Author(s): Shreenivas Londhe
Linked Author(s):
Keywords: Stream flow forecasting; Data driven modelling; M5 Model Trees
Abstract: Stream flow modelling is perhaps the most sought after research topic for hydrologists all over the world owing to its vital importance in design, construction, operation and maintenance of many hydraulic structures. Traditionally this is done by using deterministic/conceptual models based on physics of the underlying process. Owing to various difficulties involved in these modelling techniques researchers are always in search of a better modelling approach which will be easier, less time consuming, data tolerant and reasonably accurate. The present work deals with modelling of stream flow using M5 Model Trees to forecast daily average discharge one day in advance at 2 stations in Narmada river basin India. Two discharge measurements stations along river Narmada namely Rajghat and Mandaleshwar are selected to develop the stream flow models. The stream flow models are developed using the previous values of measured stream flow to forecast stream flow one day in advance (so-called temporal mapping). After observing the flow hydrographs year by year it was decided to develop separate monthly stream flow models for monsoon months of July, August, September and October and a common non-monsoon model for the months from November to June. All models provide a reasonable accuracy of forecasts as evident from their comparison with the observed values by plotting scatter plots, hydrographs and by calculating correlation coefficient between the predicted and the observe values.
Year: 2010