Rainfall dynamics drive processes that, under certain conditions, severely affect living in urban areas (risk of urban flooding) but also conditions in receiving ecosystems (risk of pollution e.g. through sewer overflows). For a model-based description of these processes, the spatial and temporal resolution of rainfall input can significantly influence the validity of modelling results. Rainfall data with high temporal resolution (in a range of 10 min) is typically required to reliably answer questions related to urban hydrology (Gujer and Krejci, 1988; Schilling, 1991; Segond et al., 2007; Fre-ni et al., 2009). On the other hand, climatic information in the required detail is rare. Spatial coverage is often low, par-ticularly in less developed regions (Onof et al., 2005). Stochastic methods have been developed to generate high-resolution rainfall series from low-resolution data. Cascade models, one method for temporal disaggregation, systemati-cally break down aggregated rainfall information; model parameters are derived from observed time series in high reso-lution. Still, applying such methods has a major drawback: high-resolution reference data are required to estimate model parameters. Aim of the paper is to develop a methodology that allows the temporal disaggregation of available rainfall in case only little or no additional information on rainfall characteristics is available. The main idea is that two stations at different geographical locations may still show similar climatic, i.e. rainfall characteristics. Assuming a sufficient degree of similarity, it may be possible to derive disaggregation model parameters from the station at which adequate data are available (high resolution data) and to use the so calibrated cascade model to disaggregate daily rainfall for the station at which data are not available. For this, a similarity analysis based on multiple regression and multidimensional scaling (MDS) of daily long-term weather data of 58 stations is combined with an established disaggregation method (random cascade model after Olsson, 1998). The practical usefulness of the approach is finally verified through a test case appli-cation. Results demonstrate that locations which are similar regarding typical climate variables also show similarity with regard to rainfall event characteristics. The Euclidian distance derived from the MDS can be used as measure of the de-gree of similarity. The performance of the disaggregation approach is (still) limited, especially regarding high disaggre-gation levels as intensities of short-term events are generally underestimated. It can be concluded that the approach has a high potential to handle data scarcity but but needs further verification and improvement, particularly with regard to the disaggregation approach.