volcanoes, earthquakes
and the hydrosphere through geodesy


[ Real-Time/High-Rate GPS | Volcano Sources | Crustal Loading ]

This is an overview of my broad research interest and respective projects I have been working on over the last years (up to 2013). It is not exhaustive and I am always interested to apply some of the techniques described here to new problems, or learn something new to look at old data in a different way. If you're interested in collaboration, send me an email!

the lab - 2021 overview

Real-Time and high-rate GPS for Early Warning and Hazard Mitigation (top)

The work focusing on producing quality real-time GPS data for near instantaneous hazard assessment and mitigation is an emerging field. It benefits tremendously from high-rate (post-processed) GPS studies, as these identify worthwhile applications at a high signal to noise ratio. Major areas of research involve earthquake early warning and rapid response, eruption early warning and (near) real-time monitoring, and tsunami monitoring. The problems in real-time applications are rather technical as methods to reduce real-time noise must be developed, and strategies for handling and analyzing the resulting large volumes of data in minimum time need to be adapted.

Currently, in my post-doc work, I focus on the integration of real-time GPS into an existing seismic system for California. We want to enable reliable magnitude estimation for large events (M>6.5), and rapid estimation of distributed slip along a fault. I use simulated static offsets and real past events to test the system and estimate the timing of warnings / rapid response.

I have been also exploring the use of sub-daily GPS for volcanic plume detection at Redoubt and Okmok volcanoes. Below I show an example for Redoubt.

Skyplot (or azimuth-elevation plot; code I used for plotting here) of phase residuals per satellite for Redoubt stations RVBM and AC17. The outer circle marks azimuths for the satellites and also indicates 0 of elevation above the horizon as seen from the station. The two inner circles mark 30, 60 of elevation. 90 of elevation is directly above the station. Thin black lines indicate the tracks of the GPS satellites. The gray traces along these lines are the time series of phase residuals (mismatch between recorded and modeled satellite signal phases) for this satellite. The red sections indicate the eruption time from 14:00-14:40 UTC on April 4, 2009. The volcano summit is about 6 km to the east of station RVBM and about 28 km southwest of site AC17. The impact of the plume is clearly seen at satellite PRN 10 for RVBM, whereas AC17 shows no misfit for this satellite.

Relevant papers:

Volcanic Source Modeling (top)

Knowledge about the plumbing system of volcanoes is important for hazard assessments and our general understanding of how these land creators work. Geodetic data (InSAR, GPS, Leveling, Tilt, Gravity) are phenomenal tools to infer knowledge about the plumbing system of volcanoes. The results depend highly on the kind of volcano (open / closed system, magma compressibility, ...), depth of the source and its volume change. The expected pattern, however, is that of inflation of the edifice when the magma source is recharged and co-eruptive deflation; much like a ballon.

Cycle of Volcano Deformation from unpressurized magma reservoir to overpressure and associated inflation to eruption with lava flow and deflation (Background: Bezymianny and Kamen with a bit of a plume from Klyuchevskoy in Kamchatka).

During my dissertation I was fortunate enough to conduct the first in-depth, GPS based geodetic studies of Redoubt Volcano in Alaska and Bezymianny Volcano in Kamchatka, Russia.

Left: Velocity field at Bezymianny volcano (horizontal: blue, vertical: red) for data from August 2005-August 2010. Lower portion is blow-up of the black rectangle in the regional map. Tectonics model (Bürgmann et al, 2005, indicated by grey rectangles in regional map) is removed from data, horizontal is with respect to ES1. Note the wide ranging subsidence signal in vertical and incoherent horizontal (Grapenthin et al., JVGR, 2013a).
Right: Co-eruptive displacement field (not velocity!) at Redoubt showing displacements from March 22-April 4 2009 relative to a site about 100 km SW (horizontal: blue, vertical: red). Note the horizontals pointing to the vent indicating the map location of the deflation source (Grapenthin et al., JVGR, 2013b).

Redoubt erupted in 2009 and was equipped with a small monitoring network. We found an interesting aseismic pre-eruptive inflation period starting about 6 months before deep long period seismicity was recorded. This magma pooled underneath residual mush from the previous eruption in 1989/90. Heat and fluids percolating through this resulted in remobilization of the residual material. The entire GPS network subsided and moved inward toward the volcano from the summer of 2008 to summer of 2009 indicating that most of the material erupted was left over from the previous event.

Cartoon illustrating the evolution of the Redoubt Volcano plumbing system as suggested by geodetic, seismic, and petrologic data. We tie deep seismicity (Power et al., 2013), petrology (Coombs et al., 2013), and our observations together by proposing a two reservoir system in the mid- to shallow crust. Material from 25 to 38 km migrated to about 13 km depth beginning as early as May 2008; reheating and remobilizing residing material in a prolate spheroid from 7 to 11.5 km. This resulted in migration to 2-4.5 km depth (Coombs et al., 2013); supported by shallow seismic tremor beginning in January/February 2009 (Buurman et al., 2013). This material extruded from 23 March 2009 on. The mix of fresh and reheated material from the deeper stages of the system replaced extruded material and made the shallow removal undetectable by geodesy (Grapenthin et al., JVGR, 2013b).

