APPlicationofIntensity Duration FrequencyCurveforKoreanRainfallDatausingcumulativeDistributionFunction

Application of Intensity-Duration-Frequency Curve
for Korean Rainfall Data

using Cumulative Distribution Function
Kewtae Kim
The Graduate School

Yonsei University
Department of Civil Engineering

A Masters Thesis
Submitted to the Department of Civil Engineering
and the Graduate School of Yonsei University
in partial fulfillment of the
requirements for the degree of
Master of Science
Kewtae Kim
July 2008
This certifies that the masters thesis

of Kewtae Kim is approved.
___________________________
Thesis Supervisor: Prof. Jun-Haeng Heo, Ph.D
___________________________
Prof. Woncheol Cho, Ph.D
___________________________
Prof. Sung-Uk Choi, Ph.D
The Graduate School

Yonsei University
July 2008
ACKNOWLEDGEMENT
I am very grateful to my advisor, Professor Jun-Haeng Heo, whose guidance, patience,

encouragement, great help, and invaluable support during the course of study possible. He
taught me hydrological statistics and his lectures were very useful. Without his guidance, I
would not have been able to finish this study. I would also like to thank Professor Woncheol
Cho, Sung-Uk Choi on my thesis committee.
Also, I am thanks professor Won-Hwan Lee who showed extraordinary appreciation
toward water. Im also thank professors Hak-Joo Whang, Bock-Mo Yeu, Keun Joo Byun, Soo
Il Kim, Youn-Kyoo Choung, Moon Kyum Kim, Kwang Baik Ko, Sanh-Hyo Kim, Ha-Won
Song, Sangseom Jeong, Sang-Ho Lee, Yun Mook Lim, Seung Heon Han, Hong-Gyoo Sohn,
Jun-Hwan Lee, Joonhong Park, Joon Heo, Hyoungkwan Kim, Tong Seok Han, Jang-Ho Kim,
and Ho Jeong Kang for my knowledge and personal development at Yonsei University.
Thank all members of Hydro-Laboratory, Yonsei University. Especially, Kyung-Duk Kim,
Kwonsu Kang, Chang-Sam Jeong, Dong-Jin Lee, Youn-Woo Kho, Kyung Hoi Heo, ChangYong Cha, Eunseok Lee, Sang-Bok Lee, Young-Seok Lee, Ji-Hoon Kim, Jiyoun Sung, MunHyoung Park, Boae Kwon, Jung-Woo Lee In-Chan Park, Haennim Park, Hyeongsik Kang,
Jung-Hwan Ahn, Taebum Kim, Donggyun Lim, Daeryoung Park, Shanghyun Ahn, Myoung Jin
Um, Wonjoon Yang, Younghoon Jung, Seungkyu Han, Sanghwa Cheong, Seonmee Jeon, HanSeong Jo, Jungwon Kim, Hee-Sun Jung, Changhun Cho, and Woonghyeon Jeon showed me
the basic attitude to the study, and many fellows including Taesoon Kim, Hongjoon Shin, Woo
Sung Nam, Sooyoung Kim, Jeong-Eun Lee, Ji Hye Kwon, and Jun Hak Lee gave me various
kinds of assistances. And I wish my graduates in the same class, Seong Teak Lim, Hae-Eun
Lee achieve their goal in life. I believe that Younghun Jung, Hye-Seon Yun, Ju-Young Shin,
Heon Cheol Jeong, Young-Il Kim, Gyeong Ho Yun, Sonkyu Yun, Jung Pyo Seo, Gian Choi, In
Young Jang, Byung Woong Choi, Won Geun Lee, Kyoungjoo Lee, Kwanghee Han, Min Hye
Yun, Young Joo Kim, Jaekook Shin, Kyungmin Sung, and Dong Seok Kim can produce good
research results.
And I am thankful to Professor Byong Ho Jun, Professor Mon mo Kim, Professor Chang
Eon Park, Je Hyeong Kim, Jong Yong Kim, member of WATERES, Suk Jin Jang, Ju Il Song,
Jae Kwon Kim, and members of K.G. civil59. Especially, I would like to thank Soomin Kim,
Sunchan Lee, Kilsoo Jung, Daham Yang, my friends, members of C.U.G. the Universe
Conquest, Jae Hwi Lee, Jae Moo Lee, Gihae Kim, Wha Jung Lee, Seong Ho Choi, Eun Sang
Lee, Ki Joo Kim, Do Won Park, Ji Hyoun Woo, So Yeon Park, Yoon Mi Lee, Jae Won Park,
and members of Jang Wi Joongang Presbyterian Church for their special affection to my study.
Finally, my sincere gratitude and appreciation go to my parents, and glory & honor to God.
TABLE OF CONTENTS
TABLE OF CONTENTS ............................................................................................................ i

LIST OF FIQURES...................................................................................................................iii
LIST OF TABLES .................................................................................................................... iv
ABSTRACT .............................................................................................................................. vi
Chapter 1. Introduction .............................................................................................................. 1
1.1 General Description.......................................................................................... 1
1.2 Research Objectives ......................................................................................... 2
1.3 Literature Review ............................................................................................. 2
Chapter 2. IDF Curve using Cumulative Distribution Function ................................................. 6
2.1 Definition of Variables, Notation and Clarification.......................................... 6
2.2 The IDF Relationship for a Specified Return Period........................................ 9
2.3 The IDF Curve using Cumulative Distribution Function................................ 12
2.4 Alternative Distribution Functions ................................................................. 14
Chapter 3. Parameter Estimation Methods ............................................................................... 17
3.1 Typical Procedure........................................................................................... 17
3.2 Robust Estimation........................................................................................... 19
3.3 One-Step Least Squares Method..................................................................... 21
3.4 Parameter Estimation Methods using Genetic Algorithms ............................. 22
Chapter 4. Application ............................................................................................................. 28
4.1 Description of Sites and Data ......................................................................... 28
4.2 Application of IDF Curve using Cumulative Distribution Function ............... 31
4.3. Parameter Estimation of Conventional IDF Curves....................................... 32
4.4. Parameter Estimation using Separation of Short- and Long-Rainfall Duration

Method ............................................................................................................ 34
Chapter 5. Comparison............................................................................................................. 36
5.1 Total Rainfall Duration................................................................................... 36
5.2 Separation of Rainfall Duration...................................................................... 39
Chapter 6. Conclusions............................................................................................................. 41
REFERENCES......................................................................................................................... 43
APPENDIX A .......................................................................................................................... 50
.................................................................................................................................. 73
ii
LIST OF FIQURES
Figure 3.1 Pareto-optimal solution indicated as "RMSE" that makes an objective function
about RMSE the smallest value and the other objective function about RRMSE
worst, and vice versa ............................................................................................... 25
Figure 3.2 Comparison of absolute bias of rainfall quantiles estimated by IDF curve whose
parameters are estimated by two extreme Pareto-optimal solutions in Figure 3.1... 26
Figure 4. 1 Location of 76 rainfall recording sites overall KMA ............................................. 28
Figure 5. 1 IDF curve in case of Modified Sherman at site code number 130 (Uljin).............. 38
Figure 5. 2 IDF curve represented Figure 5. 1 in detail at site code number 130 (Uljin) ......... 38
iii
LIST OF TABLES
Table 3. 1 Separation of short- and long-rainfall durations method ......................................... 27

Table 4. 1 Information of 76 sites controlled by Korea KMA ................................................. 29
Table 5. 1 Number of the smallest RMSE or RRMSE of total sites using 12 different IDF
curves in case of Gumbel and GEV......................................................................... 37
Table 5. 2 Number of the smallest RMSE or RRMSE of total sites using 8 different IDF curves
in case of Gumbel and GEV .................................................................................... 39
in case of Gumbel and GEV .................................................................................... 40
Table A. 1 Average RMSE and RRMSE of each site using 12 different IDF curve in case of
Gumbel .................................................................................................................... 51
Table A. 2 Average RMSE and RRMSE of each site using 12 different IDF curve in case of
GEV......................................................................................................................... 54
Table A. 3 Average RMSE and RRMSE excluded regression analysis IDF curves of each site
in case of Gumbel.................................................................................................... 57
Table A. 4 Average RMSE and RRMSE excluded regression analysis IDF curves of each site
in case of GEV ........................................................................................................ 60
Table A. 5 Average RMSE and RRMSE of IDF curves separated short- and long-rainfall
durations for each site in case of Gumbel................................................................ 63
durations for each site in case of GEV .................................................................... 65
Table A. 7 Parameters of IDF curve using CDF estimated using SEP_DUR method in case of
Gumbel .................................................................................................................... 67
iv
GEV......................................................................................................................... 70
ABSTRACT

Kewtae, Kim
Dept. of Civil Engineering
The Graduate School
Yonsei University
The conventional intensity-duration-frequency (IDF) curve has generally come from the
empirical intuition of researcher, and linear or non-linear regression analysis also has been
usually applied to estimate the parameters of IDF curves. Hence, one can say that the
conventional IDF curves hardly have statistical characteristics of rainfall data at the site of
interest.
In this study, IDF curve using cumulative distribution function (CDF) was applied to
Korean rainfall data, and the parameters of IDF curve using CDF were estimated using singleobjective genetic algorithms and multi-objective genetic algorithms (MOGA) to improve the
accuracy. And then, the parameters were estimated using separation of short- and long-rainfall
durations method (SEP_DUR) to derive more precise formulas. The parameters of
conventional IDF curves were newly estimated to compare the results each other.
As the results, root-mean-square-error (RMSE) and relative RMSE (RRMSE) of IDF curve
using CDF have the smallest values for both cases of total rainfall duration and separation of
vi
short- and long-rainfall durations. The MOGA shows the smallest RMSE and RRMSE among
the applied parameter estimation methods for separation of short- and long-rainfall durations.
It is found that the IDF curve using CDF for SEP_DUR with MOGA is the more accurate than
any other methods.
Key words : Intensity-Duration-Frequency Curve, Multi-objective Genetic Algorithms,

Separation of Short-and Long-Rainfall Durations, Cumulative Distribution Function
vii
Chapter 1. Introduction
1.1 General Description

Generally, at-site frequency analysis and regional frequency analysis are performed to
estimate rainfall quantile considering various durations and frequency (return period) at the
site of interest. In order to perform these frequency analysis methods, it should be essential to
compile rainfall data recorded enough length about demanded duration and to estimate
parameters of various probability distributions and to perform the goodness-of-fit tests
corresponding for analysis site.
Although there is an advantage that rainfall quantile can be estimated by at-site frequency
analysis and regional frequency analysis, there are defects; rainfall data should be compiled
more than 30 year and it is necessary to estimate the parameters of a given probability
distribution and to perform the goodness-of-fit test.
IDF curve has widely been used, when hydraulic structure is designed in Korea. Talbot,
Sherman, Japanese and Semi-Log type are the conventional intensity-duration-frequency
(IDF) curves that have widely been used in Korea(Lee (1997), Yoon (1998)). And Talbot,
Sherman, Japanese and Semi-Log also are modified to improve the accuracy of IDF curves.
And other IDF curves have been developed by Lee (1980), Lee (1993), Heo (1999) and
Ministry of Construction and Transportation (2000). However, the conventional IDF curves
formulation has generally come from the empirical intuition of researcher, and regression
analysis also has been usually applied to estimate the parameters of IDF curves. Hence, one
can say that the conventional IDF curves hardly have statistical characteristics of rainfall data
having rationale at the site of interest (Koutsoyiannis et al., 1998).
In this study, in order to overcome such a problem, theoretically derived IDF curve using
cumulative distribution function (CDF) is applied to Korean rainfall data. The parameters of
IDF curve using CDF were estimated using single objective genetic algorithms (SGA) and
multi-objective genetic algorithms (MOGA) to improve the accuracy
1.2 Research Objectives

The major objective of this study is to apply the IDF curve using CDF reflecting the
statistical characteristics of rainfall data at the site of interest. The specific objectives are:
(1) to collect annual maximum rainfall data corresponding to each duration for site overall
Korea Meteorological Administration (KMA).
(2) to adopt IDF curve using CDF for Korean rainfall data.
(3) to estimate the parameters of adopting IDF curve using genetic algorithms to improve
the accuracy.
(4) to newly estimate the parameters of conventional IDF curves to compare the results
each other.
(5) to compare the accuracy of total IDF curves.
1.3 Literature Review

The rainfall intensity-duration-frequency relationship is one of the most generally used
methods in urban drainage design and floodplain management. The establishment of such
relationships goes back to as early as 1928 (Meyer, 1928). After a few late Sherman (1931)
derived applicable general intensity duration formula to other localities, and Bernard (1932)
was to make available for any locality within the limits of the study, rainfall intensity formulas
for frequencies of 5, 10, 15, 25, 50 and 100 year, applicable to rainfall duration of 120 to 6000
min. Bell (1969) developed IDF relationship using formula computed the depth-duration ratio
for the U.S.S.R. Chen (1983) developed a simple method to derive a generalized rainfall
intensity-duration-frequency formula for any location in the United States using three isopluvial maps of the U.S Weather Bureau Technical Paper No.40.
In the 1990s, some other approaches mathematically more consistent had been proposed,
Burlando and Rosso (1996) proposed the mathematical framework to model extreme storm
probabilities from the scaling properties of observed data of station precipitation, and the
simple scaling and the multiple scaling conjectures was thus introduced to describe the
temporal structure of extreme storm rainfall. Also, Koutsoyiannis (1994; 1996; 1998) proposed
a new approach to the formulation and construction of the intensity-duration-frequency curves
using data from both recording and non-recording stations. More specially, it discussed a
general rigorous formula for the Intensity-Duration-Frequency relationship whose specific
forms had been explicitly derived from the underlying probability distribution function of
maximum intensities. And it also proposed two methods for a reliable parameter estimation of
the IDF relationship. Finally, it discussed a framework for the regionalization of IDF
relationships by also incorporating data from non-recording stations.
More recently, Garcia-Bartual and Schneider(2001) used statistical distribution and found
the Gumbel Extreme Value (GEV) distribution fitted to data well. Yu et al.(2004) developed
regional rainfall intensity-duration-frequency relations for non-recording sites based on scaling
theory, which uses the hypothesis of piecewise simple scaling combines with the Gumbel
distribution. Mohymont et al.(2004) assessed IDF-curves for precipitation for three stations in
Central Africa and proposed more physically based models for the IDF-curves. Di Baldassarre
et al.(2006) analyzed to test the capability of seven different depth-duration-frequency curves
characterized by two or three parameters to provide an estimate of the design rainfall for storm
durations shorter than 1 hour, when their parameterization is performed by using data referred
to longer storms. Karahan et al.(2007) estimated parameters of a mathematical framework for
IDF relationship presented by Koutsoyiannis et al. (1998) using genetic algorithm approach.
Singh and Zhang (2007) derived intensity-duration-frequency (IDF) curves from bivariate
rainfall frequency analysis using the copula method.
In Korea, Lee (1980) derived the rainfall intensity probability formulas by the least square
method and constructed the rainfall intensity-duration-frequency curves. And Lee et al. (1992;
1993) derived a typical probable rainfall intensity formula by analyzing pre-issued probable
rainfall intensity formulas over principal rainfall observation stations, and to obtain the
regional characteristics based on the rainfall patterns by evaluating probable rainfall amount.
Yoo (1995) analyzed generalization of probability IDF curve.
Han et al. (1996) and Heo et al. (1999) derived the rainfall intensity-duration-frequency
equation based on the Linearizing Method and the appropriate probability distribution.
Especially, Heo (1999) performed frequency analysis of annual maximum rainfall data for 22
rainfall gauging stations in Korea and estimated parameter using the method of moments
(MOM), maximum likelihood (ML), and probability weighted moments (PWM). And the GEV
distribution was selected as an appropriate model for annual maximum rainfall data based in
parameter validity condition, graphical analysis, separation effect, and goodness of fit tests.
For the selected GEV model, spatial analysis was performed and rainfall intensity-durationfrequency formula was derived by using linearization technique.
Seoh et al. (1999) analyzed return period due to rainfall duration in 98 Pusan rainfall
datum. Lee and Lee (1999) derived rainfall intensity formula based on the representative
probability distribution in Korea and divided whole region into five zones for 12-rainfall
durations. Choi et al. (2000) derived the probable rainfall depths and the probable rainfall
intensity formulas for Inchon. Park et al. (2000) analyzed the variation in peak discharge
according to the type of probable rainfall intensity formula for Wi stream basin. Song and
Seoh(2000) estimated probable rainfall intensity formula and applied the optimization
technique using Simplex method and Powell method. Lee et al. (2001) derived the rainfall
intensity formula by the regional frequency analysis of individual zone based in representative
probability distribution in Korea. Seoh et al. (2001) constructed data base of minutely rainfall
intensity. Yoo et al. (2001) proposed a theoretical methodology for deriving a rainfall
intensity-duration-frequency curve using a simple rectangular pulsed Poisson process model.
Heo and Chae (2002) improved reading method of rainfall recording paper to estimate
accurate probable rainfall intensity. Kim et al. (2002) estimated probable rainfall intensity
corresponding to separate of rainfall duration. Lee and Seong (2003) analyzed the smoothing
of rainfall intensity-duration-frequency relationship curve by the Box-Cox transformation. Yoo
and Rho (2004) computed probable rainfall intensity and flood discharge for culvert design.
Choi (2006) estimated real time rainfall intensity using rainfall radar and rain gauges. Han et al.
(2006) suggested probable rainfall intensity formula considering the pattern change of
maximum rainfall at Incheon city.
Recently, Kim et al. (2007a) estimated IDF curve considering climate change using GCM
model. Yoo et al. (2007) proposed and evaluated a methodology for deriving the rainfall
intensity-duration-frequency relationship for durations less than 10 minutes used for designing
drainage systems in small urban catchments and roads. Kim et al. (2007b) estimated parameter
of intensity-duration-frequency curve using genetic algorithm and compared with study of
existing estimation method. Also, Shin et al. (2007) suggested the separation of short and long
durations for estimation the parameters of IDF curve using Multi-objective Genetic Algorithm
(MOGA).
Chapter 2. IDF Curve using Cumulative Distribution
Equation Chapter 2 Section 1
Function
It is self-evident that the IDF relationship is a mathematical relationship among the rainfall
intensity i the duration d , and the return period T (or, equivalently, the annual frequency
of excess, typically referred to as frequency only). However, these terms may have different
meanings in different contexts of engineering hydrology and this may lead to confusion or
ambiguity (Koutsoyiannis et al., 1998). For the sake of a comprehensive presentation and
unambiguousness in the material that follows we include in sections 2.1 and 2.2 the definitions,
clarifications, and description of the general properties of the idf relationships. The reader
familiar with these issues may proceed directly to section 2.3.
2.1 Definition of Variables, Notation and Clarification

Let (t ) denoted the instantaneous rainfall intensity process, where t denotes time. Let
d be a selected (arbitrary) time duration (typically from a few minutes to several hours or few
days), which serves as the length of a time window over which we integrate the instantaneous
rainfall intensity process (t ) . Moving this time window along time we form the moving
average process, given by
d (t ) =
1
d
t d
( s ) ds
(2.1)
In reality, because we do not know the instantaneous intensity in continuous time, but
rather have measurements of the average intensity (t ) for a given resolution (typically 5-10
min to 1 h), Eq. (2.1) becomes
d (t ) =
(t i )
d
N 1
(2.2)
i =0
where it is assumed that the duration d is an integer multiple of the resolution , i.e.
d = N . Given the stochastic process d (t ) we can form the series of the maximum average
intensities (or simply maximum intensities) il (d ) (l = 1,..., n) , which consists of n values,
where n is the number of (hydrological) years through which we have available
measurements of rainfall intensities. This can be done in two ways. According to the first way,
we form the series of annual maxima (or annual maximum series) by
il (d ) = max
{ d (t )}
(2.3)
l <t <l
where l and l + are the beginning and end time of the l th year. According to the second
way, we form the series above threshold (also known as partial duration series or annual excess
series) by selecting those values of d (t ) that exceed a certain threshold , selected in a
manner that the series { il (d )} includes exactly n values. To ensure stochastic independence
among il (d ) , we also set a lower time limit (e.g. one or more days) for consecutive values,
thus defining the series by
{ il (d ), l = 1,..., n } = d (tl ) d > , tl > tl 1 + , d (tl ) =
max { d (t )}
tl < t < tl +
(2.4)
where the three conditions of the right-hand part must hold all together, otherwise the point tl
and the respective intensity d (tl ) are not selected for the series {il (d )} .
In practice, the construction of the series of maximum intensities is performed

simultaneously for a number k of durations d j , j = 1,..., k , starting from a minimum duration
equal to the time resolution of observations (e.g. from 5-10 min to 1 h depending on the
measuring device) and ending with a maximum duration of interest in engineering problems
(typically 24 or 48 h). Normally, all k series must have the same length n but, owing to
missing values, it is possible to have different lengths n j for different durations d j .
The above description of the construction of the series of maximum intensity allows us to
observe that the duration d is not a random variable but, rather, a parameter for the intensity.
It is not related to the actual duration of rainfall events, but is simply the length of the time
window for averaging the process of intensity. On the contrary, the series of maximum
intensities il (d ) is considered as a random sample of a random variable I (d ) (Koutsoyiannis
et al., 1998).
The return period T for a given duration d and maximum intensity i (d ) is the average
time interval between excess of the value i (d ) . It is well known (e.g. see Kottegoda (1980))
that for the annual series, under the assumption that consecutive values are independent, the
return period of an event is the reciprocal of the probability of excess of that event, i.e.
T =
1
1 F
(2.5)
where F denotes the probability distribution function of I (d ) which, of course, is

evaluated at the particular magnitude of interest. It is also known (e.g. see Raudkivi (1979))
that the return period T ' for the series above threshold is related to that of the series of
annual maxima by
T=
1
1 exp(1/ T ')
(2.6)
T '=
1
ln(1 1/ T )
(2.7)
(A good approximation of Eq. (2.6) and Eq. (2.7) with an accuracy of two decimal digits
is given by the very simple relation T = T '+ 0.5 ). Thus, the return period is always related to
the distribution function of the series of annual maxima.
Given the above clarifications, Koutsoyiannis et al.(1998) observed that the problem of the
construction of IDF curves is somehow idiosyncratic. It is not a problem of statistical analysis
of a single random variable, as it includes two variables i and d . Nor is it a problem of two
random variables, because d is not a random variable. In fact, it consists of the study of a
family of random variables I (d ) , where d takes theoretically an infinite number of values
from a real interval. This family of random variables I (d ) does not form a typical stochastic
process: I does not represent intensity at a certain time, and d does not represent time but,
rather, a time interval. Nevertheless, invoking the theory of stochastic processes is not
necessary in this problem because they were not interested in a multidimensional (i.e. of order
greater than 1) distribution of I (d ) . They were rather interested in the first-order distribution
of I (d ) , i.e. the function F (i; d ) = P ( I (d ) < i ) , which is the target of the construction of the
idf curves. Indeed, the function F (i; d ) can be directly transformed into a relationship among
the quantities i , d , T .
2.2 The IDF Relationship for a Specified Return Period

All typical idf relationships of the literature for a specific return period are special cases of
the generalized formula (Koutsoyiannis et al., 1998)
i=
(2.8)
(d + )
where , , , and are non-negative coefficients with 1 . The latter inequality is

easily derived from the demand that the rainfall depth h = id is an increasing function of d .
Equation (2.8) is not obtained by any theoretical reasoning, but is an empirical formula,
encapsulating the experience from several idf studies. In the bibliography, they found
simplified versions of Eq. (2.8), which are derived by adopting one or two of the restrictions
= 1, = 1, and = 0 .
It should be noted that considering 1 and 1 results in overparameterization of
Eq. (2.8). Indeed, the quantity 1/(d + ) can be adequately approximated by 1/(d + ') *
where ' and * are coefficients depending on and , which can be determined
numerically in terms of minimization of the root-mean-square error. Consequently,
1/(d + ) is approximated by 1/(d + ') ' , where ' = * . A numerical investigation

was done to show how adequate the approximation of 1/(d + ) by 1/(d + ') is. The
duration d was restricted between the values d min = 1/12h (i.e. 5 min) and d max = 120h , an
interval much wider than the one typically used. The parameter varied between 0 and
max = 12d min (i.e. 1 h), and the parameter between 0 and 1. The root-mean-square
standardized error (rmsse) of the approximation took a maximum value of 2.3% for = 0.55
and = max ; the corresponding maximum absolute standardized error (mase) was 4.3%. For
the most frequent case that d min , the corresponding errors are 0.7% (rmsse) and 1.3%
(rmse). These errors are much less than the typical estimation errors and the uncertainty due to
the limited sizes of the typical samples available. In conclusion, the parameter in the
denominator of Eq. (2.8) can be neglected and the remaining two parameters suffice. Hence,
hereafter we will assume that = 1 .
10
Initially, the coefficients , , and can be considered as dependent on the return

period. However, their functional dependence cannot be arbitrary, because the relationships for
any two return periods T1 and T2 < T1 must not intersect. If {1 , 1 , 1} and {2 , 2 , 2 }
are the two parameter sets for T1 and T2 respectively, then it can be shown that there exist at
least two sets of constraints leading to feasible (i.e. not intersecting) idf curves. These are
1 > 0, 2 > 0,
1 1 1 1
1
1,
,
>
2
2 2 2 2
(2.9)
and
1 0, 2 0, 1 = 2 = ,
1

> 1, 1 2
2
1
2
1
(2.10)
To both these sets, the following obvious inequalities are additional constraints
1 > 0, 2 > 0, 0 < 1 < 1, 0 < 2 < 1
(2.11)
The essential difference between the sets of constraints in Eq. (2.9) and Eq. (2.10) is that
the former does not allow to take zero value, whereas the latter does allow this special
value. Furthermore, it can be shown that, if is allowed to take zero value, then the
exponent in Eq. (2.8) must be constant and independent of the return period. Because the
case = 0 must not be excluded, it is reasonable to adopt the set of constraints of Eq. (2.10)
for the subsequent analysis. For convenience, it is reasonable to consider independent of
the return period as well, thus leading to the following final set of restrictions
1 = 2 = 0, 0 < 1 = 2 = < 1, 1 > 2 > 0
(2.12)
In this final set of restrictions, the only parameter that is considered as an (increasing)
function of the return period T is . This leads, indeed, in a strong simplification of the
problem of construction of idf curves. This theoretical discussion is empirically verified, as
11
numerous studies have shown that real world families of idf curves can be well described with
constant parameters and .
2.3 The IDF Curve using Cumulative Distribution Function

After the above discussion they could formulate a mathematical framework for IDF
relationship in the form (Koutsoyiannis et al., 1998)
i=
a(T )
b( d )
(2.13)
which has the advantage of a separable functional dependence of i on T and d . The

function b(d ) is
b(d ) = (d + )
(2.14)
where and are parameters to be estimated ( > 0 , 0 < < 1 ). The function a (T )
(which coincides with of section 2.2) is given in the bibliography (e.g. Raudkivi (1979);
Shaw (1983); Subramanya (1984); Chow (1988); Wanielista (1990); Singh (1992)) by the
following alternative relations
a (T ) = T
(2.15)
a (T ) = c + ln T
(2.16)
The first is the oldest (Bernard, 1932) yet the most common until recently (e.g. see
Kouthyari (1992) and Pagliara (1993)). These relations are rather empirical and their use has
been dictated by their simplicity and computational convenience rather than their theoretical
consistency with the probability distribution functions which are appropriate for the maximum
rainfall intensity. Chen (1983) applied a more theoretical analysis to obtain similar
12
relationships. Koutsoyiannis (1994) reported that Eq. (2.15) is inappropriate for certain issues
such as simulation (e.g. it tends to underestimate the variance). Koutsoyiannis (1996) has
demonstrated empirically that if the maximum rainfall intensity has a Gumbel distribution then
the parameters and of Eq. (2.15) are not in fact constant but they depend on the
return period T (this is also demonstrated briefly below).
In fact, there is no need to introduce a (T ) as an empirical function, as it can be
completely determined, in a theoretically consistent manner, from the probability distribution
function of the maximum rainfall intensity I (d ) . Indeed, if the intensity I (d ) of a certain
duration d has a particular distribution Fl ( d ) (i; d ) , this will also be the distribution of the
variable Y = I (d )b(d ) , which is no more than the intensity rescaled by b(d ) (with the
parameters of the latter distribution being properly rescaled). This has also been reported by
Koutsoyiannis (1994), appendix A for the Gumbel distribution, but it can be generalized for
any distribution. Mathematically, this is expressed by
P{I (d ) i} = P{I (d )b(d ) ib(d )} = P{Y y}
(2.17)
where P{ } denotes probability, or
FI ( d ) (i; d ) = FY ( yT ) = 1
1
T
(2.18)
Hence, if yT is the (l 1/ T ) -quantile of the distribution function FY , then
yT a (T ) = FY 1 (1 1/ T )
(2.19)
which proves our claim that a (T ) is completely determined from the distribution function of
intensity.
They point out that the inverse of a distribution function appearing in Eq. (2.19) generally
does not have as simple an expression as those of the empirical functions Eq. (2.15) and Eq.
13
(2.16), and in some cases FY 1 ( ) cannot be expressed with an explicit analytical equation.
However, as shown below, approximate analytical expressions and always be gotten
adequately simple and more accurate than the empirical functions Eq. (2.15) and Eq. (2.16).
In section 2.5, it is shown to examine the most typical distribution functions of maximum
intensities and to obtain for each distribution function the corresponding function a (T ) .
Notably, it is shown that the empirical functions Eq. (2.15) and Eq. (2.16) can be obtained
by our general methodology, but they correspond to distribution functions that may not be
appropriate for maximum rainfall intensities.
2.4 Alternative Distribution Functions

To better serve our purpose, the mathematical expressions of the alternative distribution
functions FY ( y ) given below may have been written intentionally in a slightly different form
from that typically used in the literature. In all distributions, and denote dimensionless
parameters whereas and c denote parameters having the same dimensions as the random
variable y (or ln y in the case of logarithmic transformation of the variable)(Koutsoyiannis
et al., 1998).
2.4.1 Gumbel Distribution Function

The type I distribution of maxima, also termed the Gumbel distribution function (Gumbel,
1958), is the most widely used distribution for IDF analysis owing to its suitability for
modelling maxima. Given that the rainfall intensity I (d ) has a Gumbel distribution for any
duration d , so will Y , and thus
14
FY ( y ) = exp(e y / + )
(2.20)
where and are the scale and location parameters respectively of the distribution
function. Combining Eqs. (2.19) and (2.20) we directly get
1
yT a (T ) = ln ln 1
T
(2.21)
which is an exact yet simple expression of a (T ) (Koutsoyiannis et al., 1998).
2.4.2 Generalized Extreme Value Distribution

This general distribution, which incorporates type I, II, and III extreme value distributions
of maxima can be written in the form
1/

FY ( y ) = exp 1 +
y ( 1/ )
(2.22)
where > 0 , > 0 , and are shape, scale, and location parameters respectively. For
= 0 the generalized extreme value (GEV) distribution turns into the Gumbel distribution;
the case where < 0 is not considered here because it implies an upper bound of the variable,
which is not the case in maximum rainfall intensity. We directly obtain from Eq. (2.22) that
yT a (T ) = +
1
ln 1 T
(2.23)
1
= ' '+ ln 1
T
15
where for simplification we have set ' = / and ' = 1 . Again we have an exact
expression of a (T ) for the GEV distribution that remains relatively simple (Koutsoyiannis et
al., 1998).
16
Chapter 3. Parameter Estimation Methods
The parameters of the IDF curve using CDF fall into two categories: those of the function
a (T ) (i.e. , , , etc., depending on the distribution function adopted) and those of the
function b(d ) (i.e. and ) (Koutsoyiannis et al., 1998). In all procedures Koutsoyiannis
at al (1998) assume that they was given groups each holding the historical intensities of a
particular duration d j , j = 1,..., k . They denoted by n j the length of the group j , and by
i jl the intensity values of this group (samples of the random variables I j = I (d j ) ) with
l = 1,..., n j denoting the rank of the value i jl in the group j arranged in descending order.
3.1 Typical Procedure

The typical parameter estimation procedure for idf curves (Raudkivi (1979); Chow (1988);
Wanielista (1990); Singh (1992)) consists of three steps. The first step consists of fitting a
probability distribution function to each group comprised of the data values for a specific
duration d j . In the second step the rainfall intensities for each d j and a set of selected return
periods (e.g. 5, 10, 20, 50, 100 years, etc.) are calculated. This is done by using the probability
distribution functions of the first step. In the third step the final idf curves are obtained in two
different ways: either (a) for each selected return period the intensities of the second step are
treated and a relationship of i as a function of d (i.e. i = iT (d ) ) is established by (bivariate)
least squares, or (b) the intensities of the second step for all selected return periods are treated
simultaneously and a relationship of i as a function of both d and T (i.e. i = i (T , d ) ) is
established by (three-variate) least squares. In case (a) different values of the parameters ,
17
and are obtained for each T . In case (b) unique values of the parameters and are
obtained, whereas is determined as a function = a(T ) The form of this function
(typically Eq. (2.15) or Eq. (2.16)) is selected a priori. In the case that a (T ) is given by the
power relationship in Eq. (2.15), the estimation procedure is simplified, because Eq. (2.13)
becomes linear by taking logarithms of both sides.
The main advantage of this parameter estimation procedure is its computational simplicity,
which in fact imposes the separation of the calculations in three steps, so that the calculations
of each step are as simple as possible. However, the procedure has some flaws, which are not
unavoidable. First, it bears the weakness of using an empirically established function a (T )
(step 3) instead of the one consistent with the probability distribution function (step 1). This
has been already discussed in section 2. Second, it is subjective, in the sense that the final
parameters depend on the selected return periods in step 2. This dependence may be essential
if the selected empirical function a (T ) departs significantly from that implied by the
probability distribution function (Koutsoyiannis, 1996). Third, it treats the three involved
variables ( i , d , T ) as having the same nature, in spite of the fact that they are
fundamentally different in nature, i.e. i represents a random variable, d is a (non-random)
parameter of this random variable, and T is a transformation of the probability distribution
function of the random variable.
In sections 3.2 and 3.3 Koutsoyiannis at al (1998) proposed two different parameter
estimation methods that are free of the flaws of the above-described typical procedure and
harmonize with the general formulation of IDF curves given in section 2. These procedures
need more complicated calculations than the typical procedure, yet remain computationally
simple. Both can be applied sing a typical spreadsheet package and do not require the
development of specialized computer programs.
18
3.2 Robust Estimation

The first proposed method estimates the parameters in two steps, the first concerning the
parameters of function b(d ) and the second those of a (T ) (Koutsoyiannis et al., 1998).
This method is based on the identity of the distribution functions of the variables Y j = I j b(d j )
of all k groups, regardless of the duration d j of each separate group. This identity leads
them to the Kruskal-Wallis statistic, which is used to test whether several sample groups
belong to the same population. Let them assume that the parameters and of b(d ) are
known. Then all values y jl = i jl b(d j ) can be found. The overall number of data values is
m = nj
k
(3.1)
j =1
Ranks rjl were assigned to all of the m data values y jl (using average ranks in the
event of ties). For each group we compute the average rank r j of the n j values of that
group. If all groups have identical distribution then each r j must be very close to (m + 1) / 2 .
This leads to the following statistic (Kruskal-Wallis) which combines the results of all groups
k KW =
k
12
m +1
n j rj
2
m(m + 1) j =1
(3.2)
The smaller the value of k KW , the greater the evidence that all groups of y values belong
to the same population. Obviously, the ranks rjl (and hence k KW ) depend on the parameters
and that were assumed as known. Consequently, the estimation problem is reduced to
an optimization problem defined as Eq.(3.2)
minimize k KW = f1 ( , )
(3.3)
19
Apparently, it is not possible to establish an analytical optimization method for our case. A
numerical search technique for optimization that makes no use of derivatives (see Pierre
(1986) and Press (1992)) is appropriate. However, it may be simpler to use a trial-and-error
method based on a common spreadsheet computer program. The advantages of the KruskalWallis statistic are its non-parametric character and its robustness, i.e. its ability not to be
affected by the presence of extreme values in the samples. They clarify, however, that the
minimum value of k KW determined by the minimization process cannot be used further to
perform the typical Kruskal-Wallis statistical test (actually, the testing is not really needed).
The reason is that this test assumes that all k groups are mutually independent. In our case, the
intensities I j of the different groups are stochastically dependent variables, as is evident from
their construction (see section 2.2). Thus, we do not know the distribution function of the
statistic k KW to perform any statistical test. Nevertheless, the minimization of its value is
achievable because the distribution function does not need to be known.
For the sake of improving the fitting of b(d ) in the region of higher intensities (and also
to simplify the calculations) it may be preferable to use in this first step of calculations a part
of the data values of each group instead of the complete series. For example, we can use the
highest 1/2 or 1/3 of intensity values for each duration.
Given the values of and , they proceed to the second step of calculations, which is
very easy. Assuming that, with these values, all groups have identical distribution, they append
all k groups of values y jl thus forming a unique (compound) sample. For this sample they
choose an appropriate distribution function, and estimate its parameters using the appropriate
estimators for that distribution (e.g. those obtained by the methods of maximum likelihood,
moments, L-moments, etc.; for a concise presentation of such estimators see Stedinger (1993)).
This defines completely the form and the parameters of a (T ) .
20
3.3 One-Step Least Squares Method

The second method estimates all parameters of both functions a (T ) and b(d ) in one
step, minimizing the total square error of the fitted idf relationship to the data (Koutsoyiannis
et al., 1998). To this aim, to each data value i jl Koutsoyiannis et al.(1998) assign an
empirical return period using, e.g. the Gringorten formula
T jl =
n j + 0.12
(3.4)
l 0.44
So, for each data value we have a triplet of numbers ( ilj , Tlj , d j ). On the other hand,
given a specific form of a (T ) , chosen among those of section 2.5 from preliminary
investigations of the type of the distribution function of intensity, they obtain the modeled
intensity
a(T jl )
$
i jl =
b( d j )
(3.5)
and the corresponding error
e jl = ln i jl ln $
i jl = ln(i jl / $
i jl )
(3.6)
where they have applied the logarithmic transformation to keep balance among the errors of
the intensities of greater durations (which are lower) and those of lower ones. The overall
mean square error is
e2 =
1
k
j =1
1
nj
e
nj
l =1
2
jl
(3.7)
Again the estimation problem is reduced into an optimization problem, defined as

minimize e = f 2 ( , , , , , ... )
(3.8)
21
A numerical search technique for optimization that makes no use of derivatives, such as the
Powell method (see Pierre(1986) and Press et al.(1992)), is appropriate for this problem.
However, it may be simpler to perform the optimization using the embedded solver tools of
common spreadsheet packages.
They note that the least squares method in fitting a theoretical to an empirical distribution
function is not a novelty of the proposed method. Rather, the innovative element of the
proposed method is the simultaneous estimation of the parameters of both the distribution
function and the duration function b(d ) .
3.4 Parameter Estimation Methods using Genetic Algorithms

Genetic Algorithms (GAs) are based on the mechanics of natural selection and natural
genetics (Goldberg, 1989). They were developed in the 1960s and refined throughout the
1970s by John Holland and his coworkers at the University of Michigan to explain and model
the adaptability of natural systems. Hollands monograph (Holland, 1975), Adaptation in
Natural and Artificial Systems, is the seminal book in this field, and Goldbergs book, Genetic
Algorithms in Search, Optimization and Machine Learning (Goldberg, 1989), deals with the
most practical optimization problems in evolutionary computation (Kim, 2005).
GAs are a very effective optimization method for solving multi-objective problems
(MOPs) because they exploit not only a single solution but also a set of many possible
solutions, namely a population, for searching the global optimum. Hence, they can easily
determine which solution is superior to the others, even though only one run of the
optimization model is performed; this is a major advantage of GAs when Pareto-optimality,
which was originally introduced by (Edgeworth, 1881) and later generalized by (Pareto, 1896),
is applied to achieve global or near-global optimum solutions for MOPs. Readers who are
22
interested in GAs and Pareto-optimality can refer to the textbooks written by (Mitchell, 1996)
and (Deb, 2001), and a number of technical papers, e.g. (Van Veldhuizen, 2000), are good
references (Kim et al., 2008a).
Nondominated sorting genetic algorithms-II (NSGA-II) (Deb et al., 2002) is widely used to
solve MOPs in the field of engineering (Deb(2003a); Deb(2003b); ISI(2004)). The key
features of NSGA-II that are different from those of the former NSGA (Srinivas, 1994) are a
reduction in time complexity, a parameterless sharing procedure that uses crowding distance
for ensuring diversity in a population, and elitism that can speed up the performance of GAs
and can also help in preventing the loss of good solutions once they are found (Deb et al.,
2002).
Genetic algorithm would be effective tool for parameter estimation of equations(Giustolisi
et al., 2006), especially for non linear relationship like IDF curves. In addition, the
compromised solutions, namely Pareto-optimal solutions, would be achieved using NSGA-II
satisfying with multi-objectives of this study (Kim et al., 2008b).
3.4.1 Parameter Estimation Method using Single-objective Genetic

Algorithms
Single-objective Genetic Algorithms (SGA) is used as the parameter estimation method. In
order to minimize deviation between objective-value and estimated-value, root-mean-squarederror (RMSE, Eq.(3.9), Objective-function-1, OF-1) and relative RMSE (RRMSE, Eq.(3.10),
Objective-function-2, OF-2) were selected as to single-objective functions of SGA (Kim et
al., 2007b).
RMSE
1 n2 n1
Qij Qij
n j =1 i =1
(3.9)
23
RRMSE =
Q
1 n2 n1 Q
ij
ij
n j =1 i =1 Qij
(3.10)
where n is the number of total data, n1 and n2 are the number of rainfall duration and
return period, Qij is the estimated rainfall quantile by IDF curve with i -th duration and j th return period, and Qij is the computed rainfall quantile by at-site frequency analysis
software (FARD2006) corresponding to i -th duration and j -th return period.
At-site frequency analysis result was selected as to objective-value (more details in the
section 4.1). The SGA source code using C language can be freely downloaded from
http://www.iitk.ac.in/kangal/soft.htm.
3.4.2 Parameter Estimation Method using NSGA-II

