Disaggregation algorithms aim to produce ensembles of rain rate time-series, with short integration times, consistent with a time-series measured with much longer integration times and some a priori known statistics. To develop and test disaggregation algorithms, it is necessary to have rain rate data measured at the finest scale considered.

Rapid response rain gauges are devices for measuring rain rates with integration times typically less than one minute and often as short as ten seconds. A widely used type of gauge measures the rain water collected in a funnel by forming equally sized drops and detecting their fall. The data recorded by the gauge is the number of drops detected in a 10 second integration time. At the end of each integration period, a partially formed drop will be present within the gauge. For the gauges built and operated by Rutherford Appleton Laboratory (Oxford, UK), each drop in a 10 s period corresponds to a collected rain height of 0.004 mm.

This paper develops a downscaling algorithm capable of producing ensembles of rain rate time-series, with integration times as short as 10 seconds, consistent with a time-series of rain rates with integration times as long as 6 hours. The algorithm is based on a stochastic multiplicative cascade using beta distributions as the random generator. The parameters are estimated from one gauge-year of rain gauge data, with a 10 second integration period, collected in the Southern UK. The statistical moments up to third order, of nine gauge-years of data, are calculated for integration times in the range 10 seconds to 6 hours. These data are compared with time-series derived by accumulating 10 second data to larger integration times, and then downscaling using the proposed algorithm.