A universal decay function based meteorologically-driven and calibration-free runoff generation module

An accurate estimate of streamflow has been a challenging task due to the complex and interconnected hydrological processes. A simple, robust and calibration-free runoff generation module is desirable in several water resource management applications. But spatial heterogeneity of, but not limited to, topography, soil and land cover makes it challenging to develop desirable runoff generation module. To address this, several runoff generations theories that apply laws of physics at the grid-scale were proposed and tested.  However, these theories have not shown a significant difference in performance on considering catchment as spatially distributed and a single unit(lumped). A typical runoff generation module follows saturation excess or infiltration excess mechanism for runoff generation. The root zone storage capacity (S_max), which controls the dynamics of water storage and partitioning of available water into different fluxes, is an important free-parameter in the saturation excess mechanism. The value of S_max needs to be estimated using observed streamflow time series. However, recent studies demonstrate that the S_max is controlled by local climate and land cover. So, in the current study, we hypothesized that runoff generation is solely governed by climate input and the amount can be estimated without explicit consideration of S_max. We tested the hypothesis using Dynamic Budyko (DB) framework, which simulates the runoff at a daily time scale using ‘instantaneous dryness index (Φ)’. We proposed a universal decay function to predict Φ using rainfall and potential evapotranspiration. The performance of proposed runoff generation module is compared with HBV and GR4J runoff generation modules for 416 MOPEX study basins. The proposed calibration free runoff generation module shows very similar performance to that of calibrated HBV and GR4J with median NSE as 0.68,0.7 and 0.68, respectively. The proposed framework can be coupled with any routing module to estimate the streamflow at basin outlet. The introduction of proposed framework address several long-term challenges in rainfall-runoff modeling. Our results suggest that more efforts should be considered in developing rainfall-runoff modeling frameworks that exploit information available in meteorological input to address streamflow dynamics.