Modelling Dam Water Level Changes in a Drought Affected Catchment Area using Neural Networks

K. Udono and R. Sitte (Australia)

Keywords

Neural networks, Rainfallrunoff model, and Damsimulation

Abstract

This paper presents the use of neural networks to model the dam level dynamics in a drought affected region with irregular rainfall. It is a part of a dynamic model aimed at modelling future water availability of the Gold Coast City, Australia and possible use of seawater desalination powered by waste incineration. The problem lies in the irregularity of rainfall, compounded with the absence of information about the catchment area. It is the absence of a regular time series pattern that makes it difficult for the NN to capture the essence of the water dynamics. We have overcome a number of modelling difficulties and been able to replicate the dam level dynamics with good accuracy. Our paper describes details of the NN model and data preparation and then results from the systematic experiments.

Important Links:



Go Back