Hybrid Data-Driven Models for Hydrological Simulation and Projection on the Catchment Scale

Changes in streamflow within catchments can have a significant impact on agricultural production, as soil moisture loss, well frequent drying and wetting, may an effect the nutrient availability of many soils. In order to predict future changes explore different scenarios, machine learning techniques been used recently hydrological sector for simulation streamflow. This paper compares use four models, namely artificial neural networks (ANNs), support vector regression (SVR), wavelet-ANN, wavelet-SVR surrogate models geophysical model simulate long-term daily water level flow River Shannon system Ireland. The performance has tested multi-lag values forecasting both short- time scales. For simulating catchment system, SVR-based performs best overall. Regarding modeling scale, hybrid wavelet-ANN among all constructed models. It is shown that data-driven methods are useful exploring large multi-station catchment, with low computational cost.