Publication detail

Comparison of Parametric and Nonparametric Estimates of Extreme Value Distribution

HOLEŠOVSKÝ, J. BLACHUT, V. FUSEK, M. MICHÁLEK, J.

Czech title

Srovnání parametrických a neparametrických odhadů rozdělení extrémních hodnot

English title

Comparison of Parametric and Nonparametric Estimates of Extreme Value Distribution

Type

abstract

Language

en

Original abstract

The presented paper is focused on comparison of different approaches to estimation of parameters of extreme value distributions. The commonly used parametric methods are compared with several nonparametric approaches, and properties of the estimates are discussed. The parametric inference is based on the partial duration series method and the generalized Pareto distribution. Unknown parameters of the distribution are estimated using the maximum likelihood method, and the method of probability weighted moments, which are often used in hydrology. The nonparametric inference is based on results presented by Gomes and Oliveira (see [1]), and the tail index of the extreme value distribution is estimated using the bootstrap methodology. The performance of estimators is compared using real and simulated data. The real data consists of historical rainfall series in the form of rainfall intensities from six stations operated by the Czech Hydrometeorological Institute in South Moravian Region in the Czech Republic. REFERENCE: [1] GOMES, M. I., OLIVEIRA, O., The Bootstrap Methodology in Statistics of Extremes – Choice of the Optimal Sample Fraction. In Extremes, Vol.4, No.4, pp. 331-358. Kluwer Academic Publishers, 2002.

Czech abstract

Příspěvek je zaměřen na srovnání různých přístupů k odhadům parametrů rozdělení extrémních hodnot. Běžně užívané parametrické metody jsou porovnány s několika neparametrickými metodami a jejich vlastnostmi. Parametrický přístup je založen na metodě Partial Duration Series a zobecněném Paretově rozdělení. Neznámé parametry tohot rozdělení jsou odhadnuty pomocí metody maximální věrohodnosti a metody pravděpodobnostně vážených momentů, jež patří mezi často užívané v hydrologii. Neparametrický přístup je založen na publikovaných výsledcích (viz. [1]), kde parametr rozdělení extrémních hodnot (tzv. EV index) je odhadnut za použití bootstrapu. Chování sledovaných odhadů je srovnáno na měřených i simulovaných datech. Měřená data sestávají z historických měření srážkových intenzit ze šesti stanic ČHMÚ situovaných v Jihomoravském kraji. REFERENCE: [1] GOMES, M. I., OLIVEIRA, O., The Bootstrap Methodology in Statistics of Extremes – Choice of the Optimal Sample Fraction. In Extremes, Vol.4, No.4, pp. 331-358. Kluwer Academic Publishers, 2002.

English abstract

The presented paper is focused on comparison of different approaches to estimation of parameters of extreme value distributions. The commonly used parametric methods are compared with several nonparametric approaches, and properties of the estimates are discussed. The parametric inference is based on the partial duration series method and the generalized Pareto distribution. Unknown parameters of the distribution are estimated using the maximum likelihood method, and the method of probability weighted moments, which are often used in hydrology. The nonparametric inference is based on results presented by Gomes and Oliveira (see [1]), and the tail index of the extreme value distribution is estimated using the bootstrap methodology. The performance of estimators is compared using real and simulated data. The real data consists of historical rainfall series in the form of rainfall intensities from six stations operated by the Czech Hydrometeorological Institute in South Moravian Region in the Czech Republic. REFERENCE: [1] GOMES, M. I., OLIVEIRA, O., The Bootstrap Methodology in Statistics of Extremes – Choice of the Optimal Sample Fraction. In Extremes, Vol.4, No.4, pp. 331-358. Kluwer Academic Publishers, 2002.

Keywords in Czech

rozdělení extrémních hodnot, IDF křivky, bootstrap

Keywords in English

extreme value ditribution, IDF curves, bootstrap

Released

24.09.2013

Location

Hejnice

ISSN

NEUVEDENO

Book

Precipitation Extremes in a Changing Climate - Book of Abstracts