Prediction of daily suspended sediment load (SSL) using new optimization algorithms and soft computing models

Abstract Accurate modeling and prediction of suspended sediment load (SSL) in rivers have an important role environmental science design engineering structures are vital for watershed management. Since different parameters such as rainfall, temperature, discharge with the lag times significant effects on SSL, quantifying understanding nonlinear interactions dynamics has always been a challenge. In this study, three soft computing models (multilayer perceptron (MLP), adaptive neuro-fuzzy system (ANFIS), radial basis function neural network (RBFNN)) were used to predict daily SSL. Four optimization algorithms (sine–cosine algorithm (SCA), particle swarm (PSO), firefly (FFA), bat (BA)) improve capability SSL models. Data from gauging stations at mouth Kasilian Talar northern Iran analysis. The selection input combinations was based principal component analysis (PCA). Uncertainty sequential uncertainty fitting (SUFI-2) performance indicators assess potential Taylor diagrams visualize match between model output observed values. Assessment predictions station revealed that ANFIS-SCA yielded best results (RMSE (root mean square error): 934.2 ton/day, MAE (mean absolute 912.2 NSE (Nash–Sutcliffe efficiency): 0.93, PBIAS: 0.12). also (RMSE: 1412.10 MAE: 1403.4 NSE: 0.92, 0.14). diagram confirmed achieved predicted values various hydraulic hydrological both stations. Further, tested Eagel Creek Basin, Indiana state, USA. indicated reduced RMSE by 15% 21% compared MLP-SCA RBFNN-SCA training phase. Comparing could decrease error ANFIS-BA, ANFIS-PSO, ANFIS-FFA, ANFIS 18%, 32%, 37%, 49% phase, respectively. integration can ability predicting Additionally, hybridization