In this method, NSGA-II, which is multi-objective genetic algorithms (MOGA), was used
to derive more precise formulas of IDF curve. In order to achieve Pareto-optimal solutions,
two multi-objective functions (Eqs. (3.9) and (3.10)) are used in this study.
RMSE represents a king of absolute squared error as compared with RRMSE. Therefore, if
the computed quantiles ( Qij ) which is true value become larger, IDF curve with a parameter
set with smaller RMSE is more accurate than that with smaller RRMSE that is a relative
squared error measure. On the contrary, if the quantile Qij become smaller, RRMSE is an
effective tool to measure the accuracy of IDF curve.
Figure 3. 1 shows Pareto-front calculated by NSGA-II with Eqs. (3.9) and (3.10) as
multi-objective functions, and the grey-colored circles are Pareto-optimal solutions. It can be
seen from the figure that a Pareto-optimal solution defined as RMSE has the smallest RMSE
24
among Pareto-front; the other extreme Pareto-optimal solution identified as RRMSE has the
smallest RRMSE.
0.0190
RMSE
RRMSE
0.0188
0.0186
0.0184
RRMSE
0.0182
0.0180
2.60
2.64
2.68
2.72
2.76
2.80
RMSE
Figure 3. 1 Pareto-optimal solution indicated as "RMSE" that makes an objective function

about RMSE the smallest value and the other objective function about RRMSE
worst, and vice versa
Figure 3. 2 depicts more clearly the characteristics of RMSE and RRMSE of Figure 3.
1. The absolute bias means the difference between the estimated rainfall quantiles by IDF
curve that is derived using RMSE Pareto-optimal solution of Figure 3. 1. On the other hand,
RRMSE means the absolute bias of the estimated quantiles using RRMSE of Figure 3. 1.
It is interesting to note that the absolute biases for RMSE and RRMSE corresponding to
the respective short- and long-rainfall duration show the different features. For example, the
absolute biases of RRMSE are smaller than those of RMSE in the short rainfall durations like
1, 2, and 3 hours. Whereas the absolutes biases of RRMSE is smaller than those of RMSE for
the long rainfall durations such as from 6 hours to 48 hours. As a result, the IDF curve derived
25
using the Pareto-optimal solution indicated RRMSE in Figure 3. 1 is more effective to

compute the rainfall quantiles for short duration. On the other hand, the IDF curve using
Absolute Bias
RMSE is more useful to calculate quantiles for long duration.
50
45
40
35
30
25
20
15
10
5
0
RMSE
RRMSE
12
15
18
24
48
Duration
Figure 3. 2 Comparison of absolute bias of rainfall quantiles estimated by IDF curve whose
parameters are estimated by two extreme Pareto-optimal solutions in Figure 3. 1
Table 3. 1 shows a method of discriminating short- and long-rainfall durations as an

example. RRMSE in the table means that the absolute bias of the rainfall quantiles using
RRMSE is smaller than that using RMSE in the Figure 3. 1, and RMSE shows that the
absolute bias using RMSE is smaller than that of RRMSE. According to the comparison
result, short-duration is by 2 hours and from 3 hours it can be said to be long-duration
(Shin(2007), Kim(2008b)). This separation of short- and long-rainfall durations method is
called SEP_DUR in this study.
26
Table 3. 1 Separation of short- and long-rainfall durations method

Return Period
(year)
2
3
5
10
20
30
50
70
80
100
200
300
500
Total
Duration(hr)
1
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
13
0
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RRMSE
RMSE
RMSE
RMSE
RMSE
RRMSE
RRMSE
RRMSE
9
4
RRMSE
RRMSE
RRMSE
RRMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RRMSE
RRMSE
RRMSE
6
7
EQUAL
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RRMSE
RMSE
RMSE
1
12
RMSE
RRMSE
RRMSE
RRMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
RMSE
3
10
RMSE
27
Chapter 4. Application
4.1 Description of Sites and Data

In this study, rainfall data of 76 rainfall recording sites overall the Korea Meteorological
Administration (KMA) were used, which the 76 sites have been controlled by KMA. 59 sites
have rainfall records at least 35 years until 2007 and the longest record is 47 years. The rainfall
data have maintained recording rain gauges well by KMA and there are almost no incidents of
incorrect measurement. Figure 4. 1 and Table 4. 1 show the location and sites information.
Figure 4. 1 Location of 76 rainfall recording sites overall KMA
28
Table 4. 1 Information of 76 sites controlled by Korea KMA

No.
Site Code
Number
Site name
First year
Last year
Record
length
Tm-X
Tm-Y
90
Sokcho
1968
2007
40
336901.3380
528692.8062
95
Cheorwon
1988
2007
20
226596.0977
516148.6695
98
Dongducheon
1998
2007
10
205268.5961
488792.3465
99
Munsan
2001
2007
179392.4277
487055.2332
100
Daegwallyeong
1981
2007
27
355053.8601
466708.3982
101
Chuncheon
1966
2007
42
264632.8290
489139.6378
102
Baengnyeongdo
2000
2007
-8264.1643
498889.6621
105
Gangneung
1961
2007
47
366590.9666
473801.8625
106
Donghae
1992
2007
16
387762.7503
447101.8778
10
108
Seoul
1961
2007
47
196910.0557
452124.6940
11
112
Incheon
1961
2007
47
166726.1348
441767.7569
12
114
Wonju
1986
2007
22
283809.5290
426585.6786
13
115
Ulleung
1961
2007
47
544819.2612
449544.5517
14
119
Suwon
1964
2007
44
198647.5245
418989.7164
15
121
Yeongwol
1995
2007
13
329356.9309
409814.2162
16
127
Chungju
1973
2007
35
284762.5150
385843.0390
17
129
Seosan
1968
2007
40
154750.0717
364036.9158
18
130
Uljin
1972
2007
36
414744.1560
390517.2335
19
131
Cheongju
1967
2007
41
239338.2020
348761.2530
20
133
Daejeon
1969
2007
39
233323.4500
319077.6045
21
135
Chupongnyeong
1961
2007
47
289344.0003
302935.0398
22
136
Andong
1983
2007
25
352745.6260
342680.1573
23
137
Sangji
2002
2007
303745.5927
323671.6535
24
138
Pohang
1961
2007
47
414413.4139
283980.4690
25
140
Gunsan
1968
2007
40
178416.1763
278352.2514
26
143
Daegu
1961
2007
47
346121.0024
266199.7322
27
146
Jeonju
1961
2007
47
213929.7374
257938.2599
28
152
Ulsan
1961
2007
47
410303.2726
231404.6510
29
155
Masan
1985
2007
23
343218.9864
186796.4925
30
156
Gwangju
1961
2007
47
190059.5180
185983.9736
31
159
Busan
1961
2007
47
385202.6520
180285.8211
32
162
Tongyeong
1968
2007
40
331231.1314
150578.2598
33
165
Mokpo
1961
2007
47
143317.2550
146646.2432
34
168
Yoesu
1961
2007
47
267755.5205
138111.6945
35
169
Heuksando
1997
2007
11
57999.9404
133181.7697
36
37
38
170
175
184
Wando
Jindo
Cheju
1973
2002
1962
2007
2007
2007
35
6
44
172513.4340
137807.2058
156231.9161
99807.1125
108427.6668
2073.4148
TM : Transverse Mercator System
29
Table 4. 1 Information of 76 sites controlled by KMA ( Continued )

No.
Site Code
Number
Site name
First year
Last year
Record
length
Tm-X
Tm-Y
39
40
41
185
187
189
Gosan
Sungsan
Seogwipo
1988
1997
1961
2007
2007
2007
20
11
47
121953.0496
188798.8911
159421.4034
-22148.3464
-11788.9053
-27665.8060
42
192
Jinju
1969
2007
39
294680.3729
185451.3788
43
201
Ganghwa
1973
2007
35
151141.4454
467677.8921
44
202
Yangpyung
1973
2007
35
243666.7505
443049.8459
45
203
Icheon
1973
2007
35
242882.7111
418114.8414
46
211
Inje
1973
2007
35
302345.2968
506990.5046
47
212
Hongcheon
1973
2007
35
277582.3317
464963.1361
48
216
Taeback
1985
2007
23
376611.5448
409468.4084
49
221
Jecheon
1973
2007
35
291050.0770
404473.6873
50
226
Boeun
1973
2007
35
265710.8250
332092.2665
51
232
Cheonan
1973
2007
35
210566.2477
364577.1789
52
235
Boryeong
1973
2007
35
160192.1333
314136.2998
53
236
Buyeo
1973
2007
35
192812.3510
307961.1641
54
238
Geumsan
1973
2007
35
243307.4232
289563.1791
55
243
Buan
1973
2007
35
174284.8680
247758.8935
56
244
Imsil
1973
2007
35
225800.7405
234746.7007
57
245
Jeongeup
1973
2007
35
187789.2871
229270.5808
58
247
namwon
1973
2007
35
230187.7254
211806.9945
59
248
Jangsu
1988
2007
20
247040.2209
239794.9323
60
256
suncheon
1973
2007
35
221740.5489
175126.6100
61
260
Jangheung
1973
2007
35
192554.5186
132252.6687
62
261
Haenam
1973
2007
35
160368.2896
117347.7116
63
262
Goheung
1973
2007
35
225217.3329
124464.0061
64
271
Bonghwa
1988
2007
20
370473.8296
384158.2743
65
272
Yeongju
1973
2007
35
335179.0246
375556.2048
66
273
Mungyeong
1973
2007
35
302681.8413
347955.6113
67
277
Yeongdeok
1973
2007
35
415702.0564
339940.2625
68
278
Uiseong
1973
2007
35
351495.3460
318565.7020
69
279
Gumi
1973
2007
35
318803.1841
293030.0127
70
281
Yeongcheon
1973
2007
35
375934.1279
276987.8510
71
284
Geochang
1973
2007
35
282410.6441
241633.2775
72
285
Hapcheon
1973
2007
35
305984.5469
230093.0786
73
288
Miryang
1973
2007
35
358196.6952
222709.9471
74
289
Sancheong
1973
2007
35
279776.1745
212955.9483
75
76
294
295
Geoje
Namhae
1973
1973
2007
2007
35
35
346607.9993
284687.9352
155551.8548
146830.8019
TM : Transverse Mercator System
30
At-site frequency analysis result was selected as to objective-value of SGA and NSGA-II.
Then, as a first step, at-site frequency analysis was performed at 72 sites recorded by KMA. It
is for this reason why analysis site should have record length more than 10 year for performing
at-site frequency analysis. Quantiles were computed about return periods (2, 3, 5, 10, 20, 30,
50, 70, 80, 100, 200, 300, 500 year) and duration (60, 120, 180, 360, 540, 720, 900, 1080,
1440, 2880 min). Gumbel distribution and GEV distribution was selected as the proper
probability distribution for the annual maximum rainfall by the goodness of fit test such as
Kolmogorov-Smirnov test, 2 -test, Cramer von Mises test, and Probability Plot Correlation
Coefficient(PPCC) test. And method of probability weighted moments (PWM) was selected as
to parameter estimation method.
In order to at-site frequency analysis, Frequency Analysis of Rainfall Data 2006
(FARD2006) program has been used (NIDP, 2007). FARD2006 was developed by National
Institute
for
Disaster
Prevention
(NIDP)
and
can
be
freely downloaded
from
http://www.nidp.go.kr.
4.2 Application of IDF Curve using Cumulative Distribution

Function
In this section, IDF curve using CDF is applied for Korean rainfall data to assess
applicable suitability. Equation form of IDF curve using CDF is Eq. (2.13) which was
proposed by Koutsoyiannis et al. (1998) (see chapter 2.4). Mostly selected probability
distributions are GEV and Gumbel distribution in Korea. Where probability distribution is
GEV and Gumbel, IDF curve using CDF is
31
ln
1
a(T )
i=
=
b( d )
( d + )
i=
a(T )
=
b( d )
( d + )
(4.1)
: GEV
ln ln 1
T
: Gumbel
(4.2)
where, , , , , are the parameters of IDF curve using CDF for the each probability
distribution function. Robust estimation (ROBUST) and One-step least squares method
(ONESTEP) are performed to estimate the parameters (see sections 3.2 and 3.3); in order to
efficiently minimize k KW and e , especially genetic algorithms was used instead of using
trial-and-error method based on a common spreadsheet computer program. Also, the
parameters of IDF curve using CDF were estimated by SGA to improve accuracy; in this study,
at-site frequency analysis result until 2007 was selected as objective-value and RMSE and
RRMSE were seleted as objective-functions. The parameters of 72 rainfall recording sites of
KMA were estimated using SGA and IDF curve using CDF. In this study, the case of using
RMSE as objective-function called IDF_CDF(OF_I) and the case of using RRMSE as
objective-function called IDF_CDF(OF_II).
4.3. Parameter Estimation of Conventional IDF Curves

Rainfall quantiles corresponding to any specific return period and rainfall duration are
essential for constructing hydraulic structures. In Korea, FARD2006 for computing rainfall
quantiles using at-site frequency analysis is already developed, but IDF curve is still very
32
useful to calculate rainfall intensity or quantile for arbitrary rainfall duration in which rainfall
data has not been recorded (Kim et al., 2008b).
The following six IDF curves are widely used in Korea. These IDF curves have been
essentially used to calculate rainfall quantiles for designing hydraulic structures in Korea.
Equations. (4.3), (4.4), (4.5) and (4.6) are Modified Talbot, Modified Sherman, Modified
Japanese and Modified Semi-Log type IDF curves; Equations (4.7) and (4.8) are developed
by Lee et al. (1993), Heo et al. (1999) and Ministry of Construction and Transportation (2000)
I (t ) =
a
+c
t +b
: Modified Talbot
(4.3)
I (t ) =
a
(t + c)b
: Modified Sherman
(4.4)
I (t ) =
: Modified Japanese
(4.5)
: Modified Semi Log
(4.6)
: LEE
(4.7)
: HEO
(4.8)
t +b
+c
I (t ) = a + b log(t + c)
I (T , t ) =
I (T , t ) =
a + b log T
tn + c
a + b ln
c + d ln
T
tn
T
+ t
t
where I (t ) and I (t , T ) are rainfall intensity (mm/hr), t is duration, T (min) is return

period (year), a, b, c, d , n are the parameters for the each recording site. The latter to IDF
curves make up for defect of the former four IDF curves that dont have return period ( T ) in
the IDF formula and show the reverse intensity as rainfall duration increases And then, the
parameters of the former four IDF curves should be estimated by regression analysis
corresponding to each return period. On the contrary, the parameters of the latter two IDF
33
curves can be estimated as one set. Therefore, the latter two IDF curves are more general than
the former four IDF curve.
In this study, in order to compare accuracy and assess applicable suitability of IDF curve
using CDF, the parameters of the six IDF curves at 72 sites of KMA were newly estimated. In
case of Modified Talbot, Modified Sherman, Modified Japanese and Modified Semi-Log, the
parameters were estimated by linear regression analysis using at-site frequency analysis result
until 2007. Moreover, in case of Lee and HEO, the parameters were estimated by SGA
(section 3.4.1); at-site frequency analysis result until 2007 was selected as objective-value and
RMSE and RRMSE as objective-functions. The case of using RMSE as objective-functions
called LEE(OF_I), HEO(OF_I) and the case of using RRMSE as objective-functions called
LEE(OF_II),HEO(OF_II).
4.4. Parameter Estimation using Separation of Short- and LongRainfall Durations Method
IDF curve using CDF was applied for 72 sites of KMA and the parameters were estimated
using SEP_DUR (section 3.4.2) to derive more precise formulas improve accuracy. SEP_DUR
uses two parameter sets for short- and long-rainfall durations, and can be employed
automatically. Object-value was selected at-site frequency analysis result and the parameters of
IDF curve using CDF were estimated using SEP_DUR (IDF_CDF(SEP_DUR)) at the 72
sites of KMA.
In order to compare accuracy, the parameters of the HEO , which is the more accurate IDF
curve among the conventional IDF curves, at 72 sites of KMA were newly estimated using the
multiple non-linear regression as conventional method separating rainfall duration
34
(HEO(REG)) (Ministry of Construction and Transportation, 2000). And the parameters of

the HEO were estimated using the SEP_DUR (HEO(SEP_DUR)).
35
Chapter 5. Comparison
5.1 Total Rainfall Duration

In this study, IDF curve using CDF was applied to Korean rainfall data and the parameters
IDF curve using CDF and conventional IDF curves were estimated using linear regression,
ROBUST, ONESTEP, and SGA. Modified Talbot, Modified Sherman, Modified Japanese and
Modified Semi-Log were estimated using linear regression analysis using at-site frequency
analysis result until 2007, and then LEE and HEO were estimated using SGA; in this study, atsite frequency analysis result until 2007 was selected as objective-value and RMSE(OF-1),
RRMSE(OF-2) as objective-functions. Lastly, the parameters of IDF curve using CDF were
estimated using ROBUST, ONESTEP, and SGA.
Intensity of each IDF curve was calculated, and at-site frequency analysis result was
transformed into intensity to compare the accuracy. Then RMSE and RRMSE between
estimated intensity ( $
i jl ) and transformed intensity ( i jl ) were calculated. Where $
i ij is the
estimated rainfall intensity by IDF curve with i -th duration and j -th return period, and iij is
the computed rainfall intensity by at-site frequency analysis software (FARD2006)
corresponding to i -th duration and j -th return period.
Table 5. 1 shows result of accuracy comparison (details in Table A. 1 and Table A. 2 of
Appendix A). It was performed to search method having smallest RMSE or RRMSE
corresponding to each site and to count number of smallest RMSE or RRMSE corresponding
to each method. And then it decided that method numerous number of smallest RMSE or
RRMSE is accurate.
The bold italic dark gray figures mean the 1st accurate method, the bold italic light gray
figures mean the 2nd accurate one and the bold italic figures mean the 3rd accurate one. Table 5.
36
1 shows that Modified Sherman is the 1st accurate IDF curve for both the Gumbel and GEV
models. And, IDF_CDF(OF_II) is the 2nd and HEO (OF_II) is the 3rd accurate one for the
Gumbel model while HEO (OF_II) is the 2nd and IDF_CDF(OF_II) is the 3rd accurate one for
the GEV model.
Table 5. 1 Number of the smallest RMSE or RRMSE of total sites using 12 different IDF
curves in case of Gumbel and GEV
IDF curves and Parameter estimation methods
Probability
Accuracy
distribution
Gumbel
GEV
LEE
HEO
Modified
SEMI
LOG
OF_I
OF_II
OF_I
OF_II
44
62
Total
119
RMSE
53
RRMSE
66
Modified
TALBOT
Modified
Modified
SHERMAN
JAPANESE
Total
106
RMSE
RRMSE
IDF_CDF
ONE
STEP
ROBUST
OF_I
OF_II
24
17
15
12
Modified Talbot, Modified Sherman, Modified Japanese and Modified Semi-Log dont
have return period ( T ) in the IDF formula. Therefore, these IDF curves should be estimated
by regression analysis corresponding to each return period. Then, these four IDF curves are not
general and useful. And alse, the four IDF curves show the reverse intensity as rainfall duration
increases. Figure 5. 1 and Figure 5. 2 well shows the one. Figure 5. 1 is a figure which is IDF
curve in case of Modified Sherman at site code number 130 (Uljin). And Figure 5. 2 represents
Figure 5. 1 in detail. Figure 5. 2 shows that a dotted line corresponding to the return period 70
year reverses intensity a bold solid line corresponding to the return period 80 year as rainfall
duration increases between duration 150 min and duration 500 min.
37
R
48
H
9H
R
12
H
R
15
H
18 R
H
R
24
H
R
6H
R
3H
R
2H
R
R
1H
100
Sherman
000130
Rainfall Intensity (mm)
Return Period
500
300
200
100
80
70
50
30
10
5
3
2
10
1
10
100
1000
10000
Duration (min)
15
H
R
12
H
R
9H
R
6H
R
3H
R
2H
R
Figure 5. 1 IDF curve in case of Modified Sherman at site code number 130 (Uljin)
50
Sherman
000130
Rainfall Intensity (mm)
40
30
20
Return Period
500
300
200
100
80
70
50
30
10
5
3
2
100
1000
Duration (min)
Figure 5. 2 IDF curve represented Figure 5. 1in detail at site code number 130 (Uljin)
38
These results lead us to the conclusion the four IDF curves should not be compared with
other IDF curves, therefore the four IDF curves should be excluded out of accuracy
comparison. Table 5. 2 shows to compare the accuracy of the rest one (details in Table A. 3
and Table A. 4 of Appendix A).
in case of Gumbel and GEV
Probability
Accuracy
distribution
Gumbel
GEV
LEE
HEO
OF_I
OF_II
OF_I
OF_II
ONE
STEP
Total
22
RMSE
13
RRMSE
Total
16
RMSE
RRMSE
ROBUST
IDF_CDF
OF_I
OF_II
109
51
58
57
69
14
31
25
26
44
Table 5. 2 shows that the bold italic dark gray figures mean the 1st accurate method, The
bold italic light gray figures mean the 2nd accurate one and the bold italic figures mean the 3rd
accurate one. IDF_CDF(OF_II) is a most accurate IDF curve for both the Gumbel and GEV
models. And, HEO (OF_II) is the 2nd accurate one and LEE (OF_II) is the 3rd accurate one.
5.2 Separation of Rainfall Duration

In order to derive more precise formulas, the rainfall durations were separated into two
such as short- and long-rainfall durations and the parameters of IDF curves were estimated for
separation of rainfall durations. The parameters of HEO were estimated using multiple nonlinear regression analysis and SEP_DUR. And, the parameters of IDF curve using CDF were
estimated by SEP_DUR.
39
Therefore, accuracy of each method was compared using RMSE and RRMSE between
estimated results and results of at-site frequency analysis. Table 5. 3 shows accuracy
comparison of the three methods (details in Table A. 5 and Table A. 6 of Appendix A. Table A.
7 and Table A. 8 of Appendix A show the parameters of IDF_CDF(SEP_DUR)).
in case of Gumbel and GEV
Probability
distribution
Gumbel
GEV
Accuracy

