3D CSEM inversion using both L2 and non-L2 measures

What is it about?

1. We developed a new 3D controlled-source electromagnetic inversion algorithm that allows general measures to be made selectively available. 2. The behaviours of general measures are examined focusing on weighting mechanism and user-defined parameters. 3. Our findings highlight the seamless integration of general measures, contributing to improved robustness against data outliers and enhanced spatial properties provided in output models.

Featured Image

Why is it important?

We wanted to show that even beginners in inversion can easily use non-L2 measures with just a small tweak to existing L2-based methods—no extra hassle required. We also aimed to clarify that outliers, where non-L2 measures shine, refer to extreme values in data fitting and sharp changes in model constraints. To make this approach immediately usable, we included practical guidance on parameter selection.