Geoscience Reference
In-Depth Information
Fig. 2.16
A na¨ve example of expectation from geostatistical forecasting - the final mapped result simply illustrates
where the wells are
reservoir model. Furthermore, geostatistical
methods need not be over-complex and are not
as opaque as sometimes presented.
refers to aspects of the model
which are specified by a random (stochastic)
outcome from a probabilistic algorithm.
To complete the terminology, a
stochastic
process
(from the Greek
stochas
for 'aiming' or
'guessing') is one whose behaviour is completely
non-deterministic. A
probabilistic
method is one
in which likelihood or probability theory is
employed.
Monte Carlo
methods, referred to
especially in relation to uncertainty handling,
are a class of algorithms that rely on repeated
random sampling to compute a
probabilistic
result. Although not strictly the same, the terms
probabilistic
and
stochastic
are often treated syn-
onymously and in this topic we will restrict the
discussion to the contrast between deterministic
and probabilistic approaches applied in reservoir
modelling.
Probability
2.5.1 Balance Between Determinism
and Probability
The underlying design issue we stress is the
balance between determinism and probability in
a model, and whether the modeller is aware of,
and in control of, this balance.
To define the terminology as used here:
Determinism is taken to mean an aspect of a
model which is fixed by the user and imposed
on the model as an absolute, such as placing a
fault in the model or precisely fixing the loca-
tion of a particular rock body;
