Skip to content

Commit

Permalink
updated paper text, Figure 2 and added database and yaml for Figure 2
Browse files Browse the repository at this point in the history
  • Loading branch information
mrosskopf committed Sep 6, 2024
1 parent 81c400f commit 2896f7f
Show file tree
Hide file tree
Showing 5 changed files with 150 additions and 14 deletions.
Binary file removed paper/DUGseisGUI_V2.JPG
Binary file not shown.
Binary file added paper/DUGseisGUI_V3.JPG
Loading
Sorry, something went wrong. Reload?
Sorry, we cannot display this file.
Sorry, this file is invalid so it cannot be displayed.
Binary file added paper/GUI_ExampleEvent_Fig2.sqlite
Binary file not shown.
69 changes: 69 additions & 0 deletions paper/GUI_exampleEvent_Fig2.yaml
Original file line number Diff line number Diff line change
@@ -0,0 +1,69 @@
version: 14

meta:
project_name: VALTER_Stimu3
project_location: Bedretto
project_description: VALTER_Stimu3

local_coordinate_system:
epsg_code: 2056
translation_vector: [2679720.696, 1151600.128, 0.0]

paths: # absolute paths
asdf_folders:
- '/PATH/TO/WAVEFORMFOLDER/.../WAVEFORMS/continuous-04-bedretto'
- '/PATH/TO/WAVEFORMFOLDER/.../WAVEFORMS/continuous-03-bedretto'
stationxml_folders:
- '/PATH/TO/STATIONXMLFOLDER/.../STATIONXMLS'
database: 'sqlite:///PATH/TO/DATABASEOUTPUT/.../GUI_ExampleEvent_Fig2.sqlite'
cache_folder: '/PATH/TO/CACHEFOLDER/.../cache'

# Must cover the whole range of the experiment, e.g. more than the expected
# start + end time for the live test case.
temporal_range:
start_time: 2023-03-14T15:50:00.000Z
end_time: 2023-03-14T15:57:00.000Z


graphical_interface:
classifications:
- passive
- relocated
- noise
- unknown
- active
pick_types:
- P
- S
uncertainties_in_ms:
- 0.00001
- 0.000025
- 0.00005
3d_view:
# Helpful to distinguish newer from older events.
color_events: plasma
size_events_in_pixel: 7
size_channels_in_pixel: 3
# Greenish color for channels.
color_channels: [0.1, 0.9, 0.0, 0.4]

location_algorithm_default_args:
velocity:
P: 5100.0
damping: 0.01
use_anisotropy: true
anisotropy_parameters:
P:
azi: 310.0
inc: 28.6
delta: 0.071
epsilon: 0.067

filters:
- filter_id: smi:local/bandpass_causal_1000_5000
filter_settings:
filter_type: butterworth_bandpass
highpass_frequency_in_hz: 1000.0
lowpass_frequency_in_hz: 5000.0
filter_corners: 4
zerophase: false
95 changes: 81 additions & 14 deletions paper/paper.md
Original file line number Diff line number Diff line change
Expand Up @@ -48,34 +48,101 @@ bibliography: paper.bib