HEO(REG)
HEO(SEP_DUR) IDF_CDF(SEP_DUR)
Total
18
121
RMSE
58
RRMSE
63
Total
16
27
101
RMSE
14
15
43
RRMSE
12
58
Table 5. 3 shows comparison of RMSE and RRMSE. The bold italic dark gray figures
mean the 1st accurate method, the bold italic light gray figures mean the 2nd accurate one and
the bold italic figures mean the 3rd accurate one. Table 5. 3 shows that IDF_CDF(SEP_DUR)
is a most accurate IDF curve for both the Gumbel and GEV models and HEO (SEP_DUR) is
the 2nd accurate one and HEO (REG) is the 3rd accurate one.
40
Chapter 6. Conclusions
In this study, IDF curve using CDF was applied to Korean rainfall data. And the
parameters of conventional IDF curves and IDF curve using CDF were estimated using linear
regression, non-linear regression, ROBUST, ONESTEP, SGA, and MOGA to improve
accuracy. And then, the accuracy of each method was compared each other. The obtained
conclusions are as follows.
Accuracy of each IDF curve was compared using RMSE and RRMSE between estimated
results and results of at-site frequency analysis. As the results, it was found that Modified
Sherman was the 1st accurate IDF curve for both the Gumbel and GEV models. And,
IDF_CDF (OF_II) was the 2nd and HEO (OF_II) was the 3rd accurate one for the Gumbel
model while HEO (OF_II) was the 2nd and IDF_CDF (OF_II) was the 3rd accurate one for the
GEV model.
Modified Talbot, Modified Sherman, Modified Japanese and Modified Semi-Log should
be estimated by regression analysis corresponding to each return period. And these four IDF
curves show the reverse intensity as rainfall duration increases. Therefore, accuracy of the rest
of the IDF curves excluding these IDF curves was compared. As the results, IDF_CDF (OF_II)
was the most accurate IDF curve for both the Gumbel and GEV models, HEO (OF_II) was the
2nd one and LEE (OF_II) was the 3rd one.
In order to derive more precise formulas, the rainfall durations were separated into two
such as short- and long-rainfall durations and the parameters of IDF curves were estimated for
separation of rainfall durations. The parameters of HEO, which is the more accurate IDF curve
among the conventional IDF curves, were estimated by multiple non-linear regression analysis
using at-site frequency analysis results and SEP_DUR. Also, the parameters of IDF curve
using CDF were estimated by SEP_DUR. And then, accuracy of each method was compared
41
using RMSE and RRMSE between estimated intensity and transformed intensity. As the results,
IDF_CDF (SEP_DUR) is the most accurate method for both the Gumbel and GEV models,
HEO (SEP_DUR) is the 2nd one and HEO (REG) is the 3rd one.
These results lead us to the conclusion that IDF curve using CDF for SEP_DUR with
MOGA is the more accurate than any other methods. Therefore, it is found that IDF curves
using CDF can be applied for Korean rainfall data.
42
REFERENCES
Bell, F.C., 1969. Generalized Rainfall-Duration-Frequency Relationship. Journal of the
Hydraulic Division, ASCE, 95(HY1), Paper 6357: 311-327.
Bernard, M.M., 1932. Formulas for rainfall intensities of long durations. Trans.ASCE, 96:
592-624.
Burlando, P. and Rosso, R., 1996. Scaling and Multiscaling Models of Depth-DurationFrequency Curves for Storm Precipitation. Journal of Hydrology, 187(1-2): 45-64.
Chen, C.I., 1983. Rainfall intensity-duration-frequency formulas. Journal of Hydraulic
Engineering, 109(12): 1603-1621.
Choi, G.W., Anhn, T.J. and Kwon, Y.S., 2000. Derivation of Probable Rainfall Intensity
Formulas at Inchon Destrict. Journal of Korea Water Resources Association, 33(2): 263276 (in Korean).
Choi, K.H., Kim, B.S., Jung, J.W. and M.S., H., 2006. Real Time Rainfall Intensity Estimation
using Rainfall Rader and Rain Gauges, Journal of Korea Water Resources Association.
Korea Water Resources Association, pp. 1511-1514 (in Korean).
Chow, V.T., Maidment, D.R., Mays, L.W., 1988. Applied Hydrology. McGraw-Hill.
Deb, K., 2001. Multi-objective Optimization Using Evolutionary Algorithms. John Wiley &
Sons, Chichester.
Deb, K., Pratap, A., Agarwal, S. and Meyarivan, T., 2002. A fast and elitist multiobjective
genetic algorithm: NSGA-. Evolutionary Computation, IEEE Trans, 6(2): 182-197.
Deb, K.J., S., 2003a. Multi-speed gearbox design using multi-objective evolusionary
algorithms. Journal of Mechanical Design, 125(3): 609-619.
Deb, K.R., A.R., 2003b. Reliable classification of two-class cancer data using evolutionary
algorithms. Biosystems, 72(1-2): 111-129.
43
Di Baldassarre, G., Brath, A. and Montanari, A., 2006. Reliability of different depth-durationfrequency equations for estimating short-duration design storms. Water Resources
Research, 42: W12501.
Edgeworth, F.Y., 1881. Mathematical Physics. Keagan Paul, London.
Garcia-Bartual, R. and M., S., 2001. Estimating maximum expected short-duration rainfall
intensityes from extreme convection storms. physics and Chemistry of the Earth, Part B:
Hydrology, Oceans, and Atmosphere, 26(9): 675-681.
Giustolisi, O., Savic, D. and Kapelan, Z., 2006. Multi-Objective Evolutionary Polynomial
Regression, 7th International Conference on Hydroinformatics (HIC 2006), Nice,
FRANCE.
Goldberg, D.E., 1989. Genetic Algorithms in search, optimization & machine learning.
Addison Wesly, Reading, Massachusetts.
Gumbel, E.J., 1958. Statistics of Extremes. Columbia University Press, New York.
Han, J.-H., Kim, K.-D., Heo, J.-H. and Cho, W., 1996. Intensity-Duration-Frequency Curve
Based in Linearizing Method, 1996' Conference of Korea Water Resources Association.
Han, M.S., Choi, G.W., Chung, Y.J. and Ahn, K.S., 2006. Suggestion of Probable Rainfall
Intensity Formula Considering the Pattern Change of Maximum Rainfall at Incheon City.
Journal of Korea Water Resources Association, 39(6): 521-531 (in Korean).
Heo, J.-H., Kim, K.-D. and Han, J.-H., 1999. Derivation of rainfall intensity-durationfrequency equation based on the appropriate probability distribution. Journal of Korea
Water Resources Association, 32(3): 247-254 (in Korean).
Heo, J.Y. and Chae, S.J., 2002. Efficiency Reading Method of Rainfall Recording Paper to
Estimate Probable Rainfall Intensity, 2002' Conference of Korea Water Resources
Association. Korea Water Resources Association, pp. 158-163 (in Korean).
Holland, J.H., 1975. Adaptation in natural and artificial systems. University of Michigan Press.
44
ISI, 2004. Fast breaking papers. http://www.esi-topics.com/fbp/, ISI.

