Professor Rhodri Davies

I am recognised for developing and integrating advanced computational tools for geodynamics with diverse observational datasets, enabling new constraints on the structure, dynamics and evolution of Earth's mantle and lithosphere. My work connects deep Earth processes to their surface expression across a wide range of spatial and temporal scales, from mantle plumes, subduction and lithosphere-asthenosphere interactions to dynamic topography, volcanism, mantle melting, sea-level change and the geodynamic environments that underpin mineral systems.

Research overview

My research focuses on the dynamics of Earth's planetary heat engine: the mantle. I investigate how mantle convection, lithospheric structure, plate motion, melting and chemical differentiation interact to shape the geological, geophysical and geochemical evolution of our planet. A central theme of my work is the link between deep Earth processes and their surface expression, including topography, volcanism, tectonics, sedimentary basin evolution, Antarctic ice-sheet change, glacial isostatic adjustment, sea-level change and the formation and preservation of mineral systems.

This research is motivated by both fundamental scientific questions and major societal challenges. Understanding how the solid Earth evolves through time is essential for explaining the distribution of continents, oceans, mountains, volcanoes, basins and mineral systems. It is also increasingly important for addressing problems of national and international significance, including sea-level change, Antarctic ice-sheet evolution, critical mineral discovery, resource security, environmental stewardship and the development of trusted digital research infrastructure.

I combine high-performance numerical simulation, inverse modelling and observational constraints from geophysics, geology and geochemistry. This integrated approach allows us to test physical mechanisms, reconcile models with data, and build quantitative reconstructions of Earth evolution through space and time. My research has contributed to fundamental understanding of mantle plumes, intraplate volcanism, subduction dynamics, dynamic topography, lithosphere-asthenosphere interactions, lower mantle structure, mantle melting, geochemical variability and the long-term thermo-chemical evolution of the mantle.

A growing part of my current work applies these same approaches to societally important Earth science problems. In sea-level science, my group is developing models of glacial isostatic adjustment and solid Earth deformation that help explain how ice sheets, the solid Earth, gravity and oceans interact. These processes are essential for interpreting past sea-level records, understanding Antarctic ice-sheet evolution and improving projections of future sea-level change. In the resources space, we are developing geodynamic frameworks for understanding the environments that underpin mineralisation, with a particular focus on how mantle flow, lithospheric architecture, plate evolution, melting and fluid pathways influence the formation, preservation and discovery of mineral systems.

My long-term vision is to help establish a new quantitative paradigm for solid Earth evolution. The overarching goal is to reconstruct the thermo-chemical structure and flow history of Earth's mantle in four dimensions. To achieve this, my group is developing inverse geodynamics: a framework that fuses observations from geophysics, geology and geochemistry with multi-resolution computational models and adjoint-based inversion. This approach aims to produce a unified understanding of plate tectonics, mantle flow, melting, chemical evolution, surface deformation and the thermo-chemical state of Earth's interior through time.

This vision is inherently collaborative. It requires numerical modellers, inverse theorists, observational geoscientists, geochemists, research software engineers and domain experts working together across traditional disciplinary boundaries. It also creates opportunities to connect fundamental geodynamics to national and international priorities in climate resilience, sea-level prediction, critical minerals, natural hazards, research infrastructure and evidence-based decision-making. I am always looking for motivated and enthusiastic students to join the group.

Key focus areas of my current and future research include:

