Watershed Hydrology

FRST 385 - FOPR 388

Lecture 13. Flood frequency analysis

 Course contents
Modules Lecture
1. Introduction to hydrology
Fundamentals of
2. Watershed concept & hydrologic cycle
Hydrology
3. Energy and mass balances
4. Precipitation
5. Soil water
6. Infiltration

Hydrologic 7. Groundwater
processes 8. Evapotranspiration
9. Runoff generation
10. Streamflow
11. Snow hydrology

Quantitative 12. Experimental design and data collection

methods 13. Flood Frequency analysis
14. Forest operations and water resources
15. Forests and peak flows: a paradigm shift
Forest Hydrology
16. Forests and water yield
17. Forests and water quality

Lecture 13 2
 Learning objectives
1. List three forestry activities that can physically impact a watershed and explain how they influence the local
hydrology.
2. Explain how soil disturbances from forestry activities can alter runoff mechanisms in a watershed.
3. Describe the effect of forest service roads on the drainage network.
4. Provide an example of how logging practices can negatively impact water quality.
5. List at least three solutions to reduce the impact of forest harvesting on water resources.
6. Define with your own words, and put into context, all the terms in the glossary at the end.

Lecture 13 3
Lecture outline
1. Objectives of frequency analysis
2. Principles and concepts
2.1 Definitions and concepts
2.2 Probability distributions
2.3 Assumptions of flood frequency analysis
3. Methods
3.1 Graphical method
3.2 Analytical method
4. Applications, examples and uses
4.1 Risk-based design
4.2 Frequency-based design
4.3 Hydrologic events
4.4 The effect of vegetation on mean and variability of events
Glossary of terms

4
 2. Forests and water – small scale
2.1 Methods to detect hydrologic change
• Paired watersheds
- Compare in natural state (calibration)
- Manage or treat one watershed (treatment)
- Compare differences (analysis)
- Annual flows (water yield), low flows, peak flows

• Hydrologic models
- Compare in natural state (calibration)
- Manage or treat one watershed (treatment)
- Compare differences (analysis)

Issues with paired watersheds:

• Representativeness
• Transferability
• Cost
• Watershed leakage
• Detecting change
• Adjustments for changing climate
• Black-box – hydrologic processes are concealed

Lecture 13 5
1. Objectives of frequency analysis
 1. Objectives of frequency analysis
• Estimate the probability and risk of damage to population
and infrastructure

• Understand the effects of weather and land use changes

on extreme events

http://www.austinsd.com/community/

www.nytimes.com/

Lecture 13 7
 1. Objectives of frequency analysis
Flooding:

• Most destructive
natural disaster
worldwide.

• Most important natural

disaster in Canada

Lecture 13 8
 1. Objectives of frequency analysis

Yukon 2.3
Cost of flooding for the
Saskatchewan 11.7
Federal Government
Quebec 301.2 (1974 – present)
Nova Scotia 2.1
http://www.ec.gc.ca/eau-
water/default.asp?lang=En&n=02A71110-1
Northwest Territories & Nunavut 1.1

Newfoundland 8.0

New Brunswick 25.9

Manitoba 229.3
Total: 713.9
British Columbia 46.6

Alberta 85.7

0 100 200 300 400

Millions of dollars [CAD$]

Lecture 13 9
2. Principles and concepts
 2. Principles and concepts
2.1 Definitions and concepts
• Frequency analysis: a procedure for estimating the frequency of occurrence or
probability of occurrence of past and/or future hydrologic events (e.g. flooding, drought,
rainfall).
• Return period (Tr): Frequency (in years) that a flow of the corresponding size or larger
will occur (only on average, not guaranteed!); or average elapsed time between
occurrences of a flood/storm with a certain magnitude or greater. Synonym: recurrence
interval.
• Exceedance probability (p): % chance that a flow of a certain magnitude is exceeded on
a given year.
• Non-exceedance probability (q): probability that a flow of a certain magnitude is not
exceeded on a given year.
• Risk (R): probability of exceeding an event at least once in the next n successive years.
• Probability density function (pdf): theoretical frequency distribution of the variable of
interest (in this case, flooding).
• Design flood: term that refers to the return period associated to a hydrologic event, in the
context of infrastructure design.
• Flood frequency curve: graph plotting return period (x) and peak discharge (y)

