Page 8 - i1052-5173-27-2
P. 8
reduction in continuity to the decrease in respective cells from the contrast analysis. times that of the depth image and the fault
contrast as a result of the depth conver- In spite of the potential different impact of placement population shows a narrower
sion. Interpreted faults tend to cross areas the two parameters on the interpreters and spread and shorter faults. A similar pattern
with lower reflection continuity. This is their relative co-dependency (i.e., enhanc- is observed for reflection continuity, where
not surprising, because faults that cut and ing the contrast can enhance the continu- high reflection continuity also results in
disrupt rock layers with the same reflec- ity), creating a combined parameter pro- a narrower fault placement spread and
tive physical properties would create low vides a general visualization of image shorter fault interpretations. The differ-
reflection continuity. The fault interpreta- quality. The results were normalized by ences in fault spread observed determine
tions coincide with where reflectors from representing the maximum value as 100 predicted fault heave, resulting, for exam-
the left join those coming from the right, and the minimum as zero. The resulting ple in regional sections, in significant dif-
at ~13–16 km along the seismic image, at merged models for the depth-converted ferences in crustal stretching predictions.
~6 km depth. The third quartile of the TWT and depth images are shown in Further work to assess the relative contri-
TWT interpretations follow this boundary. Figures 2E and 2F. butions of contrast and continuity to visual
In the depth seismic image this joining of image quality to create a single weighted
reflectors is less clear (potentially due to There is a diffuse horizontal boundary parameter would provide a fully quantified
the lack of reflectivity/continuity), and the in the merged values in the TWT image at visualization of image quality.
third quartile is more variable, especially ~4.5 km depth (Fig. 2E), marking a
in the deeper part below 5 km. The greater change from “green” and hotter colors at The two images were presented in dif-
amount of faults dipping to the right below shallower levels to lower “blue” values as ferent domains (TWT and depth), resulting
5 km depth in the depth image can be depth increases. This 4.5 km depth marks in an 18% longer vertical scale in the depth
explained by a lack of reflection continuity the point at which the distance between image that could have changed the percep-
at distances along the section line >13 km, the first and third quartiles increases from tion of the fault geometries to interpreters.
and 5 km depth (Fig. 2D). In the TWT 523 m to more than double (1234 m) at the However, our correlations suggest that
image, right-dipping faults have to be bottom of the section. This boundary also image quality had the major influence on
interpreted crossing strong, continuous coincides with the average depth of the interpretation choice. We note that the
reflections below 5 km, and most right- interpreted TWT faults, suggesting that it average depth of the faults interpreted in
dipping fault interpretations stop at shal- marks a clear increment in the uncertainty the TWT image coincides with a boundary
lower depths (from 17 fault interpretations of the image for interpretation. Faults are in depreciating image quality in the com-
at 3 km depth to only 5 at 5 km depth). In interpreted until a deeper point in the bined analysis. Although our results show
the depth image, the reflections are more depth image, potentially because this that depth conversion choices (including
discontinuous and fault interpretations boundary in image quality is less percep- the method used) change seismic image
continue to greater depths. The extent of tible. The positions of the outlier interpre- quality, all image manipulations have the
the outliers also seems to be affected by tations show a greater change, from a nar- potential to change interpretation out-
reflection continuity. In the TWT seismic row converging spread to divergent with comes. We therefore need to better under-
image for example, the right outlier line the spread increasing with depth. In the stand image perception so that such image
coincides with a break in reflection conti- case of the depth image (Fig. 2F), this manipulations do not arbitrarily influence
nuity between 4-10 km depth (Fig. 2C); in boundary is less noticeable, possibly due or bias interpretation outcome.
the depth image, the right outlier stays to to the overall low values and poor image
the left of a package with more continuous quality, although fault spread does For a fixed binary threshold, image
horizontal reflections, located at 2.5–5.5 increase with depth below 4.5 km. The contrast and continuity are associated
km depth (Fig. 2D). results suggest that there may be a con- parameters, so increasing image contrast
trast and continuity threshold within the can artificially increase continuity. This
Combined Image Analysis seismic images beyond which the fault correlation causes issues in determining
interpretations are almost unconstrained the best methods for enhancing imagery
The analysis of contrast and continuity by the data. in order to maximize interpretation effec-
highlighted spatial associations between tiveness. It also has impacts on the pro-
image quality and fault interpretation IMPACT ON INTERPRETATION cessing of seismic data and the model
(Figs. 2A–2D). In reality, the image, as chosen to create an image. Initial process-
viewed by the interpreter, is the result of The experiment outlined above shows ing models generally assume a sub-
combining both contrast and continuity. In that image contrast and the continuity of horizontal, sub-parallel reflector stratig-
an attempt to merge the results of the con- features both impact on the interpretation raphy with minimal disruption. Thus,
trast and continuity analyses, the continu- outcome of the seismic imagery. they enhance reflector continuity. Our
ity analysis was converted into a cell Interpreters are less prone to cross stratal results, albeit based on TWT and depth
model, based on the contrast grid. This reflections if they are “strong” (i.e., high imagery rather than different processing
conversion assigned the maximum conti- contrast and high continuity), and where models, show that reflector continuity is
nuity value contained within a cell to each reflections are “weak,” uncertainty in spatially related to fault placement cer-
cell in the grid. To merge the analyses, interpretation increases. In general, tainty. The processing of strong reflector
cells in the new continuity cell model were enhancing image contrast helps to con- continuity in seismic image data may
multiplied with the values from the strain the interpretation, as seen in the result in greater constraint, or certainty,
TWT image, where image contrast is three in fault placement than is warranted by
8 GSA Today | February 2017
