Forecasting Snowmelt Runoff: Probabilistic Model

A runoff forecasting technique has been developed and applied to a small watershed in Central Arizona. A synthesized precipitation sequence obtained from an event-based stochastic model provided the inputs for a deterministic watershed snowmelt model. To account for possible combinations of precipitation sequences, a number of simulations were made which resulted in a relative frequency distribution of total seasonal runoff. In general, no significant difference was found to exist between observed and simulated precipitation data. Insufficient data prevented obtaining reasonable watershed model parameters through optimization. Nevertheless, the forecasting technique appears promising to account for uncertainty in future precipitation.