Surface Processes and Coupling with Planetary Interiors

Scientists boarding D/V JOIDES Resolution off New Jersey in 1993. Sea level in an ice-free world would be 66 meters (216.5 feet) higher than now—shoulder-high to the Statue of Liberty. Image credit: Kenneth G. Miller, James V. Browning, and Gregory S. Mountain

How do surface processes shape the environment we live in, and how do they interact with the solid Earth interior, with implications for planetary evolution and habitability?

My work on surface processes and interior-surface coupling focuses on:

• Reading ice loss from sea level. The interplay among sea level change, cryosphere, and the solid Earth (Coulson et al. 2026; Bao et al. submitted). Ice-mass redistribution perturbs Earth’s gravity, rotation and solid surface and drives geographically variable sea-level change, a pattern known as a sea-level fingerprint. Building on an inversion of sea-surface height for Greenland ice mass change (Coulson et al. 2026), I developed a physics-informed neural operator (PI-DIFNO) that estimates ice loss directly from sea-level fingerprints and revealed scale-dependent limits on what can be recovered (Bao et al. submitted).

Predicted (left) and observed (right) sea levels caused by melting of the Greenland Ice Sheet (GrIS). A statistically significant correlation between the two fields (P < 0.001) provides an unambiguous observational detection of the near-field sea level fingerprint of recent GrIS melting in our warming world. Graphic: Coulson, et al., 2022 / Science

Teaching a neural network to read ice loss from the ocean. Each row starts from a sea-level pattern (left), and the network recovers the Greenland ice-thickness change that produced it (middle), compared with the truth and the error (right). Smooth, large-scale ice loss is recovered almost perfectly, while fine coastal detail is harder to see from the ocean. From Bao et al. (submitted)

• 4,000 times faster sea-level modeling. Accelerated glacial isostatic adjustment (GIA) modeling with SLEvolve (Bao et al. submitted). An exact finite-state representation of viscoelastic memory replaces the full load-history convolution in 1-D GIA solvers with recursively updated relaxation states, reducing the history-dependent cost from O(N^2) to O(N). SLEvolve matches conventional solvers to machine precision with up to 4,000-fold acceleration, opening the door to systematic exploration of ice histories and mantle viscosity.

How much faster SLEvolve runs than a conventional sea-level solver (a), and how long each takes (b), as the number of time steps grows. The speed-up keeps increasing with longer simulations, turning computations that took days into minutes. From Bao et al. (submitted)

• Sea level and the human story: coastlines, resources and migration. Collaborations on the consequences of ice loss for people and the geological record (Lucas et al. 2026; Powell et al. 2026; Wurmser et al. Under review at PNAS): Antarctic land and mineral-resource emergence in a warming world (Lucas et al. 2026), amplified sea-level rise in small island developing states (Powell et al. 2026), and early Pleistocene Red Sea shorelines linked to hominin dispersal out of Africa (Wurmser et al. Under review at PNAS).

Antarctic land that could become ice-free by 2300 under a high ice-melt scenario, shown with national territorial claims (outer arcs) and known mineral occurrences (symbols). From Lucas et al. (2026)

• Deep volatile cycles. Volatile cycling among different spheres of the Earth system over billions of years (Bao et al. In preparation)

Related Work

1.

Coulson, Sophie, Andrew Lloyd, Xiyuan Bao, Jerry X. Mitrovica, Sönke Dangendorf, Linda Pan, et al. 2026. “Inverting sea surface height data yields greenland ice mass changes (1993-2019): A proof of concept.” Geophysical Journal International

2.

Bao, Xiyuan, Sophie Coulson, Natasha Valencic, Mostafa Mousavi, Sönke Dangendorf, Andrew Lloyd, and Jerry Mitrovica. submitted. “Neural operators reveal scale-dependent limits on recovering greenland ice loss from sea-level fingerprints.” https://arxiv.org/abs/2610.23748.

3.

Bao, Xiyuan, Mostafa Mousavi, Konstantin Latychev, and Jerry Mitrovica. submitted. “An exact finite-state representation of viscoelastic memory for accelerated sea-level modeling: SLEvolve v1.0.” https://arxiv.org/abs/2610.23762.

4.

Lucas, Erica, Fred Richards, Gabriel Cederberg, Xiyuan Bao, Mark J. Hoggard, Stephen R. J. Tsuji, et al. 2026. “Emergence of antarctic mineral resources in a warming world.” Nature Climate Change

5.

Powell, Robert, Gabriel Cederberg, Sönke Dangendorf, Erica Lucas, Xiyuan Bao, Jessica Creveling, et al. 2026. “Disproportionate sea level rise in small island developing states of the pacific and indian oceans driven by greenland and antarctic ice mass loss.” Communications Earth & Environment

6.

Wurmser, Sylvie S., Manar Al Asad, Xiyuan Bao, Jessica Creveling, Jerry X. Mitrovica, and Christian A. Tryon. Under review at PNAS. “Reopening the southern gate: Early pleistocene sea level and hominin dispersal across the red sea.”

7.

Bao, Xiyuan, Zebin Cao, Kenneth Christopher Gourley, Sara Gutierrez, Yuanyuan Liang, Dongyuan Zhou, et al. In preparation. “Coupled evolution of earth’s hydrogen distribution and thermal history.”