A stochastic model for daily rainfall disaggregation into fine time scale for a large region

A robust model for disaggregation of daily rainfall data at a point within a large region to any fine timescale of choice is presented. Limited fine timescale data are required to calibrate only three parameters for the regional model, to establish monthly variation of simulation timescale lag-1 autocorrelations, and also to establish a scaling law between the simulation timescale and the 24-h aggregation levels. Site specific parameters are obtained using the 24-h statistics to disaggregate a long record of daily data by repetition and proportional adjusting techniques with capping. An Australia-wide data set has been used as a case study to illustrate the capability of the model. It has been demonstrated that the disaggregation model predicts very well the gross statistics (including extreme values) of rainfall time series down to 6-min timescale. The possibility of linking the disaggregation model to daily, or global circulation, models that can capture the inter-annual variability of the rainfall process for simulation beyond the number of years of record is being explored.