Assessing the return flow in human-induced rivers using data-driven and hydrologic models case study: Ergene River Basin

Abstract

We propose a method for distinguishing return flow from streamflow time series in rivers located in urban and industrial watersheds. The trend analysis is conducted using the Mann-Kendall method, while streamflow time series are divided into pre- and post-change sets using the Mann-Whitney change point method. The Two-Parameter Filtering (TPF) is utilized to separate baseflow before determining the return flow. By employing modified Local Minimum Flow (LMF) method, the TPF results are decomposed into groundwater baseflow and return flow. The LMF method was adjusted by modifying the time interval based on low flow during the dry season. To validate the calculated subsurface flow and runoff, the physical method is verified using the Soil and Water Assessment Tool (SWAT). After calibrating the model for the pre-change period, modeling is conducted for the entire streamflow time series. This method is applied to industrial land that has proliferated over the past two decades upstream of the Ergene River Basin in the European part of Turkiye. The proposed method estimates return flows with satisfactory accuracy and reliability, representing more than half of the baseflow. Since 1990, the return flow has increased from 5 to 38 percent of river streamflow. The research demonstrates that conventional filtering methods could not reliably separate the baseflow of human-induced rivers. A reanalysis is necessary to determine the actual groundwater baseflow and return flow based on the results obtained from these methods.