Shallow false detections about gamma HOT 5 CLOSED

Just for completeness, relocation indeed did wonders. Below is the output after using NonLinLoc on the outputs (and slightly increasing the threshold in the number of picks). Mind the different scaling of the y-axis and colorbar compared to the previous plot.

from gamma.

Hi Jannes, thanks for sharing the updates! My current test examples are quite shallow. If possible, could you contribute this example to the repository? I think others may also be testing deep earthquakes. I forgot to mention another thing you can try is the "initial_points". The default is [1,1,1] meaning the number of initialization points along x, y, z. You can try adding more points along z axis. You can pass other values to the config dict.

from gamma.

Hi Jannes, I agree this situation is common in association. One reason is the location in association is not accurate, and I initialize all event locations on the surface:

from gamma.

Hi Weiqiang,
thanks for the quick reply! I was anyhow planning to do relocation, so let's see how it'll look afterwards. I'll close the issue for now but I might update it once I got around to do relocation.

from gamma.

It's actually the same example as in the SeisBench example notebook, just for a longer time period (1 month). The NonLinLoc integration is pretty messy, so I can't really share that part for now. I'll also try playing around with further points in z direction. Thanks for the hint!

from gamma.