The report discusses the application of deterministic and stochastic approaches in daily rainfall-runoff modeling. The 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 analyzed 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 analyzed 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 the computed runoff is modified based on error term computed using the AR model.
The 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.
For meeting the increased water requirements, it is necessary that intensive and extensive investigations and studies are carried out in a systematic manner on various aspects of hydrology of water resources projects. The interest of hydrologists and water resources engineers lies in the land phase of the hydrologic cycle and especially in the disposition of water received from the rainfall.
A hydrologic model is a simplified description of the hydrologic cycle depicting an appropriate relationship between rainfall and runoff. Hydrologic models are required not only for deciding about water yields or design parameters, or computing forecasts but also for understanding and evaluating effects of development and other activities on hydrological regime of river basins.
A good hydrologic model should be capable of simulating runoff close to the observed one. The difference of historic and computed runoff from the model indicates the deviation or the error which should be minimum for a good model. This error can be analyzed and an appropriate stochastic model or distribution can be fitted to it, and then it is possible to modify the computed flows and minimize the deviations.
Download the report here:
/articles/daily-rainfall-runoff-modeling-using-deterministic-and-stochastic-approaches-research