G. Earth and Planetary Science Letters, 2017, 474: 13
[117] 欧阳自远, 邹永廖. 火星科学概论. 上海: 上海科技教育出版社, 2015: 28
[118] Neumann G A, Zuber M T, Wieczorek M A, et al. Journal of Geophysical Research: Planets, 2004, 109(E08): 1
[119] Wilhelms D E, Squyres S W. Nature, 1984, 309(5964): 138
[120] Frey H, Schultz R A. Geophysical Research Letters, 1988, 15(3): 229
[121] Leone G, Tackley P J, Gerya T V, et al. Geophysical Research Letters, 2014, 41(24): 8736
[122] Zhong S J, Zuber M T. Earth and Planetary Science Letters, 2001, 189(1-2): 75
[123] Roberts J H, Zhong S J. Journal of Geophysical Research: Planets, 2006, 111(E06): 1
[124] 陈石, 张健, 石耀霖. 中国地球物理学会第二十三届年会论文集. 青岛: 中国海洋大学出版社, 2007: 799
[125] Sclater J G, Jaupart C, Galson D. Reviews of Geophysics, 1980, 18(1): 269
[126] 张健, 石耀霖. 地球物理学报, 1998, 41(6): 763
[127] Hauber E, Broz P, Jagert F, et al. Geophysical Research Letters, 2011, 38(10): 1
[128] Kiefer W S. Meteoritics and Planetary Science, 2003, 38(12): 1815
[129] Phillips R J, Zuber M T, Smrekar S E, et al. Science, 2008, 320(5880): 1182
[130] Connerney J E P, Acuña M H, Wasilewski P J, et al. Science, 1999, 284(5415): 794
[131] Connerney J E P, Acuña M H, Ness N F, et al. Proceedings of the National Academy of Sciences of the United States of America, 2005, 102(42): 14970
[132] Loddoch A, Hansen U. Journal of Geophysical Research: Planets, 2008, 113(E09): 1
[133] Solomon S C, Aharonson O, Aurnou J M, et al. Science, 2005, 307(5713): 1214
[134] Plesa A C, Grott M, Tosi N, et al. Journal of Geophysical Research: Planets, 2016, 121(12): 2386
[135] Stein C A. Global Earth Physics A Handbook of Physical Constants. Washington: American Geophysical Union, 1995: 144
[136] Davies J H. Geochemistry Geophysics Geosystems, 2013, 14(10): 4608
[137] Ramsey M S, Harris A J L, Crown D A. Journal of Volcanology and Geothermal Research, 2016, 311: 198
[138] Acuña M H, Connerney J E P, Wasilewski P, et al. Science, 1998, 279(5357): 1676
[139] Acuña M H, Connerney J E P, Ness N F, et al. Science, 1999, 284(5415): 790
[140] 吴季, 朱光武, 赵华, 等. 空间科学学报, 2009, 29(5): 449
[141] Ernst W G. Journal of Asian Earth Sciences, 2017, 145: 334
[142] Kaula W M. Geophysical Research Letters, 1990, 17(9): 1401
[143] Kaula W M. Science, 1990, 247(4947): 1191
[144] 刘鹏雷, 吴耀, 刘强, 等. 地质科技情报, 2014, 33(3): 25
[145] Olson P, Silver P G, Carlson R W. Nature, 1990, 344(6263): 209
[146] Garwin L. Nature, 1991, 351(6329): 699
[147] Hager B H, O'Connell R J. Journal of Geophysical Research Solid Earth, 1979, 84(B3): 1031
[148] 傅容珊, 林芬, 黄建华. 地球物理学报, 1992, 35(1): 52
[149] 叶正仁, 白武明, 滕春凯. 地球物理学报, 1993, 36(1): 10
[150] 叶正仁, 滕春凯, 张新武. 地球物理学报, 1995, 38(2): 174
[151] 叶正仁, 朱日祥. 地球物理学报, 1996, 39(1): 47
[152] 朱涛. 博士论文. 北京: 中国地震局地球物理研究所, 2005: 5
[153] Richter F M. Reviews of Geophysics, 1973, 11(2): 223
[154] 李荫亭. 力学与实践, 1985, 7(04): 8
[155] 李三忠, 郭玲莉, 戴黎明, 等. 地学前缘. 2015, 22(6): 65
[156] 钟时杰. 地球物理学报, 2021, 64(10): 3478
[157] 李力刚. 科学通报, 2016, 61(13): 1395
[158] Dziewonski A M, Anderson D L. Physics of the Earth and Planetary Interiors, 1981, 25: 297
[159] Lowrie W. Fundamentals of Geophysics. 2nd Edition. Cambridge: Cambridge University Press, 2007: 247
[160] Peltier W R. Mantle Convection: Plate Tectonics and Global Dynamics. New York: Gordon and Breach Sciences Publishers, 1989: 389
[161] 石耀霖. 地球科学进展, 2001, 16(04): 496
[162] Hofmann A W. Nature, 1997, 385(6613): 219
[163] 曾融生. 中国地球物理学会第十七届年会. 昆明: 云南科技出版社, 2001: 1
[164] Lay T, Williams Q, Garnero E J. Nature, 1998, 392(6675): 461
[165] Becker T W, Kellogg J B, O'Connell R J. Earth and Planetary Science Letters, 1999, 171(3): 351
[166] Kellogg L H, Hager B H, van der Hilst R D. Science, 1999, 283(5409): 1881
[167] Gyüre B, Jánosi I M. Physical Review E, 2009, 80(4): 046307
[168] 吴翔宇, 李红谊. 国际地震动态, 2019, (5): 1
[169] Courtillot V, Davaille A, Besse J, et al. Earth and Planetary Science Letters, 2003, 205(3-4): 295
[170] Ziethe R, Seiferlin K, Hiesinger H. Planetary and Space Science, 2009, 57(7): 784