Lecture 13 11
 2. Principles and concepts
2.1 Definitions and concepts

Probability density
function (pdf)
(frequency)

Exceedance
probability (p)

Non-exceedance QTr Tr = return period

probability (q)
Lecture 13 12
 2. Principles and concepts
2.1 Definitions and concepts
p= 1
Tr
p = 50%

Tr = 1
p = 5% p

Q2 Q20

Lecture 13 13
 2. Principles and concepts
2.2 Probability distributions
14
Histogram of peak
12 Probability density function discharge for the Capilano
10 (pdf) River (1914 – 2003).
Frequency

0
100
120
140
160
180
200
220
240
260
280
300
320
340
360
380
400
420
440
460
480
3
Peak discharge (m /s)

Lecture 13 14
 2. Principles and concepts
2.2 Probability distributions

Different types of pdfs

according to number of
parameters, symmetry,
discrete vs.
continuous:
Frequency

Mean (µ)

Skewness (g) Variance (σ2)

Examples of Gumbel (left) and normal (right) pdfs

Lecture 13 15
 2. Principles and concepts
2.2 Probability distributions

Lecture 13 16
 2. Principles and concepts
2.2 Probability distributions

Lecture 13 17
 2. Principles and concepts
2.1 Definitions and concepts

Probability density function (pdf) Cumulative distribution function (cdf)

Cumulative frequency
Frequency

Normal distribution

Lecture 13 18
 2. Principles and concepts
2.2 Probability distributions
Probability density function (pdf) Cumulative distribution function (cdf)

Cumulative frequency
Frequency

Lognormal distribution
Lecture 13 19
 2. Principles and concepts
2.2 Probability distributions

Probability density function (pdf) Cumulative distribution function (cdf)

Frequency

Cumulative frequency

Gumbel distribution

Lecture 13 20
 2. Principles and concepts
2.3 Assumptions of flood frequency analysis
1. The time series of flood is a set of independent observations. This assures that a hydrologic event such as
a single large storm does not enter the data set more than once. For example, a single storm system may
produce two or more runoff peaks, only one of which should enter the data set: the largest.

2. The time series of floods is homogeneous. This assures that all of the flood
observations are from the same population, for instance:
- The stream gauge has not been relocated.
- Land use change has not occurred.
- No structures have been placed on the stream or its tributaries.
- There are no major natural lakes upstream of the stream gauge.
- Climate has not changed or changes are accounted for.
Lecture 13 21
3. Methods
 3. Methods
600
•Graphical method:

Peak discharge (m3/s)

500
using data records to 400
estimate Tr and p 300
within the observed 200
record. 100
0
0 10 20 30 40 50 60 70 80 90 100
600
•Analytical method:
Peak discharge (m3/s)
500

using a theoretical 400

pdf to extrapolate 300

beyond the observed 200

Observed records
record. 100 Log PIII pdf fit
0
0 10 20 30 40 50 60 70 80 90 100
Return period (y)

Comparison between graphical (top) and analytical

(bottom) methods of frequency analysis
Lecture 13 23
 3. Methods
3.1 Graphical method
* Lab 8
Year Qmax [m3s-1] Year Qmax [m3 s-1] Rank (m) p Tr [years]
1978 119 1983 393 1 0.1 10.0
1979 247 1981 336 2 0.2 5.0
1980 293 1980 293 3 0.3 3.3
1981 336 Sort 1979 247 4 0.4 2.5
1982 159 1984 206 5 0.5 2.0
1983 393 1986 206 6 0.6 1.7
1984 206 1982 159 7 0.7 1.4
1985 150 1985 150 8 0.8 1.3
1986 206 1978 119 9 0.9 1.1

p= m 1
• Exceedance probability (p):
n +1
or p=
Tr
• Return period (Tr): Tr = n +1 Weibull plotting
or Tr = 1
position formula
m p
Lecture 13 24
 3. Methods
3.1 Graphical method
* Lab 8
Example of flood
450 450 frequency curves using
Q = 375 m3/s

