CHAPTER 3

FREQUENCY ANALYSIS
Siti Nazahiyah Rahmat
Mohd Shalahuddin bin Adnan

1
Learning Outcomes
At the end of this chapter, students should be able to:

1 2
Determine the Apply the
types of concept of
frequency stochastic
analysis hydrology in
engineering
works
• Hydrologic systems are affected by extreme events such as severe storms, floods
and droughts.
• The frequency of occurrence of an extreme event is the inverse of its magnitude, i.e.
severe event occurs less frequent compared to moderate events.
• The objective of frequency analysis of hydrologic data is to produce a relationship
between the magnitude of extreme events to their frequency of occurrence through
the use of probability distributions.
• Frequency analysis is done based on observed hydrologic data such as streamflow
and /or rainfall records. The hydrologic data analyzed are stochastic, i.e. space-
independent and / or time-independent.
• Flood frequency analysis is useful in designing the capacities of dam, spillway,
channel, road, or railway drainage. Meanwhile, drought frequency analysis is often
used to determine the depth of a navigable channel.
3
3.1 Rainfall Frequency
Return Period

Annual maximum discharge

180000
160000
140000
Discharge (cfs)

120000
100000
80000
60000
40000
20000
0
1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980
Year
Suppose a random variable X is greater than or equal to some level xT.
The recurrence interval,  is the time between occurrences of X  xT.

4
Annual maximum discharges
Year 1930 1940 1950 1960 1970
0 55,900 13,300 23,700 9,190
1 58,000 12,300 55,800 9,740
2 56,000 28,400 10,800 58,500
3 7,710 11,600 4,100 33,100
4 12,300 8,560 5,720 25,200
5 38,500 22,000 4,950 15,000 30,200
6 179,000 17,900 1,730 9,790 14,100
7 17,200 46,000 25,300 70,000 54,500
8 25,400 6,970 58,300 44,300 12,700
9 4,940 20,600 10,100 15,200

The return period of an event of a

given magnitude may be defined
as average recurrence interval
between events equaling or
exceeding a specified magnitude

5
Probability
The chance of occurrence of a flood of a return period T, in a unit time is 1/T,
that is the probability of occurrence. It is referred to as the probability of
annual exceedence.
1
P ( X  xT ) =
T
P ( X  xT each year for N years ) = (1 − P )
N

P ( X  xT at least once in N years ) = 1 − (1 − P )

N

6
Example 3.1
Estimate the probability that the annual maximum discharge, Q will
exceed 50,000 cfs at least once during the next three years.

1
𝑃 𝑄 ≥ 50,000 = = 0.20
5.13
𝑃 𝑄 ≥ 50,000 cfs at least once during the next 3 years = 1 − 1 − 0.20 3

𝑃 𝑄 ≥ 50,000 cfs at least once during the next 3 years = 0.48

7
Example 3.2
A culvert has been designed for a 50-year exceedance interval. What is
the probability that one flood or more of the design capacity will occur in
the 100-year lifetime of the structure?
N = 100 years
1
𝑃 𝑄 ≥ design capacity = = 0.02
50

𝑃 At least once in 100 years = 1 − 1 − 0.02 100

𝑃 At least once in 100 years = 0.87

8
Calculate the probability that the annual maximum
discharge, Q on the Pahang River will exceed 25,000
cfs at least once during the next three years. Assume
that the return period T = 4.6 years.
3.2 Flood Frequency Analysis
Flood frequency analyses are used to predict design floods for sites along a river. The technique
involves using observed annual peak flow discharge data to calculate statistical information such
as mean values, standard deviations, skewness, and recurrence intervals.

These statistical data are then used to construct frequency distributions, which are graphs and
tables that tell the likelihood of various discharges as a function of recurrence interval or
exceedance probability.

Flood frequency distributions can take on many forms according to the equations used to carry
out the statistical analyses. Four of the common forms are:

i. Normal Distribution
ii. Log-Normal Distribution
iii. Extreme-Value Distribution Type I - Gumbel Distribution
Type II - Cauchy Distribution
Type III - Weibull Distribution
iv. Log-Pearson Type III Distribution

10
11
Normal Distribution
In the case of a normal distribution, the variable X’ is equal to the
standard normal deviate and has a N (0,1) (µ = 0 and σ = 1). The
frequency factor is equal to the standard normal deviate corresponding
to a given exceedance probability or return period. The frequency factor
KT as follows
𝑥𝑇 = 𝜇𝑥 + 𝐾𝑇 𝜎𝑥

2.515517 + 0.802853𝑤 + 0.010328𝑤 2

𝐾𝑇 = 𝑤 −
1 + 1.432788𝑤 + 0.189296𝑤 2 + 0.001308𝑤 3

where 1
1 2
𝑤 = ln 2 0 < p < 0.5
𝑝
Example - Normal Distribution
Annual rainfall at a given location can be described by a normal distribution
with a mean of 127 cm and a standard deviation of 19 cm. Estimate the
magnitude of the 50-years annual rainfall.
The 50-year rainfall, x50, can be written in terms of the frequency factor K50 as

