Académique Documents
Professionnel Documents
Culture Documents
Estimation
Traditional assumptions
• Annual flood peaks are statistically independent … OK
• Flood probability model is the same every year:
• In some part of Australia multi-decadal climate variability violates this
assumption
• Need to assume no major changes to catchment such as land use change affect
distribution of flood peaks
• Flood peaks are measured without error … not true when rating curve is extrapolated
• The flood probability distribution is known except for its parameters … In Australia
flood peaks are well fitted by the log Pearson 3 distribution though other distributions
8-2
such as Generalized Extreme Value fit equally well
Fitting LP3 Distribution to Gauged Data
Fitting LP3 to gauged data involves the same approach as fitting rainfall data
Step 1: Calculate estimates of the moments of the population of logarithms of floods. If Xi is the logarithm to base 10 of an
annual peak discharge Qi and N years is the length of record, then for the logarithmic values find the mean M,
standard deviation S and skewness g
1 N 1 N N
M = ∑Xi , ∑ ( Xi − M )2 , N
( )
(N −1)(N − 2)S3 ∑
S= g= −
3
X M
N −1 i=1
i
N i=1 i=1
Step 2: Check for low or high outliers – see Australian Rainfall Runoff for guidance
Step 4: At a selected AEP of 1 in Y, determine the confidence limits CL (QY) about the discharge QY using
F.δ .S
log(CL(QY )) = logQY ±
N
where F is a normal frequency factor for the desired confidence level (eg 1.645 for 5% and 95% confidence
limits),
δ is a tabulated parameter for determining the standard error of a Pearson III distribution, and
S is the standard deviation of logarithms of flows.
8-3
Fitting LP3 Distribution to Gauged Data
Table for δ for known values of exceedance probability and skewness
SKEW EXCEEDANCE PROBABILITY
0.998 0.995 0.990 0.980 0.950 0.900 0.800 0.500 0.200 0.100 0.050 0.020 0.010 0.005 0.002
1.5 3.7717 3.3897 3.0080 2.5321 1.7499 1.0823 0.6656 1.3065 1.6574 2.2544 3.3430 5.2228 6.8350 8.5168 10.9126
1.4 3.7635 3.3238 2.9061 2.4073 1.6284 1.0079 0.6933 1.2768 1.6131 2.2052 3.2347 4.9854 6.4618 8.1186 10.2530
1.3 3.7285 3.2374 2.7915 2.2792 1.5171 0.9522 0.7265 1.2482 1.5730 2.1544 3.1238 4.7498 6.1543 7.5835 9.6113
1.2 3.6683 3.1348 2.6695 2.1532 1.4198 0.9164 0.7626 1.2213 1.5361 2.1018 3.0109 4.5169 5.7936 7.1321 8.9892
1.1 3.5872 3.0220 2.5463 2.0350 1.3401 0.9000 0.7998 1.1966 1.5019 2.0475 2.8964 4.2874 5.4553 6.6928 8.3881
1.0 3.4922 2.9065 2.4287 1.9299 1.2807 0.9015 0.8370 1.1742 1.4696 1.9913 2.7808 4.0620 5.1499 6.2669 7.8094
0.9 3.3919 2.7963 2.3238 1.8434 1.2434 0.9182 0.8736 1.1544 1.4387 1.9335 2.6647 3.8416 4.8288 5.8817 7.3749
0.8 3.2967 2.7001 2.2381 1.7797 1.2286 0.9471 0.9093 1.1372 1.4086 1.8741 2.5487 3.6268 4.5181 5.4846 6.7804
