CS(AR)-24/99-2000 : Daily rainfall-runoff modelling using deterministic and stochastic approaches

Abstract

The present report discusses about the application of deterministic and stochastic approaches in daily rainfall-runoff modelling. Structure of a simple hydrologic water balance type conceptual model of daily time scale is discussed and tested on three catchments of India. Further, the deviations of historic and computed runoff series obtained by using the model, which indicate the error in the fitting of model, is analysed and used to improve the predicted flows. In the first case, the average error of a day is computed and the computed runoff as obtained by the use of water balance model is modified using this error. In the second case, the error is first analysed for its serial dependence and accordingly an AR model is fitted to the series. The water balance model is run on real time mode and computed runoff is modified based on error term computed using the AR model. Study indicates that incorporation of simple error term alone can improve the performance of a water balance type model. Also, if the model is run on real time mode and computed runoff is updated using the AR model, improvement in the efficiency of the computed flows can be achieved.