High-resolution eikonal imaging and uncertainty quantification of the Kilauea caldera
Angela F. Gao 1 , John D. Wilding 2 , Ettore Biondi 2 , Katherine L. Bouman 1 , Zachary E. Ross 2 1 Computing and Mathematical Sciences, California Institute of Technology, Pasadena, California, USA 2 Seismological Laboratory, California Institute of Technology, Pasadena, California, USA · 2025
Read High-resolution eikonal imaging and uncertainty quantification of the Kilauea caldera by Angela F. Gao 1 , John D. Wilding 2 , Ettore Biondi 2 , Katherine L. Bouman 1 , Zachary E. Ross 2 1 Computing and Mathematical Sciences, California Institute of Technology, Pasadena, California, USA 2 Seismological Laboratory, California Institute of Technology, Pasadena, California, USA (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2507-23692.txt.
Introduction
Angela Gao and co-authors image the Kilauea caldera in Hawaii in a 2025 geophysics paper. Images of Earth interior reveal how seismic and volcanic activity interact. High resolution images are hard to obtain because wave travel is nonlinear and the inverse problem is ill posed, so results can depend on starting guesses.
Their method builds a three dimensional view of the magma system at finer resolution than before. A variational approach with stochastic training gives an uncertainty estimate for each pixel, without strong prior assumptions. The work also estimates how much melt sits in the system, aiding volcano understanding.
Catalog record
| CREATOR | Angela F. Gao 1 , John D. Wilding 2 , Ettore Biondi 2 , Katherine L. Bouman 1 , Zachary E. Ross 2 1 Computing and Mathematical Sciences, California Institute of Technology, Pasadena, California, USA 2 Seismological Laboratory, California Institute of Technology, Pasadena, California, USA |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 52 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2507-23692
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
High-resolution eikonal imaging and uncertainty quantification of the Kilauea caldera Angela F. Gao 1 , John D. Wilding 2 , Ettore Biondi 2 , Katherine L. Bouman 1 , Zachary E. Ross 2 1 Computing and Mathematical Sciences, California Institute of Technology, Pasadena, California, USA 2 Seismological Laboratory, California Institute of Technology, Pasadena, California, USA Abstract keywords seismic tomography, volcano seismology, Bayesian inference, statistical methods, tomography {summary} Images of the Earth’s interior can provide us with insight into the underlying properties of the Earth, such as how seismic activity might emerge and the interplay between seismic and volcanic activity. Understanding these systems requires reliable high-resolution images to understand mechanisms and estimate physical quantities. However, reliable images are often difficult to obtain due to the non-linear nature of seismic wave propagation and the ill-posedness of the related inverse problem. Reconstructions rely on good initial estimates as well as hand-crafted priors, which can ultimately bias solutions. In our work, we present a 3D reconstruction of Kilauea’s magmatic system at a previously unattained resolution. Our eikonal tomography procedure improves upon prior imaging results of Kilauea through increased resolution and per-pixel uncertainties estimated through variational inference. In particular, solving eikonal imaging using variational inference with stochastic gradient descent enables stable inversion and uncertainty quantification in the absence of strong prior knowledge of the velocity structure. Our work makes two key contributions: developing a stochastic eikonal tomography scheme with uncertainty quantification and illuminating the structure and melt quantity of the magmatic system that underlies Kilauea. 1 Introduction Kilauea volcano in Hawai‘i is one of the most extensively monitored volcanoes on Earth. Since the onset of the decades-long Pu‘u ‘O‘o eruption in 1983, seismological, geodetic, and gravity investigations into the sustained eruptive activity at Kilauea have provided a rich understanding of the location of the volcano’s near-surface melt reservoirs, rift zone structure, and dike pathways and dynamics [ 15 , 27 , 2 , 21 ] . As a result, pre-eruptive patterns of deformation and seismicity originating within the upper kilometers of the subsurface can often be interpreted to assess the volcano’s immediate eruptive potential [ 35 , 1 , 37 , 24 ] . Our understanding of Kilauea’s longer-term eruption cycle behavior, however, may be limited by our understanding of the lower-crustal and mantle structures that supply magma to the volcano. Kilauea is supplied with magma from a zone of melting at 80-100 km depth proximal to the Hawaiian hotspot [ 57 ] . Recent work has suggested that structures within Kilauea’s mantle plumbing system can have an important influence on short-term fluctuations in magma supply rate to