LESSONS FROM COMBINED INTERPRETATIONS OF WIDE-ANGLE AND CONVENTIONAL REFLECTION DATA IN THE NORTHERN YILGARN, WESTERN AUSTRALIA

Tanya Fomin1,2,3 , Bruce Goleby1,2,3 & Malcolm Nicoll1,2

1pmd* CRC, 2GA, 3ANSIR

INTRODUCTIONWide-angle seismic vibroseis data were collected during 2001 in the northern Yilgarn Craton. This experiment focused on the Leonora-Laverton Region and was carried out coincident with the deep seismic reflection transect.Main objectives of wide-angle survey:* Test if it is possible to collect high-density wide-angle data at large offsets (60+ km), using a vibroseis source * Conduct a comparative study of seismic images of the crust, constructed from near vertical; and wide-angle data; * Supplement deep seismic reflection data with velocity information (mainly for the upper crust).

Location of wide-angle seismic line (174 km) which coincident to the deep seismic reflection transect (384 km). White dots represent vibration points. Red line (64 km) represents vibration points and receiver array.

•120 short period instruments deployed (~ 500 m apart), 10 ms sample rate •Deep seismic reflection data collected along the same line•3 vibrators used as energy sources. 80 m spacing between source points. 3 variable sweeps, each 12s long•Data were recovered from only 60 stations•Final velocity model is based on data from the 60 stations•40 common receiver gathers were used to produce the wide-angle stacked sectionRECORDER

Deployment of short period refraction instruments alongthe wide-angle survey

SOURCE

Vibratorson line

WIDE-ANGLE SURVEY DETAILS

FIELD EXPERIMENT

The vibrations from the vibroseis trucks were recorded by both surveys simultaneously. Acquisition parameters used for the wide-angle survey were selected so that it could fit into the schedule and technology of the conventional reflection survey. The major challenge of the experiment was to collect wide-angle data utilising acquisition parameters of a typical deep seismic reflection survey.

FROM SEISMIC TRACES TO VELOCITY MODEL

Shot-to-receiver distance 400 m

Raw non-correlated seismic trace recorded at 400 m offset

Seconds0 60

Sweep 12-80 Hz Sweep 8-72 Hz

Distance (km)0-25 25

0

8

Tim

e (s

ec)

Non-correlated seismic traces gathered in common receiver offsets

Cross-correlated seismic traces gathered in common receiver offsets, three sweeps stackedat every vibration point. First arrivals and reflections become quite prominent (red arrows)

0-25 25Distance (km)

Common receiver gather, seismic events picked for velocity model (red and blue dots)

Distance (km)

Tim

e-X

/8 (s

ec)

50-50

4

00 Distance (km)0 170

Tim

e-X

/8 (s

ec)

10

0

Major seismic events picked from common receiver gathers for 60 recording stations

0

0

10Distance, km0 170

4

12

Tim

e (s

ec)

Dep

th (k

m)

Ray tracing diagram and travel time plot.Travel time plot shows observed (squares) and computed (lines) travel times. The 2D velocity model was produced by forward modelingsoftware, based on the ray-tracing algorithm of Zelt and Smith (1992)

Velocity model of the upper crust based on data from the 60 recording stations: first arrivals and wide-angle reflections

Range of good matches can be achieved for velocity models within + 500 m/s deviations, which is not sufficiently accurate to define a velocity model in lower crust

Ray tracing diagram and travel time plot for two stations based on first arrivals and wide-angle reflections

Distance (km)0 170

Tim

e (s

ec)

Dep

th (k

m)

0

0

12

40

0Tim

e-X

/8 (s

ec)

10 Major seismic events picked from common receiver gathers

SPECIFICS OF THE LOWER CRUST MODEL

pmd CRC

Australian Government

Geoscience Australia

The upper crustal model consists of three layers over basement:

1. Thin upper layer (5.0 - 5.9 km/s).2. 2 km thick high velocity body (6.4 - 6.7 km/s), interpreted as mafic

rocks, restricted to the western part of the line at ~4 km depth. 3. 4 to 8 km thick relatively low velocity layer (6.0 - 6.1 km/s - granite-

gneissic composition) at ~12 km depth in the western part of the line.4. Basement ( 6.2 - 6.3 km/s).

