Super-virtual Refraction Interferometry (SVI)

Objective

In this lab we will learn how to increase the signal to noise ratio (SNR) of recorded refraction data using super-virtual refraction interferometry (SVI).

Introduction

SVI is a two-steps process (Figure 1), the first one (Figure 1a) is to create the virtual data set using correlation and the second one (Figure 1b) is to create the super-virtual data set using convolution.

For more information, read attached paper (Hanafy and Al-Hagan, NSG 2012)

Figure : The steps for creating super-virtual refraction arrivals. a). Correlation of the recorded trace at A with that at B for a source at x to give the trace with the virtual refraction having traveltime denoted by . This arrival time will be the same for all post-critical source positions, so stacking will enhance the SNR of the virtual refraction by . b). Similar to that in a) except the virtual refraction traces are convolved with the actual refraction traces and stacked for different geophone positions to give the super-virtual trace with a SNR enhanced by . Here, N denotes the number of coincident source and receiver positions that are at post-critical offset for this particular refraction.

Figure 2: a) A shot gather example of the clean field data set. b) The super-virtual shot gather

Procedure

I. Clean (high SNR) field data

1. Download the field data example (AllData.mat)

2. Download the MatLab file (sv_field_2.m) to find SVI data set

3. Download the MatLab files (BandPass.m), (wigg.m), and (matlab_2_dpik.m)

4. Copy all these files in the same directory

5. Read and understand the code sv_field_2.m, then run it

6. Compare the original shot gathers and the virtual shot gathers

7. Pick a couple of shots before and after the SVI and compare the picked times

II. Noisy Field Data

1. Download the second data set (csgall.mat), this is a noisy fielddata

2. You can use the following muting window

uperT=[1 7;11 54;27 124;35 143;54 207;81 259;120 387];

lowerT=[1 66;10 139;26 214;45 270;68 345;84 415;120 523];

3. Modify the MatLab file if required,

4. Repeat steps 5, 6, and 7 from previous example

III. Field Data Recorded inside KAUST

1. Download the third data set (CSG_Data.zip), this data is recorded near the Safaa Stadum, KAUST (Figure 3),

(48 CSGs, 48 Receivers, dg = ds = 5m., dt = 0.25 ms, No. of stacks of CSG # 1 and # 48 is 40 stacks, while other CSGs 10 stacks)

2. Download the modified SVI code (sv_field_3.zip), the upper and lower picking files for mute around first breaks (Wind_FB.zip)

3. Run the matlab code with the given muting window,

4. Run the matlab code with different muting window, what is the effect of muting window on the result?

5. Run the matlab code with NO muting, what is the effect on the result?

6. There are few bad traces in the data, mute these traces then re-run the code with a good muting window, any improvements?

7. Adjust the code to use tapper for the time-window muting.

8. Adjust the SVI code to acknowledge Iterative SVI, does the S/N ratio improved with increasing number of iterations?

9. Pick the raw data (only good traces) and the SVI data sets, compare both picking sets

10. Invert the pickings from the raw and the SVI data sets, compare the two tomograms

IV. Other Field Examples

1. Qademah Field Data (Data and code).

109 CSGs, 109 Receivers, dx = ds = 3m, dt = 0.25 ms

2. Long Qademah Field Data (Data)

228 CSGs, 228 Receivers, dx = dg = 10, dt = 2 ms

3. Nevada Field Data (Data)

240 CSGs, 240 Receivers, dx = dg = 20, dt = 1 ms

Note that, only shots # 1, 5, 9, ….., 237 are alive and others are not

Figure 3: The location of the recorded data