| Abstract: | Following the statistical analyses of long‐term rainfall‐runoff records from research basins in humid temperate latitudes, Hewlett and co‐workers extended the global challenge to disprove their findings that rainfall intensity was non‐significant. This paper responds to Hewlett's challenge as no preceding analyses have involved forested basins in a tropical cyclone‐prone area. Based on a 7 year rainfall‐runoff record, quickflow (QF), peak flow (QP) and quickflow response ratios (QRR) were regressed as dependent variables against rainfall parameters (intensity, Pi, amount, P), storm duration, D and antecedent flow, I. These data sets were categorised into total streamflow (Q) classes and stratified into three seasons, (monsoon, post‐monsoon and dry) for forested and cleared catchments. Where rainfall variable collinearity met acceptable levels, the addition of Pi to regression models including P, D, I contributed up to 9% and 66% of the respective variations in quickflow and peak flow. For the highest Q storm classes (monsoon), Pi alone accounted for up to 67% and 91% of the variation in QF and QP respectively and was the dominant influence on QP for all seasons. The very high rainfall intensities experienced in the monsoon season is a causal factor why these results differ from those of other research drainage basins. Surprisingly, Pi continued to have a significant influence on QF for dry season classes when less‐intense rainfall occurs. Further the results were similar for both catchments across all seasons. P was the dominant independent variable affecting QF above a threshold Q of 50 mm (monsoon), as rainfall contributes directly to saturation overland flow and return flow under saturated conditions. Further although QRR increased with increasing Q for each season, the regression results for that parameter were poor possibly due to the non‐linearity of the rainfall‐runoff relationship. Copyright © 2010 John Wiley & Sons, Ltd. |