LOW VELOCITY LAYER

Distance (km)0 170

0

14

Dep

th (k

m)

6.4-6.7km/sW E

6.0-6.1km/s

12

3

46.2-6.3km/s

Q1. HOW TO DESIGN A COST EFFECTIVE AND SCIENTIFICALLY PRODUCTIVEWIDE-ANGLE SURVEY THAT WOULD FIT INTO REFLECTION SURVEY?

Q2. WHAT ARE THE DEVELOPMENT CHALLENGES IN PROCESSING WIDE-ANGLEVIBROSEIS DATA ?

The Yilgarn seismic reflection transect crosses the Eastern Goldfields Terrane that is a part of Archaean Yilgarn Craton and comprises mainly granite and greenstone rocks. Wide-angle line focused on Leonora-Laverton Tectonic Region that mostly includes greenstones.

Distance (km)

Tim

e-X

/8, (

sec)

0

2

0

0

0

2

2

2

-40 40

Vibration Point interval: common receiver gathers, simply decimating traces

80 m vibration point interval utilised in the field

Minimum required: vibration point interval every 240 m

Vibration point interval every 1.5 km

Vibration point intervalevery 3 km

Velocity model : To what extent is it a function of station interval?

Increasing the recorder density does not significantly improve the velocity model in the upper crust.In the upper 4 km, models A and B differ by 100 to 300 m/s.

6.2-6.3km/s

Two important differences are:

1. The bottom of the high velocity body sits shallower (1.5 – 2.0 km) in model B, due to the increased ray coverage with the incorporation of more recorders, importantly more reflections from the bottom of this body have become available 2. Deeper coverage in model B is because important arrivals from the lower crust were missing in the data set used to develop model A

RECOMMENDATIONS FOR SURVEY PARAMETERS INTERPRETED FROM FINDINGS:

1. How dense do we need both shot and recorder intervals? • 240 m – minimum vibration point interval• 5-10 km – recorder interval - for building a velocity model of upper crust• 1-1.5 km - recorder interval - for stacking wide-angle data2. Greater source energy required to build lower crustal velocity model3. Recording instrument sample rate of at least 4-6 msec - for better resolution of stacked sections

WIDE-ANGLE AND CONVENTIONAL REFLECTION – HOW DO THEY COMPARE?

•The reflection data image structures within the upper crust quite well, but fail to detect long wavelength seismic velocity stratification•There is a low correlation between interpreted seismic reflection horizons and velocity boundaries

Comparison of the wide-angle data velocity model with the reflection data interpretation

Interpretation (lines) of reflection seismic data, superimposed on the velocity model wide-angle data0 170

TW

T, s

ec

Distance, km

0

45.5

6.06.5

5.0

The high-velocity body appears relatively transparent in the reflection seismic image. There are fewer reflective horizons interpreted in this area compared to the eastern and western flanks of the line. This correlation suggests a relatively smooth velocity variation in this area (conventional reflection wavelengths are too short to detect it). This is most likely associated with the formation of the greenstone belt, which may have significantly overprinted the original reflection structure.

Wide-angle and conventional reflection sections

Common shot gather from reflection data

Comparison of data from wide-angle and conventional reflection surveys along the same surface locations

Offset, km-5 5

Tim

e (s

ec)

Tim

e (s

ec)

0

12

13

0 0

Common receiver gather from wide-angle data

There are less reflections in the wide-angle data, in this range of offsets, due to the lower frequency range recorded by the shot period instruments

Stacked seismic section derived from wide-angle data

A basic processing stream was used to stack the wide-angle data for comparison with the conventional reflection section:

•Data was sorted into common receiver gathers;•Refraction statics were applied;•Normal moveout correction;•Sort in CDP order;•CDP stack;•Bandpass filter;•Signal enhancement;•Amplitude balance.

