Prediction of River Discharge Using Neural Networks
Abstract
Flow prediction is an essential aspect in many of the activities associated with the planning and
operation of the components of water resources system. Reliable stream flow prediction model is
important in assisting water resources managers and engineers in allocation of water to
competing users like hydropower, irrigation and domestic. Hydrologic component requires both
short term and long term forecasts of stream flow events in order to optimize the system or to
plan for future expansion or reduction (Kisi 2005).
Several models for stream flow prediction exist which according to (Talaee, 2012 and Kisi et al.,
2007) may broadly be grouped under three main categories: (1) conceptual models (2) physically
based models and (3) black-box models or data-driven models. Conceptual models rely on
simple arrangement of relatively small number of interlinked conceptual elements, each
representing a segment of land phase of a hydrologic cycle. The physically based models are
specifically designed to mathematically simulate or approximate (in some physically realistic
manner) the general internal sub-processes and physical mechanisms that govern the stream flow
process, whereas the black-box models or data-driven models are designed to identify the
connection between the inputs and the outputs, without going into the analysis of the internal
structure of the physical process. Data-driven models provide an alternative to physically based
model due to its complexity (Lin et al., 2009).