An Improved Linear Runoff Generation Model By Incorporating Potential Evapotranspiration

Abstract:

A widely used runoff generation model is the simple linear model (SLM), which is built on the concept of converting the precipitation to storm runoff by a convolution integral with response function involving a memory span. SLM and its various modifications use rainfall as a sole driver. In this study, we propose to incorporate potential evapotranspiration (PET) as a secondary factor with a separate response function in addition to the main precipitation driver, which we called two-variable linear model (TVLM). The performance of the TVLM is evaluated against the observed data from several catchments in the Poyang Lake Basin in China. In addition, we also estimate the impact of climate variability on the catchment runoff by the TVLM. We demonstrated that incorporating PET improves model performance with the same memory span in the precipitation response function. The precipitation response function is not affected by the addition of PET memory, which indicates that the PET term is indeed a secondary correction factor. The response function of precipitation is always positive, while the response function of PET is mostly negative. The annual runoff results from the SLM are always slightly higher than those from the TVLM, which is attributed to the negative effects of PET on runoff. Similar to the SLM, the proposed TVLM is also based on the linear response concept and can be considered as an extended and improved representation of runoff generation mechanism.