Ground Response in B-scan from Uneven Surface Simulation is Even

[Ismael Xique]

I have run the following simulation:

`
#title: gprMaxSim 
#num_threads: 12 
#domain: 1.00 1.00 0.70
#dx_dy_dz: 0.00500 0.00500 0.00500 
#time_step_stability_factor: 0.980 
#time_window: 1.564e-08 
#material: 4.0 0.01 1.0 0.0 ground 
#material: 16.0 0.01 1.0 0.0 target

#waveform: ricker 1.00 900000000.00 wave_src

#pml_cells: 10

#python:
tmp = current_model_run-1; 
stepSize = 0.02
offset = 0.025000 
print( '#hertzian_dipole: z {} 0.500000 0.600000 wave_src'.format( 0.200000 + stepSize * tmp ));
print( '#rx: {} 0.500000 0.600000'.format( (0.200000 + offset) + stepSize * tmp ) );
#end_python:

#fractal_box: 0.0 0.0 0.0 1.00 1.00 0.50 1.5 1 1 1 1 ground my_soil_box 1
#add_surface_roughness: 0 0 0.5 1.00 1.00 0.50 1.5 1 1 0.4 0.55 my_soil_box 1
#box: 0.45 0.45 0.20 0.55 0.55 0.30 target

#geometry_view: 0 0 0 1.00 1.00 0.70 0.00500 0.00500 0.00500 heterogeneous_soil n
`

It is supposed to produce a single B-scan with an uneven ground response, but after combining all the A-scans together the B-scan continues to have an even ground response. Does gprMax realign the A-scans to make the B-scan or am I messing up somewhere? I have attached images of the geometry and B-scan. Any help will be greatly appreciated.

[Craig Warren]

Hi Ismael,

I don't see anything immediately obvious in your input file that could cause this. Can you confirm a couple of things:

• What version of gprMax are you using?
• What tool did you use to merge the A-scans?
• Did you use the supplied MATLAB script to plot the B-scan?

I ran a B-scan of a simulation with a fractal soil, rough surface, and antenna model the other day and could see the effects of the rough surface. So the only significant difference I see between these models is that you are using a homogeneous material with a rough surface.

Kind regards,

Craig

[Ismael Xique]

Thanks for the quick reply. 

I am using version 3.0.8 

I used a custom matlab function that simply appends the a-scans, but I have also used the python code provided to merge and plot B-scans. I have included the results below. 

[Antonis Giannopoulos]

Hi Ismael,

We did have a more careful look with Craig on this and it looks like the problem should be the fact that you are looking at a vertical dipole (z) response over the interface. This should be rather insensitive to the target and the small undulation of the surface so the output is not unreasonable but it is interesting to see why. In any case for your problem most likely you want an "x" or "y" oriented dipole and corresponding output point.

So, I think 

print( '#hertzian_dipole: y {} 0.500000 0.600000 wave_src'.format( 0.200000 + stepSize * tmp ));

might do it.

Best

Antonis

 

[Craig Warren]

Hi Ismael,

A couple of further observations:

• A 5mm spatial resolution is getting a bit close to having numerical dispersion in the model. At 5mm the max. freq that can be represented is ~1.5GHz, so I'd recommend using 4mm or even 3mm.
  
• The reflections from the ground and target are going to be very small in amplitude compared with the direct wave from transmitter to receiver. 
  
• You can see in the attached A-scan (Ey.png), which is directly over the target, that with default plot scaling you only see the direct wave. By zooming in (Ey_zoom.png) you can see the reflection from the top (green highlight) and bottom (blue highlight) surfaces of the target box. 
  
• The reflection from the ground is combined with the direct wave (due to your geometry) so not clearly separable.
  
• The response between ~10ns and ~12ns is likely multiples of reflections between the ground surface and target, and top and bottom surfaces of the target.

Kind regards,

Craig