Several stochastic models of rainfall occurrence and daily rainfall am
ount have been applied to data sets, of 18-20 years duration, for 22 i
slands in the Western Pacific. The models for rainfall occurrence were
first- and second-order Markov chains, the truncated negative binomia
l distribution, and the truncated geometric distribution; each was app
lied with parameters determined independently for each month, and with
parameters assumed to vary smoothly through the year according to a F
ourier series. The models tested for daily rainfall amounts were the W
eibull distribution, a mixed exponential distribution, the two-paramet
er gamma distribution, a skewed normal distribution, the kappa distrib
ution, and the Srikanthan-McMahon model. The rainfall data were classi
fied according to the number of adjoining wet days, and the models tes
ted with all data grouped together, solitary wet days fitted separatel
y from other wet days, and each class of data fitted separately. The r
esults of these tests showed that this classification improves model f
itting, in spite of the additional number of parameters required. For
several months of the year, there are significant differences between
the distributions of rainfall amounts fdr the three classes. However,
for 18 of the 22 stations, the Srikanthan-McMahon model outperformed a
ll others, and successfully simulated the differences between the rain
fall classes.