1. Developing inverse geodynamics and data-informed models of Earth evolution, by combining adjoint methods, high-performance finite-element modelling and observational constraints to reconstruct mantle structure and flow through space and time.
2. Understanding how Earth's surface responds to mantle flow, by reconciling observational constraints on residual and dynamic topography with theoretical predictions from mantle convection models. This work provides fundamental insight into how deep and shallow mantle processes shape Earth's surface.
3. Isolating the mechanisms underpinning intraplate volcanism, by combining geodynamical models with seismological, geochemical and petrological constraints. This research has produced major results on Australian and Pacific volcanism, including studies published in Geology and Nature, and contributed to the award of the 2018 Anton Hales Medal from the Australian Academy of Science.
4. Developing quantitative models of mantle melting and geochemical variability, by coupling geodynamical simulations with melting models and geochemical observations from mid-ocean ridges, ocean islands and continental volcanic provinces. This work aims to understand how mantle flow, lithospheric thickness, source heterogeneity and melting processes combine to generate the diverse chemical signatures observed at Earth's surface.
5. Quantifying the force balance governing plate motions and tectonic change, using coupled models of mantle and lithosphere dynamics to understand how mantle flow, plumes, slabs and lithospheric structure influence surface tectonics and rapid changes in plate motion.
6. Reconciling geophysical and geochemical constraints on mantle structure, by testing synthetic models of mantle dynamics and evolution against seismic, geochemical and geological observations. This work builds on a long-standing program of theoretical, methodological and applied research into the nature of mantle heterogeneity.
7. Understanding and modelling solid Earth's controls on sea-level change, by developing models of glacial isostatic adjustment, solid Earth deformation and mantle viscosity that help constrain Antarctic ice-sheet history, present-day observations and future sea-level change.
8. Revealing the geodynamical environments underpinning mineralisation, by integrating mantle and lithosphere dynamics, plate reconstructions, melting processes, geochemistry and data-driven approaches to identify the deep-time conditions that influence mineral formation, preservation and resource prospectivity.

Fundamental progress on these goals relies on the continued development of inverse geodynamics, robust research software infrastructure and close integration between computational modelling and observational Earth science.

Research tools

My current research software development is centred on G-ADOPT, a computational platform for geodynamic simulation and inversion developed and maintained at the Research School of Earth Sciences. G-ADOPT is designed to make high-performance, adjoint-capable and data-informed geodynamical modelling more accessible, scalable and reproducible. It provides the foundation for my group's current work on inverse geodynamics, mantle evolution, glacial isostatic adjustment, sea-level change and the geodynamic context of mineral systems.

G-ADOPT is built on Firedrake, an automated finite-element system for solving partial differential equations. Firedrake separates the mathematical expression of a problem from many of the low-level implementation details, while using automatic code generation and sophisticated performance optimisation to deliver efficient and extensible research software. This approach allows geodynamicists to build transparent, flexible and reproducible models while taking advantage of modern high-performance computing.

I was also a long-standing developer and user of Fluidity, a computational modelling framework based on adaptive, anisotropic, unstructured meshes. My work with Fluidity helped establish the value of adaptive mesh methods for geodynamical simulation, particularly for problems involving strongly localised features such as subducting slabs, plume conduits, thermal boundary layers and plate interfaces. This earlier work provided an important foundation for my current focus on G-ADOPT, where the emphasis has shifted towards automated finite-element methods, adjoint-based inversion, reproducible workflows and scalable community research software.

Research highlights

Scott et al., Geoscientific Model Development (2026) - Automated forward and adjoint modelling of viscoelastic deformation of the solid Earth. This study extends G-ADOPT into one of the most societally important parts of solid Earth science: understanding how the planet responds when ice sheets grow and shrink. As ice melts, the solid Earth rebounds, the gravity field changes and sea level does not rise evenly everywhere. Capturing these effects is essential for interpreting past sea-level records and improving projections of future sea-level change. This paper shows that G-ADOPT can model this coupled ice-Earth-sea-level problem in a flexible and efficient way, while also generating the sensitivity information needed to work backwards from observations to infer ice histories and mantle properties. In doing so, it broadens G-ADOPT from a mantle convection platform into a wider framework for data-informed solid Earth and sea-level modelling.

This example illustrates the key advance enabled by adjoint-based modelling in G-ADOPT. Rather than testing one ice history or one Earth structure at a time, the method allows us to work backwards from observations and jointly recover both the unknown ice load and the viscosity structure of the solid Earth. This is central to improving glacial isostatic adjustment and sea-level models, because the sea-level signal depends on both the history of ice loading and the way the mantle deforms in response. The figure shows how the inversion progressively improves: after only a few iterations the broad pattern is recovered, and after further optimisation both the ice load and viscosity structure converge toward the target model.

Adjoint-based joint optimisation of unknown ice load and viscosity structure. Panel (a) shows the target viscosity field and surface ice load. Panels (b), (c) and (d) show the inversion results after 2, 10 and 100 optimisation iterations, respectively.

