ABSTRACT
Hydrological forecasting is the estimation of hydrological and meteorological phenomena for a special future time span. Hydrological forecasting can be categorized into two classes including short-term and long-term hydrological forecasts. Short-term hydrological forecasts often refer to a period of up to 2 days and apply flood warning systems and real-time operation of water resources systems. On the other hand, long-term hydrological forecasts refer to a period exceeding 1 week up to 1 year and apply water resources management. A hydrological forecasting service combines real-time and historical data inputs with hydrological models and modeling. Many forecasting techniques for carrying out relationships among the independent and dependent variables have been developed. Conceptual models also the same as statistical models establish relationships among the independent and dependent variables. The main difference of these models compared with statistical models is in their calibration process. The calibration process of these models is usually also performed based on experimental methods or innovation methods such as artificial intelligence (AI). The efficient techniques in AI are the artificial neural networks (ANNs), adaptive neuro-fuzzy inference system (ANFIS), and support vector machine (SVM). During the recent decades, the ANNs, ANFIS, and SVM have been applied and regarded as three of the most successful tools in the various areas of hydrology. The AI methods provide better estimations than the conventional physical methods. The AI models are suitable in various applications such as flood and drought forecasting, rainfall–runoff modeling, water quality, groundwater forecasting, and regionalization of flood and low flow.