# Summary
Detecting earthquakes and compiling these to earthquake catalogs are fundamental tasks in seismology. Acoustic emission sensors allow detecting tiniest so called picoseismic events representing fractures on mm, cm or dm scale (-6<Magnitude<0). Such picoseismic events have corner frequencies of 1kHz-1MHz and cannot be handled by standard seismic processing softwares that deal with signals <500 Hz. Other commercial software for monitoring picoseismicity for structural health monitoring applications, e.g. in mines exists, but are only trigger-based. For large-scale experiments in underground laboratories (e.g., hydraulic stimulation, earthquake nucleation, nuclear waste disposal) continuous recordings of the seismicity data streams in MHz range are needed to study the rock response in great detail. The DUGseis software package is filling this gap. It was developed to manage, process and visualize continuous, high-frequency seismic data. The package can be used to create earthquake catalogs in real-time, as well as in post-processing, and directly visualize their event waveforms and locations in a graphical interface. Since the software is python based, users can easily add their own processing routines.
Detecting earthquakes and compiling them to earthquake catalogs are fundamental tasks in seismology.
Acoustic emission sensors allow the detection of tiniest so called picoseismic events, representing fractures on the mm,
cm or dm scale (-6<magnitude<0). Such picoseismic events have corner frequencies of 1 kHz-1 MHz and cannot be
handled by standard seismic processing softwares that deal with signals <500 Hz. Other commercial software
for monitoring picoseismicity (e.g. from GMuG - Gesellschaft für Materialprüfung und Geophysik) for structural health
monitoring applications, e.g., in mines, exists but are
only trigger-based. For large-scale experiments in underground laboratories (e.g., hydraulic stimulation,
earthquake nucleation, nuclear waste disposal), continuous recordings of the seismicity data streams in the MHz
range are required to study the rock response in great detail. The DUGseis software package fills this gap.
It has been developed to manage, process and visualize continuous, high-frequency seismic data. The package can be
used to create real-time and post-processing earthquake catalogs and to directly visualize their
event waveforms and locations in a graphical interface. As the software is Python-based, users can easily
add their own processing routines.

# Statement of need
The open source, Python-based DUGseis package is designed to align with the functionalities of SeisComP [@seiscomp], a standard software used in microseismic-large scale earthquake processing (<500Hz, M>-0.5). DUGseis is tailored to picoseismic events (-6<M<0) with much higher frequency ranges (kHz-MHz), as recorded by acoustic emission sensors (AE sensors). High-frequency seismic data processing is common in mining environments to monitor tunnel stability, and became very popular in underground laboratories. Until now, these projects have employed trigger-based recordings, meaning that the incoming waveform data is only saved to disk if a pre-set trigger threshold is reached by a recorded event. One disadvantage of the triggered recording strategy is the so-called dead time. After an event is triggered, no additional event can be triggered until the processing of the first triggered event is completed. The dead time can be a multiple of the recording time, meaning important events can be missed if another event happened just before. Removing these dead times plays a significant role if high event rates are expected. With DUGseis it is possible to record and store [@DugSeisAcqui] continuous waveform data in the MHz range and directly process the data, removing these dead times. Being Python-based, DUGseis offers a high flexibility for the researchers to complement the processing with their own Python-based codes, adjusted to the project needs.
The open-source, Python-based DUGseis package is designed to align with the functionalities of
SeisComP [@seiscomp], a standard software used in microseismic-large scale earthquake processing (<500Hz, M>-0.5).
DUGseis is tailored for picoseismic events (-6<M<0) with much higher frequency ranges (kHz-MHz), as recorded by
acoustic emission sensors (AE sensors). High-frequency seismic data processing is common in mining environments to
monitor tunnel stability and has become very popular in underground laboratories. To date, these projects used
trigger-based recordings, which means that the incoming waveform data is only saved to disk when a pre-set
trigger threshold is reached by a recorded event. A disadvantage of the triggered recording strategy is the
so-called dead time. Once an event has been triggered, no further event can be triggered until the processing of
the first triggered event has been completed. The dead time can be a multiple of the recording time, meaning that
important events can be missed if another event has just occurred before. Eliminating this dead time is important
when high event rates are expected. With DUGseis it is possible to record and store [@DugSeisAcqui] continuous
waveform data in the MHz range and process the data directly, removing these dead times. As DUGseis is Python-based,
researchers have the flexibility to complement the processing with their own Python-based
codes, adapted to the need of their project.

