Development of a Geomorphology based Model for Prediction of Surface Runoff from Ungauged Watersheds

Abstract

Presence of a plethora of ungauged and inadequately gauged watersheds in India necessitates development of models for these watersheds to predict the hydrologic responses with acceptable accuracy. Besides the Curve Number (CN) approach, the geomorphologic instantaneous unit hydrograph (GIUH) models are adopted by the researchers for estimation of surface runoff from ungauged watersheds. This paper discusses the adequacy of a modified exponential distributed geomorphologic instantaneous unit hydrograph (ED-GIUH) model in generating direct runoff hydrographs (DRHs). The ED-GIUH concept was used to generate the DRHs for the Banha watersheds under Upper Damoder Valley, Jharkhand, India. The estimated runoff using the ED-GIUH concept was compared with the original Natural Resources Conservation Service Curve Number (NRCS CN) generated runoff and validated with the observed runoff data ofthe watershed. The accuracy of estimation of these approaches were compared to ascertain the applicability of ED-GIUH and CN concepts for runoff prediction from ungauged watershed in India. The model input data, including natural drainage network and Horton's morphological parameters were prepared using a watershed morphological estimation tool (WMET) interface of ArcGIS@. The path probability of channel and overland flow were estimated from the generated feature classes of watershed topology and drainage networks to derive the instantaneous unit hydrograph (IUH). Further, from the IUH, an accounting procedure was used to estimate unit hydrograph (UH) and DRHs for different rainfall events occurring over the watershed. It was observed that the ED-GIUH technique were better (coefficient of determination (R²) =0.97, model efficiency (E) =0.96) than the NRCS- CN approach (R² =0.74, E = 0.71) when compared with the observed DRH resulting from rainfall events of 6h duration or less. However, for rainfall events greater than 6h, the ED-GIUH method (R²= 0.71, E = 0.58) failed to predict the surface runoff accurately than the CN method (R² = 0.89, E = 0.84). Thus, facilitated by using ArcGIS@, the ED-GIUH model could be used to predict DRHs from small duration events more accurately than the CN approaches for ungauged watersheds having similar geomorphology as that of the Banha watershed.