|
Daily runoff prediction using artificial intelligence models
|
Vahid Moosavi *  , Negin Rashidi , Mehdi Vafakhah |
|
|
|
Abstract: (31 Views) |
Rainfall–runoff modeling is one of the common approaches used for estimating runoff and serves as an efficient tool for analyzing hydrological processes, assessing water resources, and optimizing watershed management. Therefore, employing methods that are both dynamic and conceptually clear, while being easy to develop and apply, is essential. In this study, the performance of three artificial intelligence models_Random Forest (RF), Group Method of Data Handling (GMDH), and Adaptive Neuro-Fuzzy Inference System (ANFIS)—was evaluated for daily runoff prediction in two watersheds: Telvar and Taleghan. The input data included 24-hour precipitation, temperature, relative humidity, average wind speed, total snow precipitation, antecedent precipitation index, and antecedent discharge index, collected and computed over a 13-year statistical period. The modeling results for the Taleghan watershed showed that the GMDH model achieved the highest performance, with the greatest coefficient of determination (R² = 0.8845) and Nash–Sutcliffe efficiency (NSE = 0.8836), as well as the lowest root mean square error (RMSE = 4.09), indicating the highest accuracy and lowest simulation error. The RF model, with R² = 0.8801, NSE = 0.8798, and RMSE = 4.16, ranked second and produced relatively acceptable results, although its accuracy was slightly lower than that of the GMDH model. In contrast, the ANFIS model, with R² = 0.8710, NSE = 0.8610, and RMSE = 4.32, demonstrated the weakest performance among the evaluated models. Similarly, the results for the Telvar watershed indicated that the GMDH model achieved the best performance with RMSE = 0.3631 and NSE = 0.939. The ANFIS model ranked second with RMSE = 0.3640 and NSE = 0.9392, while the RF model ranked third with RMSE = 0.4017 and NSE = 0.9260. Overall, the results demonstrated that all three models performed relatively well in simulating runoff variations, suggesting that the application of artificial intelligence models can serve as an effective and reliable tool for surface runoff estimation and hydrological process analysis.
|
Article number: 5 |
| Keywords: Artificial Intelligence, Runoff, Data-Driven Model, Modeling, Machine Learning |
|
|
|
Type of Study: Research |
Subject:
Special
Received: 2025/10/7 | Revised: 2025/11/15 | Accepted: 2025/11/15 | Published: 2025/11/15 | ePublished: 2025/11/15
|
|
|
|
|
|
|
| Send email to the article author |
|
|
|
| Add your comments about this article |
|
|
|
