• Proceedings of the Korea Water Resources Association Conference
• /
• 2009.05a
• /
• pp.1321-1326
• /
• 2009

강우는 수자원 확보 측면에서 근원이 되는 요소이다. 그러므로 정확한 확률강우량 산정은 미래의 가용 수자원량을 예측하는데 있어 중요한 사항중 하나이며 무엇보다 신중한 결정이 요구된다. 또한 하천의 범람에 의한 침수를 예방하는 수공구조물 등의 설계에 있어서는 신뢰할 수 있는 확률강우량 산정이 선행되어야 한다. 본 연구에서는 최근 우리나라 극치강우확률분포로서 많은 연구가 이루어지고 있는 GEV 분포(GEV-O)를 기반으로 위치 매개변수에 시간의 함수를 고려한 개선된 GEV 분포(GEV-A)를 이용하여 서울지점에 적용함으로서 GEV-O 분포에 의한 확률강우량과 GEV-A 분포로 산정된 확률강우량을 비교 검토하였다. 먼저 임의의 난수 발생을 통해 최우도추정법과 확률가중모멘트법으로 매개변수를 추정한 GEV-O 분포와 최우도추정법으로 매개변수를 추정한 GEV-A 분포의 상대평균제곱근오차 (R-RMSE)를 계산하여 비교함으로서 GEV-A 분포의 효율성을 판단하였다. 사례연구는 1961년부터 2008년까지 서울강우관측소에서 측정된 연최대 1일 강우량으로 하였으며 $X^2$-검정, PPCC-검정으로 적합도 검정을 실시하였다. 강우빈도분석 결과 GEV-A 분포가 GEV-O 분포로 산정된 결과 보다 대체로 재현기간 200년 이상일 경우, 과다 산정되는 경향을 보였다. 추후 개선된 GEV 분포를 서울 인근 지점에 적용함으로서 지역빈도해석(Regional Frequency Analysis)을 실행하기 위한 연구가 진행되어야 할 것이다. 또한 확률홍수량 산정 등에도 개선된 GEV 분포를 이용함으로서 보다 정확하고 신뢰성 있는 확률수문량을 예측하여야 할 것이다.

• Wind and Structures
• /
• v.9 no.3
• /
• pp.179-191
• /
• 2006

Parent wind data are often acknowledged to fit a Weibull probability distribution, implying that wind epoch maxima should fall into the domain of attraction of the Gumbel (Type I) extreme value distribution. However, observations of wind epoch maxima are not fitted well by this distribution and a Generalised Extreme Value (GEV) analysis leading to a Type III fit empirically appears to be better. Thus there is an apparent paradox. The reasons why advocates of the GEV approach seem to prefer it are briefly summarised. This paper gives a detailed analysis of the errors involved when the GEV is fitted to epoch maxima of Weibull origin. It is shown that the results in terms of the shape parameter are an artefact of these errors. The errors are unavoidable with the present sample sizes. If proper significance tests are applied, then the null hypothesis of a Type I fit, as predicted by theory, will almost always be retained. The GEV leads to an unacceptable ambiguity in defining design loads. For these reasons, it is concluded that the GEV approach does not seem to be a sensible option.

• KSCE Journal of Civil and Environmental Engineering Research
• /
• v.29 no.3B
• /
• pp.259-267
• /
• 2009

Generalized Extreme Value (GEV) distribution is recommended for flood frequency and extreme rainfall distribution in many country. L-moment method is the most common estimation procedure for the GEV distribution. In this study, the relationships between the cross-site correlations between extreme events and the cross-correlation of estimators of L-moment ratios (L-moment Coefficient of Variation (L-CV) and L-moment Coefficient of Skewness (L-CS)) for data generated from GEV distribution were derived by Monte Carlo simulation. Those relationships were fit to the simple power function. In this Monte Carlo simulation, GEV+ distribution were employed wherein unrealistic negative values were excluded. The simple power models provide accurate description of the relationships between cross-correlation of data and cross-correlation of L-moment ratios. Estimated parameters and accuracies of the power functions were reported for different GEV distribution parameters combinations. Moreover, this study provided a description about regional regression approach using Generalized Least Square (GLS) regression method which require the cross-site correlation among L-moment estimators. The relationships derived in this study allow regional GLS regression analyses of both L-CV and L-CS estimators that correctly incorporate the cross-correlation among GEV L-moment estimators.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2007.05a
• /
• pp.321-325
• /
• 2007

