Supplements to the HEC-1 hydrologic model using the Monte Carlo method and artificial neural network (Oklahoma).

摘要

The HEC-1 rainfall runoff model, developed by the U.S. Army Corps of Engineers at Davis, is a one-dimensional analysis model. This model is used for forecasting the expected runoff process for a basin. The Glover Basin, located in the southeastern portion of Oklahoma, has been difficult to model in a timely and accurate manner using the HEC-1 software. This study introduces two additional tools to supplement the HEC-1 model without much effort for the hydrologic engineer. The Monte Carlo method was used to provide a forecaster a statistically based range of peak flow discharge values for any storm event being modeled. The Monte Carlo method randomly selects various input parameter values for the HEC-1. A program was developed to select a range of input values that the basin has historically encountered since 1997.; Doppler data provided by the River Forecast Center at Tulsa were used for this study for rainfall input values to the HEC-1 model. Doppler radar data were also used in this study for input to several Artificial Neural Network models (ANN). The ANN model is capable of identifying nonlinear relationships between input and output data. This type of modeling scheme has been found to be useful in problems for which physical equations cannot describe the characteristics of the problem. A one-hour, three-hour, six-hour, and nine-hour forecast for the Glover basin was conducted using ANN models to predict discharge data for the basin. These modeling schemes do not burden a forecaster in learning new methods but enhance the forecaster with tools that can be implemented fairly easily. The Monte Carlo method aids the hydrologic engineer by producing a range of peak flow values for various average conditions expected for the basin. This method compliments the HEC-1 model by providing results of any storm event before the runoff reaches the point of interest. The ANN model also aids the engineer in predicting discrete discharge values to which the calibration of the HEC-1 model can begin. These two methods enhance the forecast modeling scheme by producing predicted information hours before the hydrograph from any storm event begins.