Eight rainfalls were intensively sampled in a subtropical rain forest
in northeastern Taiwan. This study, carried out at Fu-shan Experimenta
l Forest was designed to examine variability of throughfall characteri
stics. The volume and chemistry of the throughfall associated with ind
ividual rains have a high degree of spatial variability, with throughf
all chemistry more variable than volume. When data from all of the sto
rms were combined, there was lower variability than that seen with ind
ividual storms, largely as a result of a lack of consistent storm to s
torm spatial variation. Thus, individual storms cannot be used to repr
esent the seasonal or annual pattern of spatial variation in throughfa
ll chemistry or volume of the study forest. Twenty samples were requir
ed to estimate throughfall volume within 10% of the population mean wi
th 95% confidence for individual storms. Yet, more than 100 samples we
re required to estimate the throughfall chemistry of some ions (e.g.,
H+). When the data from all eight storms were combined, 7, 20, and 25
samples were necessary to estimate throughfall volume, ion concentrati
ons, and ion fluxes (except for Mg2+ and H+) within 10% of the populat
ion mean with 95% confidence. In the absence of estimates of spatial v
ariation, differences in sampling protocols among throughfall studies
make comparisons of throughfall characteristics difficult. In this stu
dy three funnels were used in tandem to capture small scale variabilit
y at each sample location, thus reducing spatial variability of throug
hfall chemistry.