Radial basis function artificial neural network (rbfann) model for simulating daily runoff from the himalayan watersheds
Publication details: Roorkee Indian Water Resources Society 2021Edition: Vol.41(1), JanDescription: 41-53pSubject(s): Online resources: In: Journal of indian water resource societySummary: In this paper, a Radial Basis Function Artificial Neural Network (RBFANN) model was developed based on k-means clustering algorithm to simulate the daily rainfall-runoff process in three Himalayan watersheds i.e., Naula, Chaukhutia, and Ramganga located in Uttarakhand State, India. Different network parameters such as learning rate in the function layer (ALR), learning rate in output layer (ALRG), and the number of iterations were optimized. The outcomes of the RBFANN model was evaluated by using statistical (i.e., root mean square error: RMSE, correlation coefficient: CC, and Nash-Sutcliffe efficiency: NSE) and hydrological (i.e., volumetric error: EV) indicators during calibration, cross-validation, and validation phases. The performance of the RBFANN model improved and stabilized within 500 iterations. The model was very sensitive to learning rate in the function layer (ALR), however, not in the output layer (ALRG). Overall results reveal a promising performance of the RBFANN model in simulating the daily runoff in the study catchments.| Item type | Current library | Status | Barcode | |
|---|---|---|---|---|
Articles Abstract Database
|
School of Engineering & Technology (PG) Archieval Section | Not for loan | 2023-0622 |
In this paper, a Radial Basis Function Artificial Neural Network (RBFANN) model was developed based on k-means clustering algorithm to
simulate the daily rainfall-runoff process in three Himalayan watersheds i.e., Naula, Chaukhutia, and Ramganga located in Uttarakhand
State, India. Different network parameters such as learning rate in the function layer (ALR), learning rate in output layer (ALRG), and the
number of iterations were optimized. The outcomes of the RBFANN model was evaluated by using statistical (i.e., root mean square error:
RMSE, correlation coefficient: CC, and Nash-Sutcliffe efficiency: NSE) and hydrological (i.e., volumetric error: EV) indicators during
calibration, cross-validation, and validation phases. The performance of the RBFANN model improved and stabilized within 500 iterations.
The model was very sensitive to learning rate in the function layer (ALR), however, not in the output layer (ALRG). Overall results reveal a
promising performance of the RBFANN model in simulating the daily runoff in the study catchments.
There are no comments on this title.
Click on an image to view it in the image viewer