TR-16 : Overland flow modelling

Abstract

Surface flow estimation is needed by planners and managers of water resource system. Most of the analysis are based on linear systems, wherein the watershed is assumed to be virgin. But, naturally, the watershed changes considerably due to man made activities like deforestation, urbanisation, etc. These do not permit application of a linear model like unit hydrograph. Moreover, precise information on spatial and temporal distribution of flow are needed for water quality studies. Distributed models are used in such situations. These are physically based models gradually varied unsteady flow equations by numerical techniques. Literature survey revealed that finite difference, finite element and method of characteristics are commonly used. Infiltration component is one of the important aspect of overland flow computations. Antecedent moisture content needs to be estimated accurately for the model to be realistic.

In many distributed models, the watershed is discretized into several sub-watersheds and assumed to comprise of plane surface flow and channel flow before reaching the outlet. The heterogeneity of the watershed poses a major problem in averaging the properties of the sub-watersheds. The effect of averaging and how small the sub-watershed should be (for better results) is not known at present.

A distributed model using method of characteristics and a concept of homogeneous response units was prepared. A watershed used by Ross and others in 1978 for testing their model was choosed for the application of the present model, considering the availability of data necessary for the model.

Results showed good performance of the model. The computed hydrograph nearly matches that of Ross and others.