We present a new method for calibrating a classified 3D seismic volume. The
classification process employs a Kohonen self-organizing map, a type of un
supervised artificial neural network; the subsequent calibration is perform
ed using one or more suites of well logs. Kohonen self-organizing maps and
other unsupervised clustering methods generate classes of data based on the
identification of various discriminating features. These methods seek an o
rganization in a dataset and form relational organized clusters. However, t
hese clusters may or may not have any physical analogues in the real world.
In order to relate them to the real world, we must develop a calibration m
ethod that not only defines the relationship between the clusters and real
physical properties, but also provides an estimate of the validity of these
relationships. With the development of this relationship, the whole datase
t can then be calibrated.
The clustering step reduces the multi-dimensional data into logically small
er groups. Each original data point defined by multiple attributes is reduc
ed to a one- or two-dimensional relational group. This establishes some log
ical clustering and reduces the complexity of the classification problem. F
urthermore, calibration should be more successful since it will have to con
sider less variability in the data.
In this paper, we present a simple calibration method that employs Bayesian
logic to provide the relationship between cluster centres and the real wor
ld. The output will give the most probable calibration between each self-or
ganized map node and wellbore-measured parameters such as lithology, porosi
ty and fluid saturation. The second part of the output comprises the calibr
ation probability. The method is described in detail, and a case study is b
riefly presented using data acquired in the Orange River Basin, South Afric
a.
The method shows promise as an alternative to current techniques for integr
ating seismic and log data during reservoir characterization.