search

yewsg / yews

Yews | Your Earthquake Waveform Solution
Apache License 2.0
14 stars 12 forks source link

It becomes problematic when the dynamic range of cfs gets too big #9

Open Lchuang opened 4 years ago

Lchuang commented 4 years ago

https://github.com/lijunzh/yews/blob/a3653f9d29cbeb257bdc28019ab7fbba365dec94/src/yews/cpic/picking.py#L9

lijunzh commented 4 years ago

Sorry to get back to your late. My spam filter put all the github notification into junk.

I agree with you that this will result in some missing detection when the dynamic range of large event overwhelms that of the small events. I left it in the current form since I was going to use picking.py after detection.py. This will mitigate the problem as picking is only doing in a smaller time windows. However, there needs to be better solutions. Two possible solutions I can think of.

  1. Have the range(10) being adadptive to the absolute value of the dynamic range instead of the relative value of that. This will requires some user input since CPIC will then need to know the lowest legitiment event dynamic range.

  2. Trigger increase of that 10 to larger values if we see the peak prominance exceed some predefined threadhold. I need to do some epxeriments to see what will be a reasonable value to replace 10.

If you have any good testing dataset, please share it with me so that I can some reference waveforms to work with.