A Neyman-Scott cluster process model for hourly precipitation is developed that accounts for seasonality by the model parameters being continuous sinusoidal functions of time. The process is fitted to a set of 27 years of hourly rainfall from Stuttgart, Germany. We give a detailed description of the fitting procedure which is based on a combination of the method of moments with a least squares approach. As part of the validation check we compare the second-order moments under the model with their empirical counterparts. These moments are mean, variance and lag(1)-autocorrelation of the series of hourly rainfall depths at several time scales, the probability of an interval being dry and the transition probabilities of wet intervals. Further checks are carried out on simulated time series which include the distribution of run length and of annual maximum rainfalls. The model turns out to give a good description of various features of hourly precipitation.
