A genetic programming approach to rainfall-runoff modelling

Planning for sustainable development of water resources relies crucially on the data available. Continuous hydrologic simulation based on conceptual m odels has proved to be the appropriate tool for studying rainfall-runoff pr ocesses and for providing necessary data. In recent years, artificial neura l networks have emerged as a novel identification technique for the modelli ng of hydrological processes. However, they represent their knowledge in te rms of a weight matrix that is not accessible to human understanding at pre sent. This paper introduces genetic programming, which is an evolutionary c omputing method that provides a 'transparent' and structured system identif ication, to rainfall-runoff modelling. The genetic-programming approach is applied to flow prediction for the Kirkton catchment in Scotland (U.K.). Th e results obtained are compared to those attained using two optimally calib rated conceptual models and an artificial neural network. Correlations iden tified using data-driven approaches (genetic programming and neural network ) are surprising in their consistency considering the relative size of the models and the number of variables included. These results also compare fav ourably with the conceptual models.