# Functionality and Features
The DUGseis software is a Python-based package with the main focus to process continuous high-frequency data, extract picoseismic event waveforms and create an earthquake catalog. To make its usage and its outputs more easily accessible for seismologists, some features use modules and functions of the ObsPy package [@Beyreuther2010], a popular package in seismology.\
The DUGseis software requires sensor metadata with sensor locations as StationXML files and continuous waveform data in the ASDF format [@Greenfield2015; @asdf] to run. Here, the continuous waveform data was acquired using DUGseis acquisition [@DugSeisAcqui], another software package developed specifically for the Bedretto Underground Laboratory to record and store waveform data of AE sensors in the ASDF format using specific Spectrum digitizer cards. For general usage of the here presented DUGseis package, the waveform data does not need to be acquired from DUGseis acquisition but the ASDF format can be transferred from other data formats. \
DUGseis retrieves all important information regarding data directories and processing settings from a configuration file. The configuration file can be used to open a graphical interface or can be given to the processing script. After processing the continuous waveform data, DUGseis outputs an event catalog as a database, which additionally can be saved as QuakeML or CSV files. Figure \autoref{fig:DUGseisScatch} shows the in- and output of the DUGseis processing.
The DUGseis software is a Python-based package that focuses on processing continuous high-frequency data,
extracting waveforms of picoseismic events and creating an earthquake catalog. To make its use and output more
accessible to seismologists, some features use modules and functions from the ObsPy package [@Beyreuther2010],
a popular package in seismology.\
The DUGseis software requires sensor metadata with sensor locations as StationXML files and continuous waveform
data in the ASDF format [@Greenfield2015; @asdf] to run. Here, the continuous waveform data was acquired using
DUGseis acquisition [@DugSeisAcqui], another software package developed specifically for the Bedretto Underground
Laboratory to record and store waveform data from AE sensors in the ASDF format using specific Spectrum digitizer
cards. For general use of the DUGseis package presented here, the waveform data does not need to be acquired
from the DUGseis acquisition, but the ASDF format can be transferred from other data formats. \
DUGseis retrieves all important information regarding data directories and processing settings from a
configuration file. The configuration file can be used to open a graphical interface or passed to
the processing script. After processing the continuous waveform data, DUGseis outputs an event catalog as
a database, which can also be saved as QuakeML or CSV files. \autoref{fig:DUGseisScatch} shows
the input and output of the DUGseis processing.

![In- and output of DUGseis processing. \label{fig:DUGseisScatch}](DUGseis_InputOutputScatch.png)

***Data processing***\
The processing itself occurs within Python scripts which call different DUGseis functions, enhancing flexibility in usage.\
DUGseis was mainly developed to process continuous waveform data recorded from AE sensors. @Villiger2020 used an initial version of the DUGseis software in the Grimsel Laboratory. The software was further developed for the hydraulic stimulation experiments performed in the Bedretto Underground Laboratory [@Obermann2024]. For this purpose, it was necessary to process data in real-time during the stimulations to evaluate the hazard potential of the ongoing stimulation and to gain a direct understanding of the fluid propagation in the rock volume. In addition, after the stimulation was completed, more detailed post-processing was needed to learn more about the rock-volume response. Both, real-time analysis and post-processing, can be done using the DUGseis package. During the live processing, directories are monitored in real time for new incoming data. For the post-processing, all directories with stored continuous waveform data are given to the processing script. This continuous data is then again processed in playback, which is a big advantage compared to the trigger-based software, where only the waveforms of the already triggered events could be revisited. \
The DUGseis package includes many typical seismological processing steps to create earthquake catalogs. The steps can include a detection stage to select event candidates on a number of pre-defined sensors. This step can be useful to speed up the processing, especially in real-time. Other processing steps are the picking of all traces with different pickers (e.g. STA/LTA), locating events with a basic location algorithm and magnitude estimations, which for now are based on acoustic emission sensors. For some of the processing steps, several methods are implemented, that the user can choose. Furthermore, all processing steps can be adapted to the researcher's needs. For more detailed information on which steps were used in the Bedretto Underground laboratory, we refer to @Obermann2024.\
After the processing is finished, the outputs of the events, included arrivals and picks, are stored in a database and can also be saved as QuakeML or CSV files.
The processing itself is done within Python scripts that call various DUGseis functions, increasing flexibility
of use.\
DUGseis was mainly developed to process continuous waveform data recorded by AE sensors. @Villiger2020 used
an initial version of the DUGseis software at the Grimsel Laboratory. The software was further developed for
the hydraulic stimulation experiments performed at the Bedretto Underground Laboratory [@Obermann2024]. For
this purpose, it was necessary to process data in real-time during the stimulations in order to assess the hazard
potential of the ongoing stimulation and to gain a direct understanding of the fluid propagation in the rock
volume. In addition, more detailed post-processing was required after stimulation to learn
more about the rock volume response. Both, real-time analysis and post-processing can be carried out using the
DUGseis package. During the live processing, directories are monitored in real time for new incoming data.
For the post-processing, all directories containing stored continuous waveform data are passed to the processing
script. This continuous data is then re-processed during replay, which is a major advantage over
trigger-based software, where only the waveforms of the already triggered events could be revisited. \
The DUGseis package includes many typical seismological processing steps to produce earthquake catalogs. The
steps may include a detection stage to select event candidates on a number of predefined sensors. This step
can be useful to speed up the processing, especially in real-time. Other processing steps are the picking
of all traces with different pickers (e.g. STA/LTA), the localization of events with a basic location algorithm and
the estimation of magnitudes, currently based on acoustic emission sensors. Several methods are implemented for
some of the processing steps, whcih can be selected by the user. Furthermore, all processing steps can be adapted
to the needs of the researcher. For more detailed information on which steps were used in the Bedretto Underground
laboratory, see @Obermann2024.\
After processing, the output of the events, including arrivals and picks, are stored in a
database and can also be saved as QuakeML or CSV files.

