Skip to the content.

WP4: Seafloor Signal Separation and Noise Removal

This WP applies advanced signal processing techniques to separate signal/noise sources and tests the use of a new rotational sensor.

Objectives

Increase sensitivity to low-frequency seismological signals (normal modes, teleseisms…)

Tasks

Number Title Responsable Status Project Management
T4.1 Reducing horizontal noise using a rotational seismometer @Crawford, Frederic Replaced by T4.5  
T4.2 Signal separation/removal techniques Ker, @Crawford completed  
T4.3 Separating seismological and biological signals Ker, Duval not started  
T4.4 Noise reduction challenge @Crawford, Duval, Ker underway  
T4.5 Rotational seismometer specification @Crawford underway  

Deliverables

Number Title Due Date Status
D4.1 Report on rotational seismometer integration in BBOBS M12 underway
D4.2 Open source software for noise separation and removal M24-48 Completed: Crawford, Amininan, Rebeyrol
D4.3 Catalog of seafloor noise sources M36 not started
D4.3 Scientific articles M24-48 Rebeyrol, Aminian

Details

Task 4.1: Reducing horizontal noise using a rotational seismometer

The task was planned to reducing seafloor horizontal noise levels using the iXblue blueSeis-1C rotational seismometer. iXblue was sold, renamed as eXail, and the new management removed the bleuSeis division. Task 4.5 replaces this task.

Task 4.2: Signal processing techniques for signal separation and noise removal

This task will be run in collaboration between IPGP and IFREMER with the support of a postdoctoral researcher specialised in signal analysis/processing and a broad group of signal processing experts from ESIEE, iXBlue, GEO3BCN and GIPSA-Lab.

Task 4.3: Separating seismological and biological signals

We will develop an approach to separate simultaneous seismological and whale call signals in the shared frequency band around 20 Hz, including the challenging chorus footprint (Bouffaut et al., 2018). We will use recently developed signal deconvolution/restoration techniques (SPOQ) using sparse non-convex norm-ratio penalties (Cherni et al., 2020) to characterise the overlapping signals using robust statistical measures (moments and moment ratios) to enhance their differences and assist their separation.

Task 4.4: The Bruit-FM open data noise reduction challenge

Proposed in mid-project, the challenge asks researchers to reduce noise on real and synthetic datasets, and to send their results to us. Once the results are received and rated, a community workshop will be held to compare the different methods and to draw conclusions about the most effecient existing methods and pathways for future improvements. Challenge and training datasets are available on Zenodo, and an article describing the Challenge has been submitted to Seismica.

Task 4.5: Rotational seismometer specification

After the restructuring of eXail in 2023 led to the abandonment of their BlueSeis rotational seismology division, this project decided to redirect the funds, originally planned for the addition of a BlueSeis rotational seismometer to an existing broad-band OBS, to the design of a BBOBS with rotational seismometer with enhanced rotational sensitivity, so that the seafloor instrument could have comparable horizontal channel noise levels to a buried instrument.