Building Data-Driven Hydrological Models: Data Issues

In Chapter 2 the conceptual framework for hybrid modelling has been presented. Data-driven modelling (DDM) constitutes an important part of this framework, and deserves more detailed presentation. The main objective of this chapter is to analyze the data-driven modelling process and the most important factors typically considered when building such models. The machine learning (computational intelligence) techniques used (artiflcial neural networks, Model Trees and Support Vector Machines) are characterized here, but (briefly) explained in Appendix B . Part of results presented here has been worked out and discussed in Elshorbagy et al. ( 2009a , b ).