Predictive Modeling of Reservoir Inflow for Multi-Step Ahead to Improve Management of Alavian Dam Reservoir

Nikofar, Bagher; Nourani, Vahid; Hosseini Baghanam, Aida

doi:10.22034/ceej.2024.59627.2308

Predictive Modeling of Reservoir Inflow for Multi-Step Ahead to Improve Management of Alavian Dam Reservoir

Authors

¹ Faculty of Civil Engineering, University of Tabriz, , Iran

² Faculty of Civil Engineering, University of Tabriz, Tabriz, Iran

10.22034/ceej.2024.59627.2308

Abstract

Accurate reservoir inflow forecasting is essential for efficient water resources management, particularly under increasing pressure on available water supplies. This study evaluated the performance of an Artificial Neural Network (ANN) and several data-mining models developed in the Weka environment for forecasting inflow to the Alavian Dam reservoir up to 12 months ahead. Monthly data from 1997 to 2022 were used to build the models. The input variables included reservoir inflow, temperature, evaporation, precipitation, basin snow cover, and drought indices. Model performance was assessed during the validation phase and compared across different forecasting horizons. The results showed that all models provided acceptable and reliable predictions. However, ANN, RF, and RT outperformed the other models in overall accuracy. Their average correlation coefficients were 0.85, 0.90, and 0.83, respectively. These models also showed strong agreement with observed data, especially in reproducing peak inflow values. The findings demonstrate the effectiveness of AI-based approaches for multi-step reservoir inflow prediction.

Keywords