I have RTK field data that I am using to calculate the volume of some landslide deposits. I think because I am recording an orthometric height, not a simple depth from the surface, I am running into some issues in calculating the volume.

What I've been doing is making a TIN from the surface points and a TIN from the depth points. I then convert those each to a raster, adjusting the number of observations as needed. I use raster calculator to subtract the surface from the depth, and get a difference raster.

If I just take the sum of the raster values and multiply it by the cell area, my volumes are generally too low. I found a solution which involves using the difference raster in the SurfaceVolume tool, and I can't seem to wrap my head around exactly why it works.

My solution is to take the number of cells in the difference raster (count) - sum of values in the difference raster. What I can't seem to justify or explain next is that if I divide this number by 1000 and enter this as my reference plane height in the SurfaceVolume tool, I get a volume (below the plane) that is virtually perfect to what is expected for the site.

Why do I need to divide by 1000? Count is unitless, sum would be meters. So count-sum is a value in meters. The input coordinate system for my difference raster is UTM, so the reference plane height is being read as meters. If I don't divide this by 1000, my volumes come out in the millions of cubic meters (which is not accurate at all). Is there anyone that has any ideas of why this seems to work? I am also keeping my z factor at 1.

An example with numbers:

Difference raster sum: 2406.5

Cell area = 4 m3

Sum * cell area = 8046.3 (reasonable but low).

or

Difference raster count: 3192

3192 - 2406.5 = 785.5

785.5/1000 = 0.785 or 0.8

if I put the difference raster in as my surface and I calculate volume below the plane height of 0.8, my volume is 11660, much more realistic.

if I leave the plane height as 785.5, my volume is 11374824. Massive!!! (again though if you divide this by 1000 it seems to be reasonable...?)

So why do I need to divide 785.5/1000 when it should already be in meters? What is the SurfaceVolume tool actually doing that makes this work?