September 29, 2020, 09:38:46 am

Author Topic: GPR autoleveling method A and method B?  (Read 4558 times)

William Johnston

  • Moderator
  • Member
  • *****
  • Posts: 9
GPR autoleveling method A and method B?
« on: February 04, 2011, 11:07:25 am »
The auto leveling step in the GPR processing assumes that it is desirable that the ground level be at exactly the same point in each sounding. And it assumes that first big wave has something to do with the location of the ground in the signal. So what both methods try to do is to line up the first big bump in the sounding.

Why are there two methods?
For the data sets I had, I couldn't create one algorithm that worked across all of them.  I am told that both methods will still fail on some data sets.  If this is happening to you and you send me your data set I'll probably create a Method C.

What exactly is going on?

For both method A and method B
The first thing I do is to find the average sounding across all the profiles and smooth it.
I then find the maximum value off this smoothed averaged sounding. This should correspond to the peak of the largest value.
I then find the first point that is 10 percent of the maximum value off the average value.  Hopefully this is a point along the rising slop of the first big peak.
This point is “match point” which I try to adjust all the other soundings to.

Method A

For each sounding I find the first point that matches or exceeds the value of the match point. This is the “key point.” 

Method B

For each sounding I first smooth it. Then find the first point on the smoothed sounding that is 10 percent of the smoothed averaged sounding’s maximum value off this soundings average value. This is the “key point.”

I then adjust the soundings so that the key point locations matches the match point location.
« Last Edit: February 07, 2011, 11:33:04 am by William Johnston »

Eileen Ernenwein

  • Moderator
  • New Member
  • *****
  • Posts: 1
    • Eileen's Staff Page
Re: GPR autoleveling method A and method B?
« Reply #1 on: February 04, 2011, 01:29:27 pm »
You only need to use auto-level if the scans (traces) don't seem to line up well. I often skip this step but sometimes it is really helpful. I have some examples that show slices of data that were not auto-leveled, and you can see visible breaks where clearly there was some drift or something changed to make the scans jump up or down (in time). If you run auto-level first then the slices are much cleaner. 

See attached image showing some GPR data in the GPR Loader that is much improved using Method A.
Eileen G Ernenwein, PhD
Research Associate
Center for Advanced Spatial Technologies
University of Arkansas, Fayetteville, USA
& Adjunct Faculty
East Tennessee State University
office: 423-439-7655


* Users Map

SimplePortal 2.3.3 © 2008-2010, SimplePortal