Seismicity and elastic structure of the lithosphere in the Eifel volcanic fields

This work :
- presents the state of the volcanism in the East Eifel Volcanic Field, Germany using seismic records
- introduces the seismological network DEEP-TEE, its properties, goals, limitations and sensitivity
- introduces a robust seismic event detector A6-DFMSD
- presents Low-Frequency magmatic microseismic earthquakes
- presents 3-D image of the seismically active part of the plumbing system of the Laacher See Volcano
- estimates the moved magma volume beneath the volcano within July 2014 - Dec. 2022

A volcano is a consequence of magma intrusion from the asthenosphere into the Earth’s lithosphere. Although the majority of the volcanoes are formed in vicinity of the boundaries of the Earth’s tectonic plates, there are some few volcanoes that are formed far from these boundaries. The formation of these kind of intraplate volcanoes is explained by e.g., the existence of mantle plumes. Laacher See Volcano (LSV) in the East Eifel Volcanic Field (EEVF), Germany, is one of them. Its crater has a ~2.5 km diameter and it was formed after its last massive eruption (volcanic explosivity index V-VI) about 13,006 ± 9 years ago. ... mehrLSV was and is known as a dormant volcano which still emits CO2 and the noble gases He, Ne and Ar of mantle origin.
The occurrence of two typical magmatic weak (ML 0.7 and 0.9) deep (40-43 km depth) low-frequency (~2-8 Hz) earthquakes beneath the LSV (~8 km epicentral distance from the south-east of LSV) in Sep. 2013 opened a new chapter to the state of the volcano. Low-frequency (LF) earthquakes in volcanic regions are known as the sign of mass transport and/or volumetric change and/or resonance in the plumbing system of the volcano. Hence, seismologists from the Geophysical Institute at KIT started and conducted densifying the pre-existing permanent seismological station network in the region since July 2014 to increase the chance of recording/capturing such weak seismic signals. They called this experiment Deep Eifel Earthquakes Project - Tiefe Eifel Erdbeben (DEEP-TEE). DEEP-TEE experienced several phases of reconfiguration (keeping quiet / low-noise sites) and expansion (new installation supporting the requirements of a better earthquake detection and localization). The DEEP-TEE dataset is enough sensitive to capture small magnitude (ML<1) local events to the depth of uppermost mantle (~40-50 km depth).
An active volcano is a massive natural complex source of vibration. Shortly, the complexity comes from the fact that a volcano is built up of material with the three phases of solid, liquid and gas under a wide range of temperature and pressure. Vibrations in volcanic regions are characterized by their waveform features (ground motion in time), frequency content and partly by the responsible generation mechanism. Tectonic, volcano-tectonic, hybrid, LF, very LF, tremor, tornillo and explosion events are the different types of the typically observed signals in active volcanic regions in which each of these events carries specific features in time, frequency and conceptional domain. Detection of the LF events, as the first precursor sign of a possible eruption, is important for saving time and manage the possible hazard of an eruption, especially in populated regions. On the other hand, LF events are weak and diverse and their signals are easily masked by the environmental level of noise, specially by the noise of the human activities.
The detection of the weak and diverse signals originating from volcanic activities is a challenge, especially in the presence of moderate to high levels of noise conditions (a case matching with the records of the DEEP-TEE dataset). This becomes more challenging for the widely and commonly used template matching techniques (a technique which uses the cross-correlation coefficient between a template and sections of the records as a characteristic function of detection). For that technique the following cases are blind spots: a) there is no a priory waveform information to search for (a case matching with records of newly installed seismometers) and b) the source migrates and, therefore, wavefields select a different ray path than their previously selected source-receiver ray path and this change results in changes in the waveforms (the case matching with magma transportation within the plumbing system of the volcanoes). Also, the developments in detection algorithms based on Artificial Intelligence still have a large dependency on the amount and diversity of the training dataset which makes it intractable for the case of detection of LF events in the EEVF. There only 75 examples of LF events were reported at the time of starting the development of a new detection algorithm presented in this thesis. On the other hand, the old concept of seismic signal detection, the short/long time average algorithm (STA/LTA) does not need a priory information regarding the features of the signals to search for. But STS/LTA detectors generate a large number of false detections when searching for weak signals with a low signal to noise ratio (SNR) of about 2, the goal for the EEVF. Hence, developing a new detection algorithm matching with the needs for the detection of weak and diverse signals, and matching with the properties of the DEEP-TEE dataset became the main priority of this thesis, after field work and data quality analysis/control of the DEEP-TEE dataset. The developed detection algorithm presented in this thesis, searches signals which match coincidences at neighboring recording stations in space, time and frequency domain. It combines a STA/LTA, a frequency range selection, an energy envelope matching, and a backprojection technique to find a robust detection model. It also initially classifies seismic detections based on their frequency content which plays a major role in searching LF events.
The LF events in the detection output lists are then localized after manually picking the first arrival onsets of the direct P and S waves of the events from the seismic records of the time period July 2014 – December 2022. Location of the LF events is done in several steps: a) the initial absolute hypocentral parameters of the events are obtained using the location program VELEST, b) for the EEVF region the new seismic velocity models KIT5 and KIT6 are generated matching with the observed picks of the events, c) events relocated using KIT6 models, d) the relative relocation algorithm HypoDD is applied to achieve a fine spatial hypocenter structure and e) a jackknifing validity test is used with HypoDD and 100 subsets of randomly selected events to achieve a reliable range of the hypocentral parameters. The results provide a high resolution and a well reliable 3-D projection of the 289 source locations of the LF events beneath the EEVF. These LF events are distributed vertically from 4 km to 45 km depth just southeast of Laacher See Volcano plus a distinct small cluster of LF events located ~12 km to the east of the center of the main cluster of LF events in the Neuwied Basin.
At the end of this thesis, I provide an estimation to the amount of the moved magma volume (MMV) beneath the EEVF within the time period July 2014 – Dec. 2022 using an updated version of the model presented by Aki et al. (1977) and Aki and Koyanagi (1981). This model establishes a relationship between the observed peak-to-peak amplitude of the seismic P- and S-wave pulses and the involved fluid or melt volume. The MMV value presented here is obtained by using twelve very good observations (amplitude measurements) and considering three possible scenarios for the geometry of the magma intrusions relative to the observation points (seismic stations).
The MMV in the studied time period is ~2.9 104 m3 equivalent to a cube with ca. 31 m side length when the median of the all the estimations is taken. Calculation of the uncertainty of this estimation is a challenging issue because the distribution of these estimated MMV is not Gaussian. Therefore, the uncertainty is estimated from the variation of the data points in Fig. 6.4. This gives a range of 1 104 – 3 105 m3 which seems to be a reasonable value in the context of petrological studies on the EEVF.