400 400 the graphical method to link

350 350 return period (left) and
Discharge (m /s)

Q = 273 m3/s
Discharge (m /s)
exceedance probability
3

300 300

3
250 250 (right) to maximum
200 200 discharge.
150 150
100 100
50 50
0
0
0 1 2 3 4 5 6 7 8 9 10 11
0 20 40 60 80 100
Return period (y)
Exceedance probability (%)
Tr = 8 years Ep = 34%

Lecture 13 25
 3. Methods
3.2 Analytical method
• Assigning probabilities to events, where every event has a probability of
occurrence.
• Every location has a specific probability density function (pdf), but we
don’t know it.
• Fitting and testing pdfs: we assume the overall shape of the pdf but use
the data to obtain the parameters.
• Selection of appropriate pdf: the larger the dataset, the more
parameters that can be included.

Lecture 13 26
 3. Methods
3.2 Analytical method

Selection of appropriate pdf:

• Number of parameters: the more

parameters the more precise, but also the
more likelihood of error.

• One approach is to test several pdfs and

select the best fit.

Lecture 13 27
 3. Methods
3.2 Analytical method * Lab 8
Pearson III coordinates
Example using Log Pearson III pdf

Year Qmax [m3 s-1] log(Qmax) p

1978 119 2.08
5
1979 247 2.39
1980 293 2.47
1981 336 2.53
1982 159 2.20
1983 393 2.59
1984 206 2.31
1985 150 2.18
1986 206 2.31
Mean 2.34
Standard deviation 0.17
Skewness -0.02

Lecture 13 28
 3. Methods
3.2 Analytical method
* Lab 8
Pearson III coordinates
p [%] Tr [y] ks log(Qmax) Qmax [m3 s-1]
5
0.01 100.0 2.33 2.74 550
0.05 20.0 1.64 2.62 418
0.10 10.0 1.28 2.56 363 102.56
0.30 3.3 0.52 2.43 269
0.50 2.0 0.00 2.34 219
0.70 1.4 -0.52 2.25 178
0.90 1.1 -1.28 2.12 132
0.95 1.1 -1.64 2.06 114
0.99 1.0 -2.33 1.94 87

Frequency factor equation log(Qmax) = x + ks×s

e.g.: 2.34 + 2.33×0.17 = 2.74

See previous slide for x and s values

Lecture 13 29
 3. Methods
3.2 Analytical method * Lab 8
Example of flood
Q = 504 m3/s

600 600
frequency curves using the
500 500 analytical method to link
return period (left) and

Discharge (m /s)
Discharge (m /s)

400

3
400 exceedance probability
3

(right) to maximum

Q = 241 m3/s
300 300
discharge.
200 200

100 100

0 0
0 20 40 60 80 100 0 20 40 60 80 100
Return period (y) Exceedance probability (%)

Tr = 73 years Ep = 40%
Lecture 13 30
 3. Methods
3.2 Analytical method
* Lab 8
Example using Log Pearson III pdf
600
Example of flood
500 frequency curve
(logarithmic scale).
Discharge (m 3/s)

400

300

200

100

0
1 10 100
Return period (y)

Lecture 13 31
 3. Methods
3.2 Analytical method
* Lab 8
Example using Log Pearson III pdf

600 600
Analytical - Log Pearson III Comparing the
500 500 graphical and analytical
Graphical - Weibull
Discharge (m 3/s)

methods

Discharge (m 3/s)
400 400

300 300

200 200
Analytical - Log Pearson III
100 100
Graphical - Weibull

0 0
1 10 100 0 20 40 60 80 100
Return period (y) Exceedance probability (%)

Lecture 13 32
 3. Methods
3.2 Analytical method
Normal Log-Normal

Comparing two
different pdfs for the
analytical method

Lecture 13 33
 3. Methods
3.2 Analytical method

Some considerations:
• Frequency analysis is only valid when conditions are stable.
• Changes in mean and variance caused by climate and land use can have
a significant impact on extreme events.
• Graphical method works best in the middle range of the dataset;
uncertainties grow in the extremes.
• Sample size and pdf selection are critical.
• What is the link between the number of parameters of a pdf and the
sample size?