Pilia et al., Earth and Planetary Science Letters (2025) - Ghost plumes hidden beneath Earth's continents. Mantle plumes are usually thought of as deep, hot upwellings that reveal themselves through chains of volcanoes. This study shows that some plumes may be much harder to see. In eastern Oman, the team identified the Dani plume: a hot mantle upwelling that appears to have left little or no volcanic expression at the surface because it sits beneath thick continental lithosphere. Instead of relying on volcanism, the study combines seismic observations, mantle transition-zone structure, residual topography and plate-motion evidence to reveal the plume's presence. The broader message is important: the absence of volcanoes does not necessarily mean the absence of deep mantle upwelling. This work extends my long-standing interest in mantle plumes, lithospheric controls on melting and dynamic topography by showing how mantle flow can shape the surface even when the usual volcanic fingerprints are missing.

Sternai et al., Tectonics (2025) - Raising the roof of the world: Intra-crustal Asian mantle supports the Himalayan-Tibetan orogen. The Himalaya and Tibetan Plateau are the highest and largest mountain system on Earth, but explaining how they are supported has remained difficult. The traditional picture is that the collision between India and Asia doubled the thickness of the crust beneath Tibet. This paper argues that the real architecture may be more subtle: rather than one continuous block of thickened crust, a strong slice of Asian mantle lithosphere may sit between layers of Indian and Asian crust. In simple terms, buoyant Indian crust helps lift the system, while stronger Asian mantle helps hold it up. The study combines numerical geodynamic models with geological, geochemical and geophysical observations, including seismic evidence and mantle-derived rocks brought to the surface. For my research program, the paper is important because it shows how geodynamic modelling can challenge long-standing textbook views of mountain building, and how the deep structure of the lithosphere controls the topography we see at the surface.

Ghelichkhan et al., Geoscientific Model Development (2024) - Automatic adjoint-based inversion schemes for geodynamics: reconstructing the evolution of Earth's mantle in space and time. This paper is a major milestone in my research program. It introduces G-ADOPT as a platform for moving geodynamics beyond forward modelling - where we ask what happens if a particular model is true - toward inverse modelling, where we use observations to infer the most likely history of the Earth system. The significance is that G-ADOPT automates much of the difficult machinery needed to calculate how model predictions depend on earlier conditions. That makes it possible to ask much more ambitious questions: What did the mantle look like millions of years ago? How did mantle flow evolve? Which histories are consistent with the present-day structure of the Earth? This is the technical foundation for my long-term goal of building four-dimensional, observation-constrained models of mantle evolution, dynamic topography, melting and plate-mantle interaction.

The figure below illustrates why G-ADOPT is such an important step forward. In this synthetic experiment, the model is asked to recover the earlier evolution of a mantle convection system from later observations. The top rows show the "true" reference simulation: how temperature and viscosity evolve through time in a mantle-like annulus. The lower rows show the reconstruction produced by G-ADOPT. Even though the reconstruction starts from incomplete information, the optimisation progressively recovers the main upwellings, downwellings, viscosity structure and the surface stresses associated with dynamic topography. In other words, G-ADOPT is not simply running a model forward; it is using observations to infer a physically consistent history of mantle flow and its surface expression.

Image

Adjoint-based reconstruction of mantle evolution using G-ADOPT. The figure compares a reference simulation with the best reconstruction. Temperature and viscosity fields show how mantle structure evolves through time, while the misfit panels show where the reconstruction differs from the reference model. Surface normal stresses are shown alongside the temperature fields as an indicator of the dynamic topography generated by mantle flow. The close agreement between the reference and reconstructed simulations demonstrates the potential of G-ADOPT to recover physically consistent histories of mantle dynamics and surface deformation.

Duvernay et al., Geochemistry, Geophysics, Geosystems (2024) - Coupled geodynamical-geochemical perspectives on the generation and composition of mid-ocean-ridge basalts. Mid-ocean-ridge basalts are among the most important chemical samples of Earth's mantle, but their compositions are shaped by several processes before they erupt. This work develops a more quantitative link between mantle flow, melting and the chemistry of erupted basalts. The aim is to understand how mantle temperature, source composition, melt generation, melt transport and crustal processing combine to produce the chemical signals measured at the surface. This is part of a broader direction in my group: treating geochemical observations not just as descriptive records, but as quantitative constraints on mantle dynamics and the physical conditions of melting.

