Search

A model comprising blocks of artificial neural networks (ANNs) combined in sequence was used to simulate the inflow and outflow in a water resources system under a shortage of water. We assessed the selection of appropriate input data using linear and non-linear cross-correlation functions and sensitivity analysis. The potential model inputs were flow, precipitation and temperature data from various gauging stations throughout the upper watershed of the 'Guadiana Menor' River (southern Spain), and the model considered various input time lags. The ANNs based on the selected inputs were effective relative to those with no relevant inputs, and produced more parsimonious models. We also investigated conceptual analogies inherent in the ANN models by analyzing the response profiles of the modelled variables (inflow and outflow) in relation to each of the selected input data. The results demonstrate that the neural approach approximated the behaviour of various components of the water resources system in terms of various hydrologic cycle processes and management rules. Our findings suggest that in dry periods a mean temperature increase of 1°C in low altitude locations of the region will result in a mean decrease of approximately 2% in the inflow to the water resources system, and a mean increase of approximately 12% in the outflow requirements for irrigation purposes. [ABSTRACT FROM AUTHOR]

Planning for sustainable development of water resources relies crucially on the data available. Continuous hydrologic simulation based onconceptual models has proved to be the appropriate tool for studyingrainfall-runoff processes and for providing necessary data. In recent years, artificial neural networks have emerged as a novel identification technique for the modelling of hydrological processes. However,they represent their knowledge in terms of a weight matrix that is not accessible to human understanding at present. This paper introduces genetic programming, which is an evolutionary computing method thatprovides a 'transparent' and structured system identification, to rainfall-runoff modelling. The genetic-programming approach is applied to flow prediction for the Kirkton catchment in Scotland (U.K.). The results obtained are compared to those attained using two optimally calibrated conceptual models and an artificial neural network. Correlations identified using data-driven approaches (genetic programming and neural network) are surprising in their consistency considering therelative size of the models and the number of variables included. These results also compare favourably with the conceptual models. [ABSTRACT FROM AUTHOR]