How do faults and the surrounding crust and mantle deform through the earthquake cycle, and can we forecast what they do next?
Faults slip in many ways: in seconds during earthquakes, over days to years to decades as afterslip, viscoelastic creep and slow slip events. Around them, the mantle flows and the brittle crust folds and fractures. These processes interact over timescales from seconds to millions of years, which makes them expensive to simulate. We develop efficient ways to solve the underlying ODEs and PDEs, using ideas from dynamical systems theory, reduced-order models and fast numerical methods. This lets us simulate many earthquake cycles quickly and test the models against geodetic, seismic and geological observations.
Questions we are interested in:
Why do some earthquakes grow large while others arrest prematurely? What controls whether a fault slips in earthquakes, slow slip events or steady creep, and how do these modes interact?
How do afterslip and mantle flow redistribute stress after large earthquakes, and how long do their effects last?
How does repeated slip on faults build geological structures?
What sets the horizon of predictability for earthquakes, and which observations and physical constraints are needed to forecast a fundamentally stochastic process?
We study the earthquake cycle using dynamical systems theory as well as by combining observations with numerical simulations. Top: simple models of two coupled faults show that their earthquakes can fall into step, but only rarely (Dascher-Cousineau et al., 2026, GRL). Bottom: satellite observations show folds growing at their hinges during the 1999 Chi-Chi earthquake in Taiwan and for decades after the 1978 Tabas earthquake in Iran (Mallick et al., 2021, JGR: Solid Earth).
What can surface observations tell us about the Earth at depth?
Friction on faults and the rheology of the crust and mantle cannot be measured directly at depth. They must be inferred from how the surface moves. We develop inverse methods that use geodetic and seismic observations to infer fault friction and the rheology of the crust and mantle. We also ask what these observations can and cannot resolve.
Questions we are interested in:
Why do different models of friction and rheology fit the same surface observations, and which observations can tell them apart?
Why does deformation sometimes stay localized on a narrow fault and sometimes spread across a wide zone of the surrounding crust? Similarly, what controls localization in the mantle?
What can surface observations tell us about the Earth at depth? Top: we fit earthquake cycle models to GPS time series to estimate how fault strength varies with depth, and test what the data can and cannot resolve (Köhne et al., 2025, Earth, Planets and Space). Bottom: models of the 2021 Maduo earthquake in Tibet show that when fault slip stops short of the surface, the surrounding crust yields across a zone up to ~2 km wide; deformation becomes localized on the fault only as slip reaches the surface (Antoine et al., 2026, EPSL).
How do plate boundaries accommodate plate motion, and how do they form and evolve?
We use geodesy and mechanical models to map where faults are locked, creeping or slipping slowly, and how plate motion is shared between megathrusts, strike-slip faults and distributed deformation of the crust. We also study how plate boundaries form and evolve, from subduction initiation to the growth of plateaus. Our work focuses on the Sunda megathrust, Indo-Burma margin, the Himalaya and Tibet.
A new plate boundary in the making. Beneath the Surma Basin, the edge of the Indian plate may be sinking into the mantle, a rare example of a new subduction zone forming. This pulls the Shillong Plateau up along the Dauki Thrust and pushes the Indo-Burman Wedge westward. Mallick et al. (2020), Earth and Planetary Science Letters.
What makes a slope fail?
We model landslides as coupled systems of friction, fluid pressure and a deforming mass driven by gravity. We study how slopes become unstable and how climatological effects (loss of glacial buttressing or fluid pressurization from infiltrating rain/snow melt) and earthquake shaking can trigger failure.
We study what makes slopes fail, combining mechanics and observations. Top: when a glacier retreats, it removes support from the slope below; depending on slope angle and pore pressure, the slope either settles into a new balance or fails (Mallick, Finnegan & Fielding, 2026, GRL). Bottom: during the 2018 Palu earthquake in Indonesia, slopes of only a few degrees slid up to ~10 m (and in some cases 100-1000m), breaking away in irrigated rice paddies and stopping in forest (Bradley et al., 2019, Nature Geoscience).