Himalayan watersheds are characterized by mountainous topography and a lack of available data. Due to the complexity of rainfall–runoff relationships in mountainous watersheds and the lack of hydrological data in many of these watersheds, process-based models have limited applicability for runoff forecasting in these areas. In light of this, accurate forecasting methods that do not necessitate extensive data sets are required for runoff forecasting in mountainous watersheds. In this study, multivariate adaptive regression spline (MARS), wavelet transform artificial neural network (WA-ANN), and regular artificial neural network (ANN) models were developed and compared for runoff forecasting applications in the mountainous watershed of Sainji in the Himalayas, an area with limited data for runoff forecasting. To develop and test the models, three micro-watersheds were gauged in the Sainji watershed in Uttaranchal State in India and data were recorded from July 1 2001 to June 30 2003. It was determined that the best WA-ANN and MARS models were comparable in terms of forecasting accuracy, with both providing very accurate runoff forecasts compared to the best ANN model. The results indicate that the WA-ANN and MARS methods are promising new methods of short-term runoff forecasting in mountainous watersheds with limited data, and warrant additional study.
Skip Nav Destination
Article navigation
Research Article|
October 11 2011
Comparison of multivariate adaptive regression splines with coupled wavelet transform artificial neural networks for runoff forecasting in Himalayan micro-watersheds with limited data
Jan Adamowski;
1Department of Bioresource Engineering, McGill University, 21 111 Lakeshore Road, Ste. Anne de Bellevue, QC, Canada H9X 3V9
E-mail: [email protected]
Search for other works by this author on:
Hiu Fung Chan;
Hiu Fung Chan
1Department of Bioresource Engineering, McGill University, 21 111 Lakeshore Road, Ste. Anne de Bellevue, QC, Canada H9X 3V9
Search for other works by this author on:
Shiv O. Prasher;
Shiv O. Prasher
1Department of Bioresource Engineering, McGill University, 21 111 Lakeshore Road, Ste. Anne de Bellevue, QC, Canada H9X 3V9
Search for other works by this author on:
Vishwa Nath Sharda
Vishwa Nath Sharda
2Central Soil and Water Conservation Research and Training Institute, 218 Kaulagarh Road, Dehradun 248 195, India
Search for other works by this author on:
Journal of Hydroinformatics (2012) 14 (3): 731–744.
Article history
Received:
April 08 2011
Accepted:
September 16 2011
Citation
Jan Adamowski, Hiu Fung Chan, Shiv O. Prasher, Vishwa Nath Sharda; Comparison of multivariate adaptive regression splines with coupled wavelet transform artificial neural networks for runoff forecasting in Himalayan micro-watersheds with limited data. Journal of Hydroinformatics 1 July 2012; 14 (3): 731–744. doi: https://doi.org/10.2166/hydro.2011.044
Download citation file: