Page 261 - The Geological Interpretation of Well Logs

- SEQUENCE STRATIGRAPHY AND STRATIGRAPHY -

type of highstand parasequence set appears to be the most of the seismic shows that the apparent correlations are
common or perhaps easiest to identify (the automatic false and the true cortelations between the two wells are
attribution of such sets to the highstand sea level phase very complex. The sequence stratigraphic analysis
has been questioned recently, as similar depositional pat- (Figure 15.15, left) would not be possible without the
terms develop during the lowering of sea level in forced seismic, but once in place can explain the complexities.
regession, Nummedal er ai., 1995). However, there is a The base of the sand near the bottom of the section is
tendency among many workers using sequence stratigra- clearly downlapping on the seismic: it is diachronous.
phy to simply identify the key surfaces and not add in the The shale interval (horizon 2) correlated between the two
detail of systems tracts (cf. Walker, 1992), especially as wells on seismic evidence is over an erosional sequence
the systems tracts are essentially used by Exxon. boundary and subsequently transgressed: a maximum
Two examples attempt to show in detail the application flooding surface is proposed. Above this, the section is
of a sequence stratigraphic analysis on real log data, markedly onlapping and there is no direct correlation
rather than dealing in models. The first example is in between the wells. Finally another sequence boundary
shallow marine to shelf sediments (Figure 15.14). The (horizon 3) ends the section, with erosion indicated on the
individual elements in the two selected wells are very seismic and explaining the thickness difference in the
distinct: a series of cleaning-up electrosequences with, at upper section. A transgression and flooding surface fol-
their base, organic-rich condensed sequences overlying low this sequence boundary. From this example it is seen
thin, burrowed transgressive deposits. The top of the that the lithostratigraphy cannot contend with deposition-
example section is a well marked, widespread erosional al complexity, that the well logs should not be correlated
break (sequence boundary sensu Exxon); the base is without the seismic and that tracing key surfaces is a
interpreted as an erosional, transgressive surface. The natural result of integrating the logs with the seismic.
sediments represent some 7-8 Ma. From the study of Sequence stratigraphy is very much an evolving
over 100 wells covering 23,000 km? it is known that these subject at the time of writing this book. Modifications
electro sequences can be correlated and to some extent, to the Exxon models are frequently being proposed or
dated. There are obviously changes, but the principal new explanations attempted. As a journalist would say
framework is made up of the flooding surfaces/condensed ‘the jury is sti}l out’, an ugly expression but effective.
sequences and closely preceding transgressive deposits.
Good sands tend to be very localised. In the two example Computer techniques in sequence stratigraphy
wells, when the section is broken down using the Exxon The concepts of sequence stratigraphy are very sophisti-
approach (Figure 15.14, left side), three sequences are cated and based on a huge amount of observation and
interpreted and the marked base of a fourth cuts off the experience. It is inconceivable that such concepts should
section at the top. Highstand systems tracts dominate, the not make use of the moder technology available for the
transgressive tracts are mostly very thin and a lowstand manipulation of logs. This section describes the author’s
systems tract is only interpreted in sequence 3. The attempts at using interactive computer techniques for a
boundary between sequence 1 and 2 is unsatisfactory sequence stratigraphic analysis.
although the very marked condensed sequence is The programme used is TerraStation, (from Terra
undoubted. Besides the highstand systems tracts (which Sciences). The principal routines discussed are the single
are clearly progradational), the identification of other log histogram, the interactive cross-plot and the interac-
systems tracts is more dictated by what is required by tive z-plot. The interactive cross-plot consists of a screen
the model than by a positive identification or distinctive on which both the cross-plot is displayed and also the log
features. With the genetic stratigraphic sequence traces themselves. Any two logs may be displayed, at any
approach (Figure 15.14, right side) four transgressive- scale over any selected interval. The cross-plot is interac-
regressive sequences afe very satisfactorily interpreted. tively linked to the log curve display so that groups of
However, the presence of sand in well A and not well B points on the cross-plot outlined by a mouse driven
is not well brought out and the status of the surface at the rubber band are immediately identified on the logs: con-
top of the section is not clear. versely intervals identified by a mouse defined box on the
The second example (Figure 15.15) is from a log curves are highlighted as values on the cross-plot. In
Palaeogene shelf edge area in the North Sea, for which other words, points on the cross-piot can be identified on
good quality seismic is available. The two wells are 2.5 the curves and intervals on the curves can be identified on
km apart. A shale section occurs at both the bottom and the cross-plot. The z-plot has the same facilities but the
at the top of the selected interval. The well logs do not third dimension is added to the plot field.
show distinctive depositional patterns (as seen in the pre- The use of these capabilities will be illustrated using
vious example) and the principal feature is a thick sand real examples.
section with more or less prominent, but thin shale inter- The first step in dividing a well into sequence strati-
vals. The wells appear reasonably similar. An established gtahic units is to identify the key surfaces. As far as the
lithostratigraphy (Knox and Cordey, 1992) can be applied logs are concermed, as has been described, key surfaces
to both (Figure 15.15, right). However, the line diagram tend to have extreme or even anomalous log responses.