A special study submitted in partial fulfillment of the requirements for the degree of Master of Science, School of Engineering and Technology

Special Studies Project Report (M. Sc.) - Asian Institute of Technology, 1982

Daily river streamflows at one station (Vientiane) along the Mekong River are studied using a historical approach to obtain a long range flood forecast, with the objective of over-coming short-term hydrological uncertainty and the narrow limitations in time of process- based models. Only daily discharge data are used in the present approach . Sixty years of data are considered as one continuous time series; a suitable linear filter is applied to separate short period variations due to rain storms from the main hydrograph . Cross-correlation coefficients between all years are computed and arranged in the upper triangle of a matrix. This matrix is then used to identify the groupings of years of closest correlation or 'fit' to each other. Given the first few months of a year, the group which best fits a current flood season can be identified by used cross-correlation coefficients . It is proposed to use the average filtered hydrograph of that groups to forecast the future of the on-going season. Apart from the mean curve, the short period residues, filtered from the original curves, are analyzed statistically . The frequencies and return periods of these residue heights are used for short term forecasting.