Abstract: High-resolution precipitation data are crucial for continuous hydrological modelling of the urban and small catchments with a short time of runoff concentration. In this study, we simulate hourly precipitation fields for nine meteorological stations surrounding Nanjing, using three daily-to-subdaily precipitation disaggregation models－i.e. single-site, regional, and multi-site disaggregation models－to disaggregate the daily precipitation data of the flood seasons (May to September) of 1983-2012. To evaluate the performance of these models, we apply multiple performance measures, including distributional statistics, extreme value distribution, spatial correlation, and intra-day wet and dry spell characteristics. Results show all the models give satisfactory simulations of distributional statistics, precipitation extremes, and intra-day wet and dry spell characteristics. Among them, only the multi-site model successfully simulates the spatial correlation of the observed hourly precipitation field, and it is more appropriate for distributed hydrological modelling. The results are useful for urban rainfall-runoff modelling and analysis.

Key words: precipitation simulation, hourly precipitation, typical disaggregation model, stochastic resampling, spatial correlation