A comparative study of honey-bee mating optimization algorithm and support vector regression system approach for river discharge predictionCase study: Kashkan River Basin

Accurate forecasting of river flow is an important aspect of water resource planning and management especially in water scarce areas of developing countries. In this study, application of the honey bee mating optimization algorithm in comparison with a robust data-driven method such as support vector regression (SVR) assessed and proved to be an efficient forecasting tool for developing water resource programs. In order to achieve this purpose, four key factors such as antecedent daily river flow, soil infiltration rate, daily potential evapotranspiration and average daily precipitation were used to develop the models. The training and verification data sets were allocated respectively from 75% and 25% of an 18 year river flow, precipitation and temperature data series which were available from years 1993 to 2011. Three quantitative standard statistical performance evaluation measures of root mean squared error (RMSE), correlation coefficient (CC) and regressive coefficient (R2) were employed to evaluate the performances of the models. The best-fit model performance indicators, RMSE, CC and R2 returned values within the ranges 18.05–32.24, 0.86–0.96 and 0.6–0.89 respectively, considering both calibration and verification data for the two mentioned methods. A comprehensive comparison of the overall performance indicated that the HBMO model performed better than SVR in river flow forecasting for the verification period

A comparative study of honey-bee mating optimization algorithm and support vector regression system approach for river discharge predictionCase study: Kashkan River Basin Keywords:

River discharge prediction , Honey-bee mating optimization , Support vector regression , Kashkan River