Over the 5 years of recording many small eruptions at Bezymianny we found only very small co-eruptive activity. The interesting observation here is, however, that the entire network subsides at rates up to 15 mm/yr. Neither subduction zone tectonics nor adjustment to volcanic loads seem to explain these observations. We infer that a a deep deflating sill beneath Klyuchevskoy at ~33 km depth can induce a signal very similar to what we record. However, to confirm this model, a much broader GPS network must be deployed. Bezymianny

Left: Sill model (white rectangle, double line indicates down dip end) inferred from seismicity below 22.5 km (black circles). Velocity predictions relative to ES1 for this model, assuming a closing rate 0.22 m/yr for the sill, are shown as white (horizontal) and black (vertical) vectors.
Right: (A) Continuous site BZ08 in summer of 2010. In the background: Kluchevskoy with a small ash plume to the left, Kamen in the middle, and Bezymianny to the right and degassing. (B) Campaign site BEZR with spike mount setup and Trimble NetRS receiver in 2010. (C) Continuous site BZ06 with solar setup installed in 2010. Bezymianny's dome steams in the background, which is the normal state. (Grapenthin et al., JVGR, 2013a).

Other studies, I was involved in focused, for example, on resolving a broad, dough-nut shaped, inflation pattern with central subsidence in InSAR data at Hekla volcano, Iceland. While we explained the inflation with a deep magma chamber and the central subsidence with adjustment to lava loading, other work suggests that a vertical cigar shaped model could induce the same pattern. Exciting stuff!

Relevant papers:

Crustal Loading Studies (top)

I have a great interest in how surface load dynamics affect the state of the crust and reveal details of the rheology of underlying materials, as well as how we can use crustal dynamics to reconstruct past loading histories.

Conceptual model of an Earth-load-response-system. The load (input, solid black rectangle) is applied to the pointed black line which represents the initial state of the Earth's surface. The Earth (filter) responds to the mass force (little black arrows) through surface displacement, changing its state to the one denoted by the solid black line (output). The load, however, may vary over time which is denoted by a load history. At first it might raise to a maximum level which results in maximum displacement and then it could drop to a minimum represented by the dashed gray lines that limit or extend the load box. Depending on the load history, the displacement might alternate between the upper and the lower dashed, gray surface lines, linked to minimum and maximum load, respectively. In addition to these elastic properties, another important property is that of time dependent stress relaxation or material creep which is the adjustment to a new stress state by ductile material flow, which in the depicted system is a filter property (Grapenthin, Computers & Geosciences, 2014).

As a visiting student in Iceland I worked on seasonal variations in continuous GPS time series. We linked this signal to changes in snow load on Iceland's major ice caps (Grapenthin et al., 2006). To facilitate this work, I wrote software that simulates the crustal response to surface load changes. This code eventually turned into my diploma thesis in computer science (German M.Sc. equivalent) and has since been called CrusDe, and developed quite a bit further (Grapenthin, 2013). CrusDe's key capability is the implementation of an abstract relationship between Earth model and surface load, which enables an easy (plug-in based) exchange of problem specific model realizations. Some of my work uses CrusDe, for example, to address the bias surface loads (lava flows, glaciers) can introduce in magma source inversions at volcanoes (Grapenthin et al. 2010, 2013). I am currently working on including a spherical Earth geometry and, of course, applying it to exciting loading problems.

Left: A map of Iceland and its largest ice caps (V: Vatnajökull, L: Langjökull, M: Myrdalsjökull, and H: Hofsjökull). The red dots represent the CGPS stations in Iceland's ISGPS network as in 2005 and used in this study. The colors represent calculated absolute peak-to-peak seasonal displacement due to maximum winter load using E = 40 GPa. (a) Vertical displacement with a peak of 37 mm under the center of Vatnajökull. (b) Vector lengths of horizontal displacements with a maximum of 6 mm east and north of Vatnajökull (displacement towards load during loading, opposite direction during unloading).
Right: Comparison of predicted and observed results of temporal modeling using a harmonic load at four CGPS stations: HOFN, SAUD, SKRO, and SOHO, in east, north and up component over the years 1999-2006. The detrended CGPS time series are shown by green dots. The red line is the best fit to the time series. The blue line is the modeled displacement using E = 40 GPa. Time series, best fit, and modeled displacement are relative to station REYK. (Grapenthin et al., GRL, 2006).

Relevant papers:

| Last modified: May 25 2023 16:19.