📄 earth science

Experimental Investigation of Dike Propagation Influenced by Graben Topography and Magma Viscosity Using Analog Modeling

This study utilizes analog modeling to demonstrate that dike propagation and volcanic vent locations in graben settings are controlled by the interplay between magma viscosity, topographic stress fields, and crustal mechanical layering, which collectively determine whether intrusions arrest, transition between geometries, or reach the surface.

Original authors: Khaled Rahimi, Marcus Bursik, Janine L. Kavanagh

Published 2026-09-17

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Khaled Rahimi, Marcus Bursik, Janine L. Kavanagh

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Volcanoes are not random explosions; they are the surface expression of a deep, pressurized struggle between molten rock and the solid ground above it. To understand where a volcano might erupt next, scientists must track the path of a dike, which is simply a crack filled with magma that pushes its way upward from a deep reservoir. The path a dike takes is not a straight line. It is dictated by the forces pressing on the rock around it, specifically the stress field, which changes depending on the shape of the land and the layers of rock the magma encounters. If the ground above is heavy, the magma pushes straight up. But if the ground has a low spot, like a valley or a rift, the weight of the rock is reduced in that area, creating a zone of "unloading" that can pull the magma sideways. Furthermore, the magma itself is not a uniform substance; it ranges from runny, fast-moving lava to thick, sluggish sludge. This difference in thickness, or viscosity, changes how the magma reacts to the stress of the surrounding rock. Predicting exactly where a volcano will open its mouth is a critical challenge for safety, yet the complex interplay between the shape of the land, the layers of the crust, and the nature of the magma has remained difficult to pin down.

To untangle these variables, researchers Khaled Rahimi, Marcus Bursik, and Janine L. Kavanagh turned to a laboratory that mimics the Earth's crust using a simple, transparent material: gelatin. They chose the Hat Creek Graben in northeastern California as their real-world guide, a long, deep valley formed by the Earth pulling apart, where different types of lava have erupted in the past. In their lab, they built a clear tank filled with layers of gelatin to represent the Earth's upper crust. They carved a long, narrow depression into the top of the gelatin to simulate the graben. Then, they injected fluids from the bottom of the tank to act as rising magma. They used two distinct fluids: water to represent runny basaltic magma, and a thick, gel-like solution made from xanthan gum to represent the much stickier basaltic andesite magma. By watching how these fluids moved through the gelatin, the team could see exactly how the shape of the valley and the thickness of the magma altered the path to the surface.

The experiments revealed a striking difference in behavior based on how thick the fluid was. When the researchers injected the thin, watery fluid, it surged upward with speed and determination. It rose almost straight through the gelatin, barely noticing the stress changes caused by the valley above. When it reached the surface, it erupted directly inside the depression, staying within a short distance of the center. The water moved too fast and with too much momentum to be easily swayed by the changing forces of the surrounding rock. In contrast, the thick, xanthan gum fluid behaved very differently. As it rose, it slowed down, giving the surrounding gelatin time to influence its path. Instead of going straight up, the thick fluid began to bend and deflect at a significant depth, curving toward the steep walls of the valley. It eventually broke through the side of the valley wall rather than the center, erupting through the fault scarp. The viscosity of the fluid essentially determined whether the magma ignored the valley's shape or followed its pull.

The researchers also explored what happens when the crust is not a single block of rock but is made of different layers with different strengths. In some experiments, they created a weak boundary between a lower layer and an upper layer of gelatin. When the magma hit this weak spot, it stopped trying to go up and instead spread out sideways, forming a flat, pancake-like intrusion known as a sill. If the injection continued, this flat sheet would thicken and dome upward, creating a mushroom-shaped structure called a laccolith. Interestingly, the presence of the valley above influenced this process; the stress field created by the valley encouraged the sill to grow into a laccolith directly beneath the depression, causing the floor of the valley to lift up. However, when the boundary between the layers was strong, the magma was able to punch through the interface and continue its journey to the surface, regardless of the layering.

Finally, the team investigated what happens when two streams of magma rise at the same time, a scenario common in volcanic fields where multiple cracks open simultaneously. They found that the two rising columns did not ignore each other. As they pushed upward, they created their own zones of pressure that interfered with one another. In one specific case, the presence of a spreading sheet of fluid from one injection caused the second, nearby injection to bend away, pushed outward by the stress of its neighbor. This interaction showed that the path of a single dike is not just about the rock above it, but also about the presence of other nearby intrusions. The study suggests that forecasting where a volcano will erupt requires a precise understanding of three things: the shape of the land, the mechanical layers of the crust, and the thickness of the magma itself. Without knowing all three, predicting the exact location of a vent remains a difficult puzzle, but these experiments provide a clear map of how these forces compete and cooperate deep underground.

Technical Summary: Experimental Investigation of Dike Propagation Influenced by Graben Topography and Magma Viscosity Using Analog Modeling

Problem Statement
Predicting volcanic vent locations during unrest is a critical challenge in volcanology, particularly in tectonically active regions characterized by topographic lows such as grabens. While regional stress fields generally govern dike trajectories, local perturbations caused by topography (specifically "unloading" stresses) and crustal heterogeneity can significantly alter propagation paths. Furthermore, the rheological properties of magma, particularly viscosity, fundamentally control ascent dynamics. Previous studies have addressed topographic unloading and mechanical layering separately, but there is a need to understand how these factors interact with magma viscosity to determine whether a dike will erupt within a graben, deflect to a fault scarp, or arrest as a sill or laccolith. This study addresses these coupled mechanisms using the Hat Creek Graben (HCG) in northeastern California as a natural analog.

