Nakao Lab - Research

Development of Data Assimilation Methods for Reconstructing Thermal Convection

Mantle convection is a fundamental process in the Earth’s interior. Part of this motion appears at the surface as plate tectonics, and it plays a key role in driving earthquakes and volcanic eruptions. However, because the mantle lies deep inside the Earth, we cannot observe it directly. In addition, the history of the mantle spans 4.6 billion years, and only limited records of its past remain. As a result, many aspects of the mantle’s internal structure and long-term evolution are still poorly understood.

Our group develops mathematical methods to quantitatively reconstruct the history of mantle temperature, flow, and material circulation. To do this, we combine multiple types of indirect evidence—such as seismic tomography, plate motions, and geological rock records—with fluid dynamics theory, using an approach called data assimilation. In this sense, our research is like putting together pieces of a puzzle to reveal how the Earth’s interior has evolved over time.

Students in our group also study related phenomena and systems, including flows in oceans and lakes, as well as volcanic eruption processes.

Related works

Nakao (2026, ESS Open Archive). https://doi.org/10.22541/essoar.177248990.09713169/v1

Nakao et al. (2025, Sci Rep). https://doi.org/10.1038/s41598-025-19724-x

Nakao et al. (2025, JGR:ML). https://doi.org/10.1029/2024JH000288

Nakao et al. (2024, GJI). https://doi.org/10.1093/gji/ggad417

Learning dynamics of a Physics-Informed Neural Network learns about mantle convection.

Simultaneous Data Analysis of Earthquakes, Volcanoes, and Plate Tectonics

Large earthquakes and explosive volcanic eruptions are major natural phenomena that characterize subduction zones, such as those around the Japanese Islands. However, the processes that lead to these events are highly complex.

Our group uses data-driven approaches to analyze various types of plate tectonic data observed in subduction zones around the world. Through this work, we aim to identify the key parameters that control these complex phenomena and to develop predictive models.

For example, our previous analyses have shown that regions where large earthquakes occur tend to have several common features: (1) a thick crust in the overriding continental plate, (2) the presence of geological structures called accretionary prisms, (3) thick sediment accumulation in the trench, (4) deep trenches, and (5) a large radius of curvature of the subducting oceanic plate.

In the future, we aim to build more accurate predictive models by increasing the number of observation sites and variables, and by improving our analytical methods.

Related works

Ueki et al. (2025, PEPI). https://doi.org/10.1016/j.pepi.2025.107464

Nakao et al. (2023, EPS). https://doi.org/10.1186/s40623-023-01839-y

Nakao et al. (2022, FES). https://doi.org/10.3389/feart.2022.1008058

Page updated

Google Sites

Report abuse