Novel Airborne EM Image Appraisal Tool for Imperfect Forward Modeling

Round 1

Reviewer 1 Report

Dear authors, I was very interested to read your research. I think you have suggested a useful tool for checking the quality of an inversion.

 

However, if you don't mind, could you explain in the final version why the observed and calculated HM data always have a mismatch in the last channel, even in synthetic cases, even quite far from the "step"?

Best wishes

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 2 Report

After going through this paper, I find that the technical content of this paper is not sound. The following are my comments.

1.      I simply cannot understand what the goal of this paper is. Do you want to do the inversion? Or do you want to evaluate the inversion results? Or do you want to find the survey sites where multi-dimensional inversions are inappropriate?

2.      I  cannot understand whether you are dealing with 1D, 2D, 2.5D, 3D, or multi-dimensional problems? Do you assume a layered model, or a 2D model? Why do you use a 3D mesh? Do you use 1D modeling method or 2.5D modeling method?

3.      In the field data example, there are large gap and chaotic areas or blurred results, what can we learn from these results?

4.      All figures have not been clearly explained either in the text or in the figure captions. Do all these figures truly support the points and conclusions of this paper?

5.      The 1D inversions in AEM is pretty useful, as it is quick and working in most cases. For the cases of rugged terrain or complex underground structures, the existing 3D modeling and inversions are also working well. With limited computation condition, one can run 3D AEM inversions. Thus, I don’t think the study in this paper by only introducing a normalized gradient will be of any help to the interpretation of AEM data.

In summary, I  have to reject this paper for lacking of technical content. 

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 3 Report

The manuscript presents an appraisal tool for analysing quasi-2D and 3D AEM models. The method is based on a so-called normalized gradient which is calculated using a multidimensional forward modelling. The authors argue that two forward models can be used in the calculation of the appraisal tool: the first, the high-fidelity (HF) model uses a full 2.5D approach with a fine mesh; the second one, the medium-fidelity (MF), uses a coarse mesh in forward calculations. The HF model is computationally very demanding, so the authors pretend that the use of MF models is enough for the appraisal objectives.

The manuscript has potential interest for the users of AEM. At my knowledge it is the first time an appraisal tool like this is presented. So, there a merit of innovation in the manuscript. My main questions concern the “definition” of the MF model an its errors and limitations. What is a coarse mesh? In general, even for HF models, the discretization of the subsurface is not the same over all volume of interest. It is denser in the areas with large resistivity contrasts and less dense (coarse) in areas with less resistivity contrasts. The large contrast zones are the areas of interest in this work. What happen if we use a coarse mesh here? The calculations will be wrong, the gradient will be wrong, and the results will be not good. What are the limits of the “coarse” mesh. I guess that a general definition is not an easy task. Anyway, this problem should be address by the authors.

Although the method can be relatively “cheaper” in terms of computation it still is very expensive for most of the EM (ground or airborne) users. 3D and 2D models (even when free) are not commonly used. I hope this work can contribute for a more utilization of those tools.

There are also some minor questions:

-Fig1 presents results of a “discrepancy principle”. Such principle is not mentioned (defined) in the text.

-it is not clear (at least for me) if the authors used 3D modelling or 2D modelling in all calculations. This should be clearly expressed.

 

It is my opinion that the manuscript can be accepted after a revision.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf