H43I-1676
Assessment of future change in streamflow through identification of its association with precipitation in Mahanadi Basin, India under data scarce scenario

Thursday, 17 December 2015
Poster Hall (Moscone South)
Aditya Gusain1, Mohit Prakash Mohanty1,2, Vittal H1, Subhankar Karmakar1, Subimal Ghosh3 and Srinivasa Rao G4, (1)Indian Institute of Technology Bombay, Centre for Environmental Sciences and Engineering, Mumbai, India, (2)Indian Institute of Technology Bombay, Mumbai, India, (3)Indian Institute of Technology Bombay, Department of Civil engineering, Mumbai, India, (4)National Remote Sensing center, Hyderabad, India
Abstract:
It is evident that change in climate alters the distribution of rainfall at different spatial and temporal scales all over the world. Quantification of change in streamflow through an understanding of the changes in rainfall pattern is challenging, particularly under data scarce situation. In order to accomplish this goal, we studied the statistical association between observed rainfall and streamflow in a large river basin, prone to various hydrologic extremes, particularly floods and droughts. One of the flood prone river basins in India, Mahanadi basin (covering an area of 0.14 million sq. km.) was selected and divided into number of sub-catchments. A set of statistical relationships between regional rainfall and streamflow for each sub-catchment was derived, which was assumed to be remained unaltered in future time period. Based on a comprehensive literature survey, 7 GCMs were selected for rainfall projection, which were reported as suitable particularly for the Indian subcontinent. General Circulation Models (GCMs) can be used to enumerate rainfall by considering various predictors such as temperature, humidity, wind velocity, geo-potential height etc. However, the difficulty lies in the coarser resolution of GCMs output, which ultimately restricts their application at a regional scale. Under such circumstances, downscaling techniques are used to project the rainfall at a required finer catchment scale. In the current study, a well recognized statistical downscaling technique utilizing classification and regression tree, and kernel regression was used. For validating purpose, the statistically derived streamflows were compared with the streamflow outputs simulated through the Variable Infiltration Capacity (VIC) model. The proposed methodology for assessment of changing pattern of streamflow under data scarce situation will be useful for sustainable management of water resources and future flood characterization.