Daily Streamflow Simulation Using Recession Characteristics

Time series of daily streamflows are studied with particular emphasis on the behavior of recessions. Long-term and periodic characteristics are modeled through spectral analysis. Gamma autoregressive models are used to simulate time intervals D between consecutive low flows and the conditional duration d of recessions. A similar model is also used for generating the magnitudes of low flows, conditional on D. A nonnormal ARIMA process is adopted for simulating a recession backward in time over a time interval d commencing with the associated low flow. Transformation to stationarity of the recession process is by means of a difference-ratio, which is the difference between the current and antecedent streamflows divided by their average. The rising limb of the hydrograph is simulated using a ranked uniform random number generator. Application is made to four river basins ranging in size from the Po at Pontelagoscuro (70,090 km^2) in northern Italy to the Arzino (123 km^2) in northeast Italy. Hydrograph characteristics and the distributions of average, low, and peak flows are maintained.