GMSYS 3D Structural Inversion/Upward Continue Question
When performing a 3D Structural Inversion of gravity data (original data was collected over a land area of varying topography), do I need to upward continue my Observed Gravity Grid to be the same elevation as the highest survey elevation? What would happen if I set up the Calculation Elevation of my 3D model to a constant elevation equal to my highest surveyed elevation, but did not upward continue my observed gravity grid to match that same constant elevation? Would I be implying that my observed gravity anomaly is greater than what I have actually measured if I did not upward continue but still used the highest survey elevation as my calculation elevation?
Answers

Hi Nick,
Thanks for the great question!
There are a few different ways you could go about this which we could discuss in detail, however, this would be my approach (assuming you have access to the 2D Filtering extension, and Oasis montaj 2023.1.1).
It is important to compare the response of your model (Calc) with your survey response (Obs) that are collected at the same reference elevation.
When setting up your Gravity survey, you also set up the Survey Elevation. Here I would set the original survey elevation, if it is a ground survey it would generally be Topography+~1 to accommodate for the height of your gravity station above topo.
If you are using the upward continued grid as the input survey data, you can apply that same filter .con file on the forward computation process (see attached image). The resulting forward response will then have the same filter applied as the input survey data. This is my preferred route because it retains the original survey spec and removes some human error when recalculating the correct reference level to go along with your upward continued grid.
I'll attempt to answer each question:
 "do I need to upward continue my Observed Gravity Grid to be the same elevation as the highest survey elevation?"
Not necessarily. I would recommend using the original survey elevation in this scenario. If you did want to go down this route to run the calculation on a flat surface (which may save some computation time), there is a "Draped to Level" tool available through the GMSYS 3D Tools menu.
 "What would happen if I set up the Calculation Elevation of my 3D model to a constant elevation equal to my highest surveyed elevation, but did not upward continue my observed gravity grid to match that same constant elevation? "
If the topography is highly variable you will notice differences in the frequency and magnitude of your calculated and observed responses. To ensure you're comparing apples to apples, it would be best that the model response is calculated at the same reference elevation as your survey was collected.
 "Would I be implying that my observed gravity anomaly is greater than what I have actually measured if I did not upward continue but still used the highest survey elevation as my calculation elevation?"
Yes, in areas where the survey elevation was raised the observed gravity anomaly would be 'higher' than what you would have measured at the new, higher height.
Hope this helps clear things up Nick! let me know if you have any followup questions.
0 
Josh,
Thank you so much for the thorough answer. It has given me some things to think about and try for sure. I am working on a very dated version of the software and one that might not have all the functionality of the latest release.
But just to clarify, in the latest version of GMSYS 3D, would I be able to perform a structural inversion on my observed gravity grid (residual CBA) using a varying elevation grid (the elevations from my gravity survey) to invert for depth to basement? In the version of GMSYS 3D that I am familiar with, I have always needed to set a constant elevation (equal to just higher than my highest survey elevation) and drape to level my residual CBA to upward continue to that constant highest elevation. If I didn't do that, I would receive an error stating "Inversion is not implemented for draped target surface." Again, that might be a limitation of my license/version of Oasis Montaj GMSYS 3D.
Is there any reason that the forward calculation or structural inversion routines would need to have a flat datum in order to run, and not simply just use the varying topography on which the data was collected in the first place?
0 
Hi Nick,
No worries, the product team just told me that we rewrote a lot of GMSYS 3D around 20122013, so it is likely there is quite a bit of added functionality (in fact, I did not even know where to find the initial screenshot you sent through!)
In the latest version, you can indeed run a structural (or physical property) inversion and forward model using a draped (varying elevation) grid as the input survey elevation. I think this was implemented in the 2012 rewrite.
I understand it was likely a limitation on computation. The draped calculations are significantly more computationally expensive than the flat surface calculations, and only implemented about 10 years ago.
It would be good to show you the new tools so you can see all the recent advancements in GMSYS 3D.
0