A Hybrid Model for Forecasting Daily Urban Water Demand

Water as one of the most important human needs plays a vital role in everyday life. Therefore, awareness of the required amount of water demand is of particular importance to policy-making in line with demand management. In this study, a hybrid model (a combination of linear and non-linear models) has been designed to predict short-term urban water demand which mathces with the climatic conditions and structure of Tehran and variables affecting the water consumption. With using the model, daily urban water demand for the next 10 days was predicted based by ARIMA, Artificial Neural Netwoks (ANN) and Wavelet Transform hybrid models. Then the forecasted values of mentioned models were evaluated by MAPE and R2 criteria in step-by-step and full 10 days predictions. Finally, wavelet transform hybrid model with low error (high prediction accuracy) was selected as an optimal model for predicting daily water demand of Tehran.