Attending: Daniel, Gary, Pat, Ross, Victor, Vuk, Andrew, Levi, Michael
Agenda:
Next Meeting: May 4
Data analysis topics
Data corrections, frame production (Gary/Daniel/Pat)
Daniel got data from Gary, now Pat is copying the data at UMN. Pat plans to start making frames in the coming week. Daniel has the complete dataset, checked a few teleseisms and the data look correct.
Event catalog (Gary/Ross)
no news, but will try to make a nice figure for the paper.
Investigated SVD approach to inverting the gamma-matrix in the radiometer algorithm. It offers a way of trading off the uncertainty in the recovery amplitude with the uncertainty in the direction.
Gary: could use the matrix norm as a measure of its size/amplitude. This number tells us how the variance changes as the spot size changes. This would quantify how the tradeoff between amplitude and direction uncertainties happens.
Will try to repeat the study using different wave types, so far used only P-waves.
Daniel/Gary/Pat: this only includes the errors introduced by the matrix inversion, which depends only on the gamma-factors, i.e. data is not included.
Repeated the studies of noise variations over the entire data set, now that it is available. Find similar trends to Tanner's, horizontal directions on the surface are much noisier due to tilt noise (weather, temperature etc). Also find that all channels on the surface are noisier than underground, due to anthropogenic sources. Observe a bump at 2 Hz in all channels, with some variation over time - not sure what causes this.
Also looked at a teleseismic event in more detail: using N-1 stations to predict the incident waveform, and then compute the expected waveform at the N-th station and compare with the observed waveform at the N-th station. Assume simple reflection off of a plane, no coupling P-S waves, assumes known isotropic velocity etc. Still gets a very good agreement, but can improve it in multiple ways.