𝑥𝑇 = 𝜇𝑥 + 𝐾𝑇 𝜎𝑥 𝑥50 = 𝜇𝑥 + 𝐾50 𝜎𝑥

From the given data µx=127 cm , δx=19 cm, the exceedance probability p is given

1 1
𝑝= = = 0.02
𝑇 50
The intermediate variable, w, is
1 1
1 2 1 2
𝑤 = ln 2 = ln = 2.797
𝑝 0.022
The frequency factor, K50, as

2.515517 + 0.802853𝑤 + 0.010328𝑤 2

𝐾𝑇 = 𝑤 −
1 + 1.432788𝑤 + 0.189296𝑤 2 + 0.001308𝑤 3

2.515517 + 0.802853(2.797) + 0.010328(2.797)2

𝐾50 = 2.797 −
1 + 1.432788𝑤 + 0.189296(2.797)2 + 0.001308(2.797)3

K 50 = 2.054

The 50 year rainfall is therefore given by

𝑥50 = 𝜇𝑥 + 𝐾50 𝜎𝑥

𝑥50 = 127 + 2.054 19 = 166 𝑐𝑚

Extreme-value Type 1 Gumbel Distribution

Gumbel (extreme value type I) distribution is another probability distribution function useful for
hydrologic analysis, particularly rainfall analysis. The probability distribution function is presented
in Table 10.1.2 and the CDF is

1 𝑥− 𝑥−
𝑓(𝑥) = exp ± − exp ± , −∞ < 𝐱 < ∞
  
−∞ <  < ∞,  > 𝟎

where  and  are scale and location parameters,  is the mode (point of maximum
probability) of the distribution and the minus of the ± used for maximum values.
Type I distribution applied to maximum values are

𝜇𝑥 = 𝑏 + 0.577𝑎 𝜎𝑥 2 = 1.645𝑎2 𝑔𝑥 = 1.1396

and for the Type I distribution applied to minimum values are

𝜇𝑥 = 𝑏 − 0.577𝑎 𝜎𝑥 2 = 1.645𝑎2 𝑔𝑥 = −1.1396

By using the transformation
𝑥−𝑏
𝑦=
𝑎

The extreme value Type I distribution can be written in the form

𝑓(𝑦) = exp ±𝑦 − exp ±𝑦

Which yield the following cumulative distribution functions

𝐹(𝑦) = exp − exp −𝑦 (maxima)

𝐹(𝑦) = 1 − exp − exp 𝑦 (minima)

These cumulative distribution are most useful in determining the return periods of extreme events such
as flood flow (annual-maximum flows), maximum rainfall, and maximum wind speed (Gumbel 1954). In
UK this method is used extensively for flood studies. It also can be used in analyzing annual-maximum
precipitation amounts for a given durations
Example 5.4
The annual–maximum discharge between 1935 and 1978 show a mean of
8.11 m3/s and a standard deviation of 8. 51 m3/s. Assuming that the annual-maximum flows are
described by an extreme-value Type I (Gumbel) distribution, estimate the annual-maximum flowrate
with the return period of 100 years.

Solution

From the question;

𝜇𝑥 = 8.11 𝑚3 Τ𝑠 𝜎𝑥 = 8.51 𝑚3 Τ𝑠

Type I distribution applied to maximum values are

𝜇𝑥 = 𝑏 + 0.577𝑎 𝜎𝑥 2 = 1.645𝑎2 𝑔𝑥 = 1.1396

𝜎𝑥 8.51
𝑎= = = 6.64 𝑚3 Τ𝑠
1.645 1.645

𝑏 = 𝜇𝑥 − 0.577𝑎 = 811 − 0.577(6.64) = 4.278 𝑚3 /𝑠

Example 5.4
These parameters are used to transform the annual maxima, X to the normalized variable Y, where

𝑋 − 𝑏 𝑋 − 4.278
𝑌= =
𝑎 6.64
For a return period of 100 years, the exceedance probability is 1/100 = 0.01 and the cumulative
probability, F(y), is 1-0.01= 0.99. The extreme-value Type I probability distribution

𝐹(𝑦) = exp − exp −𝑦

0.99 = exp − exp −𝑦100

Where y100 is the event with a return period of 100 years.
𝑦100 = − ln( − ln 0 . 99) = 4.60

The annual-maximum flow with the return period of 100 years

𝑥100 − 𝑏 𝑥100 − 4.28 𝑥100 = 4.28 + 6.64(4.6)

𝑦100 = =
𝑎 6.64 = 34.82 𝑚3 Τ𝑠
1. Based on Table Q1, estimate the probability that the annual maximum discharge Q
will exceed 10 m3/s at least once within the next 6 years.
(7 marks)
Table Q1 Discharge data (m3/s)
Year 1970 1980 1990 2000 2010
0 5.90 13.31 14.56 9.19
1 4.55 5.80 12.88 5.58 9.74
2 5.65 8.93 3.54 10.88 5.85
3 6.55 7.71 9.45 4.10 13.34
4 3.5 1.34 8.56 5.73 2.53
5 3.45 2.25 4.95 15.2 3.33
6 4.34 11.79 11.73 9.79 14.23
7 17.67 4.67 28.5 7.55
8 10.22 6.97 5.83 4.430
9 6.87 12.45 10.17 15.20

