Civil Engineering - Theses

Permanent URI for this collection

Search Results

Now showing 1 - 1 of 1
  • Item
    Thumbnail Image
    A methodology for estimating yield from small ungauged rural catchments
    Jayasuriya, Lekamge Niranjali Nelunika ( 1991)
    An estimate of streamflow yield from ungauged catchments is often required by water management Authorities. A survey of Authorities indicated that the techniques used in computing streamflow yield from small ungauged rural catchments were limited to a range of empirical methods with questionable accuracy. There is clearly a need to develop a more reliable simplified methodology. The proposed methodology is based on the calibration of a rainfall-runoff model using a number of gauged catchments and the development of regression relationships between catchment physiographical characteristics and model parameters. The parameters are then used in conjunction with the rainfall-runoff model and meteorological observations to estimate streamflow yield from ungauged catchments. The 2-parameter rainfall-runoff model (MOSAZ) was developed only after a detailed study of actual evapotranspiration and parameter optimisation. Morton's model, based on Bouchet's complementary theory, proved to be an adequate method to calculate actual evapotranspiration from catchments. The model was tested with data from forested, native pasture and irrigated wheat catchments. The model is considered to be superior to methods based on pan evaporation as the actual evapotranspiration predicted using Morton's model is independent of the catchment cover and the prevailing soil moisture conditions in the catchment. A number of optimisation techniques based on direct search and gradient methods were tested for accuracy. From the methods tested, the pattern search and the Gauss-Marquardt algorithms proved to be superior. The interaction between satisfactory compliance with simple least squares error assumptions and the goodness-of-fit between observed and predicted streamflow was also studied. An important feature of the proposed methodology is the use of a multi-dimensional plotting routine termed Andrews' curves to separate the 184 catchments in the study region into a number of hydrologically homogeneous groups based on catchment physiographical characteristics which are related to MOSAZ model parameters. Regression relationships between MOSAZ model parameters and catchment physiographical characteristics were developed for one homogeneous group of catchments based on information from 17 catchments. The developed regression relationships were used to calculate streamflow from four test catchments to demonstrate the applicability of the developed methodology. As the results appeared promising, it is possible to extend the developed methodology to other homogeneous catchment groups.