Assessment of the effective width of riparian buffer strips to reduce suspended sediment in an agricultural landscape using ANFIS and SWAT models

Abstract

Riparian buffers are important features that help to mitigate suspended sediment loads within rivers running through agricultural landscapes. Evaluating their effectiveness for sediment control by different modelling approaches can help direct beneficial management practices. The Soil and Water Assessment Tool (SWAT) model and the Adapted Neuro-Fuzzy Inference System (ANFIS) based model were used for prediction of suspended sediment concentrations (SSC) and sediment loads in the Mill River watershed (PEI, Canada). Those models were then used to assess the impact of riparian buffer widths in reducing sediment loads. The ANFIS model predicted measured SSC more accurately than the SWAT model. The relationship between buffer width and sediment reduction was asymptotic, and the relationship begins to plateau when the width reaches 50 m. Increasing the buffer width from 15 to 100 m led to an increase in sediment loads retention of 30.5% and 36.2% of the total stream sediment load for the SWAT and ANFIS models, respectively. This study highlighted that a data-driven ANFIS based model can be used to simulate the impact of land use changes on the sediment delivery in a river.