Davies et al., Elsevier book chapter (2023) - Observations and Models of Dynamic Topography: Current Status and Future Directions. Dynamic topography is the part of Earth's surface relief that is pushed up or pulled down by flow in the mantle beneath. It matters because these vertical motions influence coastlines, sedimentary basins, sea level, continental flooding, landscape evolution and ice-sheet stability. This review brings together observational and modelling work on dynamic topography and shows how a long-standing mismatch between observations and models has begun to be resolved. The key insight is that both deep mantle flow and shallower interactions between the asthenosphere and lithosphere are needed to explain Earth's surface response. The chapter also sets out the next major challenge: reconstructing how dynamic topography has changed through geological time. That challenge connects directly to G-ADOPT and inverse geodynamics, because reconstructing past surface motion requires models that can combine mantle dynamics, plate reconstructions and geological observations in a physically consistent way.

Davies et al., Geoscientific Model Development (2022) - Towards Automatic Finite Element Methods for Geodynamics via Firedrake. This paper laid the computational foundations for much of my current work. It showed that Firedrake, an automated finite-element system, could be used to build accurate, efficient, flexible and reproducible geodynamical models. The key idea is to separate the scientific description of a problem from many of the technical details of implementation. That separation makes it easier to build, test, extend and share complex models. This work provided the foundation for G-ADOPT and for a shift in my research program from bespoke forward simulations toward scalable, community-facing, data-informed and adjoint-capable geodynamic modelling.

Present-day thermal structure predicted from a global mantle convection simulation in Firedrake, where the geographic distribution of mantle heterogeneity is dictated by 230 Myr of imposed plate motion history from Muller et al. (2016). Each image includes a radial surface immediately above the core-mantle boundary, a cross-section and transparent isosurfaces at temperature anomalies away from the radial average. Blue and red isosurfaces highlight downwelling slabs and upwelling mantle plumes, respectively. Continental boundaries provide geographic reference. Panel a provides an Africa-centred view; panel b is centred on the Pacific Ocean and includes surface glyphs showing imposed plate velocities.

Davies et al., Nature Geosciences (2019): we combined observational constraints on residual topography beneath the world's oceans with a statistical approach to spherical harmonic analysis, generating a robust estimate of the power spectrum of Earth's oceanic residual topography. We showed that theoretical predictions can only be reconciled with the observed spectrum when they include both lithospheric structure and its contribution to global mantle flow. This study helped resolve a long-standing debate about the spatial distribution, wavelength and amplitude of present-day dynamic topography, and demonstrated that both deep mantle and shallow lithospheric processes are essential to Earth's surface response.

Inferred oceanic residual topography from a spherical harmonic model, regularised using automatic relevance determination, of an updated compilation of oceanic point-wise and ship-track residual topography measurements. The left panel displays the long-wavelength components of this model.

Iaffaldano et al., Nature Geoscience (2018): plate motions are well studied, but we still lack a complete physical understanding of how changes in plate motion influence surface deformation, especially where plate boundaries are diffuse. In this study, we identified geologically recent plate-motion changes in the Indian Ocean realm and linked the resulting deformation of the Indian Ocean floor to the Reunion mantle plume. The results highlight the role of deep mantle processes in shaping plate motion, intraplate deformation and the evolution of Earth's surface.

Jones et al., Nature (2017): this interdisciplinary study demonstrated that geographically and geochemically distinct double volcanic hotspot tracks on the Pacific plate, including Hawaii, Samoa, Society, Foundation and Marquesas, emerged concurrently at around 3 Ma. We proposed a dynamical explanation linking their emergence to a recent and previously unidentified change in Pacific plate motion, and showed how this change could generate systematic geochemical variations observed at Earth's surface.

Left: bathymetric map of recent Hawaiian volcanism, highlighting the Loa and Kea tracks. Right: schematic diagram of the tilted Hawaiian plume, the overlying Pacific plate and associated surface volcanism.

Animation highlighting the emergence of a double volcanic hotspot track in response to a plate-motion change.

Jones et al., Earth and Planetary Science Letters (2016): we quantitatively demonstrated how deep-mantle heterogeneities can be transported into a plume conduit, and identified the conditions under which deep-mantle structure can be inferred from geochemical observations at Earth's surface.

