BRUIT-FM challenge dataset

This dataset contains synthetic and real seafloor seismological data with noise at frequencies below 0.1 Hz.  The BRUIT-FM Challenge asks participants to remove as much noise as possible in order to 1) obtain the cleanest waveform possible for earthquake identification and evaluation; 2) obtain the cleanest waveform for seafloor compliance calculation; and 3) obtain the best possible estimate of seafloor compliance: the ratio of seafloor motion to seafloor pressure at frequencies between approximately 0.003 and 0.03 Hz.

The dataset consists of two "challenge" datasets (one real, one synthetic), one "main training" dataset (synthetic) and 100 'mass training" datasets.  Each training dataset provides "solution" output waveforms and compliances.