0.7 3.2170 2.6258 2.1777 1.7422 1.2352 0.9856 0.9441 1.1226 1.3790 1.8135 2.4332 3.4184 4.2506 5.1044 6.2723
0.6 3.1634 2.5809 2.1471 1.7328 1.2613 1.0311 0.9781 1.1105 1.3496 1.7519 2.3189 3.2169 3.9620 4.7655 5.8442
0.5 3.1449 2.5710 2.1493 1.7516 1.3042 1.0818 1.0113 1.1007 1.3202 1.6895 2.2062 3.0232 3.6961 4.3972 5.3892
0.4 3.1156 2.5652 2.1638 1.7867 1.3600 1.1365 1.0416 1.0918 1.2905 1.6227 2.0915 2.8423 3.4724 4.1386 5.0602
0.3 3.1997 2.6434 2.2394 1.8607 1.4292 1.1937 1.0744 1.0866 1.2609 1.5610 1.9852 2.6656 3.2365 3.8397 4.6726
0.2 3.3323 2.7599 2.3451 1.9557 1.5074 1.2529 1.1067 1.0830 1.2309 1.4989 1.8815 2.4986 3.0175 3.5656 4.3219
0.1 3.5128 2.9129 2.4782 2.0692 1.5930 1.3133 1.1385 1.0808 1.2006 1.4367 1.7810 2.3425 2.8168 3.3183 4.0101
0.0 3.7212 3.0983 2.6363 2.1988 1.6845 1.3748 1.1698 1.0801 1.1698 1.3748 1.6845 2.1988 2.6363 3.0983 3.7212
-0.1 4.0101 3.3183 2.8168 2.3425 1.7810 1.4367 1.2006 1.0808 1.1385 1.3133 1.5930 2.0692 2.4782 2.9129 3.5128
-0.2 4.3219 3.5656 3.0175 2.4986 1.8815 1.4989 1.2309 1.0830 1.1067 1.2529 1.5074 1.9557 2.3451 2.7599 3.3323
-0.3 4.6726 3.8397 3.2365 2.6656 1.9852 1.5610 1.2609 1.0866 1.0744 1.1937 1.4292 1.8607 2.2394 2.6434 3.1997
-0.4 5.0602 4.1386 3.4724 2.8423 2.0915 1.6227 1.2905 1.0918 1.0416 1.1365 1.3600 1.7867 2.1638 2.5652 3.1156
-0.5 5.3892 4.3972 3.6961 3.0232 2.2062 1.6895 1.3202 1.1007 1.0113 1.0818 1.3042 1.7516 2.1493 2.5710 3.1449
-0.6 5.8442 4.7655 3.9620 3.2169 2.3189 1.7519 1.3496 1.1105 0.9781 1.0311 1.2613 1.7328 2.1471 2.5809 3.1634
-0.7 6.2723 5.1044 4.2506 3.4184 2.4332 1.8135 1.3790 1.1226 0.9441 0.9856 1.2352 1.7422 2.1777 2.6258 3.2170
-0.8 6.7804 5.4846 4.5181 3.6268 2.5487 1.8741 1.4086 1.1372 0.9093 0.9471 1.2286 1.7797 2.2381 2.7001 3.2967
-0.9 7.3749 5.8817 4.8288 3.8416 2.6647 1.9335 1.4387 1.1544 0.8736 0.9182 1.2434 1.8434 2.3238 2.7963 3.3919
-1.0 7.8094 6.2669 5.1499 4.0620 2.7808 1.9913 1.4696 1.1742 0.8370 0.9015 1.2807 1.9299 2.4287 2.9065 3.4922
-1.1 8.3881 6.6928 5.4553 4.2874 2.8964 2.0475 1.5019 1.1966 0.7998 0.9000 1.3401 2.0350 2.5463 3.0220 3.5872
-1.2 8.9892 7.1321 5.7936 4.5169 3.0109 2.1018 1.5361 1.2213 0.7626 0.9164 1.4198 2.1532 2.6695 3.1348 3.6683
-1.3 9.6113 7.5835 6.1543 4.7498 3.1238 2.1544 1.5730 1.2482 0.7265 0.9522 1.5171 2.2792 2.7915 3.2374 3.7285
-1.4 10.2530 8.1186 6.4618 4.9854 3.2347 2.2052 1.6131 1.2768 0.6933 1.0079 1.6284 2.4073 2.9061 3.3238 3.7635 8-4
-1.5 10.9126 8.5168 6.8350 5.2228 3.3430 2.2544 1.6574 1.3065 0.6656 1.0823 1.7499 2.5321 3.0080 3.3897 3.7717
Fitting LP3 Distribution: Example
Styx River at Jeogla From Table in section 4 (Class Notes)
A tributary of the Macleay River on the mid north coast of Mean M = 2.06695
N.S.W. Standard deviation S = 0.47708 Q y = M + K y ( g )S
Skewness g = -0.47309
National number 206001.
