The thresholds considered in this study are total
nine cases; the mean, the meanG0.25stdv (stdv:
standard deviation), the meanG0.5stdv, the meanG
0.75stdv, and the meanG1.0stdv. As the thresholds
are represented by the use of the mean and standard
deviation, we may better compare the data collected
before and after the long dry period. Also, we may
overcome the difference in their means of the old
Chukwooki and modern flip-bucket precipitation
measurements. The observed number of occurrences,
and the observed return and duration characteristics
below (or above) these thresholds are summarized in
Tables 2 and 3, respectively.
The decision of proper thresholds was done by
evaluating the occurrence characteristics of dry years.
For a given threshold the number of dry yearsrecorded until time t from the beginning was counted,
and, by dividing it by the total number of dry years for
the whole period, we could estimate the relative
frequency at time t. Then, the Kolmogorov-Smirnov
test was applied to decide if Eq. (1) is valid or not
(Clark and Disney, 1985). The results are summarized
in Table 4.
As can be seen from Table 4, the proper threshold
to confirm the independent occurrence of dry years
may be the level of meanG0.75stdv. However, as the
level of meanG0.5stdv was rejected from the
Kolmogorov-Smirnov test by a narrow margin for
the 5% significance level (but accepted for the 1%
significance level), the maximum threshold would be
around the level of meanG0.5stdv. Further analysis in
this study is based on the thresholds of meanG
0.5stdv, meanG0.75stdv, and meanG1.0stdv.
|