Nonparametric Kernel Estimation of Annual Maximum Stream Flow Quantiles

Authors

  • Ani Shabri

DOI:

https://doi.org/10.11113/matematika.v18.n.123

Abstract

Kaedah Nonparametric Kernel dicadangkan dan dinilaikan perlaksanaannya dalam menganggar kuantil aliran tahunan maksimum. Penganggar bagi bandwidth dianggarkan menggunakan teknik optimal dan `cross-validation'. Hasil keputusan menggunakan data sebenar yang terhad dari Malaysia, menunjukkan bahawa penganggar kuantil berdasarkan model nonparametric menggunakan kedua-dua teknik ini menghasilkan nilai punca min ralat kuasa dua dan punca min ralat mutlak yang kecil. Berdasarkan ujian pekali korelasi menunjukkan bahawa pendekatan model nonparametric adalah tepat, seragam dan ianya boleh dijadikan sebagai kaedah alternatif bagi model parametric dalam analisis frekuensi banjir. Katakunci: Bandwidth; `cross-validation'; kernel; pekali korelasi. A nonparametric kernel methods is proposed and evaluated performance for estimating annual maximum stream flow quantiles. The bandwidth of the estimator is estimated by using an optimal technique and a cross-validation technique. Results obtained from a limited amount of real data from Malaysia show that quantiles estimated by nonparametric method using these techniques have small root mean square error and root mean absolute error. Based on correlation coefficient test shown that the nonparametric model approach is accurate, uniform and flexible alternatives to parametric models for flood frequency analysis. Keywords: Bandwidth; cross-validation; kernel; correlation coefficient.

Downloads

Published

2002-12-01

Issue

Section

Mathematics