Lecture 13 34
 3. Methods
3.2 Analytical method
• Different distributions
may give equally good
fit to the same sample of
observed floods, but
may have different
extrapolation
capabilities leading to
significantly different
estimates of extreme
floods, such as 100 year
event.

Lecture 13 35
4. Applications, examples and uses
 4. Applications, examples and uses
4.1 Risk-based design
Risk (R): probability of exceeding an event at least once in the next n
successive years.

Probability of non-exceedance (q):

q=1–p

Probability of non-exceedance in n successive years:

qn = (1 – p)n

Risk (R):

1−1− 1 
n
1 – (1 – p)n =
 Tr 

Lecture 13 37
 4. Applications, examples and uses
4.1 Risk-based design

Example: the lifetime of a stream

crossing is 25 years; the stream
R=1−1−  =1−1−  = 0.222 = 22.2%
n 25
1 1 crossing was originally designed
 Tr   100 based on the 100-yr design flood.
What is the risk of failure of the stream
Design return period for various project lives and risks of failure crossing in its lifetime?

Lecture 13 38
 4. Applications, examples and uses
4.1 Risk-based design
• Projects involve risks to
environment, property
and life.

• When danger to human

life is absent, the design
will focus on minimizing
the total expected cost.

Principles of economic risk analysis for structure size

selection

Lecture 13 39
 4. Applications, examples and uses
4.2 Frequency-based design

• Selection of a frequency or return period for the

design of a structure is most often based on
potential damage to the environment, property,
danger to life and economic losses.

• For example, the Forest Practices Code requires

all stream crossings under forest roads be
designed to convey the 100-year design flood.

Lecture 13 40
 4. Applications, examples and uses
4.3 Hydrologic events
Flooding Land use change impacts
Landslides

Storms

cse.niaes.affrc.go.jp/miwa/esid/highlight/

deahlinsurance.com

sfgate.com/blogs/

carletong.files.wordpress.com

Lecture 13 41
 4. Applications, examples and uses
4.4 The effect of vegetation on mean and variability of events

Schematic showing the effect on floods when

mean and/or variance increases after forest
harvesting (Alila et al., 2009)

Flood threshold

Lecture 13 42
 4. Applications, examples and uses
4.4 The effect of vegetation on mean and variability of events

Lecture 13 43
 Glossary of terms

Analytical method Mean

Continuous function Non-exceedance probability
Cumulative distribution function Probability density function
Design flood Probability distributions
Discrete function Return period
Exceedance probability Risk
Flood frequency curve Risk-based design
Frequency analysis Skewness
Frequency-based design Symmetry
Graphical method Variability

Lecture X 44
 Acknowledgements
Image sources
• All images used in this document are either originals by the author or comply with fair dealing under Canadian Copyright Act.
• Unless explicit, click on each image to identify its source and acknowledge copyright holder or author.
• This document cannot be distributed or copied partially or as a whole for purposes other than personal study of students registered in the course.

Contributors
• Dr. Younes Alila.
• Dr. Andres Varhola.

Bibliography

Brooks K.N., Ffolliot P.F., Gregersen H.M, DeBano L.F. 2003. Hydrology and the management of watersheds. Iowa State Press. 574 p.

Brutsaert W. 2005. Hydrology: An introduction. Cambridge University Press. 605 p.

Chang M. 2006. Forest Hydrology: An introduction to water and forests. Second Edition. CRC Press. 474.

Dingman S.L. 2002. Physical Hydrology. Second Edition. Prentice Hall. 646 p.

.
Lecture 13 45
Thank you!!!