Data driven models such as Artificial Neural Networks (ANNs) became a very popular tool in hydrology for a long time, especially in rainfall–runoff modelling. However, it does not have common usage in mountainous catchments, where snowmelt plays an important role, due to lack of continuous snow observations. In order to improve the accuracy of snowmelt modeling, recently available satellite snow products are considered as an alternative input to these models. In this study, two different ANN models are employed and compared with each other using novel MODIS satellite snow covered area products as an alternative input into climatic data based models. Firstly, flows are modelled with Multi-Layer Perceptron (MLP) network using gradient-based Levenberg–Marquardt algorithm. Secondly, Radial Basis Function (RBF) network is developed. Both models are performed to estimate the daily flows of Karasu River in the Upper Euphrates Basin, Turkey using 2002 – 2011 data. The main difference between the RBF network and MLP network is in the nature of the nonlinearities associated with hidden nodes. The nonlinearity in MLP is implemented by a fixed function such as a sigmoid. On the other hand, the RBF method bases its nonlinearities on the training set data. In the study the determination of model architectures, optimization algorithms and methods to avoid overfitting are elaborately investigated
Optimal control of reservoirs is a challenging task due to conflicting objectives, complex system structure, and uncertainties in the system. Real time control decisions suffer from streamflow forecast uncertainty. This study aims to use Probabilistic Streamflow Forecasts (PSFs) having a lead-time up to 48 h as input for the recurrent reservoir operation problem. A related technique for decision making is multi-stage stochastic optimization using scenario trees, referred to as Tree-based Model Predictive Control (TB-MPC). Deterministic Streamflow Forecasts (DSFs) are provided by applying random perturbations on perfect data. PSFs are synthetically generated from DSFs by a new approach which explicitly presents dynamic uncertainty evolution. We assessed different variables in the generation of stochasticity and compared the results using different scenarios. The developed real-time hourly flood control was applied to a test case which had limited reservoir storage and restricted downstream condition. According to hindcasting closed-loop experiment results, TB-MPC outperforms the deterministic counterpart in terms of decreased downstream flood risk according to different independent forecast scenarios. TB-MPC was also tested considering different number of tree branches, forecast horizons, and different inflow conditions. We conclude that using synthetic PSFs in TB-MPC can provide more robust solutions against forecast uncertainty by resolution of uncertainty in trees.