Flood Event Forecasting: A Complementary System-Theoretic Modelling

The main requirements for cm operational conceptual hydrologic model are function development, model calibration, and validation followed by production of forecast runs and analysis of results. Taking into cognizance the complexity of physical systems the models represent the most dominant processes since they are mere simplifications of the physical reality and thus concomitantly subject to a whole lot of significant uncertainties. As such in this study, attempt was made at coupling the artificial neural network (ANN) with the Xinanjiang conceptual model with the view to enhancing the quality of its flow forecast. The approach uses latest observations and errors/residuals in runoff/discharge forecasts from the model. The two complementary models are used in such a way that errors of the model are forecasted by a neural network model so that flow forecasts can be improved as new observations come in. Results obtained show that there is a substantial improvement in the accuracy of the forecasts when the complementary model was operated on top of the Xinanjiang conceptual model compared to the results of the model alone. Further analysis on the model forecast residuals indicate that the residual time series does not follow a normal distribution but rather exhibits non-gaussianity; similarly too, the existence of persistent pattern in the error/residual structure is discernibly evident.