Methodology
The authors employed analog laboratory modeling to simulate dike propagation in a graben setting.

Experimental Setup: Experiments were conducted in a transparent Perspex tank (40 × 40 × 30 cm) filled with gelatin, which serves as an elastic crust analog with tunable Young's moduli (5–10 kPa). A specialized mold created an elongated topographic low (graben) with flanks inclined at 75°, mimicking the geometry of the HCG (widening from 4 cm to 8 cm along the strike).
Magma Analogs: Two fluids with contrasting viscosities were injected from the base of the gelatin slab:
- Water: Representing low-viscosity basaltic magma (Newtonian, $\mu \approx 10^{-3}$ Pa.s).
- Xanthan Gum (XG) Solution (0.15 wt%): Representing higher-viscosity basaltic andesite (shear-thinning, effective viscosity $7.0 \times 10^{-2}$ to $22 \times 10^{-1}$ Pa.s).
Experimental Variables: The study varied fluid viscosity, injection methods (sequential vs. simultaneous), and crustal layering (single-layer homogeneous vs. two-layer systems with varying interface strengths).
Imaging and Analysis: Subsurface stress fields were visualized using photoelasticity (polarized light), while dike geometry and surface deformation were recorded via high-definition cameras. Data were processed using Tracker software to quantify dike length, dip, and tip position.
Scaling and Modeling: Results were scaled to natural conditions using dimensionless parameters ( $\Pi_1$ : ratio of overpressure to viscous resistance; $\Pi_2$ : ratio of topographic unloading to viscous resistance). Additionally, 2D plane-strain finite element models (COMSOL Multiphysics) were constructed to characterize the stress field orientation beneath the graben.

Key Results

Viscosity-Dependent Trajectories:
- High-Viscosity (XG) Dikes: These dikes exhibited significant deflection. They propagated sub-vertically initially but deflected toward the graben fault scarp at greater depths (approx. 12 cm in the model) where the maximum compressive stress ( $\sigma_1$ ) rotated. They ultimately erupted through the fault scarp, often with a horizontal distance of >3 cm from the graben centerline.
- Low-Viscosity (Water) Dikes: These dikes showed minimal deflection. Instead, they tended to rotate about their vertical axis in the upper crust, maintaining a near-vertical trajectory. They predominantly erupted within the graben (in-graben), with vent locations typically <2 cm from the centerline.
Mechanical Layering and Intrusion Geometry:
- Weak Interfaces: In two-layer systems with weak mechanical coupling, dikes arrested at the interface, transitioning laterally to form sills. Continued injection caused these sills to thicken and evolve into laccoliths beneath the graben, coinciding with the uplift of the graben floor.
- Strong Interfaces: When the interface between layers was strong (high rigidity ratio), dikes penetrated the interface and propagated into the upper layer, eventually erupting.
- Graben Influence on Laccoliths: The presence of the graben modified the local stress field, promoting the transition from sill to laccolith and controlling the laccolith's elliptical geometry, with the long axis aligned with the graben.
Dike-Dike Interactions:
- Sequential Injection: The central dike had negligible effect on a subsequently injected offset dike.
- Simultaneous Injection: Stress fields interacted significantly. In cases where a central dike developed a "canoe-like" sheet (lateral propagation), it induced local compressional stress that caused an adjacent offset dike to bend outward, away from the graben.
Stress Field Rotation: Numerical models confirmed that $\sigma_1$ rotates from vertical at depth to sub-horizontal beneath the graben floor. High-viscosity dikes, moving slowly, closely followed this rotating stress trajectory, leading to deflection. Low-viscosity dikes, moving rapidly, were less sensitive to this rotation, maintaining a more vertical path until shallow depths where rotation caused outward deflection.

Significance and Claims
The authors claim that this study advances the understanding of the mechanisms controlling dike propagation and volcanic eruption in graben settings. The primary contribution is the demonstration that forecasting vent locations in such environments requires knowledge of both the regional/local stress field (induced by topography) and the specific rheological properties of the magma.

Specifically, the paper asserts that:

Vent location is not solely determined by graben geometry but is a function of the interplay between magma viscosity and the stress perturbation caused by topographic unloading.
High-viscosity magmas are more likely to be deflected to fault scarps, while low-viscosity magmas are more likely to erupt within the graben, often with rotated orientations.
Mechanical layering acts as a critical control on whether intrusions arrest as sills, evolve into laccoliths, or reach the surface.
The study provides a framework for interpreting volcanic unrest signals, suggesting that the likelihood of eruption in-graben versus off-graben depends on the coupling of magma composition, crustal layering, and topographic stress fields.

The authors conclude that while these experimental insights improve the prediction of eruption sites in graben settings, further investigation is needed to fully integrate these mechanisms into regional tectonic and volcanic risk forecasting models.

More like this