Use of weighted integral variables to determine the relation between rainfall intensity and storm flow and peak flow generation

Abstract

Hewlett's (1977) work using his storm flow separation technique has established that peak rainfall intensities have little predictive value in the determination of storm flow depth or storm peak flow from small upland streams emanating from forested catchments. A logical but disconcerting inference is that storm rainfall intensities have little influence on either the volume of storm flow or peak flow. To obtain additional information a variant of the technique was applied to excellent records from two separate, steep, upland catchments in southern Australia. The catchments were regarded as having quite distinct rainfall response characteristics by hydrologists involved in data collection. Antecedent flow, storm rainfall, and measures of maximum rainfall intensity were supplemented by weighted integrals of rainfall intensity with time to test whether they gave additional insight into the influence of rainfall intensity. As expected, regression analysis showed that knowledge of maximum intensities gave little increase in predictive ability over storm rainfall. However, the time integral of rainfall intensity to the power of 1.5 was better correlated with peak flow than any other measure on both catchments. The results suggest that rainfall intensities are more important in the generation of peak flows than indicated by regression analysis using only peak rainfall intensities. Comparison of independent measurements of the variables using separated rain gauges showed that independent measures of the weighted variables were better correlated than measures of storm maximum intensities, indicating that they are also a more precise measure.