Skip to the content.

The Noise Reduction Challenge

We will propose two datasets as a challenge to all researchers to 1) minimize non-seismological noise and 2) calculate the seafloor compliance (seafloor motion divided by pressure as a function of frequency). The first dataset consists of true seafloor data recorded on the Mid-Atlantic Ridge, the second of synthetic data. For the first dataset, we will show our processing and calculated results. The second is a blind test.

All researchers are invited to process these data and send us their results. All participants will be invited to be co-authors of a community paper comparing the different methods and results.

The datasets

ARC-EN-SUB station 8

8 days of data, sampled at 1 sps, from near the RAINBOW hydrothermal field. Lots of earthquakes and a relatively weak infragravity wave signal make this a hefty challenge. Data, our processing codes (using tiskitpy and the bruit-fm toolbox and results are here.

Here are plots of the data and our results: can you do better?

run_obspy.py

Waveform plot

Waveforms

Probabilistic Power Spectral Density

Waveforms

run_tiskitpy.py

Waveforms (original, rotated, and rotated + transfer function noise removal)

Automatic Waveforms

Power spectral densities of the above three waveforms

Automatic PSDs

Pressure-acceleration coherence of cleaned data

Automatic Coherence

Compliance of cleaned data (amplitude problem, probably using COUNTS)

Automatic Compliance

Cheating!

We get a better result if we manually identify glitches and other anomalous noise:

Waveforms

Manual Waveforms

Power spectral densities

Manual PSDs

Pressure-acceleration coherence

Manual Coherence

Compliance (amplitude problem, probably using COUNTS)

Manual Compliance

Synthetic data

Coming!