Lat.30° 36.4', long. 152° 09.8'. AEP K
Y
K S
Y
log Q
Y
Q
Y
Catchment area 163 km2 . % = M + K YS m3/s
Annual floods for the 47 years 1939-1985 99 -2.667 -1.272 0.795 6.23
95 -1.768 -0.844 1.223 16.7
90 -1.321 -0.630 1.437 27.3
Rank Year Q AEP Rank Year Q AEP 80 -0.810 -0.386 1.681 47.9
m m3/s % m m3/s % 50 0.078 0.037 2.104 127.1
20 0.856 0.408 2.475 299
1 1954 878 1.27 25 1973 117 52.1 10 1.220 0.582 2.649 446
2 1950 541 3.39 26 1984 111 54.2 5 1.493 0.712 2.779 602
3 1959 521 5.51 27 1961 108 56.4 2 1.792 0.855 2.922 836
4 1963 513 7.63 28 1982 105 58.5 1 1.975 0.942 3.009 1020
5 1967 436 9.75 29 1972 92.2 60.6
6 1962 411 11.9 30 1951 88.6 62.7
F .δ .S
7 1946 405 14.0 31 1943 79.9 64.8 FS
δ log(CL (QY )) = log QY ±
8 1980 315 16.1 32 1981 74.0 67.0 N N
9 1952 309 18.2 33 1953 71.9 69.1
10 1985 300 20.3 34 1960 62.6 71.2
11 1945 294 22.5 35 1979 61.2 73.3 AEP QY log QY δ 0.114 δ log CL5 CL5 log CL95 CL95
12 1971 258 24.6 36 1947 60.3 75.4 3 3
13 1956 255 26.7 37 1939 58.0 77.5 % m /s =(3)+(5) m /s =(3)-(5) m3/s
14 1978 235 28.8 38 1969 53.5 79.7 (1) (2) (3) (4) (5) (6) (7) (8) (9)
15 1949 221 30.9 39 1955 39.1 81.8
16 1977 220 33.1 40 1958 26.7 83.9 99 6.23 0.795 3.636 0.415 1.210 16.2 0.380 2.4
17 1975 206 35.2 41 1942 26.1 86.0 95 16.7 1.223 2.175 0.248 1.471 29.6 0.975 9.4
18 1976 196 37.3 42 1970 23.8 88.1 90 27.3 1.437 1.672 0.191 1.628 42.4 1.246 17.6
19 1974 194 39.4 43 1957 22.4 90.3 80 47.9 1.681 1.312 0.150 1.831 67.7 1.531 34.0
20 1964 190 41.5 44 1941 22.1 92.4 50 127 2.104 1.098 0.125 2.229 170 1.979 95.3
21 1965 186 43.6 45 1983 18.6 94.5 20 299 2.475 1.019 0.116 2.591 390 2.359 229
22 1948 177 45.8 46 1940 13.0 96.6 10 446 2.649 1.097 0.125 2.774 594 2.524 334
23 1944 164 47.9 47 1966 8.18 98.7 5 602 2.779 1.319 0.150 2.929 850 2.629 426
24 1968 126 50.0 2 836 2.922 1.761 0.201 3.123 1330 2.721 526
1 1020 3.009 2.153 0.245 3.254 1800 2.764 581
8-5
Fitting LP3 Distribution: Example
8-6
Use of Pre-Gauged Data
Prior to gauging valuable historic information may be available about large floods
Use of such data with gauged data requires use of a Bayesian approach Æ need software
6.00
Gauged flow
Exp parameter quantile
Hunter River at Singleton
5.00 Expected prob quantile 31 years gauged flows 1938 to 1969
90% limits
An examination of historic records indicates that
during the ungauged period 1820 to 1937 there
Log10 Flow
4.00
4.00
Gauged + historic
1 in Y Quantile qY Quantile confidence
3.00 Gauged + limits
historic AEP 5% limit 95% limit
10 3,281 2,188 5,006
2.00
50 9,351 5,786 16,416
100 13,535 7,751 27,100
500 28,615 12,790 87,327
1.00
1.5 2 5 10 20 50100200 500 8-7
1 in Y AEP
Peak-Over-Threshold Series
POT analysis is likely to give a more accurate estimate of flows with ARIs less
than 5 years than an analysis using annual maximum flows.
However, extracting POT data is much more difficult than extracting annual
maximum data.
Need to ensure each peak above a threshold is independent of other peaks –
consult ARR.