Davies et al., Nature (2015): we identified the full extent of Earth's longest continental hotspot track: the Cosgrove track in eastern Australia. This study showed a clear relationship between lithospheric thickness, volcanic outcrop and magma composition, indicating that lithospheric thickness exerts a dominant control on the volume and chemistry of plume-derived magmas and placing an observational constraint on sub-continental melting depths.

In Davies et al. (2015), we identified the Cosgrove hotspot track in eastern Australia, extending from Cape Hillsborough to Cosgrove. Magma composition and volcanic outcrop correlate with lithospheric thickness along this track, constraining the sub-continental melting depth of mantle plumes. The figure shows the distribution of eastern Australian Cenozoic volcanic centres and their relationship to lithospheric thickness.

Davies & Rawlinson, Geology (2014): this interdisciplinary study combined high-resolution seismic data from the transportable WOMBAT array with 3-D geodynamical models to show that lithospheric steps can localise mantle melting beneath the Newer Volcanics Province of Victoria and South Australia. The study helped address the broader question of why step changes in lithospheric thickness, common at craton edges and passive margins, produce volcanism only in some locations. We showed that both lithospheric structure and the orientation of mantle flow are critical.

Variations in P-wave velocity at 100 km depth beneath the Newer Volcanics region of southeastern Australia. The horizontal extent of volcanic outcrop is denoted by a dashed line. This interdisciplinary seismological and geodynamical study demonstrated that recent volcanism in southeastern Australia is linked to edge-driven convection and provides a potential explanation for why lithospheric thickness contrasts produce isolated volcanism in some settings.

Garel et al., Geochemistry, Geophysics, Geosystems (2014): we showed that competition between overriding-plate strength and tensile stresses in the subducting plate controls both trench migration and slab sinking rates, with important consequences for slab interaction with the mantle transition zone.

A 2-D thermo-mechanical dynamic subduction simulation from Fluidity. Temporal snapshots show temperature, viscosity, dominant deformation mechanism and the underlying computational mesh. In this class of model, slab buoyancy and rheology arise self-consistently through variations in temperature, pressure and strain rate. The computational mesh adapts during the simulation, concentrating resolution in dynamically important regions such as the slab core and the interface between the subducting and overriding plates.

Davies et al., Earth and Planetary Science Letters (2012): synthetic seismic structures from global mantle circulation models showed that thermal heterogeneity alone can explain key observed characteristics of the deep mantle. This result has important implications for the nature, amplitude and distribution of chemical heterogeneity in Earth's interior.

Shear-wave velocity perturbations beneath Africa from a tomographic model, a purely thermal model and a thermo-chemical model. By accounting for the geographic bias, smearing and damping inherent to tomographic models, the study enabled direct comparison between geodynamical predictions and seismic tomography, showing that thermal heterogeneity alone can reproduce key deep-mantle seismic characteristics.

Davies et al., Geochemistry, Geophysics, Geosystems (2011): we presented a new computational framework for geodynamical simulation built around parallel, anisotropic, adaptive, unstructured mesh techniques. This approach enabled simulations of mantle and lithospheric dynamics with an unprecedented combination of local resolution, geometric flexibility and computational efficiency.

Anisotropic and isotropic adapted unstructured meshes from a Fluidity convection simulation. The comparison illustrates how anisotropic mesh adaptation can efficiently concentrate resolution within thermal boundary layers and other dynamically important regions while avoiding unnecessary refinement elsewhere.

Davies & Davies, Solid Earth (2010): we presented a substantially improved estimate of Earth's surface heat flux using GIS methodologies and an enhanced global heat-flow dataset. This work provides an important constraint on Earth's thermal budget and long-term evolution.

Davies & Davies, Earth and Planetary Science Letters (2009): we used global mantle convection models to show that mantle plumes can reconcile a wide range of hotspot observations, providing quantitative support for the mantle plume hypothesis.

Temporal evolution of a high-Rayleigh-number global mantle convection simulation. Snapshots show thermal structure through time, including hot upwelling plumes rising from the deep mantle. The simulations demonstrate that plumes can be long-lived, mobile, ephemeral, pulsed or merging, and can reproduce a wide range of observed hotspot characteristics.

Davies et al., Geochemistry, Geophysics, Geosystems (2007): this was the first study to demonstrate the applicability and benefits of adaptive mesh refinement techniques for geodynamical simulations. These schemes use intelligent algorithms to modify the computational grid automatically, enhancing resolution where required by the evolving solution.