3개의 매개변수(location, scale, shape)로 이루어진 GEV와 GLO 분포는, 미국의 공식적인 홍수빈도 분포인 Log Pearson Type III와 함께 수문분야에서 중요한 위치를 차지하고 있다. 본 연구에서는 Monte Carlo 실험을 이용하여 GEV와 GLO 분포에서 서로 다른 두 지점의 유출량 자료를 생성하여 L-CV(L-moment Coefficient of Variation; $\tau_2$)와 L-CS(L-moment Coefficient of Skewness; $\tau_3$)를 추정하였으며, L-moment 추정값들 간의 교차상관$(\tau_2-\tau_2,\;\tau_3-\tau_3,\;\tau_2-\tau_3)$과 유출량 자료간의 교차상관의 관계를 Simple Power 함수를 이용하여 유도하였다. 실험 과정에서 GEV와 GLO 분포가 비현실적인 음수 유출량을 생성하여, 실험 결과에 큰 영향이 있음을 확인하여, 두 분포에서 생성된 유출량 자료에서 음수값을 제외한 GEV+와 GLO+ 분포를 이용하여 관계식을 유도하고 이를 GEV와 GLO 분포의 결과와도 비교하였다. 본 연구에서 도출된 관계식은 향후 Generalized Least Square 회귀식을 이용하여 홍수분포의 지역 매개변수를 추정하기 위해 활용성이 클 것으로 기대한다.

• The Korean Journal of Applied Statistics
• /
• v.23 no.3
• /
• pp.471-485
• /
• 2010

VaR is used extensively as a tool for risk management by financial institutions. For convenience, the normal distribution is usually assumed for the measurement of VaR, but recently the method using extreme value theory is attracted for more accurate VaR estimation. So far, GEV and GPD models are used for probability models of EVT for the VaR estimation. In this paper, the PP model is suggested for improved VaR estimation as compared to the traditonal EV models such as GEV and GPD models. In view of the stochastic process, the PP model is regarded as a generalized model which include GEV and GPD models. In the empirical analysis, the PP model is shown to be superior to GEV and GPD models for the performance of VaR estimation.

• Journal of Wetlands Research
• /
• v.20 no.4
• /
• pp.338-344
• /
• 2018

In this study, the surface air temperature (SAT) and the dew-point temperature (DPT) are applied as the covariance of the location parameter among three parameters of GEV distribution to reflect the non-stationarity of extreme rainfall due to climate change. Busan station is selected as the study site and the monthly maximum daily rainfall depth from May to October is used for analysis. Various models are constructed to select the most appropriate co-variate(SAT and DPT) function for location parameter of GEV distribution, and the model with the smallest AIC(Akaike Information Criterion) is selected as the optimal model. As a result, it is found that the non-stationary GEV distribution with co-variate of exp(DPT) is the best. The selected model is used to analyze the effect of climate change scenarios on extreme rainfall quantile. It is confirmed that the design rainfall depth is highly likely to increase as the future DPT increases.

• Proceedings of the Korean Society of Agricultural Engineers Conference
• /
• 2002.10a
• /
• pp.225-228
• /
• 2002

This research seeks to derive the design rainfalls through the L-moment with the test of homogeneity, independence and outlier of data on annual maximum daily rainfall in 38 Korean rainfall stations. To select the fit appropriate distribution of annual maximum daily rainfall data according to rainfall stations, applied were Generalized Extreme Value (GEV), Generalized Logistic (GLO) and Generalized Pareto (GPA) probability distributions were applied. and their aptness was judged Dusing an L-moment ratio diagram and the Kolmogorov-Smirnov (K-S) test, the aptitude was judged of applied distributions such as GEV, GLO and GPA. The GEV and GLO distributions were selected as the appropriate distributions. Their parameters were estimated Targetingfrom the observed and simulated annual maximum daily rainfalls and using Monte Carlo techniques, the parameters of GEV and GLO selected as suitable distributions were estimated and. dDesign rainfallss were then derived, using the L-moment. Appropriate design rainfalls were suggested by doing a comparative analysis of design rainfall from the GEV and GLO distributions according to rainfall stations.

