Evaluating the impact of subsurface hydraulic barriers on Qanat flow rates using quantile regression forest

Qanats, as hydraulic innovations, enabled the sustainable extraction and distribution of groundwater for irrigation and domestic use during history. This study presents a data-driven modeling framework that implements Quantile Regression Forest (QRF), Random Forest (RF), and Support Vector Regression (SVR) to predict Qanat discharge under altered subsurface conditions. Using field data from the Dirsak Qanat in northern Iran, a traditional drainage system recently enhanced by the construction of a subsurface dam (SD), we investigate the dam's effect on discharge potential. The modeling framework incorporates determination of multiple hydro-meteorological inputs including precipitation, temperature, evaporation, humidity, runoff depth, infiltration depth and groundwater levels observed at three monitoring wells. A binary (dummy) variable was also introduced to represent the presence or absence of the SD, thereby capturing the associated changes in boundary conditions. The analysis further revealed that the SD and evaporation are the most influential factors, highlighting the combined effects of anthropogenic modifications and climatic variations on the discharge behavior of the Qanats. It was also concluded that the QRF model with a Nash-Sutcliffe Efficiency (NSE) of 0.818, demonstrate strong predictive capability in capturing complex and non-linear hydrological interactions.