# Depth-Conversion of Seismic Reflection Data in the Eromanga Basin, South Australia

**Timothy A. Rady, 2006Bachelor of Science (Petroleum Geology & Geophysics)Australian School of PetroleumThe University of Adelaide**

## Abstract

Depth-conversion of seismic reflection data in the Eromanga Basin is traditionally performed using

an average velocity model from datum to target horizon. Two-way time and depth data for the target

horizon at each well are used to calculate a pseudo-velocity, which is then gridded and contoured to

produce a velocity map. The average velocity map is then combined with an inte1preted two-way time

map to yield a depth map for the target horizon. The average velocity model is only controlled at each

well and relies on interpolation and extrapolation of values between wells to produce a velocity map,

hence any velocity variation between wells due to gross changes in lithology, pressure or burial depth

are not accounted for.

The top McKinlay Member and top Hutton Sandstone horizons were depth-converted using the

average velocity, interval velocity layer cake, constant velocity layer cake and processing velocityderived

techniques. Each technique was selected to attempt to improve on the currently widely-used

average velocity technique by accounting for variation of geological factors, such as lithology,

thickness and burial depth in their velocity models. Velocity survey data is not widely m'ailable for the

study area, hence analytic velocity functions such as V0-k and other sonic log-derived models for

depth-conversion were not considered. As a result, the depth-conversion techniques utilized in this

study rely on velocity models derived by pseudo-velocities. A second type of input velocity data -

seismic processing velocities - were used to construct a velocity model from datum to each target

horizon for use in the fourth depth-conversion technique.

The average velocity technique was used as a reference with which to compare the other three

techniques. A layer-cake technique comprising methodology similar to the average velocity technique

was undertaken to take variation in lithology and formation thickness into account. A second layer

cake technique using a constant velocity to derive an approximate depth from a twt thickness map was

undertaken to improve the confidence in depth prediction between wells. This technique also

considers the effect of lithology and thickness variation on velocity. For both layer cake techniques,

the sequence from datum to top Hutton Sandstone was split into five layers for the depth conversion:

mean sea level to top Oodnadatta Formation (layer 1), top Oodnadatta Formation to top Coorikiana

Sandstone (layer 2); top Coorikiana Sandstone to top Cadna-Owie Formation (layer 3); top CadnaOwie

Formation to top McKinlay Member (layer 4) and top McKinlay member to top Hutton

Sandstone (layer 5).

The average velocity method predicts depth to the top McKinlay Member with a mean error of 2.1m

and standard deviation of the error of 4.9111. The mean and standard deviation of the errors predicted

by each tested depth-conversion technique are similar to average velocity technique at both target

horizons. Mean errors for the four techniques range between 0. 7- 2.5 m at top McKinlay Member

and 2.2- 4.8m at top Hutton Sandstone. The standard deviation ofthe errors for the four techniques

ranges between 4- 6.6m for top McKinlay Member and 4.9- 8.5 m for top Hutton Sandstone. The

maximum error and range of error are more significant in determining the success of each depthconversion

technique as they give an insight into the potential error bar of depth prognoses in future

drilling. The lowest maximum error at top McKinlay Member is 9.9 m and lowest range of error (the

difference between minimum and maximum errors) is 14.9 m, both of which are recorded by the

average velocity depth-conversion technique. At top Hutton Sandstone, the lowest maximum error is

13.5 m and lowest range of error is 18m, also recorded by the average velocity depth-conversion

technique.

The layer cake techniques show no significant improvement over the currently widely-used average

velocity technique. This suggests that the identified geological factors have a negligible effect on

velocity variation or that the increased source of error in these multilayer velocity models oufll!eighs

their potential for improvement. Depth-conversion utilizing seismic velocities produces relatively

accurate depth prognoses, uses a densely-sampled and geologically-feasible velocity model and, as a

result, greatly increases the confidence in depth prognosis away from existing well control. In light of

these results, future depth prognoses in the study area and geologically-similar areas should be

calculated using the average velocity depth-conversion technique. Additionally, the careful use of

seismic velocities can be successful in deriving accurate depths from seismic reflection data in

regions of little well control or greater distances befll!een wells.