I am working with inversion of magnetic data and one colleague of my supervisor has said this about subtracting the IGRF for inversions:
"””
He needs to subtract a value from them to make the dataset closer to 0 centred. The inversion is very sensitive to that number. If you get it wrong it will put material in the padding or at the bottom of the model to compensate.
We find dipole anomalies that appear to be only induced and gauge the zero level based on a synthetic model at that latitude.
If he needs a quick synthetic modelling package we use pblock - written by the same block who writes potent. Its free and runs on Windows and Wine. https://www.geoss.com.au/downloads.html
The average rarely cuts it, you'd likely get closer by using the IGRF value but still not close enough.
They'll need to estimate it more closely. If they are a geophysicist then just tell them to estimate the zero level for the anomaly and they should know what to do if they have been taught properly. If they are a data scientist pretending to be a geophysicist have them generate a synthetic model for a simple body at the same magnetic latitude as the area to be inverted. Measure the amplitude of the high and the low to get the ratio of positive/negative.
Now find a dipole which appears to only have an induced component and measure the range between the high and low and using the ratio from the synthetic model estimate zero. Do it on as many dipoles as possible to get a feel for the average value of zero.
"""
This guy just told me that you can find this method in any geophysics textbook. However, I am unable to find this technique in the literature and he did not develop any further his suggestion (also my background is in geology). Any ideas or suggestions of publications that explain what is he talking about? I looked in textbooks such as Hinze, Dentith and Mudge, Blakely, Isles and Rankin, etc, and could not find it.
I was first doing the subtraction of the average value from each magnetic observation to get the data zero centred. I know that I can remove the IGRF from a software process packaging, like Oasis Montaj, but it seems that he is saying to first create synthetic models to then remove the regional signal, which I am unaware of.
Thank you in advance.