• Proceedings of the Korean Society of Agricultural Engineers Conference
• /
• 2003.10a
• /
• pp.443-446
• /
• 2003

This study was conducted to draw design rainfall for the regional design rainfall derived by the optimal distribution and method of frequency analysis. The design rainfalls were calculated by the regional and at-site analysis for Log-Pearson type III and GEV distributions and were compared with Relative efficiency(RE) which is ratio of Relative root-mean-square error(RRMSE) by the regional and at-site analysis for Log-Pearson type III and GEV distributions. Consequently, optimal design rainfalls following the regions and consecutive durations were derived by the regional frequency analysis for GEV distribution and design rainfall maps were drawn by GIS techniques.

• Proceedings of the Korean Society of Agricultural Engineers Conference
• /
• 1999.10c
• /
• pp.479-485
• /
• 1999

This study was conducted to derived design floods by Generalized Extreme Value(GEV) distributiion for the annual maximum series at ten watersheds along Han, Nagdong, Geum , Yeongsan and Seomjin river systems. Adequency for the analysis of flood data used in this study was established by the test of Independence, Homogeneity , detection of Outliers. Coefficient of variation , skewness and kurtosis were calculated by the L-Moment, and LH-Moment ratio respectively. Parameters were estimated by the Method of L-Method of LH-Moment. Design floods obtained by Method of L-Moments and LH-Moments using different methods for plotting positions in GEV distributions and were compared with those obatined using the Method of L-Moments and LH-Moments by the Relative Mean Errors and Realtive Absoulte Errors. It was found that desgin floods derived by the method of L-Moments and LH-Moments using Cunnane plotting position foumula in the GEV distribution are much closer to those of the observed data in comparison with those obtained by methods of L-moments and LH-moments using the other formula for poltting postions from the viewpoint of Relative Mean Errors and Relative Absoulte Errors. In view of the fact that hydraulic structures indcluding dams and levees are generally usiong design floods with the return period of two hundred years or so, design floods derived by LH-Moments are seemed to be more reasonable than those of L-Moments in the GEV distribution.

• Proceedings of the Korea Water Resources Association Conference
• /
• 2016.05a
• /
• pp.201-201
• /
• 2016

The asymptotic extreme value distributions of maxima are a natural choice when designing against future extreme events like flood peaks or wave heights, given a stationary time series. The generalized extreme value distribution (GEV) is often utilised in this context because it is seen as a convenient single expression for extreme event analysis. However, the GEV has a drawback because the location of the distribution bound relative to the data is a discontinuous function of the GEV shape parameter. That is, for annual maxima approximated by the Gumbel distribution, the data is also consistent with a GEV distribution with an upper bound (no lower bound) or a GEV distribution with a lower bound (no upper bound). A more consistent single extreme value expression for design purposes is proposed as the Weibull distribution of smallest extremes, as applied to transformed annual maxima. The Weibull distribution limit holds here for sufficiently large sample sizes, irrespective of the extreme value domain of attraction applicable to the untransformed maxima. The Gumbel, Type 2, and Type 3 extreme value distributions thus become redundant, together with the GEV, because in reality there is only a single asymptotic extreme value distribution required for design purposes - the Weibull distribution of minima as applied to transformed maxima. An illustrative synthetic example is given showing transformed maxima from the normal distribution approaching the Weibull limit much faster than the untransformed sample maxima approach the normal distribution Gumbel limit. Some New Zealand examples are given with the Weibull distribution being applied to reciprocal transformations of annual flood maxima, where the untransformed maxima follow apparently different extreme value distributions.