Ormsbee (1989) proposed a simple approach for disaggregation of hourly precipitation data into synthetic high frequency data, as is often needed in urban hydrology studies. Ormsbee presented algorithms for “discrete” disaggregation into 20-minute data, and for “continuous” disaggregation into data of any divisor of 60 (typically 5-minute or 15-minute data). Deterministic and stochastic algorithms were presented for implementing both the discrete and continuous methods. The discrete deterministic approach disaggregates a single hour’s data into three 20-minute increments based on linear scaling of the distribution of rainfall among the preceding, current, and immediately subsequent hour of the rainfall time series. For example, if rainfall totals in three successive hours were 10, 30, and 20 millimeters, the hour with 30 mm could be disaggregated into 20-minute increments with respective rainfall depths of 5, 15, and 10 mm (i.e. successive intensities of 15, 45, and 30 millimeters per hour). The deterministic continuous disaggregation algorithm could be used to break this hour of rainfall into synthetic 15-minute increments of 3.75, 8.75, 10, and 7.5 millimeters (intensities of 15, 35, 40, and 30 millimeters per hour).

Ormsbee’s algorithms have subsequently been shown to be reasonably effective, though yielding a negative bias when compared with actual high frequency data (Durrans et al., 1999). Improved methodologies have been proposed using artificial neural networks (Burian et al., 2001), and using a regression model to improve the estimation of maximum short duration rainfall intensity (Cowpertwait, 2001).

While the recent investigations establish promising alternatives to Ormsbee’s method, NetSTORM incorporates Ormsbee’s algorithms with two minor improvements. The simplicity of the original approach facilitates its explanation to the public, thus making it more readily accepted for inclusion in engineering studies. The NetSTORM implementation of the continuous disaggregation algorithm allows stochastic disaggregation to either maintain the existing numerical significance of the original data, or to further resolve it by the ratio of the original time series interval to the new interval. For example, if hourly data were measured in millimeters, NetSTORM could create a 6-minute output time series resolved to millimeters, or could produce synthetic data with 0.1 mm discretization. (For precipitation data measured in tenths of inches, the output can be in tenths of inches using the original algorithm, or resolved to 0.01 in using the modification implemented in NetSTORM). NetSTORM’s second improvement to the original algorithm allows the disaggregation of sub-hourly data to a finer scale, such as conversion of 15-minute data to 5- or 1- minute resolution.