3RD INTERNATIONAL CONGRESS ON TECHNOLOGY - ENGINEERING & SCIENCE - Kuala Lumpur - Malaysia (2017-02-09)

Applying Statistical Methods For Predictions In Ungauged Catchments
Streamflow data are crucial for water resources management. They are necessary for most applications in water resources such as drainage structure design, flood forecasting and water allocation. Accurate determination of streamflow is therefore critical for water resources planning and management. Managing natural disasters such as floods and droughts is particularly challenging if streamflow data are not available or of poor quality. While realising the importance of streamflow data, many catchments around the worlds especially in developing and least-developed countries remain ungauged. This is due to a number of reasons e.g. unfavourable topographical conditions for installing a flow gauge and high investment and maintenance costs. To address the issues of ungauged catchments, methods of predictions in ungauged catchments or regionalization techniques have been developed[2]. This research aims to assess the applicability of well-established methods to tropical catchments where hydrological characteristics are highly variable.The overall methodology includes data quality check, data analysis, regression analysis, rainfall-runoff modeling, and performance assessment[1]. A regression method, i.e. simple linear regression and multiple linear regression, which has been widely applied to a number of catchments was chosen for this study. Physical catchment properties such as rainfall, land use, and soil type were included in the regression model as independent variables. Streamflow indices such as runoff coefficient, baseflow index, and rainfall-runoff elasticity were estimated through the regression and then transformed into streamflow time-series via a rainfall-runoff model[3]. The performance of the regression model in predicting streamflow data was evaluated by comparing the simulated to observed streamflow time series which was set aside during model development. Performance indices used were correlation (r) and Nash-Sutcliffe Efficiency (NSE). This study was demonstrated through a case study of the upper Ping catchment in northwest of Thailand. For this study,the proposed regression model was able to predict streamflow with acceptable accuracy. This offers useful information that is fundamental to water resources development in the upper Ping catchment.
Nutthanon Sa-ngonsub, Supattra Visessri, Pisit jarumaneeroj