8-8
Rational Method: Conceptual Foundation
When whole catchment is contributing
to outlet, a steady state is reached
CiA
Discharge
Time
tc, time of
concentration
Q Constant
infiltration rate φ
Rational formula Q = C i A
Discharge from First derived in mid-1800s and still used
catchment Q
throughout world 8-9
Rational Method: Application
Simple to apply deterministic rational method:
• Measure catchment area
• Estimate time of concentration tc eg kinematic
Design rainfall
wave tc
• Estimate runoff coefficient C
rate i
• Estimate design rainfall intensity i by setting
storm duration to time of concentration Storm
tc, time of
concentration duration td
• Estimate peak flow Q = C i A
Units Q = C i A
m3/s m/s m2
Q = 0.2778 C i A
m3/s mm/hr km2
8-10
Rational Method: Accuracy
In practice large standard errors of 30 to 40% arise from use of rational method:
• Rainfall rate is not constant
• Rainfall is not spatially uniform
• Antecedent wetness which affects infiltration ignored
• Difficult to estimate runoff coefficient
discharge Q
QY
Annual max
Q = C i A
CY = QY
A I t c ,Y
Catchment area
Y ARI, years
tc is a time of concentration or response
time of catchment. Use ARR regional IFD procedure
to estimate average design
In eastern NSW tc(hr) = 0.76 A(km2)0.38
rainfall intensity Itc,Y for duration
tc and ARI of Y years
Using gauged site C10 values draw a contour map of C10 values.
8-12
Map of C10 Runoff Coefficients for Eastern NSW
8-13
Source: Australian rainfall Runoff Vol 2 (1987))
Regionalisation: Estimation at Ungauged Catchments
C10 map enables estimation of C10 at ungauged catchments.
Frequency factor
Use ARR regional IFD procedure
to estimate average design
In eastern NSW tc = 0.76 A0.38
rainfall intensity Itc,Y for duration
tc and ARI of Y years
ii. For this duration tc and the selected design ARI Y years, the design rainfall intensity Itc,Y is obtained by ARR
regional IFD analysis. No reduction of this point intensity should be made for the size of the catchment.
iii. Read the runoff coefficient C10 for an ARI of 10 years from the map in ARR Vol 2.If the site is east of the line
joining Ashford, Tamworth, Bathurst, Yass, Tumut and Jingellic, this value is adopted as C10. If the site is west
of the line and the catchment size is greater than 100 km2 (but smaller than the recommended maximum of 250
km2), the map value is again adopted as C10 if it is greater than 0.10, or this latter value is used if the map value
is less. For catchments smaller than 100 km2 west of the line, the procedure for determining values of C10 is as
follows:
• if the value read from Figure 5.1 (Volume 2) is greater than or equal to 0.40, it should be used but should
not be adjusted for area as described below.
• if it is less than 0.40, the value from Figure 5.1 should be multiplied by (100/A)0.15 where area A is in km2.
• if this adjusted value is greater than 0.40, then 0.40 should be used.
• if the adjusted value is less than 0.10, a value of 0.10 should be used.
iv. For ARIs other than 10 years, an appropriate frequency factor FFY is found from Table for the relevant zone.
For locations within 25 km of a zone boundary, FFY should be determined for each zone and the design value
found by linear interpolation from the two values based on actual location within the 50 km border strip. The
runoff coefficient CY for the adopted ARI is found by:
CY = C10 × FFY
v. Design flood magnitude is
QY = CY Itc,Y A
8-15
Map of FFY Zones
8-16
Source: Australian rainfall Runoff Vol (1987))
Table of Frequency Factors FFy
Table PRM-1. Frequency factors FFY for Rational Method in Eastern New South Wales.
FF1 0.67 0.50 0.62 0.57 0.62 0.89 0.43 0.37 0.38 0.52 0.66 0.69
FF2 0.81 0.66 0.74 0.70 0.78 0.92 0.58 0.53 0.54 0.64 0.74 0.77
FF5 0.92 0.85 0.88 0.86 0.90 0.95 0.80 0.77 0.78 0.82 0.87 0.89
FF10 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00 1.00
FF20 1.07 1.14 1.12 1.14 1.10 1.05 1.20 1.25 1.26 121 1.15 1.10
FF50 * 1.32 * 1.33 * 1.17 1.54 1.74 1.71 1.52 1.39 1.26
FF100 * 1.46 * 1.50 * 1.24 1.80 2.20 2.14 1.78 1.60 1.34
Zones
Below FF50 FF100
500m
A 0.366I12,50 0.588I12,50
FF50 = 1.90 − FF100 = 2.45 −
I12,2 I12,2
B 0.366I12,50 0.588I12,50
FF50 = 1.99 − FF100 = 2.57 −
I12,2 I12,2
C 0.366I12,50 0.588I12,50
FF50 = 1.97 − FF100 = 2.54 −
I12,2 I12,2
8-17
Source: Australian rainfall Runoff Vol (1987))
Probabilistic Rational Method: Example
Estimate the 100-year flood peak at Newcastle, NSW for a 20 km2 catchment
8-18
From Lecture 4
8-19