search

daniel-koehn / DENISE-Black-Edition

2D time-domain isotropic (visco)elastic FD modeling and full waveform inversion (FWI) code for P/SV-waves
GNU General Public License v2.0
121 stars 66 forks source link

L5 (global correlation norm) misfit is negative. How to normalize it? #36

Open pplotn opened 2 years ago

pplotn commented 2 years ago

Hello, I am using the L5 or (global correlation) norm. As mentioned in the manual its implementation follows (Application of multi-source waveform inversion to marine streamer data using the global correlation norm, Choi,Alkhalifah, 2012).

Originally it is negative.

I have a question: how to normalize it?

I attached the misfit file. And plots of the misfit functions on attached picture.

seis_fwi_log_L5.txt marm_PICTURE