Page 35 - Statistics and Data Analysis in Geology
P. 35

Statistics and Data Analysis in  Geology - Chapter 2



























                                                 Number of successes
              Figure 2-7.  Discrete distribution for the probability  of n discoveries in drilling four out of
                   ten  prospects when four prospects contain oil.


             two outcomes, hit porous sandstone and dril2 into an anticline, are not mutually ex-
             clusive as we hope that both can occur simultaneously.  Since the presence  of  a
             sandstone is governed by factors that operated at the time of  deposition, and since
             the occurrence of an anticlinal fold is presumed to be related to tectonic conditions
             at a later time, the two outcomes are unrelated, or independent. If  two events are
             not mutually exclusive but  are independent,  the joint probability  that they will
             occur simultaneously is the product of  their separate probabilities  of  occurrence.
             That is, p (hit sandstone  and drill anticline)  =  p (hit sandstone) x  p (drill anticline).
             This is the muZtipZicative rule of probability.
                 Two events may be related in some way, so that the outcome of one is depen-
             dent in part on the outcome of  the other.  The joint probability of  such events is
             said to be conditional.  Such events are extremely important in geology, because
             we may be  able to observe one event directly, but the other event is hidden.  If
             the two are conditional, the occurrence of  the observable event tells us something
             about  the likely state of  the hidden event.  For  example, the upward movement
             of  magma in chambers beneath a volcano such as Mt. St. Helens in Washington is
             believed to cause a harmonic tremor, a particular type of  earthquake.  We  cannot
              directly observe an active magma chamber, but we can observe and record the seis-
             mic activity associated with a volcano. If a conditional relationship exists between
              these two events, the occurrence of harmonic tremors may help predict eruptions.
             If  p(tremor) is the probability that a harmonic tremor  occurs  and p(eruption) is
              the probability of  a subsequent volcanic eruption, then p (tremor and eruption) #
              p (tremor) x  p (eruption) if the two events have a conditional relationship.
                  The conditional probability that an eruption will occur, given that harmonic
              tremors have been recorded, is denoted p (eruption 1  tremor). In this instance the
              conditional probability of an eruption is greater than the unconditional probability,
              or p (eruption), which is simply the probability that an eruption will occur without
              any knowledge of  other events. Other conditional probabilities may be lower than

              22
   30   31   32   33   34   35   36   37   38   39   40