Kilauea, which may in turn influence its eruptive behavior on timescales of weeks to years [ 54 ] . During an observed “magma surge” episode in 2003-2007, the rate of magma supply to Kilauea increased by a factor of roughly two [ 42 ] ; other work has suggested that such surges may be enabled by feedback loops between tectonic and magmatic processes within mantle magma-bearing structures [ 54 ] . Furthermore, temporal correlations between eruptions at Kilauea and earthquake swarms at >30 km depth have been interpreted to evidence an “open” fluid connection between the surface and the mantle through which pressure gradients can rapidly propagate [ 56 ] . While density anomalies constrained by gravity modeling [ 10 ] and concentrations of long-period seismicity [ 57 , 3 ] suggest the existence of a columnar magma conduit extending downwards from Kilauea’s shallow reservoirs, the extent, geometry, and compositional melt fraction of this structure is still unknown. Improving our understanding of the mid-to-lower crustal structures that supply magma to Kilauea would allow us to better understand the features that enable short-term magma surges to the volcano, as well as the correlations between surface unrest and mantle seismicity. Numerous seismic tomography models have been developed to study subsurface structure at Kilauea [ 11 , 51 , 38 , 16 , 39 , 31 ] but there is still significant uncertainty about the overall magma storage beneath the Hawai‘i volcanic system. Prior models typically resolve features along Kilauea’s rift zones, corresponding to basaltic dikes, as well as anomalies at depths of 1-12 km beneath the caldera that are usually interpreted as corresponding to the deposition of ultramafic cumulates and a magma column beneath the summit [ 17 , 47 ] . As the amount of high-quality first-arrival pick data collected at Kilauea has grown, the resolution of resolved models has steadily increased (e.g., [ 31 ] ). However, more detailed conclusions from such models are often hindered by challenges associated with the tomographic inverse problem. It is well known that seismic tomography is ill-posed and non-unique [ 43 ] , and these properties result from several factors including the sparsity of noisy observations and the non-linearity of the forward problem. Consequently, inversion results are sensitive to the initial values and rely on strong regularization (e.g., smoothing) [ 43 ] . Seismic imaging studies typically utilize dampening as a form of regularization, which relies on the assumption that the initial 1-D velocity model captures the first-order structure of the underlying region of interest (e.g., [ 50 ] ). However, an incomplete or inaccurate initialization can cause the inversion to be trapped in local minima due to the non-linearity of the optimization problem; thus, solving for a single reconstruction may provide a limited or biased view of the structures that are actually constrained by the available travel-time data [ 12 ] . The uncharacterized uncertainty of the velocity, which results from the non-linear and non-unique tomography problem, limits our ability to draw robust inferences about the geometry and melt fraction of melt storage regions. For example, existing tomographic models of Kilauea generally resolve a single, undifferentiated velocity anomaly that is interpreted as a melt storage region below the caldera surface, although evidence from geodetic studies suggests the presence of multiple, distinct chambers [ 42 , 2 ] . Without quantitative information about model uncertainty, it is difficult to assess these models’ ability to differentiate between a single spatially distributed melt storage region or multiple neighboring chambers. Furthermore, placing confidence bounds on melt fraction values calculated from velocity values within imaged melt storage regions (e.g., [ 32 ] ) requires an estimate of uncertainty in the model parameters, which is not available in typical tomographic workflows. In this study, we develop an eikonal tomography procedure and apply it to produce a high-resolution tomographic image of the crust beneath the Kilauea summit. Our technique leverages advances in statistical inference to produce first-order uncertainty estimates through variational inference on recovered tomographic models. In particular, we use stochastic optimization through Monte Carlo sampling, which enables stable inversion in settings with weak or no priors. We validate our technique on realistic test cases and then apply our technique to travel-time data recorded at Kilauea volcano and identify a distinctly high- V_{P} / V_{S} region beneath the summit that we interpret as the magma-bearing column connecting the surface reservoirs to magma supply from the mantle. We take advantage of our derived uncertainties to estimate the size of this column and constrain a melt fraction of roughly 15% within this column. Our results suggest the existence of