Is it possible to get the maximum and minimum values of Normalized Misfit are larger than the Observed or Predicted Gravity anomaly values?

[sant_63]

Hello, everyone

Myself Santosh Kumar, PhD Student at Department of Applied Geophysics, IIT(ISM), Dhanbad, India.

I congratulate to all SimPEG Contributors. SimPEG is amazing !! I am using SimPEG since few months ago.

I have a question about Sparse Norm Inversion of Gravity Anomaly Data on Tensor Mesh (Sparse Norm Inversion of Gravity Anomaly Data — SimPEG 0.15.2 documentation).

Is it possible to get the maximum and minimum values of Normalized Misfit are larger than the Observed or Predicted Gravity anomaly values?.

#1. The upper figure correspond to the area with 27 km length in horizontal direction, and 29 km length in vertical direction. The range of normalized misfit values are more than +40 to -40. While the range of observed or Predicted Bouguer Gravity anomaly values lie in between +10 to -10 mGal.

#2. The lower figure correspond to the area with 60 km length in horizontal and vertical directions. The range of normalized misfit values are more than +20 to -20. While the range of observed or Predicted Bouguer Gravity anomaly values lie in between +30 to -30 mGal.

[lheagy]

Hi @sant_63,

Thanks for your question and apologies for our late reply!

This looks like an interesting example. For the normalized misfit, are you dividing by the true data? If so, you would need to be careful with regions where this is close to zero, as dividing by zero (or a small number close to that) will cause the value to increase substantially. It might be more straightforward to just look at the misfit, without normalizing.