Karahan, H., Ceylan, H. and Tamer Ayvaz, M., 2007. Prediction rainfall intensity using a
genetic algorithm approach. Hydrological Processes, 21: 470-475.
Kim, B.S., Kyoung, M.S., Lee, K.H. and Kim, H.S., 2007a. The estimation of IDF Curve
Considering Climate Change, 2007' Conference of Korea Water Resources Association.
Kim, S.T., Kim, B.S., Kim, H.S. and Seoh, B.-H., 2002. Estimation of Probable Rainfall
Intensity corresponding to Separate of Rainfall Duration 2002' Conference of Korea Water
Resources Association. Korea Water Resources Association, pp. 215-220 (in Korean).
Kim, T., 2005. Multireservoir System Optimization Using Multi-objective Genetic Algorithms,
PhD dissertation, Yonsei University, Seoul.
Kim, T., Heo, J.-H., Bae, D.-H. and Kim, J.-H., 2008a. Single-reservoir operating rules for a
year using multiobjective genetic algorithm. Journal of Hydroinformatics, 10(2): 163-179.
Kim, T., Shin, J., Kim, K. and Heo, J.-H., 2008b. Improving Accuracy of IDF Curves Using
Long- and Short- duration Separation and Multi-Objective Genetic Algorithm. 2008 EWRI
Congress.
Kim, T., Shin, J., Kim, S. and Heo, J.-H., 2007b. Parameter Estimation of Intensity-DurationFrequency Curve Using Genetic Algorithm (I): Comparison Study of Existing Estimation
Method. Journal of Korea Water Resources Association, 40(10): 811-821 (in Korean).
Kottegoda, 1980. Stochastic Water Resources Technology. Macmillan Press, London.
Kouthyari, U.C., Garde, R.J., 1992. Rainfall intensity-duration-frequency formula for India.
Journal of Hydraulic Engineering, 188(2): 323-336.
Koutsoyiannis, D., 1994. A Stochastic Disaggregation Method for Design Storm and Flood
Synthesis. Journal of Hydrology, 156(1-4): 193-225.
Koutsoyiannis, D., 1996. Statistical Hydrology. National Technical University Press, Athens
(in Greek).
45
Koutsoyiannis, D., Kozonis, D. and Manetas, A., 1998. A mathematical framework for
studying rainfall intensity-duration-frequency relationships. Journal of Hydrology, 206(12): 118-135.
Lee, H.C. and Seong, K.W., 2003. The Smmothing of Rainfall Intensity-Duration-Frequency
Relationship curve by the Box-Cox Transformation. Journal of Korean Society of Civil
Engineers, 36(2): 153-159 (in Korean).
Lee, J.J. and Lee, J.S., 1999. A Derivation of Rainfall Intensity-Duration-Frequency
Relationship for the Design of Urban Drainage System in Korea. Journal of Korea Water
Resources Association, 32(4): 403-415 (in Korean).
Lee, J.J., Lee, J.S. and Park, J.Y., 2001. Derivation of Probable Rainfall Intensity Formula of
Individual Zone to Estimate the Design Rainfall. Journal of Korean Society of Civil
Engineers, 21(1-B): 1-10 (in Korean).
Lee, W.-H., 1980. A Stochastic Analysis on Determination of Design Precipitation for
Planning of Urbanized Stream & Sewerage. Journal of Korean Society of Civil Engineers,
28(4): 81-93 (in Korean).
Lee, W.-H., 1997. Hydrology. MunUnDang, Seoul, Korea (in Korean).
Lee, W.-H., Park, S.-D. and Choi, S.-Y., 1992. The Derivation of Representative Probable
Rainfall Intensity Equation, 1992' Conference of Korean Society of Civil Engineers, pp.
135-138 (in Korean).
Lee, W.-H., Park, S.-D. and Choi, S.-Y., 1993. A deviation of the typical probable rainfall
intensity formula in Korea. Journal of The Korean Society of Civil Enginners, 13(1): 115120 (in Korean).
Meyer, A.F., 1928. The Elements of Hydrology. Willey, New York.
Ministry of Construction and Transportation, 2000. 1999 Water resources management
technique development report, Volume I:Rainfall quantile atlas of Korea (in Korean).
Korea Institute of Construction Technology, Ilsan, Kyonggi-Do.
46
Mitchell, M., 1996. An Introduction to Genetic Algorithms. MIT Press, Cambridge, MA.
Mohymont, B., Demaree, G.R. and Faka, D.N., 2004. Establishment of IDF-curves for
precipitation in the tropical ara of Central Africa - comparison of techniques and results.
Natural Hazards and Earth System Sciences, 4: 375-387.
NIDP, 2007. Improvement and Application of Rainfall Analysis Program. National Institute
for Disaster Prevention (NIDP).
Pagliara, S., Viti, C., 1993. Discussion: rainfall-duration-frequency formula for India by U.C.
Kothyari and R.J. Garde. Journal of Hydraulic Engineering, 119(8): 962-967.
Pareto, V., 1896. Cours d'economic politique. F. Rough, Lausanne.
Park, J.Y., Lee, J.S. and Lee, J.J., 2000. The Variation in Peak Descharge according to the
Type of Probable Rainfall Intensity Formula - Application to Wi Stream Basin -, 2000'
Conference of Korea Water Resources Association. Korea Water Resources Association,
pp. 203-208 (in Korean).
Pierre, D.P., 1986. Optimization Teory With Applications. Dover, New York.
Press, W.H., Teukolsky, S.A., Vetterling, W.T. and Flannery, B.P., 1992. Numerical Recipes
in C. Cambridge University Press, Cambridge.
Raudkivi, A.J., 1979. Hydrology, An advanced Introduction to Hydrological Processes and
Modelling. Pergamon, Oxford.
Seoh, E.S., Song, B.H. and Kim, M.J., 2001. Construction of Minutely Rainfall Intensity Data
Base in Korea, 2001' Conference of Korea Water Resources Association, Korea Water
Resources Association, pp. 134-139 (in Korean).
Seoh, K.W., Song, I.J., Lim, J.Y. and Goo, B.S., 1999. Analysis of Return Period due to
Rainfall Durations in '98 Pusan Rainfall Datum, 1999' Conference of Korea Water
Shaw, E.M., 1983. Hydrology in Practice. Berkshire, UK.
47
Sherman, C.W., 1931. Frequency and intensity of excessive rainfalls at Boston, Massachushtts.
ASCE Transactions, 95(Paper No.1780: 951-960.
Shin, J., Kim, T., Kim, S. and Heo, J.-H., 2007. Parameter Estimation of Intensity-DurationFrequency Formula Using Genetic Algorithm(): Separation of Short and Long Durations.
Journal of Korea Water Resources Association, 40(10): 823-832 (in Korean).
Singh, V.P., 1992. Elementary Hydrology. Prentice-Hall, New Jersey.
Singh, V.P. and Zhang, L., 2007. IDF Curves Using the Frank Archimedean Copula. Journal
of Hydraulic Engineering, 12(6): 651-662.
Song, T.J. and Seoh, B.-H., 2000. An Estimation of Probable Rainfall Intensity Formula by the
Optimization Technique, 2000' Conference of Korean Society of Civil Engineers. Korean
Society of Civil Engineers, pp. 99-102 (in Korean).
Srinivas, N.D., K., 1994. Multiobjective optimization using nondominated sorting in genetic
algorithms. Evolutionary Computation, 2(3): 221-248.
Stedinger, J.R., Vogel, R.M. Foufoula-Georgiou,A., 1993. Frequency analysis of extreme
events. In: D.R. Maidment (Editor), Handbook of Hydrology, Chapter 18. McGraw-Hill,
New York.
Subramanya, K., 1984. Engineering Hydrology. Tata/McGraw-Hill, New Delhi.
Van Veldhuizen, D.A.L., G.B., 2000. Multiobjective evolutionary algorithms: analyzing the
state-of-the-art. Evolutionary Computation, 8(2): 125-147.
Wanielista, M., 1990. Hydrology and Water Quality Control. John Wiley, New York.
Yoo, C., Kim, N.-W. and Jung, K., 2001. A Point Rainfall Model and Rainfall IntensityDuration-Frequency Analysis. Journal of Korea Water Resources Association, 34(6): 577586 (in Korean).
Yoo, C., Park, C., Kim, K. and Jun, K.S., 2007. Derivation of Minutely Rainfall IntensityDuration-Frequency Relationships by Applying the Moupfouma Distribution. Journal of
Korea Water Resources Association, 40(8): 643-654 (in Korean).
48
Yoo, D.H., 1995. Generalization of Probability IDF curve, 1995' Conference of Korea Water
Yoo, D.H. and Rho, J.S., 2004. Computation of Probable Rainfall Intensity and Flood
Discharge for Culvert Design, 2004' Conferece of Korea Water Resources Association.
Korea Water Resources Association, pp. (in Korean).
Yoon, Y.-N., 1998. Engineering Hydrology. Cheong Moon Gak Publishers, Seoul, Korea (in
Korean).
Yu, P.-S., Yang, T.-C. and Lin, C.-S., 2004. Regional rainfall intensity formulas based on
scaling property of rainfall. Journal of Hydrology, 295: 108-123.
49
APPENDIX A
50
Table A. 1 Average RMSE and RRMSE of each site using 12 different IDF curve in case of Gumbel
No.
SITE
NAME
Number of
the Smallest
RMSE or
RRMSE
TALBOT
RMSE
RRMSE
SHERMAN
RMSE
RRMSE
JAPANESE
RMSE
106
SEMI_LOG
RRMSE
RMSE
LEE (OF_I)
RRMSE
RMSE
RRMSE
LEE (OF_II)
RMSE
RRMSE
HEO (OF_I)
RMSE
RRMSE
HEO (OF_II)
RMSE
RRMSE
ONESTEP
RMSE
ROBUST
RRMSE
RMSE
IDF_CDF (OF_I)
IDF_CDF (OF_II)
RRMSE
RMSE
RMSE
RRMSE
1
RRMSE
24
44
62
17
0.561
0.023
90
5.524
0.107
0.375
0.015
3.557
0.061
2.046
0.147
3.884
0.078
1.454
0.044
2.512
0.078
1.044
0.041
13.017
0.326
31.923
1.000
1.505
0.039
95
7.807
0.140
1.041
0.022
2.831
0.073
3.002
0.184
7.170
0.135
2.544
0.072
1.051
0.026
0.911
0.024
37.465
0.940
35.602
1.000
2.011
0.059
1.657
0.046
98
8.994
0.108
2.649
0.056
4.693
0.107
4.277
0.196
4.393
0.069
2.578
0.057
7.501
0.149
2.771
0.078
87.955
1.679
52.368
1.000
2.931
0.059
2.550
0.057
100
1.656
0.037
1.330
0.029
1.046
0.042
0.777
0.023
1.700
0.041
1.581
0.040
7.942
0.206
1.986
0.069
20.802
0.455
36.652
1.000
6.340
0.115
2.259
0.064
101
6.846
0.140
0.504
0.016
3.120
0.080
2.890
0.216
6.137
0.127
2.425
0.073
0.646
0.025
0.471
0.018
10.122
0.340
29.691
1.000
1.512
0.046
0.949
0.032
105
3.130
0.060
1.215
0.031
2.338
0.082
1.252
0.096
3.236
0.061
1.362
0.036
6.904
0.195
1.805
0.078
8.489
0.162
38.042
1.000
2.878
0.054
1.177
0.038
106
6.283
0.114
1.349
0.050
4.536
0.118
3.533
0.227
4.049
0.079
2.648
0.072
6.271
0.176
3.141
0.090
47.396
1.305
38.388
1.000
2.065
0.058
1.538
0.053
108
9.891
0.131
1.139
0.023
6.586
0.179
13.235
0.725
3.892
0.065
1.289
0.046
6.011
0.107
3.192
0.064
7.109
0.165
44.400
1.000
1.547
0.032
1.007
0.028
112
5.662
0.090
0.731
0.017
4.444
0.143
5.035
0.467
2.804
0.056
1.377
0.041
6.185
0.134
2.581
0.069
9.998
0.168
38.788
1.000
1.744
0.041
1.227
0.030
10
114
8.972
0.149
1.334
0.042
3.694
0.092
5.337
0.360
7.649
0.129
4.498
0.093
2.387
0.049
1.804
0.042
25.955
0.692
32.033
1.000
4.100
0.078
1.828
0.053
11
115
4.138
0.079
0.690
0.020
4.851
0.313
7.714
0.806
3.094
0.074
2.487
0.072
4.925
0.154
2.311
0.092
3.469
0.109
28.522
1.000
1.262
0.026
0.874
0.022
12
119
6.225
0.098
0.431
0.014
4.421
0.133
4.105
0.368
2.564
0.048
1.151
0.035
5.847
0.128
2.299
0.064
5.601
0.178
39.382
1.000
1.599
0.033
0.854
0.022
13
121
9.487
0.191
1.679
0.052
3.055
0.087
4.358
0.307
9.561
0.188
3.889
0.121
2.668
0.068
1.694
0.057
27.861
0.838
29.879
1.000
1.895
0.064
1.612
0.056
14
127
8.729
0.152
0.871
0.026
10.862
0.203
10.546
0.674
6.438
0.114
2.670
0.078
2.901
0.072
1.833
0.048
12.142
0.459
33.107
1.000
0.959
0.029
0.782
0.027
15
129
6.682
0.105
1.716
0.045
5.497
0.148
4.522
0.373
5.283
0.089
3.309
0.069
3.322
0.101
1.179
0.054
15.300
0.325
33.304
1.000
3.970
0.072
1.888
0.051
16
130
4.958
0.112
1.130
0.041
4.757
0.141
2.570
0.201
4.811
0.111
2.889
0.080
2.010
0.085
0.861
0.043
12.881
0.470
24.231
1.000
2.760
0.068
1.333
0.047
17
131
8.403
0.141
0.336
0.014
7.790
0.215
10.006
0.745
5.090
0.089
2.431
0.065
3.119
0.076
1.621
0.051
5.663
0.167
32.895
1.000
0.593
0.018
0.433
0.016
18
133
5.678
0.097
0.681
0.016
4.588
0.161
7.399
0.597
2.260
0.049
1.215
0.038
5.892
0.122
2.755
0.068
14.895
0.360
36.550
1.000
1.455
0.034
1.211
0.027
19
135
4.104
0.097
0.748
0.027
4.235
0.154
2.184
0.238
3.394
0.081
1.829
0.058
2.512
0.094
0.875
0.051
6.510
0.194
25.105
1.000
1.671
0.044
0.833
0.031
20
136
6.851
0.167
1.234
0.055
4.844
0.179
7.594
0.673
6.878
0.169
4.386
0.130
2.335
0.068
1.139
0.059
10.663
0.571
20.485
1.000
4.151
0.114
1.200
0.071
21
138
4.945
0.089
1.177
0.036
4.870
0.148
3.117
0.260
4.166
0.074
2.668
0.060
4.554
0.152
1.436
0.074
1.786
0.077
32.162
1.000
2.227
0.045
1.225
0.036
22
140
5.370
0.108
0.347
0.015
3.895
0.148
5.173
0.533
2.667
0.066
0.892
0.046
4.909
0.116
2.282
0.067
10.304
0.216
31.639
1.000
1.265
0.029
0.854
0.025
23
143
5.853
0.122
0.651
0.023
7.069
0.217
8.682
0.699
5.055
0.104
2.697
0.079
2.492
0.089
1.046
0.050
4.213
0.166
27.610
1.000
1.838
0.042
0.663
0.026
24
146
6.154
0.094
1.263
0.026
6.333
0.202
10.113
0.801
3.581
0.056
2.610
0.049
5.184
0.133
2.020
0.077
6.818
0.173
35.143
1.000
2.438
0.037
1.399
0.027
25
152
3.448
0.061
1.407
0.031
3.006
0.105
2.087
0.179
3.404
0.063
1.619
0.038
5.975
0.176
1.737
0.080
3.297
0.080
33.910
1.000
1.718
0.039
1.212
0.036
26
155
7.139
0.110
0.706
0.020
5.311
0.168
11.009
0.732
1.509
0.039
1.069
0.033
7.490
0.155
3.674
0.088
18.445
0.492
39.206
1.000
0.905
0.024
0.833
0.023
27
156
6.651
0.115
0.462
0.012
6.447
0.189
10.237
0.693
3.724
0.068
1.753
0.049
4.236
0.104
1.748
0.059
10.117
0.216
33.949
1.000
0.999
0.023
0.735
0.018
28
159
6.425
0.086
1.259
0.026
5.331
0.154
7.644
0.553
3.100
0.046
1.929
0.037
5.904
0.140
1.954
0.070
5.869
0.207
39.671
1.000
2.761
0.042
1.491
0.029
51
Table A. 1 Average RMSE and RRMSE of each site using 12 different IDF curve in case of Gumbel (Continued)
IDF_CDF (OF_I)
IDF_CDF (OF_II)
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
162
8.193
0.107
1.671
0.034
6.260
0.164
11.951
0.638
4.790
0.064
3.477
0.054
5.193
0.140
1.864
0.072
7.389
0.268
38.584
1.000
3.311
0.044
2.343
0.038
165
6.366
0.121
0.453
0.018
3.378
0.085
6.911
0.594
4.944
0.099
2.237
0.066
2.621
0.080
1.055
0.042
12.896
0.328
28.944
1.000
1.647
0.041
0.727
0.027
168
6.103
0.101
0.812
0.019
4.570
0.154
3.911
0.383
2.578
0.047
1.342
0.035
4.913
0.120
2.029
0.065
2.201
0.091
34.741
1.000
0.936
0.022
0.815
0.020
169
10.225
0.175
1.802
0.050
8.833
0.114
9.928
0.759
6.380
0.123
2.768
0.089
4.095
0.090
2.824
0.071
44.460
1.280
33.858
1.000
2.583
0.059
1.681
0.051
170
3.554
0.057
0.756
0.017
2.562
0.103
2.034
0.170
2.197
0.046
1.472
0.033
7.784
0.172
2.526
0.076
25.438
0.496
41.591
1.000
2.179
0.036
1.414
0.032
184
4.470
0.075
1.435
0.037
3.708
0.113
2.697
0.211
3.906
0.071
1.994
0.048
5.267
0.151
1.551
0.070
22.869
0.593
35.318
1.000
1.819
0.045
1.356
0.042
185
3.876
0.082
1.105
0.036
5.485
0.184
3.576
0.397
3.080
0.070
1.746
0.054
4.149
0.126
1.362
0.067
36.576
1.124
30.299
1.000
1.751
0.050
1.248
0.044
187
5.349
0.083
0.912
0.020
4.359
0.146
3.898
0.364
2.591
0.047
1.178
0.030
5.259
0.124
1.811
0.063
14.622
0.305
36.921
1.000
1.489
0.031
1.605
0.030
189
5.652
0.095
0.693
0.015
4.484
0.153
4.839
0.456
1.996
0.037
1.161
0.030
5.262
0.125
2.015
0.067
16.537
0.404
34.908
1.000
2.095
0.040
1.201
0.029
192
6.341
0.083
0.830
0.016
4.122
0.132
6.167
0.461
2.179
0.029
1.690
0.027
11.646
0.193
4.966
0.096
24.244
0.348
51.642
1.000
1.025
0.020
0.881
0.019
201
9.545
0.146
0.829
0.029
8.187
0.152
5.569
0.418
7.058
0.111
2.634
0.073
2.375
0.063
1.575
0.041
12.051
0.267
39.512
1.000
2.518
0.027
1.357
0.021
202
9.398
0.141
1.937
0.057
5.525
0.137
5.927
0.362
8.054
0.125
5.374
0.099
3.714
0.076
2.451
0.056
12.068
0.357
33.935
1.000
1.507
0.043
0.994
0.035
203
7.734
0.147
0.912
0.025
2.458
0.079
3.474
0.223
7.516
0.145
3.444
0.085
1.100
0.034
0.916
0.029
9.780
0.346
30.108
1.000
6.014
0.100
2.296
0.067
211
10.211
0.165
0.788
0.027
2.818
0.077
5.571
0.368
9.493
0.153
4.642
0.100
2.272
0.041
1.188
0.033
11.267
0.345
34.360
1.000
1.464
0.045
1.132
0.035
212
8.866
0.158
0.680
0.023
3.475
0.085
3.859
0.259
7.625
0.132
3.223
0.079
1.216
0.041
0.986
0.037
23.839
0.740
32.896
1.000
2.452
0.055
0.971
0.038
216
7.494
0.131
0.487
0.018
2.855
0.044
3.534
0.310
5.244
0.098
1.639
0.058
3.036
0.070
1.536
0.041
9.923
0.290
34.981
1.000
0.719
0.030
0.765
0.026
221
6.435
0.099
0.987
0.019
6.009
0.167
11.466
0.701
3.576
0.058
2.261
0.048
6.090
0.149
2.254
0.076
6.971
0.282
38.484
1.000
1.125
0.037
0.842
0.029
226
3.852
0.061
1.231
0.024
4.379
0.161
10.376
0.710
2.049
0.034
1.229
0.026
6.780
0.167
2.408
0.088
15.404
0.379
36.031
1.000
1.760
0.029
1.088
0.022
232
7.959
0.114
0.878
0.017
6.124
0.181
11.681
0.764
3.759
0.066
1.811
0.050
5.825
0.118
2.639
0.068
24.224
0.565
40.423
1.000
1.536
0.027
1.290
0.025
235
8.043
0.099
1.835
0.035
7.451
0.197
12.079
0.788
4.023
0.053
2.990
0.047
5.780
0.133
2.077
0.074
11.837
0.339
40.231
1.000
1.511
0.033
1.267
0.027
236
7.349
0.131
0.464
0.021
5.685
0.167
9.568
0.739
4.517
0.080
2.245
0.060
2.968
0.078
1.638
0.052
14.417
0.601
31.048
1.000
3.633
0.047
2.116
0.036
238
7.217
0.126
0.642
0.024
5.209
0.187
9.581
0.749
2.482
0.058
1.074
0.046
5.813
0.131
2.955
0.081
13.726
0.315
34.405
1.000
1.109
0.027
0.532
0.022
243
4.014
0.083
1.169
0.025
6.335
0.214
5.663
0.568
2.359
0.051
1.686
0.044
4.618
0.128
1.644
0.072
18.350
0.445
29.987
1.000
1.245
0.029
0.607
0.024
244
11.447
0.173
1.036
0.023
2.135
0.031
10.165
0.829
7.817
0.123
4.218
0.096
2.538
0.060
1.867
0.046
14.633
0.467
33.850
1.000
1.733
0.039
1.310
0.033
245
8.665
0.158
0.345
0.019
7.980
0.189
9.659
0.735
5.997
0.111
2.838
0.079
2.706
0.071
1.288
0.046
17.243
0.474
30.933
1.000
0.797
0.027
0.782
0.026
247
8.375
0.178
0.392
0.016
3.402
0.088
9.112
0.649
7.544
0.156
3.197
0.098
0.630
0.023
0.538
0.022
22.335
0.724
26.575
1.000
0.487
0.021
0.413
0.020
248
8.028
0.122
0.327
0.011
5.446
0.148
11.540
0.692
3.162
0.053
1.599
0.041
5.712
0.125
2.681
0.069
12.716
0.308
38.685
1.000
0.573
0.028
0.486
0.025
256
3.952
0.069
0.499
0.013
2.682
0.098
2.023
0.181
1.969
0.043
1.275
0.030
6.670
0.151
2.262
0.069
28.625
0.614
38.363
1.000
0.414
0.013
0.506
0.016
260
4.838
0.081
1.589
0.037
3.589
0.108
2.541
0.191
3.803
0.073
2.087
0.051
4.613
0.128
1.297
0.059
9.732
0.241
35.279
1.000
1.517
0.030
1.165
0.026
261
7.672
0.095
0.707
0.018
5.172
0.137
5.646
0.424
2.618
0.041
1.396
0.033
7.988
0.156
3.212
0.075
19.229
0.421
47.585
1.000
2.156
0.051
1.751
0.047
262
5.678
0.097
0.681
0.016
4.588
0.161
7.399
0.597
2.260
0.049
1.215
0.038
5.892
0.122
2.755
0.068
14.895
0.360
36.550
1.000
1.356
0.025
0.848
0.020
271
4.496
0.080
1.514
0.031
4.729
0.138
3.455
0.330
3.198
0.061
2.088
0.048
5.440
0.144
1.597
0.069
40.194
0.861
35.572
1.000
2.052
0.044
1.704
0.042
SITE
NAME
TALBOT
SHERMAN
JAPANESE
SEMI_LOG
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
52
HEO (OF_II)
ONESTEP
ROBUST
Table A. 1 Average RMSE and RRMSE of each site using 12 different IDF curve in case of Gumbel (Continued)
IDF_CDF (OF_I)
IDF_CDF (OF_II)
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
272
6.851
0.126
0.548
0.020
4.913
0.137
3.979
0.381
5.445
0.101
2.557
0.069
2.603
0.081
0.979
0.041
14.131
0.499
30.901
1.000
1.723
0.039
0.754
0.025
273
8.118
0.173
0.537
0.018
6.918
0.175
8.415
0.703
7.090
0.149
3.204
0.099
1.241
0.040
0.930
0.032
7.771
0.396
25.528
1.000
0.411
0.026
0.377
0.023
277
4.109
0.086
1.558
0.053
4.135
0.140
2.899
0.258
3.979
0.083
2.818
0.070
3.540
0.140
1.101
0.073
10.367
0.396
26.797
1.000
2.228
0.058
1.636
0.054
278
6.037
0.132
0.314
0.015
7.257
0.228
7.851
0.732
4.914
0.107
2.513
0.079
2.408
0.079
1.281
0.047
18.615
0.659
25.374
1.000
1.462
0.034
0.349
0.016
279
7.022
0.149
0.967
0.031
7.571
0.182
8.290
0.681
6.383
0.133
3.760
0.097
1.079
0.046
0.739
0.033
9.017
0.435
24.247
1.000
2.356
0.051
1.085
0.034
281
6.011
0.126
0.674
0.025
6.448
0.218
7.864
0.727
4.811
0.099
2.829
0.077
2.285
0.083
1.040
0.051
8.109
0.423
25.000
1.000
1.861
0.039
0.879
0.028
284
6.674
0.111
1.281
0.040
6.591
0.183
10.034
0.718
4.879
0.077
3.578
0.068
3.543
0.117
1.253
0.067
14.273
0.578
31.265
1.000
3.279
0.054
1.555
0.040
285
6.408
0.110
0.832
0.017
6.230
0.182
10.350
0.742
3.101
0.060
1.700
0.050
6.059
0.139
3.035
0.080
28.693
0.739
35.857
1.000
1.036
0.024
0.875
0.019
288
7.638
0.114
0.649
0.022
6.116
0.163
10.911
0.715
5.086
0.076
3.047
0.060
4.076
0.115
1.694
0.062
29.649
0.776
34.993
1.000
1.922
0.030
0.921
0.024
289
5.370
0.078
0.510
0.013
3.732
0.120
3.067
0.229
2.272
0.037
0.804
0.020
7.579
0.165
2.773
0.078
24.439
0.517
42.369
1.000
1.096
0.020
0.506
0.014
294
6.657
0.085
1.243
0.027
4.507
0.113
3.575
0.223
3.183
0.047
1.498
0.031
6.127
0.137
1.925
0.065
24.293
0.561
42.867
1.000
2.505
0.040
1.614
0.031
295
3.672
0.059
0.838
0.016
2.565
0.101
2.266
0.180
1.339
0.025
1.344
0.025
9.322
0.183
3.327
0.086
23.413
0.496
43.803
1.000
2.212
0.030
1.434
0.025
SITE
NAME
TALBOT
SHERMAN
JAPANESE
SEMI_LOG
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
53
HEO (OF_II)
ONESTEP
ROBUST
Table A. 2 Average RMSE and RRMSE of each site using 12 different IDF curve in case of GEV
No.
SITE
NAME
Number of
the Smallest
RMSE or
RRMSE
TALBOT
RMSE
SHERMAN
RRMSE
RMSE
53
66
6.550
0.113
1.605
RRMSE
JAPANESE
RMSE
SEMI_LOG
RRMSE
RMSE
0.036
2.295
119
LEE (OF_I)
RRMSE
RMSE
0.051
2.524
0.150
LEE (OF_II)
RRMSE
RMSE
6.812
HEO (OF_I)
RRMSE
RMSE
0.121
3.266
HEO (OF_II)
RRMSE
RMSE
12
0.074
5.578
0.105
2.633
RRMSE
ONESTEP
RMSE
ROBUST
IDF_CDF (OF_I)
IDF_CDF (OF_II)
RRMSE
RMSE
RRMSE
RMSE
0.249
37.112
1.000
1.840
0.052
1.688
0.040
RRMSE
RMSE
0.064
8.835
15
RRMSE
7
90
95
7.597
0.130
2.885
0.051
5.294
0.125
3.970
0.242
8.624
0.200
5.364
0.138
6.490
0.173
4.013
0.092
34.956
0.825
39.070
1.000
7.954
0.189
5.154
0.122
98
17.075
0.165
3.888
0.068
8.399
0.111
7.682
0.216
21.760
0.264
6.401
0.111
8.705
0.185
4.054
0.087
82.619
1.530
58.115
1.000
8.287
0.176
4.292
0.098
100
2.090
0.046
1.723
0.032
2.222
0.050
1.823
0.041
6.239
0.116
4.173
0.078
18.193
0.355
6.492
0.128
16.518
0.381
44.173
1.000
14.834
0.224
5.346
0.100
101
4.697
0.115
0.674
0.017
3.358
0.088
2.001
0.183
4.356
0.111
2.296
0.071
2.749
0.056
1.399
0.034
11.025
0.323
29.058
1.000
2.760
0.071
2.297
0.059
105
4.705
0.066
4.062
0.043
5.664
0.080
5.757
0.108
9.182
0.189
11.815
0.142
24.203
0.358
12.159
0.157
15.042
0.236
54.382
1.000
11.244
0.138
6.195
0.073
106
11.846
0.147
4.025
0.066
8.218
0.170
10.623
0.656
8.863
0.213
9.423
0.139
12.603
0.304
8.931
0.158
41.870
1.383
45.458
1.000
9.164
0.216
8.395
0.127
108
9.861
0.131
2.357
0.046
6.003
0.172
8.032
0.543
9.768
0.154
3.864
0.089
3.111
0.068
2.561
0.057
8.367
0.171
44.043
1.000
5.769
0.110
3.554
0.078
112
5.857
0.088
1.291
0.022
4.008
0.136
4.771
0.455
6.877
0.126
2.586
0.076
6.882
0.122
2.680
0.070
8.415
0.186
42.152
1.000
3.641
0.077
2.854
0.057
10
114
9.410
0.148
2.236
0.054
4.523
0.111
6.098
0.383
10.435
0.186
5.762
0.120
3.103
0.071
2.838
0.063
23.990
0.607
34.249
1.000
7.016
0.134
2.879
0.083
11
115
5.429
0.086
1.442
0.029
5.471
0.396
8.947
0.787
5.322
0.116
4.253
0.108
4.831
0.142
2.058
0.084
2.416
0.096
30.830
1.000
2.165
0.043
1.609
0.034
12
119
4.521
0.083
1.602
0.026
3.601
0.085
4.453
0.384
5.479
0.108
3.769
0.087
12.424
0.192
5.573
0.104
5.614
0.167
46.792
1.000
3.733
0.052
3.066
0.047
13
121
17.795
0.277
5.406
0.128
10.873
0.168
12.632
0.602
20.036
0.278
11.511
0.192
7.410
0.142
7.031
0.122
20.661
0.721
39.123
1.000
9.516
0.152
8.023
0.132
14
127
9.258
0.148
2.780
0.045
7.768
0.188
11.505
0.705
7.447
0.116
4.909
0.099
7.979
0.129
5.139
0.081
8.196
0.373
38.600
1.000
3.466
0.053
3.034
0.047
15
129
11.492
0.159
2.623
0.063
4.967
0.145
11.913
0.541
10.658
0.165
6.325
0.119
5.650
0.093
2.813
0.075
14.482
0.354
34.506
1.000
4.828
0.099
2.666
0.082
16
130
5.838
0.122
1.861
0.061
4.133
0.148
3.463
0.203
5.961
0.141
4.012
0.104
2.825
0.075
2.464
0.063
13.902
0.513
23.603
1.000
4.031
0.103
2.344
0.080
17
131
5.628
0.126
1.758
0.052
8.151
0.236
5.642
0.540
4.781
0.123
3.530
0.097
7.749
0.138
3.872
0.093
7.620
0.180
32.994
1.000
5.070
0.100
4.029
0.087
18
133
4.374
0.087
1.819
0.030
3.991
0.150
4.932
0.460
3.969
0.086
3.459
0.070
7.838
0.137
3.797
0.079
15.610
0.335
36.568
1.000
4.476
0.076
3.802
0.069
19
135
2.622
0.079
0.789
0.029
3.834
0.141
1.873
0.258
2.647
0.082
1.655
0.062
5.180
0.154
1.712
0.080
6.123
0.166
25.923
1.000
2.497
0.062
1.672
0.054
20
136
10.390
0.192
3.577
0.088
6.049
0.199
8.966
0.692
10.170
0.189
7.868
0.157
3.863
0.091
3.163
0.084
10.177
0.579
22.949
1.000
4.024
0.100
4.173
0.100
21
138
4.922
0.084
2.526
0.045
4.382
0.125
3.454
0.236
8.262
0.169
7.892
0.146
10.231
0.209
4.304
0.114
12.212
0.251
41.614
1.000
3.159
0.058
2.602
0.053
22
140
4.839
0.110
1.821
0.048
3.323
0.133
4.219
0.456
4.965
0.137
2.953
0.096
8.797
0.152
4.247
0.096
9.605
0.203
34.431
1.000
5.077
0.100
3.742
0.080
23
143
4.432
0.106
1.330
0.036
5.996
0.193
6.811
0.633
3.432
0.081
2.442
0.073
3.396
0.122
1.203
0.074
5.043
0.183
26.803
1.000
2.145
0.058
1.222
0.046
24
146
5.358
0.088
2.827
0.046
6.464
0.181
10.342
0.817
3.891
0.061
3.540
0.058
6.403
0.166
2.461
0.098
6.590
0.172
35.829