[171] 欧阳自远. 月球科学概论. 北京: 中国宇航出版社, 2005: 3
[172] Zhong S J, Parmentier E M, Zuber M T. Earth and Planetary Science Letters, 2000, 177(3-4): 131
[173] Stegman D R, Jellinek A M, Zatman S A, et al. Nature, 2003, 421(6919): 143
[174] 支羽萧, 石耀霖. 地球物理学报, 2023, 66(02): 685
[175] Spohn T, Konrad W, Breuer D, et al. Icarus, 2001, 149(1): 54
[176] Zhang N, Parmentier E M, Liang Y. Journal of Geophysical Research: Planets, 2013, 118(9): 1789
[177] 陶莎, 石耀霖, 朱伯靖. 地球物理学报, 2023, 66(09): 3730
[178] Tackley P J. Journal of Geophysical Research: Planets, 2001, 106(E12): 32971
[179] Moore W B. Journal of Geophysical Research: Planets, 2003, 108(E08): 5096
[180] Breuer D, Moore W B. Planets and Moons, 2007, 10: 299
[181] O'Reilly T C, Davies G F. Geophysical Research Letters, 1981, 8(4): 313
[182] Tackley P J, Schubert G, Glatzmaier G A, et al. Icarus, 2001, 149(1): 79
[183] Shahnas M H, Pysklywec R N, Peltier W R. Icarus, 2013, 225(1): 15
[184] Barr A C, Pappalardo R T. Journal of Geophysical Research-Planets, 2005, 110(E12): 1
[185] Kivelson M G, Khurana K K, Russell C T, et al. Science, 2000, 289(5483): 1340
[186] Spohn T, Schubert G. Icarus, 2003, 161(2): 456
[187] Freeman J. Planetary and Space Science, 2006, 54(1): 2
[188] Grasset O, Sotin C, Deschamps F. Planetary and Space Science, 2000, 48(7-8): 617
[189] Tobie G, Grasset O, Lunine J I, et al. Icarus, 2005, 175(2): 496
[190] Hemingway D, Nimmo F, Zebker H, et al. Nature, 2013, 500(7464): 550
[191] Mitri G, Showman A P. Icarus, 2008, 193(2): 387
[192] Reh K, Lunine J I, Cable M L, et al. 2016 IEEE Aerospace Conference. New York: IEEE, 2016: 1
[193] Mann A. Proceedings of the National Academy of Sciences of the United States of America, 2017, 114(18):
[194] Chandrasekhar S. Hydrodynamic and Hydromagnetic Stability. Oxford: Oxford University Press, 1961:
[195] 傅容珊, 黄建华. 地球动力学. 北京: 高等教育出版社, 2001: 226
[196] Zhong S J, McNamara A, Tan E, et al. Geochemistry Geophysics Geosystems, 2008, 9: 32
[197] 白武明. 地球动力学. 北京: 地震出版社, 2003: 132
[198] Gurnis M, Davies G F. Geophysical Journal of the Royal Astronomical Society, 1986, 85(3): 523
[199] 王青平, 白武明, 王洪亮. 地球物理学报, 2011, 54(6): 1566
[200] Jarvis G T, Peltier W R. Mantle convection, plate tectonics and global dynamics. New York: Gordon and Breach Sciences Publishers, 1989: 479
[201] 谢鸿森. 地球深部物质科学导论. 北京: 科学出版社, 1997: 250
[202] Richter F, McKenzie D. Journal of Geophysics-Zeitschrift Fur Geophysik, 1978, 44(5): 441
[203] McKinnon W B. Geophysical Research Letters, 1999, 26(7): 951
[204] Deschamps F, Sotin C. Journal of Geophysical Research: Planets, 2001, 106(E03): 5107
[205] Glikson A, Vickers J. Earth and Planetary Science Letters, 2006, 241(1-2): 11
[206] Travis B, Olson P. Geophysical Journal International, 1994, 118(1): 1
[207] Yoshida M, Kageyama A. Journal of Geophysical Research: Solid Earth, 2006, 111(B03): 1
[208] Fowler A C, Howell P D, Khaleque T S. Geophysical and Astrophysical Fluid Dynamics, 2016, 110(2):
[209] Leitch A M, Yuen D A, Lausten C L. Journal of Geophysical Research: Planets, 1992, 97(E12): 20899
[210] Travnikov V, Zaussinger F, Beltrame P, et al. Physical Review E, 2017, 96(2): 023108
[211] Guerrero J M, Lowman J P, Deschamps F, et al. Journal of Geophysical Research: Planets, 2018, 123(7):
[212] Sülle B. Acta Geodaetica Et Geophysica, 2015, 50(4): 403
[213] Schubert G, Bercovici D, Glatzmaier G A. Journal of Geophysical Research, 1990, 95(B09): 14105
[214] Lowman J P, King S D, Trim S J. Journal of Geophysical Research: Solid Earth, 2011, 116(B12): 1
[215] 朱涛. 地学前缘, 2017, 24(5): 192
[216] Steinbach V, Yuen D A. Geophysical Research Letters, 1992, 19(22): 2243
[217] Schubert G, Solomatov V S, Tackley P J, et al. Venus II, 1997: 1245
[218] 杨安, 翁辉辉, 黄金水. 中国科学: 地球科学, 2015, 45(5): 611
[219] 杨安. 博士论文. 合肥: 中国科学技术大学, 2014: 1
[220] 傅容珊, 刘斌. 固体地球物理学基础. 合肥: 中国科学技术大学出版社, 2009: 224
[221] 龚盛夏, 黄乘利. 天文学进展, 2013, 31(04): 391