The TMTS clipThreshold parameter was set to 8*10^6 on 01/09/2023 for the real-time TMTS on ucbrt ( /home/ncss/run/params/tmts.cfg ) and also it was set to 2^30 on 01/04/2023 for the web based TMTS on rodgers and ucbpp ( /home/ncss/run/tmts/conf/tmts.cfg , tmts1.cfg, and ect)
# Clip threshold: maximum absolute value of raw trace data above which data is# considered unusable. 8388608 = 2**23, maximum counts on 24-bit digitizer# original value#clipThreshold 6710886.0# 01/04/2023 - 2^30clipThreshold 1073741824.0The reason of the parameter modification was that seismologists noticed that data was not always available in the moment tensor interface when they was computing a moment tensor.
Removing station data was confirmed by the following log messages:
Peggy proposed to increase the clip threshold. Below is her email
Hi Folks,
after I looked at one of the quakes in the Ferndale sequence, Andrei wrote to ask me about the clipping limit in TMTS. Currently, both the automatic system and the review system are set to exclude stations that have more than around 6 million counts on one of the component. This is the reason that quite a few stations have been excluded in the TMTS interface when we have tried to work on events in the recent sequence.
Please let me know what you think of my suggestion at the end of this email. There are a variety of factors that impact the value that we should choose for the clipping threshold.
- Sometimes BB sensors go non-linear: This is correct, however, it is more a question of high frequencies, than of the absolute value of the signal. And this is likely to happen close to an event (where the definition of “close” depends on the magnitude of the event). In general, most BB sensors we have installed actually have a higher dynamic range than the limits of the 24-bit digitizer.
- The actual maximum number of counts for a 24-bit digitizer is 8,388,608 counts.
- Some of the digitizers (Q330HR and Q8) can actually do 26-bit (or 25-bit).
- The “intelligence” of the automatic system is not as capable as our reviewers.
So, here is my suggestion: Automatic TMTS: we set the clipping threshold to 8,000,000 counts. This system has a set of distance ranges for choosing stations that will hopefully prevent it from choosing stations that might have gone non-linear. We have a set of stations that it will automatically exclude (Taka has been managing this list), because they are noisy.
Manual TMTS review system: Here, we set the clipping threshold to 31-bits (effectively, no clipping threshold). In this case, it is very important for the reviewers to exercise their intelligence – look at the data to decide if the station is acceptable for use. The capability to plot the raw data and the filtered data exist in TMTS. Looking at these data are an important step in the review process.
Please let me know what you think,
Peggy
Here is an example from Doug Dreger how data looks when BB sensor go non-linear: hops_example.pdf