1. The experiment has demonstrated the effectiveness of using a vibroseis source for collecting wide-angle data simultaneously with reflection surveying; 2. Velocity model for the upper crust demonstrates the possibility of differentiating high velocity mafic rocks from low velocity rocks;3. Wide-angle reflections from the lower crust are not sufficient to build an accurate velocity model, lower crustal refractions in the first arrivals are required to better constrain this model but unfortunately the signal was to weak at large offsets to achieve this;4. Reflection horizons and changes in reflectivity patterns mapped by the reflection technique will not necessarily coincide with velocity boundaries imaged by wide-angle reflection technology due to different wavelengths and different interference styles; 5. The stacked wide-angle seismic section has some similarities with the conventional reflection section. However, more work is needed to identify if the wide-angle stacked section is of further use;6. A more accurate geological interpretation is possible using a combination of wide-angle and reflection seismic techniques.

AcknowledgementsThe authors are grateful to Prof. Stewart Greenhalgh, Adelaide University, Prof. Brian Kennett, Australian National University and ANSIR for providing wide-angle recording equipment to collect the data and to the ANU staff for participation in the field work. We also thank Leonie Jones, Geoscience Australia, for providing the seismic reflection data and Yilgarn Y2 Project Team for providing the seismic interpretation of reflection data.

Average velocity along wide-angle line to constant depth levels in the crust

V average to 0.5 kmV average to 1 kmV average to 1.4 kmV average to 2 kmV average to 3 kmV average to 4 kmV average to 5 kmV average to 10 km

Distance, km0 170

5.4

5.6

5.8

6.0

6.4

6.2Vel

ocity

, km

/sec

The seismic velocity model derived from the wide-angle data was used to depth-convert the seismic reflection data. There is no substantial difference in the appearance of two-way time and depth scale images. This is because of very small velocity variations along vertical profiles through the crust within the area of the study. Average seismic velocity along the seismic line, used for depth conversion, varies from 5.5 to 5.7 km/s to a constant depth level of 0.5 km. Similar velocity variations are estimated for deeper parts of the crust. Average velocity along the seismic line varies from 6.2 to 6.4 km/s for 4 km constant depth.

0 170

Dep

th (k

m)

Distance (km)

0

12

Velocity model in depth scale

10

Stacked data (different offsets in two data sets)

Potential for AVO (amplitude versus offset) analysis ?

Similarities and differences existDifferences may be due to:

1. Lower frequency cut-offin wide-angle

2. Lower fold in wide-angle3. Different NMO4. Acquisition methodology differences

Stacked section for area in red box from figure above

TW

T (s

ec)

0

CDP5530 6200 CDP5530 6200

Wide-angle Conventional reflection

10

7

0 1707

5 recording stations with average spacing of 20 km

0

6.4-6.7 km/s

5.0-5.9km/s

Distance (km)0 170

Dep

th (k

m)

7A

6.2-6.3 km/s

60 recording stations with average spacing of 1 km

170Distance (km)014

0

LOW VELOCITY LAYER

6.0-6.1 km/s

B

7

Velocity (km/sec)

6.4-6.7 km/s

5.6

6.1

6.5

5.3

6.76.3

5.95.8

5.15.5

5.0

CONCLUSIONS

Q3. CAN WE REDUCE THE AMOUNT OF DATA REQUIRED TO DEVELOP A VELOCITY MODEL?

Q4. USING WIDE-ANGLE VELOCITY INFORMATION TO IMPROVE THE GEOLOGICALMODEL. COMPARISON OF WIDE-ANGLE AND REFLECTION POST STACK DATA .

WHAT DOES IT TELL US?