نوع مقاله : مقاله پژوهشی
نویسندگان
1 دانش‌آموخته دکتری، گروه آبخیزداری، دانشگاه علوم کشاورزی و منابع طبیعی ساری، مازندران، ایران
2 استاد گروه مرتع و آبخیزداری، دانشگاه فردوسی مشهد، مشهد، ایران
3 دانشیار گروه آبخیزداری، دانشگاه علوم کشاورزی و منابع طبیعی ساری، مازندران، ایران
چکیده
کلیدواژه‌ها
عنوان مقاله [English]
نویسندگان [English]
In fact determine a reliable models and selection of inputs with proper temporal delay for river flow forecasting is a key topic for watershed managers, hydrologists, river engineers. In recent decades use of intelligent algorithms and fuzzy collection theories for modeling of hydrological phenomena that have complexity and uncertainly, has been noticed by researchers. In this regard, in the present study was used adaptive neuro-fuzzy inference system (ANFIS) and different input patterns of flow discharge (with daily time steps up to 7 days ago) in order to river flow forecast of Kasilian catchment. Then in order to further investigate of this process artificial neural network (ANN) model was used and the results were evaluated using quantitative statistics including coefficient of determination (R2) and root mean square error (RMSE). In ANFIS model the results of river flow prediction was improved up 4-day step and afterward decreased while in ANN up 5-day step improved. Evaluation of quantitative statistics values for the best patterns of two models applied during validation period indicated that ANFIS (R2 = 0.60, RMSE = 0.64) had high accuracy than ANN (R2 = 0.51, RMSE = 1.74) in river flow forecasting.
کلیدواژه‌ها [English]
)