Sequential Generation of Rainfall and Runoff Data

A practical procedure is demonstrated by applying sequential generation techniques to rainfall and runoff data for stochastic, hydrological analysis of drainage basin systems. In this method, the stochastic process is formulated by several major components including the hourly annual storm rainfalls, the abstractions, the routing model, the baseflow, the direct runoff, and the total runoff. As a practical example, 28 annual storms and the corresponding runoff data recorded in the French Broad River Basin at Bent Creek, North Carolina, are used in the analysis. Conventional models are adopted for separations of the abstractions and the baseflow. The hourly storm rainfalls are represented by a Markov-chain model. The basin system is simulated by a series of equal linear reservoirs. The nonlinearity of the system is considered by varying system parameters. From the historical data, 1,000 annual storms are generated sequentially by the Monte Carlo methods and then routed through the simulated basin system to produce 1,000 generated floods which are represented by stochastic flow-duration curves for use in water resources planning and design.