Verification and Design Application of Long-term Rainfall Disaggregation Procedures

Bill James and Rob James


Urban drainage problems are often designed against hypothetical rain storms derived from long-term observations, and associated with a frequency computed by a combination of the recurrence of their total rainfall and their duration. Commonly, the general method is called wet-event or design-storm methodology. Technical problems associated with this method include: water quality processes are not simply associated with wet events, but more importantly with the preceding dry event; water and moisture levels in the drainage system generally at the onset (start-up) of the event are more important in determining the runoff response than the rain itself; and the derived time-distribution of the design storm oftentimes has a shape that seems remote from reality. Consequently what is preferred is a continuous modeling approach, in which dry-weather processes are modeled in the correct sequence together with the wet-weather processes. However, because of the fast response of urban drainage systems, this continuous methodology requires a long-duration high-time-resolution (LDHTR) input rainfall datafile, usually not available in the existing record, whereas long-duration low-time-resolution (LDLTR) and short-duration high-time-resolution (SDHTR) records are widely available in most hydro-meteorological regions in North America. PCSWMM uses statistical techniques to disaggregate these two types of record into a credible long-duration high time-resolution record for replacing or validating design storms. The investigation reported here first disaggregates real 5y 15-minutely SDHTR and a 60y hourly LDLTR record into a 60y 15-minutely DIS-LDHTR (called the base HR record). Secondly, the study reconstitutes from this a 60y hourly LDLTR (called the base LR record), and then disaggregates the base LDLTR into a derived DER-HR record. The various synthesized records (DER-HR vs DIS-HR; base LDLTR and LDLTR) are compared statistically, and conclusions drawn regarding their applicability. If time permits the performance of various design storms derived from these disaggregated records are compared with that of a continuous model for designs of conveyance (culvert diameter) and flood attenuation (required storage).

Permanent link: