search

emanuelhuber / RGPR

Ground-penetrating radar (GPR) data processing and visualisation: a free and open-source software package (R language)
http://emanuelhuber.github.io/RGPR/
162 stars 49 forks source link

Possible misunderstranding in the tutorial #35

Closed wilhem closed 4 years ago

wilhem commented 4 years ago

Thank for your reply. That's clear the question.

But I found another possible misunderstanding regarding the function dewow() In the wiki it is reported that:

Dewow

Remove the low-frequency components (the so-called “wow”) of the GPR record using:

a running median filter (type = "runnmed")
a running mean filter (type = "runmean")
a Gaussian filter (type = "Gaussian")

For the two first cases, the argument w is the length of the filter in time units.

But looking at the help

?dewow

I get:

w [numeric(1)] If type = runmed, MAD or runmean, window length of the filter in trace unit; If type = Gaussian, standard deviation in trace unit. If w = NULL, w is estimated as five times the wavelength corresponding to the maximum frequency of x (estimated with spec)

Now it is not clear to me what w should be.

Let's say I have in my function w = 5 (time unit [ns]). Does it mean that the window length of the filter goes from 0 ns to 5 ns all over the traces? Or does it mean, that the window length of the filter for every n trace [n-5, n-4, n-3, ...0, n+1, n+2,.., n+5] goes from 0 ns till the end of the vertical time windows?

Regards

Originally posted by @wilhem in https://github.com/emanuelhuber/RGPR/issues/33#issuecomment-667663670

emanuelhuber commented 4 years ago

Hello Wilhem

Thank you for noticing this inconsistency. The help page of the function dewow() is the most correct. The argument w has to be given in trace unit. Though trace unit may not be the best term, it means either time unit (if the data are in time unit) or depth unit (e.g. after a migration or time-to-depth conversation). The unit is returned by depthunit (also not the best term because depthunit can return a time unit, e.g., ns).

This inconsistency will be corrected in the develop version of RGPR. I will update the tutorial as well as the help page of dewow().

Does it mean that the window length of the filter goes from 0 ns to 5 ns all over the traces?

Yes (if your data were collected as a function of time).