1.000
3.764
0.057
2.624
0.049
25
152
3.357
0.066
2.491
0.047
2.623
0.082
1.841
0.157
4.212
0.104
3.193
0.085
9.142
0.203
2.965
0.101
4.676
0.148
38.433
1.000
5.494
0.089
3.006
0.071
26
155
5.420
0.093
1.911
0.045
5.180
0.164
10.441
0.791
5.830
0.126
5.593
0.093
15.313
0.299
8.554
0.163
12.269
0.450
47.491
1.000
5.660
0.128
5.011
0.092
27
156
4.913
0.097
1.107
0.022
5.869
0.175
4.950
0.482
4.583
0.097
2.300
0.066
5.257
0.113
1.792
0.064
10.024
0.188
34.565
1.000
2.986
0.064
2.337
0.051
28
159
6.575
0.091
3.609
0.065
5.978
0.153
6.153
0.452
7.858
0.124
4.308
0.081
5.390
0.132
2.562
0.075
7.028
0.199
39.254
1.000
4.918
0.090
3.680
0.080
54
Table A. 2 Average RMSE and RRMSE of each site using 12 different IDF curve in case of GEV (Continued)
IDF_CDF (OF_I)
IDF_CDF (OF_II)
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
162
10.966
0.122
1.983
0.043
5.547
0.151
8.098
0.484
12.366
0.168
6.197
0.100
3.241
0.091
1.932
0.065
6.050
0.219
42.842
1.000
6.169
0.099
3.099
0.065
165
5.460
0.119
0.466
0.018
3.221
0.111
3.741
0.447
7.621
0.200
2.919
0.111
2.369
0.083
1.006
0.046
14.632
0.348
27.981
1.000
5.168
0.153
2.862
0.096
168
5.976
0.106
1.562
0.031
4.419
0.167
4.212
0.403
3.825
0.078
2.005
0.052
4.089
0.091
2.310
0.056
2.297
0.087
34.011
1.000
2.287
0.051
1.995
0.044
169
11.655
0.214
4.113
0.107
86.355
2.522
7.947
0.635
14.682
0.303
4.344
0.151
6.664
0.199
5.200
0.119
41.544
1.092
37.197
1.000
7.730
0.207
4.700
0.122
170
2.518
0.050
0.799
0.018
2.346
0.105
1.955
0.158
3.645
0.090
2.836
0.069
13.142
0.212
4.882
0.102
20.188
0.395
47.945
1.000
6.387
0.072
3.711
0.057
184
7.876
0.121
1.769
0.051
4.598
0.137
4.352
0.257
6.400
0.114
3.457
0.085
4.148
0.108
2.254
0.069
23.429
0.645
34.968
1.000
3.910
0.083
1.957
0.067
185
5.673
0.112
0.958
0.038
4.581
0.170
3.664
0.387
3.875
0.088
1.674
0.065
3.281
0.102
1.877
0.065
36.702
1.169
30.200
1.000
1.442
0.050
0.968
0.044
187
3.948
0.070
1.421
0.029
3.341
0.118
3.972
0.384
5.279
0.110
2.853
0.076
9.700
0.174
3.778
0.096
9.811
0.224
43.044
1.000
4.959
0.085
3.928
0.074
189
3.885
0.080
1.889
0.033
4.559
0.149
4.541
0.438
2.505
0.059
1.892
0.050
7.914
0.192
2.821
0.106
16.793
0.402
34.816
1.000
3.412
0.059
2.381
0.049
192
4.974
0.073
1.265
0.017
4.023
0.111
6.047
0.447
7.219
0.135
7.510
0.111
21.119
0.249
10.365
0.137
15.789
0.286
66.562
1.000
2.496
0.058
1.890
0.050
201
13.392
0.182
2.934
0.057
5.277
0.077
6.793
0.442
13.371
0.189
4.539
0.110
3.262
0.064
3.289
0.061
9.355
0.247
42.745
1.000
5.078
0.043
4.186
0.040
202
7.160
0.124
2.750
0.067
5.830
0.136
5.017
0.365
7.653
0.150
4.713
0.109
3.579
0.090
2.582
0.068
10.882
0.297
34.927
1.000
3.568
0.090
2.592
0.065
203
15.804
0.215
2.045
0.063
6.905
0.120
11.851
0.483
17.184
0.228
9.735
0.141
3.976
0.084
3.079
0.072
8.279
0.315
37.800
1.000
5.166
0.110
3.684
0.092
211
15.820
0.197
4.122
0.058
7.716
0.112
14.132
0.508
16.316
0.209
9.382
0.132
5.361
0.074
2.640
0.055
10.462
0.329
38.612
1.000
7.229
0.100
5.532
0.086
212
11.912
0.191
1.514
0.045
5.368
0.116
6.944
0.321
12.764
0.213
5.165
0.113
3.844
0.085
1.429
0.057
23.003
0.740
33.994
1.000
6.083
0.082
4.525
0.063
216
6.888
0.125
0.975
0.019
3.548
0.076
3.403
0.304
4.320
0.086
1.641
0.054
3.739
0.083
1.971
0.047
10.220
0.292
34.710
1.000
3.435
0.086
1.603
0.055
221
5.491
0.088
2.760
0.028
7.843
0.164
14.223
0.731
10.217
0.225
14.749
0.174
19.604
0.297
12.563
0.157
16.586
0.300
55.981
1.000
1.565
0.035
1.411
0.031
226
5.862
0.083
3.253
0.060
5.714
0.169
11.090
0.698
4.444
0.071
3.380
0.063
6.277
0.162
2.685
0.093
16.220
0.405
35.667
1.000
4.647
0.071
4.209
0.053
232
9.370
0.132
3.031
0.046
6.282
0.313
8.707
0.632
10.138
0.172
3.523
0.103
7.643
0.109
4.797
0.079
21.005
0.464
43.965
1.000
3.692
0.064
3.492
0.062
235
7.504
0.105
4.797
0.071
6.633
0.167
11.547
0.690
9.283
0.176
8.268
0.142
11.813
0.218
5.843
0.134
8.357
0.267
50.498
1.000
5.401
0.099
4.238
0.076
236
9.891
0.154
0.606
0.021
7.741
0.210
9.397
0.794
5.382
0.091
4.180
0.080
3.719
0.101
2.212
0.070
14.353
0.700
32.090
1.000
6.375
0.096
5.052
0.086
238
7.282
0.119
1.459
0.043
5.738
0.179
10.117
0.795
3.598
0.080
3.818
0.071
9.480
0.202
4.741
0.118
9.450
0.256
39.423
1.000
1.782
0.062
2.115
0.059
243
3.461
0.093
2.077
0.051
6.611
0.200
3.963
0.445
4.523
0.108
3.697
0.097
8.533
0.215
3.412
0.124
19.972
0.462
29.509
1.000
2.873
0.069
2.719
0.059
244
16.144
0.220
5.123
0.084
14.046
0.162
10.552
0.832
10.484
0.166
6.823
0.135
7.185
0.102
7.177
0.096
10.771
0.391
39.020
1.000
6.693
0.140
3.875
0.103
245
8.290
0.174
1.635
0.063
9.367
0.222
6.868
0.591
8.082
0.180
3.465
0.109
2.243
0.077
1.793
0.067
20.167
0.534
28.502
1.000
6.216
0.089
5.624
0.084
247
7.402
0.166
0.562
0.022
4.190
0.099
8.503
0.644
5.736
0.126
2.861
0.089
1.175
0.045
0.913
0.038
24.165
0.819
24.945
1.000
3.337
0.106
2.597
0.084
248
11.899
0.135
1.344
0.025
8.693
0.178
11.362
0.732
6.959
0.125
7.548
0.091
10.608
0.219
6.576
0.120
8.763
0.288
46.445
1.000
2.004
0.072
1.740
0.067
256
4.981
0.075
3.033
0.046
3.481
0.098
2.760
0.169
6.633
0.136
3.608
0.083
7.224
0.142
2.667
0.077
25.678
0.523
41.812
1.000
6.462
0.121
6.172
0.079
260
3.532
0.072
1.735
0.040
2.569
0.082
2.862
0.295
4.769
0.127
3.505
0.097
8.622
0.159
3.117
0.085
8.208
0.218
38.718
1.000
5.241
0.093
3.696
0.078
261
7.588
0.091
3.060
0.035
5.452
0.135
6.770
0.441
7.946
0.107
4.002
0.070
10.966
0.157
5.707
0.086
14.636
0.338
52.996
1.000
6.356
0.112
4.147
0.089
262
5.429
0.086
1.442
0.029
5.471
0.396
8.947
0.787
5.322
0.116
4.253
0.108
4.831
0.142
2.058
0.084
2.416
0.096
30.830
1.000
4.188
0.059
3.770
0.047
271
3.592
0.075
1.634
0.033
3.975
0.113
3.671
0.359
4.696
0.117
3.563
0.096
10.618
0.186
4.157
0.099
36.218
0.729
40.719
1.000
5.955
0.088
4.099
0.075
SITE
NAME
TALBOT
SHERMAN
JAPANESE
SEMI_LOG
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
55
HEO (OF_II)
ONESTEP
ROBUST
Table A. 2 Average RMSE and RRMSE of each site using 12 different IDF curve in case of GEV (Continued)
IDF_CDF (OF_I)
IDF_CDF (OF_II)
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
RMSE
RRMSE
272
7.831
0.135
0.884
0.036
6.039
0.146
4.850
0.412
8.006
0.149
3.747
0.093
1.618
0.051
1.065
0.039
13.731
0.472
31.507
1.000
4.123
0.090
1.501
0.053
273
7.821
0.188
0.913
0.037
6.384
0.186
7.967
0.686
7.364
0.179
2.543
0.107
1.183
0.051
1.092
0.049
9.249
0.452
23.977
1.000
1.445
0.063
1.279
0.058
277
4.038
0.082
1.865
0.051
4.757
0.158
2.955
0.246
5.553
0.109
4.018
0.092
5.512
0.166
1.822
0.086
6.785
0.300
30.998
1.000
2.190
0.059
1.854
0.054
278
5.457
0.121
1.697
0.047
7.512
0.224
6.360
0.613
5.709
0.129
3.381
0.095
1.816
0.071
1.210
0.051
19.214
0.635
24.996
1.000
3.791
0.091
1.920
0.062
279
5.156
0.133
0.785
0.029
6.311
0.178
4.154
0.479
5.176
0.135
2.373
0.090
3.016
0.069
1.704
0.050
8.448
0.359
24.385
1.000
2.218
0.061
1.934
0.053
281
6.572
0.136
0.924
0.038
1.232
0.038
7.660
0.740
4.493
0.097
3.092
0.083
2.381
0.094
1.364
0.065
9.072
0.498
24.312
1.000
1.783
0.060
1.228
0.056
284
4.201
0.089
1.670
0.040
5.551
0.176
9.054
0.724
2.635
0.065
2.323
0.060
5.467
0.164
2.082
0.096
14.431
0.583
30.777
1.000
2.272
0.061
1.564
0.052
285
3.812
0.091
1.195
0.026
5.264
0.165
7.246
0.630
2.020
0.062
1.942
0.057
8.875
0.206
3.599
0.118
29.355
0.722
35.333
1.000
2.560
0.062
1.958
0.055
288
8.276
0.116
2.137
0.039
6.709
0.171
10.940
0.735
4.642
0.072
4.041
0.065
5.264
0.145
2.201
0.082
29.193
0.789
35.636
1.000
3.065
0.049
2.405
0.044
289
5.952
0.099
2.103
0.062
4.860
0.130
3.889
0.220
7.402
0.123
2.676
0.072
5.566
0.142
2.424
0.081
26.210
0.552
40.861
1.000
3.249
0.078
2.348
0.069
294
9.675
0.122
2.810
0.057
5.128
0.119
5.351
0.210
11.826
0.213
5.113
0.121
4.446
0.105
4.057
0.080
26.541
0.575
41.329
1.000
8.485
0.173
4.299
0.117
295
6.219
0.085
3.100
0.043
4.692
0.116
4.514
0.222
4.730
0.063
3.361
0.047
12.716
0.214
6.602
0.108
20.030
0.461
47.508
1.000
5.178
0.069
3.367
0.045
SITE
NAME
TALBOT
SHERMAN
JAPANESE
SEMI_LOG
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
56
HEO (OF_II)
ONESTEP
ROBUST
Table A. 3 Average RMSE and RRMSE excluded regression analysis IDF curves of each site in case of Gumbel
No.
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
HEO (OF_II)
ONESTEP
ROBUST
IDF_CDF (OF_I) IDF_CDF (OF_II)
SITE
NAME RMSE RRMSE RMSE RRMSE RMSE RRMSE RMSE RRMSE RMSE RRMSE RMSE RRMSE RMSE RRMSE RMSE RRMSE
Number of
the Smallest
RMSE or
RRMSE
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
90
95
98
100
101
105
106
108
112
114
115
119
121
127
129
130
131
133
135
136
138
140
143
146
152
155
156
159
22
109
13
51
58
3.884
7.170
4.393
1.700
6.137
3.236
4.049
3.892
2.804
7.649
3.094
2.564
9.561
6.438
5.283
4.811
5.090
2.260
3.394
6.878
4.166
2.667
5.055
3.581
3.404
1.509
3.724
3.100
0.078
0.135
0.069
0.041
0.127
0.061
0.079
0.065
0.056
0.129
0.074
0.048
0.188
0.114
0.089
0.111
0.089
0.049
0.081
0.169
0.074
0.066
0.104
0.056
0.063
0.039
0.068
0.046
1.454
2.544
2.578
1.581
2.425
1.362
2.648
1.289
1.377
4.498
2.487
1.151
3.889
2.670
3.309
2.889
2.431
1.215
1.829
4.386
2.668
0.892
2.697
2.610
1.619
1.069
1.753
1.929
0.044
0.072
0.057
0.040
0.073
0.036
0.072
0.046
0.041
0.093
0.072
0.035
0.121
0.078
0.069
0.080
0.065
0.038
0.058
0.130
0.060
0.046
0.079
0.049
0.038
0.033
0.049
0.037
2.512
1.051
7.501
7.942
0.646
6.904
6.271
6.011
6.185
2.387
4.925
5.847
2.668
2.901
3.322
2.010
3.119
5.892
2.512
2.335
4.554
4.909
2.492
5.184
5.975
7.490
4.236
5.904
0.078
0.026
0.149
0.206
0.025
0.195
0.176
0.107
0.134
0.049
0.154
0.128
0.068
0.072
0.101
0.085
0.076
0.122
0.094
0.068
0.152
0.116
0.089
0.133
0.176
0.155
0.104
0.140
1.044
0.911
2.771
1.986
0.471
1.805
3.141
3.192
2.581
1.804
2.311
2.299
1.694
1.833
1.179
0.861
1.621
2.755
0.875
1.139
1.436
2.282
1.046
2.020
1.737
3.674
1.748
1.954
0.041
0.024
0.078
0.069
0.018
0.078
0.090
0.064
0.069
0.042
0.092
0.064
0.057
0.048
0.054
0.043
0.051
0.068
0.051
0.059
0.074
0.067
0.050
0.077
0.080
0.088
0.059
0.070
13.017
37.465
87.955
20.802
10.122
8.489
47.396
7.109
9.998
25.955
3.469
5.601
27.861
12.142
15.300
12.881
5.663
14.895
6.510
10.663
1.786
10.304
4.213
6.818
3.297
18.445
10.117
5.869
0.326
0.940
1.679
0.455
0.340
0.162
1.305
0.165
0.168
0.692
0.109
0.178
0.838
0.459
0.325
0.470
0.167
0.360
0.194
0.571
0.077
0.216
0.166
0.173
0.080
0.492
0.216
0.207
31.923
35.602
52.368
36.652
29.691
38.042
38.388
44.400
38.788
32.033
28.522
39.382
29.879
33.107
33.304
24.231
32.895
36.550
25.105
20.485
32.162
31.639
27.610
35.143
33.910
39.206
33.949
39.671
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.505
2.011
2.931
6.340
1.512
2.878
2.065
1.547
1.744
4.100
1.262
1.599
1.895
0.959
3.970
2.760
0.593
1.455
1.671
4.151
2.227
1.265
1.838
2.438
1.718
0.905
0.999
2.761
0.039
0.059
0.059
0.115
0.046
0.054
0.058
0.032
0.041
0.078
0.026
0.033
0.064
0.029
0.072
0.068
0.018
0.034
0.044
0.114
0.045
0.029
0.042
0.037
0.039
0.024
0.023
0.042
0.561
1.657
2.550
2.259
0.949
1.177
1.538
1.007
1.227
1.828
0.874
0.854
1.612
0.782
1.888
1.333
0.433
1.211
0.833
1.200
1.225
0.854
0.663
1.399
1.212
0.833
0.735
1.491
0.023
0.046
0.057
0.064
0.032
0.038
0.053
0.028
0.030
0.053
0.022
0.022
0.056
0.027
0.051
0.047
0.016
0.027
0.031
0.071
0.036
0.025
0.026
0.027
0.036
0.023
0.018
0.029
57
Table A. 3 Average RMSE and RRMSE excluded regression analysis IDF curves of each site in case of Gumbel (Continued)
No.
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
HEO (OF_II)
ONESTEP
ROBUST
SITE
162
165
168
169
170
184
185
187
189
192
201
202
203
211
212
216
221
226
232
235
236
238
243
244
245
247
248
256
260
261
262
271
4.790
4.944
2.578
6.380
2.197
3.906
3.080
2.591
1.996
2.179
7.058
8.054
7.516
9.493
7.625
5.244
3.576
2.049
3.759
4.023
4.517
2.482
2.359
7.817
5.997
7.544
3.162
1.969
3.803
2.618
2.804
3.198
0.064
0.099
0.047
0.123
0.046
0.071
0.070
0.047
0.037
0.029
0.111
0.125
0.145
0.153
0.132
0.098
0.058
0.034
0.066
0.053
0.080
0.058
0.051
0.123
0.111
0.156
0.053
0.043
0.073
0.041
0.056
0.061
3.477
2.237
1.342
2.768
1.472
1.994
1.746
1.178
1.161
1.690
2.634
5.374
3.444
4.642
3.223
1.639
2.261
1.229
1.811
2.990
2.245
1.074
1.686
4.218
2.838
3.197
1.599
1.275
2.087
1.396
1.377
2.088
0.054
0.066
0.035
0.089
0.033
0.048
0.054
0.030
0.030
0.027
0.073
0.099
0.085
0.100
0.079
0.058
0.048
0.026
0.050
0.047
0.060
0.046
0.044
0.096
0.079
0.098
0.041
0.030
0.051
0.033
0.041
0.048
5.193
2.621
4.913
4.095
7.784
5.267
4.149
5.259
5.262
11.646
2.375
3.714
1.100
2.272
1.216
3.036
6.090
6.780
5.825
5.780
2.968
5.813
4.618
2.538
2.706
0.630
5.712
6.670
4.613
7.988
6.185
5.440
0.140
0.080
0.120
0.090
0.172
0.151
0.126
0.124
0.125
0.193
0.063
0.076
0.034
0.041
0.041
0.070
0.149
0.167
0.118
0.133
0.078
0.131
0.128
0.060
0.071
0.023
0.125
0.151
0.128
0.156
0.134
0.144
1.864
1.055
2.029
2.824
2.526
1.551
1.362
1.811
2.015
4.966
1.575
2.451
0.916
1.188
0.986
1.536
2.254
2.408
2.639
2.077
1.638
2.955
1.644
1.867
1.288
0.538
2.681
2.262
1.297
3.212
2.581
1.597
0.072
0.042
0.065
0.071
0.076
0.070
0.067
0.063
0.067
0.096
0.041
0.056
0.029
0.033
0.037
0.041
0.076
0.088
0.068
0.074
0.052
0.081
0.072
0.046
0.046
0.022
0.069
0.069
0.059
0.075
0.069
0.069
58
7.389
12.896
2.201
44.460
25.438
22.869
36.576
14.622
16.537
24.244
12.051
12.068
9.780
11.267
23.839
9.923
6.971
15.404
24.224
11.837
14.417
13.726
18.350
14.633
17.243
22.335
12.716
28.625
9.732
19.229
9.998
40.194
0.268
0.328
0.091
1.280
0.496
0.593
1.124
0.305
0.404
0.348
0.267
0.357
0.346
0.345
0.740
0.290
0.282
0.379
0.565
0.339
0.601
0.315
0.445
0.467
0.474
0.724
0.308
0.614
0.241
0.421
0.168
0.861
38.584
28.944
34.741
33.858
41.591
35.318
30.299
36.921
34.908
51.642
39.512
33.935
30.108
34.360
32.896
34.981
38.484
36.031
40.423
40.231
31.048
34.405
29.987
33.850
30.933
26.575
38.685
38.363
35.279
47.585
38.788
35.572
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
3.311
1.647
0.936
2.583
2.179
1.819
1.751
1.489
2.095
1.025
2.518
1.507
6.014
1.464
2.452
0.719
1.125
1.760
1.536
1.511
3.633
1.109
1.245
1.733
0.797
0.487
0.573
0.414
1.517
2.156
1.356
2.052
0.044
0.041
0.022
0.059
0.036
0.045
0.050
0.031
0.040
0.020
0.027
0.043
0.100
0.045
0.055
0.030
0.037
0.029
0.027
0.033
0.047
0.027
0.029
0.039
0.027
0.021
0.028
0.013
0.030
0.051
0.025
0.044
2.343
0.727
0.815
1.681
1.414
1.356
1.248
1.605
1.201
0.881
1.357
0.994
2.296
1.132
0.971
0.765
0.842
1.088
1.290
1.267
2.116
0.532
0.607
1.310
0.782
0.413
0.486
0.506
1.165
1.751
0.848
1.704
0.038
0.027
0.020
0.051
0.032
0.042
0.044
0.030
0.029
0.019
0.021
0.035
0.067
0.035
0.038
0.026
0.029
0.022
0.025
0.027
0.036
0.022
0.024
0.033
0.026
0.020
0.025
0.016
0.026
0.047
0.020
0.042
Table A. 3 Average RMSE and RRMSE excluded regression analysis IDF curves of each site in case of Gumbel (Continued)
No.
61
62
63
64
65
66
67
68
69
70
71
72
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
HEO (OF_II)
ONESTEP
ROBUST
SITE
272
273
277
278
279
281
284
285
288
289
294
295
5.445
7.090
3.979
4.914
6.383
4.811
4.879
3.101
5.086
2.272
3.183
1.339
0.101
0.149
0.083
0.107
0.133
0.099
0.077
0.060
0.076
0.037
0.047
0.025
2.557
3.204
2.818
2.513
3.760
2.829
3.578
1.700
3.047
0.804
1.498
1.344
0.069
0.099
0.070
0.079
0.097
0.077
0.068
0.050
0.060
0.020
0.031
0.025
2.603
1.241
3.540
2.408
1.079
2.285
3.543
6.059
4.076
7.579
6.127
9.322
0.081
0.040
0.140
0.079
0.046
0.083
0.117
0.139
0.115
0.165
0.137
0.183
0.979
0.930
1.101
1.281
0.739
1.040
1.253
3.035
1.694
2.773
1.925
3.327
0.041
0.032
0.073
0.047
0.033
0.051
0.067
0.080
0.062
0.078
0.065
0.086
59
14.131
7.771
10.367
18.615
9.017
8.109
14.273
28.693
29.649
24.439
24.293
23.413
0.499
0.396
0.396
0.659
0.435
0.423
0.578
0.739
0.776
0.517
0.561
0.496
30.901
25.528
26.797
25.374
24.247
25.000
31.265
35.857
34.993
42.369
42.867
43.803
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.723
0.411
2.228
1.462
2.356
1.861
3.279
1.036
1.922
1.096
2.505
2.212
0.039
0.026
0.058
0.034
0.051
0.039
0.054
0.024
0.030
0.020
0.040
0.030
0.754
0.377
1.636
0.349
1.085
0.879
1.555
0.875
0.921
0.506
1.614
1.434
0.025
0.023
0.054
0.016
0.034
0.028
0.040
0.019
0.024
0.014
0.031
0.025
Table A. 4 Average RMSE and RRMSE excluded regression analysis IDF curves of each site in case of GEV
No.
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
HEO (OF_II)
ONESTEP
ROBUST
SITE
Number of
the Smallest
RMSE or
RRMSE
1
90
2
95
3
98
4
100
5
101
6
105
7
106
8
108
9
112
10
114
11
115
12
119
13
121
14
127
15
129
16
130
17
131
18
133
19
135
20
136
21
138
22
140
23
143
24
146
25
152
26
155
27
156
28
159
16
57
69
14
31
26
25
44
6.812
8.624
21.760
6.239
4.356
9.182
8.863
9.768
6.877
10.435
5.322
5.479
20.036
7.447
10.658
5.961
4.781
3.969
2.647
10.170
8.262
4.965
3.432
3.891
4.212
5.830
4.583
7.858
0.121
0.200
0.264
0.116
0.111
0.189
0.213
0.154
0.126
0.186
0.116
0.108
0.278
0.116
0.165
0.141
0.123
0.086
0.082
0.189
0.169
0.137
0.081
0.061
0.104
0.126
0.097
0.124
3.266
5.364
6.401
4.173
2.296
11.815
9.423
3.864
2.586
5.762
4.253
3.769
11.511
4.909
6.325
4.012
3.530
3.459
1.655
7.868
7.892
2.953
2.442
3.540
3.193
5.593
2.300
4.308
0.074
0.138
0.111
0.078
0.071
0.142
0.139
0.089
0.076
0.120
0.108
0.087
0.192
0.099
0.119
0.104
0.097
0.070
0.062
0.157
0.146
0.096
0.073
0.058
0.085
0.093
0.066
0.081
5.578
6.490
8.705
18.193
2.749
24.203
12.603
3.111
6.882
3.103
4.831
12.424
7.410
7.979
5.650
2.825
7.749
7.838
5.180
3.863
10.231
8.797
3.396
6.403
9.142
15.313
5.257
5.390
0.105
0.173
0.185
0.355
0.056
0.358
0.304
0.068
0.122
0.071
0.142
0.192
0.142
0.129
0.093
0.075
0.138
0.137
0.154
0.091
0.209
0.152
0.122
0.166
0.203
0.299
0.113
0.132
2.633
4.013
4.054
6.492
1.399
12.159
8.931
2.561
2.680
2.838
2.058
5.573
7.031
5.139
2.813
2.464
3.872
3.797
1.712
3.163
4.304
4.247
1.203
2.461
2.965
8.554
1.792
2.562
0.064
0.092
0.087
0.128
0.034
0.157
0.158
0.057
0.070
0.063
0.084
0.104
0.122
0.081
0.075
0.063
0.093
0.079
0.080
0.084
0.114
0.096
0.074
0.098
0.101
0.163
0.064
0.075
8.835
34.956
82.619
16.518
11.025
15.042
41.870
8.367
8.415
23.990
2.416
5.614
20.661
8.196
14.482
13.902
7.620
15.610
6.123
10.177
12.212
9.605
5.043
6.590
4.676
12.269
10.024
7.028
0.249
0.825
1.530
0.381
0.323
0.236
1.383
0.171
0.186
0.607
0.096
0.167
0.721
0.373
0.354
0.513
0.180
0.335
0.166
0.579
0.251
0.203
0.183
0.172
0.148
0.450
0.188
0.199
37.112
39.070
58.115
44.173
29.058
54.382
45.458
44.043
42.152
34.249
30.830
46.792
39.123
38.600
34.506
23.603
32.994
36.568
25.923
22.949
41.614
34.431
26.803
35.829
38.433
47.491
34.565
39.254
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.840
7.954
8.287
14.834
2.760
11.244
9.164
5.769
3.641
7.016
2.165
3.733
9.516
3.466
4.828
4.031
5.070
4.476
2.497
4.024
3.159
5.077
2.145
3.764
5.494
5.660
2.986
4.918
0.052
0.189
0.176
0.224
0.071
0.138
0.216
0.110
0.077
0.134
0.043
0.052
0.152
0.053
0.099
0.103
0.100
0.076
0.062
0.100
0.058
0.100
0.058
0.057
0.089
0.128
0.064
0.090
1.688
5.154
4.292
5.346
2.297
6.195
8.395
3.554
2.854
2.879
1.609
3.066
8.023
3.034
2.666
2.344
4.029
3.802
1.672
4.173
2.602
3.742
1.222
2.624
3.006
5.011
2.337
3.680
0.040
0.122
0.098
0.100
0.059
0.073
0.127
0.078
0.057
0.083
0.034
0.047
0.132
0.047
0.082
0.080
0.087
0.069
0.054
0.100
0.053
0.080
0.046
0.049
0.071
0.092
0.051
0.080
60
Table A. 4 Average RMSE and RRMSE excluded regression analysis IDF curves of each site in case of GEV (Continued)
No.
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
HEO (OF_II)
ONESTEP
ROBUST
SITE
162
165
168
169
170
184
185
187
189
192
201
202
203
211
212
216
221
226
232
235
236
238
243
244
245
247
248
256
260
261
262
271
12.366
7.621
3.825
14.682
3.645
6.400
3.875
5.279
2.505
7.219
13.371
7.653
17.184
16.316
12.764
4.320
10.217
4.444
10.138
9.283
5.382
3.598
4.523
10.484
8.082
5.736
6.959
6.633
4.769
7.946
7.447
4.696
0.168
0.200
0.078
0.303
0.090
0.114
0.088
0.110
0.059
0.135
0.189
0.150
0.228
0.209
0.213
0.086
0.225
0.071
0.172
0.176
0.091
0.080
0.108
0.166
0.180
0.126
0.125
0.136
0.127
0.107
0.116
0.117
6.197
2.919
2.005
4.344
2.836
3.457
1.674
2.853
1.892
7.510
4.539
4.713
9.735
9.382
5.165
1.641
14.749
3.380
3.523
8.268
4.180
3.818
3.697
6.823
3.465
2.861
7.548
3.608
3.505
4.002
4.909
3.563
0.100
0.111
0.052
0.151
0.069
0.085
0.065
0.076
0.050
0.111
0.110
0.109
0.141
0.132
0.113
0.054
0.174
0.063
0.103
0.142
0.080
0.071
0.097
0.135
0.109
0.089
0.091
0.083
0.097
0.070
0.099
0.096
3.241
2.369
4.089
6.664
13.142
4.148
3.281
9.700
7.914
21.119
3.262
3.579
3.976
5.361
3.844
3.739
19.604
6.277
7.643
11.813
3.719
9.480
8.533
7.185
2.243
1.175
10.608
7.224
8.622
10.966
7.979
10.618
0.091
0.083
0.091
0.199
0.212
0.108
0.102
0.174
0.192
0.249
0.064
0.090
0.084
0.074
0.085
0.083
0.297
0.162
0.109
0.218
0.101
0.202
0.215
0.102
0.077
0.045
0.219
0.142
0.159
0.157
0.129
0.186
1.932
1.006
2.310
5.200
4.882
2.254
1.877
3.778
2.821
10.365
3.289
2.582
3.079
2.640
1.429
1.971
12.563
2.685
4.797
5.843
2.212
4.741
3.412
7.177
1.793
0.913
6.576
2.667
3.117
5.707
5.139
4.157
0.065
0.046
0.056
0.119
0.102
0.069
0.065
0.096
0.106
0.137
0.061
0.068
0.072
0.055
0.057
0.047
0.157
0.093
0.079
0.134
0.070
0.118
0.124
0.096
0.067
0.038
0.120
0.077
0.085
0.086
0.081
0.099
61
6.050
14.632
2.297
41.544
20.188
23.429
36.702
9.811
16.793
15.789
9.355
10.882
8.279
10.462
23.003
10.220
16.586
16.220
21.005
8.357
14.353
9.450
19.972
10.771
20.167
24.165
8.763
25.678
8.208
14.636
8.196
36.218
0.219
0.348
0.087
1.092
0.395
0.645
1.169
0.224
0.402
0.286
0.247
0.297
0.315
0.329
0.740
0.292
0.300
0.405
0.464
0.267
0.700
0.256
0.462
0.391
0.534
0.819
0.288
0.523
0.218
0.338
0.373
0.729
42.842
27.981
34.011
37.197
47.945
34.968
30.200
43.044
34.816
66.562
42.745
34.927
37.800
38.612
33.994
34.710
55.981
35.667
43.965
50.498
32.090
39.423
29.509
39.020
28.502
24.945
46.445
41.812
38.718
52.996
38.600
40.719
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
6.169
5.168
2.287
7.730
6.387
3.910
1.442
4.959
3.412
2.496
5.078
3.568
5.166
7.229
6.083
3.435
1.565
4.647
3.692
5.401
6.375
1.782
2.873
6.693
6.216
3.337
2.004
6.462
5.241
6.356
4.188
5.955
0.099
0.153
0.051
0.207
0.072
0.083
0.050
0.085
0.059
0.058
0.043
0.090
0.110
0.100
0.082
0.086
0.035
0.071
0.064
0.099
0.096
0.062
0.069
0.140
0.089
0.106
0.072
0.121
0.093
0.112
0.059
0.088
3.099
2.862
1.995
4.700
3.711
1.957
0.968
3.928
2.381
1.890
4.186
2.592
3.684
5.532
4.525
1.603
1.411
4.209
3.492
4.238
5.052
2.115
2.719
3.875
5.624
2.597
1.740
6.172
3.696
4.147
3.770
4.099
0.065
0.096
0.044
0.122
0.057
0.067
0.044
0.074
0.049
0.050
0.040
0.065
0.092
0.086
0.063
0.055
0.031
0.053
0.062
0.076
0.086
0.059
0.059
0.103
0.084
0.084
0.067
0.079
0.078
0.089
0.047
0.075
Table A. 4 Average RMSE and RRMSE excluded regression analysis IDF curves of each site in case of GEV (Continued)
No.
61
62
63
64
65
66
67
68
69
70
71
72
LEE (OF_I)
LEE (OF_II)
HEO (OF_I)
HEO (OF_II)
ONESTEP
ROBUST
SITE
272
273
277
278
279
281
284
285
288
289
294
295
8.006
7.364
5.553
5.709
5.176
4.493
2.635
2.020
4.642
7.402
11.826
4.730
0.149
0.179
0.109
0.129
0.135
0.097
0.065
0.062
0.072
0.123
0.213
0.063
3.747
2.543
4.018
3.381
2.373
3.092
2.323
1.942
4.041
2.676
5.113
3.361
0.093
0.107
0.092
0.095
0.090
0.083
0.060
0.057
0.065
0.072
0.121
0.047
1.618
1.183
5.512
1.816
3.016
2.381
5.467
8.875
5.264
5.566
4.446
12.716
0.051
0.051
0.166
0.071
0.069
0.094
0.164
0.206
0.145
0.142
0.105
0.214
1.065
1.092
1.822
1.210
1.704
1.364
2.082
3.599
2.201
2.424
4.057
6.602
0.039
0.049
0.086
0.051
0.050
0.065
0.096
0.118
0.082
0.081
0.080
0.108
62
13.731
9.249
6.785
19.214
8.448
9.072
14.431
29.355
29.193
26.210
26.541
20.030
0.472
0.452
0.300
0.635
0.359
0.498
0.583
0.722
0.789
0.552
0.575
0.461
31.507
23.977
30.998
24.996
24.385
24.312
30.777
35.333
35.636
40.861
41.329
47.508
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
1.000
4.123
1.445
2.190
3.791
2.218
1.783
2.272
2.560
3.065
3.249
8.485
5.178
0.090
0.063
0.059
0.091
0.061
0.060
0.061
0.062
0.049
0.078
0.173
0.069
1.501
1.279
1.854
1.920
1.934
1.228
1.564
1.958
2.405
2.348
4.299
3.367
0.053
0.058
0.054
0.062
0.053
0.056
0.052
0.055
0.044
0.069
0.117
0.045
durations for each site in case of Gumbel
No.
SITE
NAME
Number of the Smallest

