Flood Risk Assessment: Flood Frequency Analysis Flashcards

1
Q

What is a flood frequency curve?

A

A plot relating peak discharge to a measure of frequency.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are the two key questions flood frequency estimation answers?

A
  1. How rare was that flood?
  2. How big a flood should infrastructure be designed for?
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

POT → __
AM (Annual Maxima) → __

A
  • Values over a threshold, more data
  • One peak per year, simpler
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

True/False
Annual maxima series includes all floods per year.

A

False (only the largest)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What is the hydrological year start month in the UK?

A

October

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

If record < 14 years and POT = AM length, use ____ to estimate Q.

A

POT

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

True/False
Annual maxima should be used if you have long records.

A

True

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

What are the two rules for independent POT events?

A
  1. Two indepedent peaks must be separated by at least 3× average time to rise
  2. The minimun flow between two peaks must be < 2/3 of the first peak
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

If average rise time is 15 hrs, peaks must be ≥ __ hrs apart to be independent.

A

45 hrs

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Constructing the Flood Frequency Curve (n large)
Formula :
P(q > qd) ≈

A

1 − m/n
Where: m = cumulative frequency, n = total samples

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

CDF is obtained by dividing cumulative frequency by __.

A

sample size (n)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

Formula
Return period: T =

A

1 / P(q > qd)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What is the probability a 10-year flood does not occur in 1 year?

A

(1-1/T)^N =>
(1-1/10) = 0.9

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

P(no flood in N years) =

A

(1 − 1/T)^N

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

P(flood at least once) =

A

1 − (1 − 1/T)^N

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

Formula
P(x floods in n years) =

A

nCx · p^x · (1 - p)^(n - x)
Where p = P(q > qd)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
17
Q

For POT sampling:
T = _____

A

(n / t) × 1 / (1 − F(qd))
Where: n = years, t = POT samples

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
18
Q

If F(qd) = 0.9, t = 40, n = 20, what’s the return period?

A

5 years

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
19
Q

What problems arise with small n?

A

Flat sections between points, P(q > qmax) = 0

20
Q

Why not use P = m/N directly?

A

Leads to T = ∞ for largest value → physically meaningless

21
Q

True/False
Plotting positions give a continuous probability curve.

22
Q

What is a limitation of plotting positions?

A
  • Give discrete estimates not continuous frequency curve
  • Cannot extrapolate beyond observed max value
23
Q

What distribution is fitted for flood frequency extrapolation?

A

The Gumbel distribution (Extreme Value Type I)

24
Q

To estimate floods with return periods longer than your record, you must ______ data.

A

extrapolate

25
What are possible distributions for extrapolation?
Log-normal, Beta, Gamma
26
What distribution is typically used for extreme floods?
Gumble
27
Transformation Gumbel variate:
z = -ln( -ln[ F(qₐ) ] ) ⇒ qₐ = bz + a
28
What is the Gumbel variate z used for?
It linearises the distribution for regression
29
List the 4 Gumbel parameter estimation methods.
Linear regression, Moment matching, L-moment matching, Maximum likelihood
30
True/False Hydrologists typically prefer maximum likelihood methods.
False (they prefer L-moments)
31
What plotting position formula is used to estimate F(qd)?
Gringorten formula
32
Formula Sample mean:
µ = (1 / N) × Σ qₐ
33
Formula Sample variance:
s² = (1 / (N - 1)) × Σ (qₐ - m)²
34
Why is the L-moment method more robust?
It reduces sensitivity to outliers by avoiding squaring deviations.
35
True/False Outliers can significantly bias Gumbel parameters.
True
36
List 3 checks to ensure data quality before fitting a distribution.
- Inspect time series for errors - Remove serial dependence - Plot fitted CDF and remove outliers
37
What two methods should be used to estimate Gumbel parameters>
Method of moments and L-moments
38
What two approaches help when flow data is scarce?
1. Use regional flood frequency analysis 2. Generate synthetic flow using hydrological models
39
Regional Flood Frequency Analysis Formula
0.0213 × AREA^0.94 × STMFRQ^0.27 × S1085^0.16 × SOIL^1.23 × RSMD^1.03 × (1 + LAKE)^-0.85
40
Regional Flood Frequency Analysis RSMD → S1085 → LAKE →
- wetness index - channel slope - % of catchment routed through lake/reservoir
41
What is the "index flood" in regional analysis?
The median or mean peak flow, used to scale the growth curve.
42
The Flood Estimation Formula
Q_Med = 3.4451 × AREA^0.3062 × SAAR^0.1536 × FARL^0.0460 × BFIHOST^0.8510
43
SAAR → FARL → BFIHOST →
- average annual rainfall - fraction of area through lakes/reservoirs - baseflow index
44
Rainfall frequency is summarised using ______ curves.
Depth-Duration-Frequency (DDF)
45
What two factors define a rainfall event's rarity?
Depth and duration
46
What approach is used when no reliable flow data exist?
Use of historical maximum flood or regional/global datasets
47
Why might L-moment estimates be preferred?
More robust to outliers and better suited for skewed data