Hydrologic Modeling with Transfer Function Based Approach: A Comparative Study over Godavari River Basin

Abstract

Hydrologic modeling (HM), which involves forming a nexus between two important components of hydrologic cycle viz. rainfall and runoff, is one of the most important steps, which provides realistic inputs for water distribution policies. A plethora of HM methodologies, categorized as ‘conceptual’ and ‘empirical’, have been proposed to estimate the fraction of rainfall, which would be available as surface runoff. However, intensive data requirements, mathematical interpretations of complicated physical processes, and assumptions that are likely to get violated under changing climatic and land-use land cover conditions render the application of conceptual models a cumbersome task. Under such conditions, empirical models play a crucial role in estimating the runoff with minimal data availability. Here, we use two transfer function based approaches viz. linear regression (LR) and kernel regression (KR) for estimating runoff over Godavari river basin. We find that LR outperforms its non-parametric counterpart in capturing long-term properties of observed runoff.