That depends on what feature you want to enhance. The best option is the trial and error. Window width normally starts with 1 [for most cases this is good enough], else you can increase slightly and see.  The scale is 0-1 and try with 1 again and gradually reducing as needed. Max gain value you can start with 500 and go up and down as you need to enhance the refection.

Dear Pranjal, 

In the version of ReflexW that I am most familiar with, the Energy Decay gain function that you are using does not require a window length to be specified. The gain function is simply applied to the whole trace.

Where the window is required is in the alternate "AGC" gain function, in which it fundamentally affects the way the gain is defined. Too short a window will boost up high frequencies, often in a way that obscures and damages the underlying data. Too long a window will have very little effect. The reason for this is that AGC calculates trace scalars based on a comparison of the RMS amplitude in the whole trace compared with that in the window.

There's probably no such thing as a "correct" gain function (unless you are using a formal spherical divergence correction and have good velocity control) but just one that performs the best for the features you want to bring out in your data. In practice, I've found that "energy decay" gain in ReflexW works nicely for very shallow targets, whereas AGC often balances the amplitudes of deeper reflections more effectively. But, it's all quite subjective and I'm sure that other practitioners will have had other experience.

I hope this helps answer those three questions, 

Best regards, 

Adam Booth 

I only use AGC to review the data; it is good for that, just try a window that includes several wavelengths, but shorter than large changes in the envelope.  Ever since Relative Amplitude Processing started in the early 1970s, AGC has been seen as a Bad Thing in the seismic world as it destroys relative amplitude and so information.

For GPR in Reflex-W, I use 'energy decay' and experiment with the scaling value by comparing the original vs 'filtered' trace displays in the process design window.  It is quite sensitive. The original processing in our Gaping Gill paper and slides showed signal down to a few metres depth.  I tweaked the scaling value by very little to show signal down to over 30 m.   

It is hard to go further unless you show your sections after background removal.

Dear Pranjal,

This is the first and simplest answer as explained in Reflex Help about the time window of the AGC gain  (content search = AGC):

The filter parameter window length determines the window size in the given time dimension. The program calculates at first an average amplitude over the total time range for each trace. After that the program scales each amplitude value in that way that the mean amplitude has the same value for each selected window around the current value with in a trace. The size of the window determines the kind of amplitude equality distribution. A window size of only one sample increment means that each time sample within one trace receives the same amplitude value (no reasonable choice), a window size of the whole trace length causes no modification of the amplitude. Small window sizes cause a strong equality distribution, large windows a weak.

Hope this answer be helpful,

Best regards,

The AGC gain may be seen as way to provide either a cheap spectral equalization or a  way to manually emphasize  the key reflectors as you may consider significant. The attached paper discussed such this matter (figure 6 and discussion below), when varying the AGC length allows to discriminate between complex stratigraphic features.

Best regards.

I do no suggest AGC, since it destroy the reflectivity information (relative reflection amplitude). It can be used for simple examination of acquired data in field. For refined processing and analysis, I prefer spherical and exponential correction (SEC) which accouns for the spherical spreading and intrinsic attenuation of refelcted electromagnetic waves. On the selection of the attenuation factor for SEC, please try and error.

Thank you all for valuable & helpful answers.

I have also applied deconvolution but did'nt get any satisfactory result.

Please suggest which type of Decon. should I choose? or any idea about filters?

Hi,

Deconvolution is only successful for GPR in some very specific applications (e.g., in ice and snow); the GPR wavelet violates a lot of the fundamental decon assumptions (stationary wavelet, minimum phase, etc) therefore you can damage your data by applying it. Consequently, decon is not often included in a GPR processing flow.

Where it is successful, the results can be very good - but it's usually the case that GPR data are not suitable for deconvolution.

Best regards, 

Adam 

Thank you Sir

Dear Adam Sir,

In Reflexw, the steps Dewow & Energy decay are avaialble in Plot options as well as in Processing option. Can I run these two steps in processing if these are already checked in the Plot option?

Yes, but they won't "do" anything to the actual data - they are quite literally just display parameters. If you only apply the operations in the Plot window and (e.g.) Export the data, the operations will not be applied in the exported file.

In any case, applying two gains and/or decons (particularly if they are designed in the same way) might not be appropriate.

Best regards, 

Adam 

Dear Sir,

Please have a look on the attached sections. I am getting a thin reflectivity from surface to bottom(circled in image) in few parallel lines around same position but there was nothing exposed on surface. What could be possible reasons for such anomaly?

Hi again Pranjal

*022.05T is an interesting profile:

• you may have a channel of some sort between stations 32 and 42
• However, the ellipse you drew is around a gain problem

Why not offer to send the raw data for section 022 to those of us who wish to play ? There are many parameter options and this correspondence is getting over-long.  Also, several if not all of your correspondents use Reflex-W, so you might have a fair-test comparison of the various workflows !

It would also help if you provided some information on the substrate you are trying to image, esp. as regards moisture, clay and salinity.  Plus it would help to know the GPR kit you used to record the data !

Regards

Robin 

Hi Pranjal,

Sometimes Energy Decay Gain can be too aggressive and over gain any skew or low frequency  noise variations between traces. Try using a gain function to see if this provides any improvement. Use a negative exponent to avoid overgaining the noise at the bottom of the range. Change the other parameters by trial and error until you get a suitable gain distribution. Use the example trace on the gain window as a guide. Also, try leaving out the Subtract-Mean (Dewow) prior to the gain function as any small artifacts it may introduces will be amplified. This should not affect the result provided any dc-shift is constant across all traces. Note that the bandpass filter applied after the gain function will remove any dc-shift any way. Move the background filter to the last processing step. As this is a 2-D filter, it is applied between traces. Any processing step that applies to individual trace like a bandpass filter may affect the result of the background removal. 

With regards to high amplitude bands in your data it may be an artifact introduced in your processing flow or may be due to the signal ringing from your equipment. Difficult to say without seeing your data.

Regards,

Val

Thank you Val Donazzolo. For gain I still don't understand why you use the gain function and why not AGC. can be explained in detail how the software works and how it is mathematically related, so that you are sure the gain function is the best choice for the data we have

Best regards,

Gatri