RMSE or RRMSE
1
90
95
98
100
101
105
106
108
112
10
114
11
115
12
119
13
121
14
127
15
129
16
130
17
131
18
133
19
135
20
136
21
138
22
140
23
143
24
146
25
152
26
155
27
156
28
159
29
162
30
165
31
168
32
169
33
170
34
184
35
185
36
187
37
189
38
192
39
201
40
202
HEO (REG)
RMSE
HEO (SEP_DUR)
RRMSE
RMSE
RMSE
58
63
1.517
0.119
1.765
0.038
0.545
0.025
3.030
0.153
2.834
0.063
2.460
0.055
0.995
0.024
0.923
2.837
0.101
1.410
2.463
0.929
1.827
5.746
5.969
4.085
1.946
2.745
3.517
2.221
3.212
1.410
1.224
2.499
3.983
1.033
1.065
1.512
3.835
1.050
2.128
1.748
3.289
2.061
2.120
2.341
2.032
2.690
3.551
2.854
1.572
1.366
7.558
2.186
2.035
7.797
2.998
IDF_CDF (SEP_DUR)
RRMSE
18
0.070
2.136
0.140
4.566
0.063
0.103
3.246
0.328
0.375
2.730
0.272
2.866
0.122
1.829
0.341
0.218
2.256
2.425
0.098
1.662
0.151
1.072
0.216
1.627
0.227
0.146
1.703
0.084
1.746
1.948
0.066
1.082
0.025
0.076
0.049
0.102
0.070
0.055
1.547
1.214
1.665
0.736
0.795
1.887
2.824
0.079
0.316
0.083
0.152
0.205
3.249
0.164
1.701
0.208
1.933
0.218
1.911
1.932
1.136
0.264
2.506
0.165
3.865
0.118
2.069
0.119
1.833
0.493
6.899
0.177
2.132
0.155
1.963
0.374
4.600
0.176
1.421
63
0.048
0.019
0.018
0.064
0.027
0.045
0.040
0.016
0.049
0.745
0.025
0.063
2.202
0.028
0.359
2.241
2.096
1.674
0.048
0.060
0.063
1.102
0.208
0.045
1.606
1.392
0.033
0.032
1.172
0.133
0.044
1.123
0.038
1.180
0.055
0.075
0.776
0.075
0.215
0.046
0.786
0.110
0.167
121
0.053
0.238
RRMSE
1.239
1.256
0.024
0.071
0.065
1.239
0.033
0.058
0.638
0.023
0.079
0.068
0.093
0.063
0.063
0.061
0.870
1.204
1.747
0.833
0.708
1.150
2.272
0.044
0.630
0.066
1.713
0.064
0.073
0.065
0.064
0.106
0.054
0.061
0.100
0.047
0.788
1.384
1.493
1.188
1.578
0.931
0.635
1.344
0.965
0.021
0.022
0.040
0.023
0.018
0.022
0.035
0.027
0.019
0.048
0.028
0.038
0.037
0.028
0.024
0.015
0.020
0.031
durations for each site in case of Gumbel (Continued)
HEO (REG)
SITE
NAME
RMSE
RRMSE
41
203
2.422
0.144
42
211
43
212
1.358
0.065
44
216
45
221
46
226
47
232
48
235
49
236
50
238
51
243
52
244
53
245
54
247
55
248
56
256
57
260
58
261
59
262
60
271
61
272
62
273
63
277
64
278
65
279
66
281
67
284
68
285
69
288
70
289
71
294
72
295
No.
1.344
1.446
2.663
2.190
2.698
3.230
2.159
1.784
3.555
1.385
3.182
2.554
1.700
3.802
2.310
1.480
4.689
1.588
1.550
1.965
1.228
1.171
1.083
1.129
1.424
4.134
1.996
4.426
2.093
4.094
HEO (SEP_DUR)
RMSE
2.657
IDF_CDF (SEP_DUR)
RRMSE
RMSE
RRMSE
0.073
2.265
0.055
0.032
1.041
0.086
0.793
0.031
0.105
0.820
0.033
0.707
0.030
0.080
0.842
0.018
0.922
0.161
1.708
0.154
2.119
0.254
2.328
0.042
0.076
0.872
1.120
0.232
2.174
0.139
1.609
0.056
105.476
0.121
1.670
0.068
1.144
0.182
0.291
2.865
2.680
0.279
2.580
0.260
1.311
0.178
0.248
0.106
0.207
1.044
1.088
0.151
1.084
0.155
0.926
0.148
1.863
0.107
0.114
1.122
1.278
0.287
2.668
0.169
1.866
0.275
3.110
0.142
2.117
0.229
4.102
64
0.052
0.065
2.227
0.142
0.060
3.401
3.188
0.104
0.087
0.022
2.263
0.243
0.075
0.561
2.432
0.138
0.066
0.978
1.224
1.776
0.038
0.040
0.029
0.022
0.027
0.031
2.658
0.610
0.024
0.796
0.026
0.416
0.549
0.030
0.020
0.028
0.068
0.414
0.013
0.051
1.780
0.043
1.861
0.040
1.142
0.024
0.088
0.824
0.043
0.601
0.060
1.697
0.051
0.929
0.028
1.639
0.037
1.136
0.023
0.068
0.036
0.445
0.050
0.303
0.051
0.700
0.092
0.908
0.082
0.481
0.033
0.072
0.062
0.050
0.072
1.356
1.301
0.018
0.025
0.025
0.014
0.024
0.022
0.013
0.027
0.020
durations for each site in case of GEV
No.
SITE
NAME
Number of the Smallest