***Graphical Interface***\
Another functionality of DUGSeis is to visualize the recorded waveform data and to allow manual repicking and relocation, both in real-time and in post-processing. Within the graphical interface, not only waveforms are displayed but also the output of the processing stage, such as origin and picktimes of an event. Additionally, the event and sensor locations are shown in a 3D visualization. Figure \autoref{fig:GUI_example} shows the layout of the graphical user interface.\
The graphical interface provides the opportunity to inspect each event and display the channels that recorded the event. Additionally, manual repicking and relocating can be done here.
Another feature of DUGSeis is its ability to visualize the recorded waveform data and to allow manual repicking and
relocating, both in real-time and in post-processing. The graphical interface (GUI) displays not only waveforms, but also
the output of the processing stage, such as the origin and pick times of an event. Additionally,
the event and sensor locations are shown in a 3D visualization. \autoref{fig:GUI_example} shows the
layout of the graphical user interface.\
The GUI provides the ability to inspect each event and view the channels that recorded
the event. In addition, manual repicking and relocating can be done here.

![Graphical interface of GUI with waveforms and picks of an event and the 3D plot of the events in the database. \label{fig:GUI_example}](DUGseisGUI_V2.JPG)
![Graphical interface of GUI with waveforms and picks of an event and the 3D plot of the events in the database. \label{fig:GUI_example}](DUGseisGUI_V3.JPG)

# Usage
Since 2021 dozens of hydraulic stimulation experiments have been performed in the Bedretto Underground Laboratory in Switzerland [@Ma2022; @Plenkers2023; @Obermann2024]. In this context, the DUGseis package was used to detect picoseismicity by processing the incoming high-frequency waveform data in real-time and in post-processing mode [@Obermann2024]. A small seismic dataset of one of the hydraulic stimulations can be found in [@DUGseisExample].
Since 2021 dozens of hydraulic stimulation experiments have been performed at the Bedretto Underground
Laboratory in Switzerland [@Ma2022; @Plenkers2023; @Obermann2024]. In this context, the DUGseis package
was used to detect picoseismicity by processing the incoming high-frequency waveform data in real-time and
in post-processing mode [@Obermann2024]. A small seismic dataset from one of the hydraulic stimulations can be
found in [@DUGseisExample].

# Acknowledgements
The submitting author, Martina Rosskopf, is funded by SNF Project “Characterizing and understanding Enhanced Geothermal Systems (EGS) - novel tools and applications in a deep underground laboratory” (200021_192151). The BedrettoLab is financed by ETH Zürich and the Werner Siemens Foundation. This paper is BULGG publication BPN_023.
The submitting author, Martina Rosskopf, is funded by SNF Project “Characterizing and understanding Enhanced
Geothermal Systems (EGS) - novel tools and applications in a deep underground laboratory” (200021_192151). The
BedrettoLab is financed by ETH Zürich and the Werner Siemens Foundation. This paper is BULGG publication BPN_023.

# References

0 comments on commit 2896f7f

Please sign in to comment.