Analysis of rainfall time series with regard to temporal disaggregation for the use in urban hydrology

The temporal resolution of rain data recorded by national weather services is often lower (10-min resolution) than the data needed for rainfall-runoff simulations (1-5 min resolution). However, the time series are sufficiently long for statistical analysis. The goal of the project presented here is to develop a procedure that disaggregates the 10-min rainfall data of the Swiss Meteorological Institute (SMI) to 1-min time resolution. The aim is to implement the procedure in a computer program which can be used by practitioners for generating high resolution rainfall time series at a specific location. In a first step the 1-min data and the 10-min data of two rainfall time series in Switzerland (Heiden, Lucerne) were compared to find characteristic patterns and correlations. The position of 1-min peak within the 10-min interval, the ratio of 1-min peak to 10-min intensity as well as the distribution of the 1-min values within the 10-min interval were investigated. The analysis showed that the positions of the 1-min peaks within a 10-min interval were not uniformly distributed. The distribution depended on the temporal trend of the 10-min values. The ratio of 1-min peak to 10-min intensity seemed to tend towards a constant ratio for high intensities. The results for the two series (Heiden and Lucerne) were not significantly different. These first findings are encouraging with regard to developing a disaggregation to 1-min values which satisfy the accuracy needed in urban hydrology. (C) 1998 Published by Elsevier Science Ltd. All rights reserved.