RMSE or RRMSE
1
90
95
98
100
101
105
106
108
112
10
114
11
115
12
119
13
121
14
127
15
129
16
130
17
131
18
133
19
135
20
136
21
138
22
140
23
143
24
146
25
152
26
155
27
156
28
159
29
162
30
165
31
168
32
169
33
170
34
184
35
185
36
187
37
189
38
192
39
201
40
202
HEO (REG)
RMSE
HEO (SEP_DUR)
RRMSE
RMSE
14
4.122
IDF_CDF (SEP_DUR)
RRMSE
RMSE
15
12
43
58
0.203
2.505
0.062
1.664
0.043
5.312
0.146
6.344
0.180
8.951
6.969
1.037
16
6.207
9.008
16.767
16.856
5.445
3.676
3.450
2.816
10.152
11.180
7.098
2.124
2.218
10.763
17.030
1.373
1.937
4.112
8.464
2.018
2.844
2.842
8.851
1.501
3.577
2.941
27
0.221
0.750
12.465
0.510
2.613
0.060
0.182
3.300
0.083
0.975
0.155
7.788
4.738
2.337
20.357
5.555
9.674
3.064
2.244
0.547
5.618
0.566
8.589
0.901
0.219
4.848
2.992
0.179
6.091
0.065
0.064
0.181
0.033
0.115
0.127
0.091
0.072
0.171
3.424
0.317
2.845
0.583
0.106
0.388
0.170
9.870
2.243
0.442
5.913
0.168
0.169
2.179
1.941
0.045
0.100
65
3.865
2.857
0.105
2.773
0.063
2.312
0.169
4.990
3.689
0.075
0.049
0.058
0.082
0.045
0.067
0.076
2.840
0.060
0.054
1.914
0.040
0.108
3.570
0.048
0.203
0.074
0.067
0.097
3.224
0.100
1.272
2.706
0.256
4.032
0.064
0.060
0.186
9.873
3.704
0.046
0.057
0.899
0.075
2.451
4.050
3.863
0.072
0.152
0.685
0.382
2.478
0.048
0.048
0.083
5.868
2.590
0.125
1.610
3.034
2.648
2.982
0.044
0.077
0.066
0.195
7.821
0.083
2.329
2.618
2.975
0.062
4.081
0.102
1.136
2.601
0.098
4.595
0.157
0.976
0.067
1.454
0.097
5.019
0.109
0.089
4.821
0.345
3.435
0.051
0.090
0.237
8.304
0.053
2.897
5.088
2.587
2.239
0.075
3.281
0.162
0.174
0.071
0.064
4.424
0.104
4.405
1.916
1.812
4.926
0.132
0.293
1.169
2.553
3.426
0.358
7.139
2.508
0.160
7.746
0.077
6.462
101
0.435
1.676
2.997
5.639
RRMSE
0.100
0.170
0.064
2.841
4.783
1.893
0.954
23.565
0.082
0.106
0.063
0.042
0.338
2.369
0.043
3.957
0.036
2.480
2.565
0.058
0.061
durations for each site in case of GEV (Continued)
No.
SITE
NAME
41
203
42
211
43
212
44
216
45
221
46
226
47
232
48
235
49
236
50
238
51
243
52
244
53
245
54
247
55
248
56
256
57
260
58
261
59
262
60
271
61
272
62
273
63
277
64
278
65
279
66
281
67
284
68
285
69
288
70
289
71
294
72
295
HEO (REG)
HEO (SEP_DUR)
RMSE
RRMSE
RMSE
RRMSE
5.180
0.229
2.852
0.066
2.757
0.143
4.686
0.065
4.413
2.140
2.568
64.059
3.700
9.459
4.849
4.566
6.500
3.541
11.339
3.456
2.621
9.075
2.817
2.890
8.216
4.907
2.070
2.170
1.957
2.441
12.864
1.645
1.744
35.750
3.350
3.608
3.694
15.940
0.185
5.876
0.179
0.132
1.625
2.432
4.580
13.740
1.300
3.802
0.368
0.301
2.469
9.827
0.466
3.946
0.318
5.763
0.662
1.669
4.074
6.817
IDF_CDF (SEP_DUR)
RMSE
3.481
RRMSE
0.076
0.091
5.236
0.084
0.054
1.535
0.050
0.201
4.283
0.054
0.081
0.086
4.491
1.490
3.412
4.317
0.240
4.258
0.123
2.725
0.093
0.125
0.103
1.992
3.374
5.662
0.032
0.057
0.061
0.067
0.066
0.060
0.056
0.086
0.084
0.451
2.015
0.073
0.516
8.747
0.141
6.060
0.068
0.107
3.974
0.076
0.319
0.850
0.156
3.781
0.262
5.182
0.518
0.679
5.648
4.968
0.039
0.084
2.563
0.065
1.826
3.451
0.090
3.645
1.396
0.114
3.976
0.084
0.069
0.065
0.042
0.067
0.253
1.014
0.038
0.123
2.078
0.084
1.995
0.054
0.050
1.972
0.049
1.627
0.048
0.233
1.105
0.050
1.446
0.391
1.165
0.050
0.216
1.779
0.078
1.135
0.124
2.540
2.753
1.946
0.180
1.991
0.692
3.350
0.271
2.948
0.279
2.910
0.304
3.716
0.777
6.499
66
0.088
0.084
0.075
0.090
0.094
1.834
2.644
2.181
3.943
3.450
0.043
0.064
0.050
0.055
0.062
0.042
0.060
0.099
0.044
Gumbel
No
Site
Name
Separation
Duration (hr)
90
95
98
100
101
105 12
106 18
108 18
112
10
114
11
115
12
119 12
13
121
14
127
15
129
16
130
17
131
18
133
19
135
20
136
21
138
22
140 12
23
143
24
146
25
152
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Parameters
16.907680
17.929280
15.124890
14.568200
35.206150
39.888640
21.773480
29.783880
11.526070
12.269390
30.824920
35.126620
25.602050
30.146780
25.455190
24.437910
29.382630
33.223270
10.271060
14.018110
24.519580
24.600370
27.002460
30.053530
8.992331
9.099454
14.839130
14.349950
17.661010
26.574980
11.255550
16.139720
15.375610
15.379480
25.505930
26.151220
14.128510
18.888840
5.686926
6.837303
23.372430
29.084120
21.949280
20.725270
13.664670
16.856040
22.481590
26.136130
29.805020
32.063890
2.681748
2.472322
2.458819
2.025692
2.136945
2.028684
2.136930
1.935616
3.016656
2.629152
1.831536
1.751608
2.068170
2.225598
2.568212
2.422519
2.358412
2.168147
3.243475
2.815265
2.417428
2.417284
2.464939
2.336677
3.221230
2.804292
2.764263
2.644944
2.696438
2.436389
2.862787
2.624187
3.101831
3.025210
2.790994
2.621575
3.088988
2.919036
4.000000
4.000000
2.177133
2.148547
2.761784
2.581740
2.849599
2.754718
2.823798
2.834390
2.255630
2.132600
1.825464
2.121756
0.883378
0.614778
2.062149
2.567642
3.000000
3.000000
0.736819
0.890298
3.000000
3.000000
1.943962
2.836549
1.207390
1.058305
2.041592
2.542579
0.185276
1.271845
2.028388
2.061704
1.905250
2.352457
0.037414
0.000000
0.729144
0.588358
1.308731
3.000000
1.277459
3.000000
0.818831
0.805145
1.908369
1.985941
1.765059
3.000000
0.000297
0.660045
2.069786
2.999998
1.906470
1.664553
0.982368
1.738282
1.446954
1.956803
3.000000
3.000000
0.617284
0.620356
0.541043
0.503002
0.693220
0.717775
0.555368
0.641626
0.558554
0.555049
0.692005
0.727943
0.700715
0.748990
0.734956
0.713689
0.783514
0.803368
0.534425
0.600383
0.881360
0.881342
0.748336
0.768624
0.517970
0.496680
0.655798
0.639018
0.679534
0.773007
0.608419
0.690189
0.713094
0.708904
0.790140
0.786780
0.698416
0.766439
0.560682
0.614422
0.745195
0.800117
0.783042
0.756550
0.694062
0.746516
0.811797
0.853366
0.775911
0.791257
67
Gumbel (Continued)
No
Site
Name
Separation
Duration (hr)
26
155
27
156
28
159
29
162
30
165
31
168
32
169 12
33
170
34
184 12
35
185 12
36
187
37
189
38
192
39
201
40
202 15
41
203
42
211
43
212
44
216
45
221
46
226
47
232
48
235
49
236
50
238
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Parameters
25.180870
23.874210
20.914890
21.309910
27.258740
34.480810
26.991780
27.041760
14.602140
15.299070
22.739020
22.739080
16.087650
13.841020
37.067530
39.707470
28.631800
32.388800
21.440460
26.963210
28.821170
26.941820
25.238480
30.631300
22.398300
24.593470
48.043690
45.920270
18.177420
19.510750
11.871840
19.586390
11.368810
11.755810
11.376430
12.932080
12.988100
12.222890
18.226010
17.406770
27.412340
31.463590
29.993560
34.361060
26.714600
26.722670
24.191510
28.762820
44.091910
44.094000
2.988671
3.108185
2.729796
2.653439
2.563562
2.449227
2.374743
2.416060
3.002281
2.797334
2.948128
2.864389
2.768753
2.620330
2.162431
1.985875
2.235154
2.075961
2.759033
2.511855
2.866600
2.978406
2.972816
2.776952
3.050821
2.981275
2.106659
2.045053
2.427364
2.216738
2.863721
2.502486
2.603192
2.306505
2.709210
2.391522
2.691676
2.556178
2.610396
2.354917
2.166102
2.156910
2.911700
2.883626
2.372981
2.227545
2.869345
2.847924
3.746221
3.697925
1.699660
1.516210
1.565366
1.658208
1.851502
2.826071
1.875273
2.021661
1.150413
1.296602
1.939905
1.971889
0.725437
0.257545
3.000000
3.000000
2.619811
3.000000
2.192021
3.000000
0.635171
0.481579
2.160885
2.999987
1.877708
2.229735
2.512358
2.295358
0.740510
0.979833
0.292272
2.391089
0.478000
0.521342
0.058468
0.414518
0.561212
0.314449
1.251650
1.038407
1.717060
2.249191
2.526337
3.000000
1.397831
1.396121
1.345300
1.959998
0.014716
0.018327
0.791838
0.783693
0.750367
0.750137
0.770049
0.824685
0.770784
0.770777
0.695899
0.697065
0.764680
0.759035
0.713870
0.659258
0.770133
0.780552
0.759581
0.783136
0.764696
0.811825
0.799499
0.785511
0.766890
0.806116
0.767548
0.788919
0.835995
0.820269
0.626381
0.632208
0.547871
0.664276
0.533908
0.526191
0.523373
0.542193
0.573055
0.547282
0.658949
0.630222
0.770059
0.806328
0.846522
0.881365
0.783392
0.773466
0.791639
0.836820
0.709097
0.725446
68
Gumbel (Continued)
No
Site
Name
51
243 3
52
244 2
53
245 2
54
247 1
55
248 1
56
256 3
57
260 9
58
261 9
59
262 3
60
271 2
61
272 2
62
273 1
63
277 2
64
278 2
65
279 2
66
281 3
67
284 2
68
285 1
69
288 2
70
289 3
71
294 3
72
295 6
Separation
Duration (hr)
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Parameters
21.317950
18.895150
22.458430
27.215360
12.352380
12.016880
12.977850
12.906540
7.910077
7.686191
20.759220
20.755870
32.775370
34.084720
26.694500
30.162940
34.939480
38.518550
29.113290
32.997520
14.910040
17.011690
8.282735
8.056210
17.053540
22.959670
11.290470
12.235860
8.336968
9.606198
10.588400
11.791150
15.466160
19.733080
24.594680
24.594700
20.802480
21.445650
34.139400
37.910570
29.612430
34.679960
36.928690
39.582480
2.970186
2.984548
3.067889
2.941711
3.335097
3.278128
3.204884
3.111086
4.000000
3.887858
2.757735
2.810095
2.427325
2.243066
2.405553
2.159263
2.061033
1.992318
2.289575
2.133105
2.761717
2.590939
3.991068
3.814651
2.718795
2.616200
3.477136
3.434688
3.729458
3.583830
3.797562
3.829471
3.177084
3.283681
2.551646
2.604421
2.626616
2.631321
2.364419
2.308310
2.707706
2.566512
2.710985
2.590501
1.502160
1.098317
2.267442
2.999991
0.239265
0.154304
0.594645
0.561198
0.153095
0.073743
1.195177
1.155920
3.000000
3.000000
2.632926
3.000000
1.842689
2.247042
2.612006
3.000000
1.008023
1.504777
0.303217
0.183555
1.940597
2.999993
0.870140
1.135851
0.395663
0.832891
0.883245
1.263551
1.126177
2.067488
1.667612
1.727094
1.382464
1.597092
2.533259
2.999835
2.279808
3.000000
3.000000
3.000000
0.791207
0.757850
0.821332
0.865098
0.691878
0.680503
0.684124
0.677139
0.572212
0.557867
0.713974
0.717938
0.771659
0.772763
0.734010
0.752287
0.760877
0.781853
0.770211
0.794463
0.658931
0.685094
0.624204
0.607521
0.730312
0.804429
0.705083
0.725468
0.620834
0.654613
0.702445
0.733819
0.718978
0.789941
0.793637
0.795337
0.748033
0.754005
0.775966
0.799583
0.741009
0.773509
0.787825
0.803431
69
GEV
No
Site
Name
Separation
Duration (hr)
90
95
18
98
100
101 15
105 15
106 12
108 18
112 15
10
114 15
11
115
12
119
13
121 24
14
127
15
129 15
16
130
17
131
18
133 15
19
135
20
136
21
138 12
22
140 12
23
143
24
146
25
152 12
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Parameters
15.558760
15.553190
13.522260
7.948117
22.198680
9.555777
18.982950
32.150350
17.217460
17.268290
22.907180
32.862120
26.388480
20.180400
22.579300
18.400250
24.515400
24.434500
9.251785
14.027940
20.376150
20.373980
27.747110
28.299270
9.391945
7.742805
12.916640
12.916680
11.806460
14.292660
9.135473
14.015210
19.808160
14.676460
28.128920
25.965650
17.892760
18.613780
5.864888
5.953167
17.429070
20.211890
21.569830
14.539990
14.995030
22.715420
28.552680
33.692410
23.449020
23.965980
2.983281
2.628806
2.395129
2.616283
2.166086
2.765898
2.402636
2.270290
2.697005
2.567646
2.386916
2.331009
2.030603
3.213949
2.490978
2.794774
2.596345
2.645920
3.160796
3.150471
2.574057
2.601643
2.921838
2.895622
3.071369
3.047280
3.408093
3.440648
2.802276
2.512132
3.084173
2.615979
3.291702
3.560367
2.861157
2.856694
3.203289
3.276567
4.000000
4.000000
2.863700
2.910723
3.120263
3.136115
3.126664
3.124192
2.812889
2.888247
2.524774
2.596723
2.023687
1.871199
1.344697
1.087233
0.832826
0.000000
3.000000
3.000000
2.275608
2.874271
3.000000
3.000000
1.448755
0.602480
1.187802
1.518396
2.128608
2.796070
0.068545
3.000000
1.873501
2.121639
2.889559
3.000000
0.000000
0.000000
0.840810
0.606701
0.160987
0.904584
0.913189
3.000000
2.360517
1.789148
2.807024
2.811107
3.000000
3.000000
0.000000
0.000000
2.324998
3.000000
2.816234
1.801662
1.442803
2.999998
2.040032
2.615792
3.000000
3.000000
0.634760
0.614936
0.511129
0.408379
0.574285
0.396347
0.565374
0.727153
0.644981
0.644407
0.709349
0.817887
0.739000
0.727929
0.687568
0.662980
0.759978
0.767062
0.501333
0.627760
0.848038
0.851611
0.811600
0.821730
0.544198
0.479731
0.681010
0.679455
0.569015
0.607535
0.555002
0.646322
0.794783
0.739805
0.816627
0.804489
0.774868
0.799707
0.573399
0.578603
0.749058
0.797506
0.808456
0.713545
0.739735
0.850813
0.882178
0.929818
0.745002
0.773277
0.153051
0.145441
0.141065
0.267384
0.104672
0.247518
0.165877
0.173716
0.000031
0.039000
0.289740
0.280982
0.000001
0.000001
0.043152
0.130416
0.134021
0.173408
0.088573
0.152758
0.100171
0.117454
0.208144
0.224227
0.092911
0.129869
0.160367
0.146148
0.000001
0.014447
0.000001
0.011068
0.115179
0.179537
0.063700
0.103140
0.075939
0.102309
0.000001
0.000001
0.275385
0.293971
0.163143
0.212181
0.000001
0.000001
0.000001
0.000019
0.165891
0.209842
70
GEV (Continued)
No
Site
Name
Separation
Duration (hr)
26
155 15
27
156
28
159 12
29
162 24
30
165 18
31
168 15
32
169 24
33
170
34
184
35
185 12
36
187 12
37
189 12
38
192
39
201
40
202 24
41
203
42
211 48
43
212
44
216 24
45
221
46
226 18
47
232
48
235 15
49
236
50
238
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Parameters
39.656040
38.017330
20.852070
23.021070
26.931930
29.919800
16.454530
15.137100
13.766420
7.737710
21.652700
18.724760
14.216620
5.365621
29.907000
31.553940
17.407230
24.532640
16.579310
18.234770
16.579310
18.234770
24.276410
23.903640
31.960360
34.922260
38.725670
37.605010
15.743050
12.429970
13.258240
18.322020
10.093660
9.081422
11.386450
10.468120
11.137650
9.079310
18.012600
18.012170
27.399660
35.494240
27.101210
32.294420
22.082980
14.998660
17.901260
23.760160
12.128930
11.668450
2.822812
3.344611
2.868024
3.098466
2.494859
2.630884
2.646274
3.000723
2.884647
3.425017
2.873815
2.962726
2.983305
3.519702
2.426898
2.429727
2.575533
2.353839
3.136115
3.028276
3.136115
3.028276
3.595105
3.592953
3.053175
3.067046
2.714562
2.746757
2.432729
2.408653
2.921715
2.812352
2.688872
2.619949
2.674059
2.449499
2.624923
2.651128
2.730438
2.473022
2.715031
2.760093
3.168204
3.022199
2.593489
2.695999
3.871064
4.000000
4.000000
4.000000
2.536203
2.061257
2.033276
2.995933
2.113772
3.000000
0.987638
1.606147
1.591497
1.116256
1.929106
1.782478
1.028057
0.000002
3.000000
3.000000
1.154713
2.872815
1.501053
1.952466
1.501053
1.952466
2.984376
2.999945
3.000000
3.000000
3.000000
3.000000
0.388462
0.000000
1.112066
2.999998
0.038711
0.000000
0.000000
0.000000
0.057912
0.000000
1.321606
1.195528
2.593131
2.999961
2.461944
3.000000
1.503845
0.899231
1.699204
3.000000
0.587990
0.422433
0.944773
0.961588
0.758242
0.804476
0.752698
0.802499
0.658277
0.664276
0.659851
0.552620
0.737530
0.709817
0.702279
0.469081
0.750242
0.780140
0.644904
0.717862
0.718571
0.735405
0.718571
0.735405
0.812432
0.814577
0.868965
0.898175
0.857535
0.857850
0.591573
0.528301
0.587574
0.681076
0.532047
0.484834
0.529926
0.490762
0.520408
0.467547
0.663047
0.647543
0.867186
0.947764
0.830591
0.869691
0.754767
0.663269
0.795510
0.891054
0.713597
0.708593
0.075581
0.068967
0.078128
0.124019
0.017698
0.080093
0.143167
0.203190
0.063036
0.199127
0.000001
0.040658
0.164257
0.308284
0.168347
0.197077
0.000001
0.000001
0.000001
0.000001
0.000001
0.000001
0.209287
0.229956
0.000262
0.002255
0.272226
0.289888
0.053101
0.091297
0.093999
0.140817
0.086804
0.080358
0.034305
0.050153
0.000002
0.043471
0.004083
0.000001
0.302589
0.302584
0.000001
0.000001
0.143019
0.201045
0.297403
0.342532
0.000001
0.000001
71
GEV (Continued)
No
Site
Name
Separation
Duration (hr)
51
243 15
52
244
53
245
54
247 15
55
248
56
256
57
260 12
58
261 15
59
262 18
60
271
61
272 15
62
273
63
277
64
278
65
279
66
281
67
284 12
68
285
69
288
70
289 15
71
294 24
72
295
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Short-Duration
Long-Duration
Parameters
27.299670
27.302770
25.628700
27.380620
12.278870
10.842000
11.348020
8.214817
7.759542
8.000373
24.722200
23.400100
26.345520
23.632960
22.539400
20.292520
32.276540
34.763510
24.661810
23.850490
12.933980
14.689850
7.611614
6.607781
15.914010
19.013680
11.074190
14.856020
10.097180
10.096060
9.684486
10.377130
20.318880
27.389140
35.082280
38.376040
20.478470
21.139820
26.368660
30.922350
18.924700
15.102460
33.908940
32.886660
2.786221
2.826703
2.898349
3.358997
3.389271
3.210846
3.454208
3.396137
4.000000
4.000000
2.580675
2.831446
2.563616
2.680012
2.554010
2.682230
2.263433
2.323489
2.542574
2.631746
2.714620
2.746753
4.000000
4.000000
3.161633
3.126304
3.387914
3.709049
3.811150
3.624656
4.000000
4.000000
3.224596
3.491084
2.500540
2.705130
2.802195
3.075201
2.563698
2.644836
2.450080
3.163171
2.805049
2.779884
1.907334
1.651763
3.000000
3.000000
0.303274
0.000045
0.667933
0.124630
0.237341
0.404390
1.224739
0.710896
3.000000
3.000000
3.000000
3.000000
2.199969
2.999999
3.000000
3.000000
0.695874
1.803963
0.244976
0.000000
2.343316
3.000000
1.021345
2.886665
1.306449
1.454453
0.741353
0.957980
1.944915
3.000000
2.793340
3.000000
1.350305
1.531864
2.005762
2.999955
1.049241
2.999838
2.610207
2.060821
0.870526
0.869603
0.838247
0.910628
0.708571
0.659721
0.645636
0.560477
0.579762
0.593304
0.779970
0.771607
0.726180
0.722111
0.705570
0.706481
0.769034
0.796393
0.757979
0.771069
0.611431
0.655957
0.604073
0.558614
0.760000
0.807569
0.684886
0.789046
0.681597
0.678245
0.700376
0.724497
0.802570
0.895763
0.889759
0.932984
0.760615
0.782184
0.711237
0.763925
0.579042
0.582943
0.782499
0.774381
0.059348
0.042300
0.022362
0.107795
0.097651
0.103450
0.000001
0.070284
0.000001
0.000001
0.033978
0.000004
0.124794
0.187442
0.152432
0.220829
0.137339
0.165381
0.181468
0.226227
0.028698
0.073725
0.000001
0.000001
0.172022
0.178809
0.023558
0.085566
0.080086
0.102139
0.000001
0.000001
0.000001
0.000001
0.007452
0.025826
0.000001
0.000001
0.000001
0.037597
0.000001
0.175695
0.042069
0.028465
72

(linear or non-linear
regression analysis)

.
(intensity-durationfrequency curve using cumulative distribution function) ,
(single-objective genetic algorithms)
(multi-objective
genetic
algorithms)
.
(separation of short- and long-rainfall durations method) .
.
.
root-mean-squared-error(RMSE) relativeRMSE(RRMSE) . .
73
RMSE RRMSE .
.
.
: , , . ,
74

APPlicationofIntensity Duration FrequencyCurveforKoreanRainfallDatausingcumulativeDistributionFunction

Încărcat de

Informații document

Descriere originală:

Titlu original

Drepturi de autor

Formate disponibile

Partajați acest document

Partajați sau inserați document

Opțiuni de partajare

Vi se pare util acest document?

Este necorespunzător acest conținut?

Drepturi de autor:

Formate disponibile

APPlicationofIntensity Duration FrequencyCurveforKoreanRainfallDatausingcumulativeDistributionFunction

Încărcat de

Drepturi de autor:

Formate disponibile

Application of Intensity-Duration-Frequency Curve

for Korean Rainfall Data

The Graduate School