Runoff Quantity Prediction of Potomac River based on Neural Networks Method

Nian Zhang and Shuhua Lai

Keywords

runoff prediction, water quantity prediction, neural networks, backpropagation learning algorithm

Abstract

In urban areas, the impervious surfaces created by buildings and pavement cause rainwater and snowmelt to flow quickly over the landscape, rather than soaking naturally into the soil or being absorbed by plants. This can change stream flows, increase flooding, endanger private and public infrastructure, erode stream banks and channels, and destroy fish habitat. Development activities like clearing vegetation, mass grading, removing and compacting soil, and adding impervious surfaces have increased stormwater runoff in the Chesapeake Bay watershed. Therefore, it is very important to evaluate the stormwater runoff to enhance the performance of an assessment operation and develop better water resources management and plan. In order to accomplish the goal, we proposed a predictive model based on recurrent neural networks trained with the Levenberg-Marquardt backpropagation learning algorithm to forecast the runoff discharge using the past runoff discharge. This computational intelligence modeling tool explored the impact of discharge and gage height to the long-run discharge forecast accuracy. Based on the excellent experimental results including the training, validation and testing errors, error autocorrelation function analysis, regression analysis, and time series response, it showed that the proposed learning algorithm proved to be successful in training the recurrent neural network for the runoff prediction.

Important Links:

Go Back