were monitored on each of the five main terraces that were instrumented for 1992
and 1993; and 34 to 41 events on the other four terraces, which were instrumented
for 1993 only (Table 3). The low (in monsoon terms) mean intensity of many of
the sampled events can be misleading, as most of the larger storms did include
periods of high-intensity rain. This was overcome in the analysis by using a number
of indices of rainfall intensity, including mean intensities and maximum 1-, 5-, 10-
and 15-minute intensities, and an EI15 index. The last was a modified form of the
EI30 index (Wischmeier & Smith, 1958) that incorporated the kinetic energy equation
of Hudson (1965). This was believed to be more applicable to tropical rains
of high intensity. The maximum 15-minute rainfall intensity was used because it has
been found to be more appropriate where the land is moderately-to-sparsely vegetated,
such as that most at risk in the Middle Hills (Stocking & Ewell, 1973). The
small number of high-intensity events each month had important implications for
field sampling and data capture. In 1993, approximately half the 90 events where
rainfall exceeded 5 mm were included in the samples on most plots; in 1992 between
25 and 40% were sampled. For all sites, either the highest or the second-highestmagnitude
rainfall event of the season, as well as a broadly comparable range of
events, were captured each year