Performance comparison of Adoptive Neuro Fuzzy Inference System (ANFIS) with Loading Simulation Program C++ (LSPC) model for streamflow simulation in El Niño Southern Oscillation (ENSO)-affected watershed

Suitable selection of hydrological modeling tools and techniques for specific hydrological study is an essential step. Currently, hydrological simulation studies are relied on various physically based, conceptual and data driven models. Though data driven model such as Adoptive Neuro Fuzzy Inference System (ANFIS) has been successfully applied for hydrologic modeling ranging from small watershed scale to large river basin scale, its performance against physically based model has yet to be evaluated to ensure that ANFIS are as capable as any physically based model for simulation study. This study was conducted in Chickasaw Creek watershed, which is located in Mobile County of South Alabama. Since adequate rain gauge stations were not available near the watershed proximity, and also the study area was affected with the El Niño Southern Oscillation (ENSO), the sea surface temperature (SST) and sea level pressure (SLP) were additionally incorporated in the ANFIS model. The research concluded that ANFIS model performance was equally comparable to a physically based watershed model, Loading Simulation Program C++ (LSPC), especially when rain gauge stations were not adequate. Additionally, the research concludes that ANFIS model performance was equally comparable to that of LSPC no matter whether SST and SLP in ANFIS input vector was included or not.