Statistical Flood Frequency Analysis Using Short Term Data

Abstract

The most suitable probability distribution for estimating flood risks using short-term data is presented in this paper. A simulation method using autoregressive process to represent daily flows are used. Several distributions were fitted to the annual peaks of autoregressive process. The levels determined using the distributions were compared with the levels from empirical distribution determined using ordered peaks of 1000 autoregressive process. It was found that the Lognormal III, Pearson Type III and the Weibull gave good risks estimates of flood of high return period. Keywords: Autoregressive process; return period; empirical distribution Beberapa taburan kebarangkalian yang sesuai untuk menganggarkan kejadian banjir menggunakan data terhad dijelaskan dalam kertas ini. Kaedah simulasi berdasarkan proses autoregresi digunakan untuk menggambarkan aliran harian. Beberapa taburan dipadankan kepada puncak maksimum bagi proses autoregrasi. Paras tertentu ditetapkan menggunakan taburan telah dibandingkan dengan paras menggunakan taburan empirik menggunakan puncak tertib bagi 1000 proses autograsi. Hasil analisis didapati bahawa taburan Lognormal III, Pearson Jenis III dan Weibull memberikan anggaran terbaik untuk menganggar banjir bagi tempoh ulangan yang tinggi. Katakunci: Proses Autoregresi; tempoh ulangan; taburan empirik