DEVELOPMENT OF SCS-CN METHOD AND AUTOREGRESSIVE TIME SERIES MODEL FOR THE ESTIMATION OF RUNOFF OF UPPER SEWANI WATERSHED OF DAMODAR CATCHMENT, JHARKHAND, INDIA

GLADWIN CUTTING NIKHIL¹, SANDEEP KUMAR PANDEY²*, MAHESH PRASAD TRIPATHI³, M. IMTIYAZ^4
1Department of Soil Water Land Engineering and Management, Sam Higginbottom Institute of Agriculture, Technology and Sciences, Allahabad, U.P., India
²Department of Soil and Water Engineering, Punjab Agricultural University, Ludhiana, Punjab, India
³College of Agricultural Engineering, Jawaharlal Nehru Krishi Vishwa Vidyalaya, Jabalpur, Madhya Pradesh, India
⁴Vaugh School of Agricultural Engineering and Technology, Sam Higginbottom Institute of Agriculture, Technology and Sciences, Allahabad, U.P., India
* Corresponding Author : pandey.vikku@gmail.com

Received : 18-03-2016     Accepted : 05-05-2016     Published : 03-09-2016
Volume : 8     Issue : 32       Pages : 1668 - 1672
Int J Agr Sci 8.32 (2016):1668-1672

Keywords : Curve Number Method, Time Series Model, Rainfall-Runoff
Academic Editor : Dr Jagvir Dixit
Conflict of Interest : None declared
Acknowledgements/Funding : None declared
Author Contribution : None declared

The present experiment was conducted the prime objective, to generate and estimate the curve number for runoff and also to develop autoregressive (AR), for the prediction of rainfall, from the study area and estimate the parameters of runoff. Autoregressive (AR) models of orders 0, 1 and 2 were tried for annual stream flow series. Parameters were estimated by the general recursive formula proposed by [24]. The adequacy of models and goodness of fit were tested by Box-Pierce Portmanteau test, Akaike Information Criterion (AIC) and by comparison of historical and predicted correlogram. The AIC value for AR (1) model (141.855) was lying between AR (0) (142.764) and AR (2) (152.749) which is satisfying the selection criteria. The mean forecast error was also very less. On the basis of the statistical test, Akaike Information Criterion, AIC the AR (1) model with estimate model parameters, estimated for the best future predictions in Upper Sewani watershed. This is graphical representation between historical and generated correlogram, where in runoff there was a very close agreement. The performance comparison of both the models was made with the coefficient of determination (R2) which was 0.988 in case of SCS- Curve Number and 0.946 in case of Autoregressive Time Series Model. On further comparison it shows that autoregressive is giving much better results than the Curve Number method, so it can be more trust worthy.