Disaggregation model for synthetic stream-flow generation

To capture the complexity of a water resources system, synthetic data generation is an essential component. Frequently, the data generation is done on an annual basis and disaggregated to smaller time scales. A generalised disaggregation framework is presented to generate seasonal stream-flows from any annual autoregressive process. A new periodic disaggregation scheme is proposed for further disaggregation into sub-seasonal flows from seasonal flows generated with a periodic autoregressive (PAR) model of any order. The new model preserves the first and second moments and has been applied to the Ganges river at Farakka in India for generation of decadal (10-day) flows from monthly flows; the 10-day period being the discrete time interval identified in the Ganges Water Treaty. The results demonstrate that the proposed coupled modelling scheme works very well and provides a flexible choice in synthetic hydrology.