2. Annual rainfall at a given location can be described by a normal distribution with a

mean of 107 cm and a standard deviation of 21 cm. Estimate the magnitude of the
50-years annual rainfall.
(6 marks)
3.3 Drought
Drought Frequency Analysis

 Meteorological drought
◼ Palmer Drought Severity Index (PDSI) (Palmer, 1965)

◼ Standardized Precipitation Index (SPI) (McKee et al., 1993)

Three drought variables are used to study the drought characteristics.

1. Drought Severity
2. Drought Duration
3. Drought Intensity
Palmer Drought Severity Index (PDSI)

The Palmer Drought Severity Index (PDSI) is calculated based on precipitation and
temperature data, as well as the local Available Water Content (AWC) of the soil.
From the inputs, all the basic terms of the water balance equation can be
determined, including evapotranspiration, soil recharge, runoff, and moisture loss
from the surface layer. The PDSI has been reasonably successful at quantifying
long-term drought.

By using surface air temperature and a physical water balance model, the PDSI
takes into account the basic effect of global warming through potential
evapotranspiration
Application of Palmer Drought Severity Index (PDSI)
Most widely used index to characterize meteorological droughts. The PDSI based on monthly hydrologic
accounting using the following equation;-

P = ET + R + Q − L
where P = precipitation
ET = evapotranspiration
R = recharge
Q = runoff
L = moisture loss from the soil

Divided the soil into two layer

(i) upper layer 2.54 cm or 1 in

(ii) lower layer

It is assumed that evapotranspiration takes place at the potential rate from the upper soil until all of the available
moisture has been removed. No further recharge until the surface layer has been brought to field capacity.
Application of Standard Precipitation Index (SPI)

There are several methods that can be used to calculate the drought frequency. For
instance, one of the most common drought indices, the Palmer Drought Index, requires
precipitation, evapotranspiration, runoff, and soil moisture data, while, the SPI only
requires precipitation (Guttman, 1998).

SPI is the standardization of the probability of an event occurring, and, assuming that the
precipitation data is normally distributed, it is calculated with the formula below:
Pit − Pi
SPI it =
s pi
Where;
Pit = precipitation measured at gauge i during period t
Pi= sample mean value of precipitation at gauge i
spi= sample standard deviation of precipitation values at gauge i
The SPI value is essentially a student t-statistic, showing how many sample standard deviations
each value deviates from the sample mean.

The way to read the SPI value is that the farther below zero, the higher the drought. The value has
been somewhat arbitrarily divided into the categories shown in Table 1 below, ranging from mild
drought to extreme drought.
Table 1: Drought Categories for SPI Values (McKee, Doesken, & Kleist, 1993)
SPI values Drought category
0 ～ –0.99 Mild drought
–1.00 ～ –1.49 Moderate drought
–1.50 ～ –1.99 Severe drought
 –2.0 Extreme drought

A typical way of using the SPI method is to calculate the SPI at each particular gage location and
then take the average of all the SPI values (Hallack-Alegria & Watkins, 2005).
Methods
➢ A gamma probability density function (PDF) is fitted to the precipitation data cumulative
distribution of precipitation (CDF) standard normal distribution (Normal CDF) SPI was
estimated.
1 1

0.8 0.8

Cumulative probability
0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 200 400 600 800 -3 -2 -1 0 1 2 3
Cumulated 12-month
SPI
precipitation (mm)

SPI is the transformation of the precipitation time series into a

standardized normal distribution.
28
Software
 3

SPI value
-1

-2

-3

-4
Jan-00 Sep-13 May-27 Jan-41 Oct-54 Jun-68 Feb-82 Oct-95 Jul-09

Date

Floods/Droughts can be characterised by its SPI values Category

2 or more Extremely wet
✓ Duration : taken as the no. of consecutive
intervals where SPI remains above/below 1.5 to 1.99 Very wet
threshold value 1.0 to 1.49 Moderate wet
✓ Severity : cumulative values of SPI within the
drought duration 0 to -0.99 Near Normal
✓ Intensity -1.0 to -1.49 Moderate drought
-1.5 to -1.99 Severe drought
-2 and less Extreme drought
3.4 Stochastic Hydrology - Applications
 RESERVOIR
DESIGN AND
OPERATION
Medium term forecasts
for hydropower/
irrigation/ water
supply

Short term forecasts

for flood control

32
 Joint variation of rainfall and
streamflow in a catchment

Rainfall-Runoff relationships

REAL-TIME FLOOD FORECASTING

To forecast water levels at ʻAʼ, with sufficient lead time.
Water level at A: Function of rainfall in the catchment upstream,
evaporation, infiltration, storage, vegetation and other
catchment characteristics.
Terima kasih