Computational approaches for annual maximum river flow series
Studies of annual peak discharge and its temporal variations are widely used in the planning and decision making process of water resources management. Very recently, soft computing techniques are gaining ground for time series analysis of hydrological events such as rainfall and runoff. In this stu...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2017-03-01
|
| Series: | Ain Shams Engineering Journal |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2090447915001422 |
| Summary: | Studies of annual peak discharge and its temporal variations are widely used in the planning and decision making process of water resources management. Very recently, soft computing techniques are gaining ground for time series analysis of hydrological events such as rainfall and runoff. In this study Artificial Neural Network (ANN) has been used in combination with wavelet to model the annual maximum flow discharge of rivers. The results of ANN-Wavelet (WANN) model indicate overall low coherence (R2 = 0.39) better than ANN (R2 = 0.31) in isolation. In the present analysis, the authors also conceded a probabilistic distributional analysis of river flow time series which has greater potential to better reflect peak flow dynamics. The results highlight that the overall performance of probability distribution models is superior to WANN model. Instead of that WANN is better than probabilistic models to find the global maxima of the series. |
|---|---|
| ISSN: | 2090-4479 |