Polynomial-based disaggregation of hourly rainfall for continuous hydrologic simulation

Hydrologic modeling of urban watersheds for designs and analyses of stormwa ter conveyance facilities can be performed in either an event-based or cont inuous fashion. Continuous simulation requires, among other things, the use of a time series of rainfall amounts. However, for urban drainage basins, which are typically small, the temporal resolution of the rainfall time ser ies must be quite fine, and often on the order of 5 to 15 minutes. This pos es a significant challenge because rainfall-gauging records are usually kep t only for hourly or longer time steps. The time step sizes in stochastic r ainfall generators are usually also too large for application to urban runo ff modeling situations. Thus, there is a need for methods by which hourly r ainfall amounts can be disaggregated to shorter time intervals. This paper presents and compares a number of approaches to this problem, which are bas ed on the use of polynomial approximating functions. Results of these evalu ations indicate that a disaggregation method presented by Ormsbee (1989) is a relatively good performer when storm durations are short (2 hours), and that a quadratic spline-based approach is a good choice for longer-duration storms. Based on these results, the Ormsbee technique is recommended becau se it provides good performance, and can be applied easily to long time ser ies of precipitation records. The quadratic spline-based approach is recomm ended as a close second choice because it performed the best most consisten tly, but remains more difficult to apply than the Ormsbee technique. Result s of this study also indicate that, on average, all of the disaggregation m ethods evaluated introduce a severe negative bias into maximum rainfall int ensities. This is cause for some well-justified concern, as the characteris tics of runoff hydrographs are quite sensitive to maximum storm intensities . Thus, there is a need to continue the search for simple yet effective hou rly rainfall disaggregation methods.