Regional Regression Model of Mean Annual Streamflow For Ungauged Catchments

No Thumbnail Available

Date

2019-10

Journal Title

Journal ISSN

Volume Title

Publisher

Addis Ababa University

Abstract

Water resource developers and modelers focus on sustainable water resources management and availability of watercourses into account during design of hydraulic structures such as small dams for use at present and the future. Since most of the rivers in Ethiopia are ungauged, knowledge of predicting annual streamflow to be used for water resource development and other multiple purposes requires a study on stream flow of ungauged catchments. In this study an attempt was made to develop multiple linear regression model of mean annual stream flow for ungauged catchments of Awash River Basin using some statistical tests to secure the performance of the model. First the annual rainfall homogeneity test, trend analysis, and monthly and annual rainfall characteristics and variation were analyzed, based on this study the mean monthly rainfall of Awash River basin for the 19 stations varied from 3.15 mm to 411.8 mm in the period of 19812013 G.C and the mean annual rainfall distribution varies from 352.9 mm to 1766.4 mm, the mean annual flow varies from 36.68 MCM to 2026.82 MCM. Second development of equation for each gauged catchments based on different forms of equations (i.e. power, polynomial and linear functions) were performed and only one form of equation was chosen depending on model performance indices like R 2 , NSE and RMSE results. The majority of the results show remarkably good fits with R 2 -values ranging from 0.51 to 0.72 and NSE ranging from 0.53 to 0.7. The third and main task of this thesis work was developing a Regression model that could be used for estimation of mean annual flow at ungauged catchments, which was trained for Region-I using 9 gauging stations and for Region-II using 8 gauging stations. The performance and model fit was checked by statistical indicators and graphical methods. For Region-I it is found that the model with prediction variables Catchment area, mean annual rainfall and mean slope result the best fit giving (R 2 =0.89), (NSE=0.94) and (RMSE=107.24), Similarly For Region-II it is found that the model with prediction variables Catchment Area, mean annual Rainfall, mean Elevation and mean Slope result the best fit giving (NSE=0.88) and (RMSE=48.51). These model evaluation indicators were under a Very Good performances for Region-I and Good performances for Region-II.

Description

Keywords

Ungauged Catchments, River basin, regression analyses, multiple linear regression (MLR)

Citation