QUESTIONS
In statistical analysis, what are the measures of central tendency? Explain.
In statistical analysis, the measures of central tendency are: (1)
arithmetic mean, (2) geometric mean, (3) median, and (4) mode.
The arithmetic mean is the first moment about the origin; it expresses the
distance from the origin to the centroid of the distribution. The
geometric mean is the nth root of the product of n terms. The median is
the value of the variable that divides the probability distribution into
two equal portions. The mode is the value of the variable that occurs
most frequently.
|
What is skewness? A distribution with a long tail on the right side has positive or negative skewness?
Skewness is the third moment of a statistical distribution about the
mean.
A distribution with a long tail on the right side has positive
skewness.
|
What are the parameters of the gamma distribution? How are the gamma and Pearson Type III distributions related?
The parameters of the gamma distribution are β and γ.
The Pearson Type III distribution has three parameters: β, γ, and xo . For xo = 0, the
Pearson Type III distribution reduces to the gamma distribution.
Therefore, the gamma distribution is a special case of the Pearson Type
III distribution.
|
What is the parameter that distinguishes the three extreme value distributions? What is the limiting value of the mean of the Gumbel variate?
The parameter k distinguishes the three extreme value distributions.
For k = 0, the extreme value distribution reduces to the Gumbel or Type I
distribution; for k less than 0, it is the Frechet or Type II
distribution; for k greater than 0, it is the Weibull or Type III
distribution.
The limiting (i.e., maximum) value of the mean of the Gumbel variate,
applicable to an infinite record length (n = ∞), is equal to the Euler
constant (0.5772).
|
What is the difference between the annual exceedance series and the annual maxima series? What is risk in the context of frequency analysis?
The annual exceedence series takes into account all extreme events above
a certain base value, regardless of when they occurred. The number of
events in the series is equal to the number of years of record. In the
annual maxima series, every year of record contributes only one value to
the extreme value series.
Risk is the probability that a certain event will occur at least once in
n successive years.
|
How is an extreme value probability paper constructed? What type of probability paper is used in the log Pearson Type III method?
Extreme value probability paper has an arithmetic scale of Gumbel y
variate in the abscissas, and an arithmetic scale of flood discharges in
the ordinates. To facilitate the plotting of return periods and
exceedence probabilities, a scale of return period T and/or exceedence
probability P is superimposed on the arithmetic scale of y.
The log Pearson II method uses log probability paper, a graph paper
featuring a logarithmic scale in the abscissas and a normal probability
scale in the ordinates.
|
What is the difference between the Weibull, Blom, and Gringorten plotting position formulas?
The Weibull, Blom, and Gringorten plotting position formulas are special
cases of Eq. 6-27. For the Weibull formula, a = 0; for the Blom formula,
a = 0.375; for the Gringorten formula, a = 0.44.
|
How is skewness variability accounted for in the log Pearson III method?
In the log Pearson III method, the variability of skewness is accounted
for by using a weighted value of skew. Station and regional skews are
weighted in inverse proportion to their mean square errors using Eq.
6-33.
|
When are high outliers considered part of historical data? When is it necessary to perform a historically weighted computation?
High outliers are treated as part of historical data when there is
sufficient evidence to indicate that the high outlier is a maximum in an
extended period of time. Otherwise, it is retained as part of the flood
series. A historically weighted computation is necessary when historical
data is available which can be used to extend the record to a period much
longer than that of the flood series.
|
Why are two-parameter distributions such as the Gumbel distribution appropiate for use in connection with short record lengths?
Experience has shown that the use of two-parameter distributions (such as
the Gumbel method) in connection with short record lengths often leads to
results which are more sensible than those obtained by fitting
three-parameter distributions (such as log Pearson III). A
three-parameter distribution fitted to a small sample may in some cases
imply that there is an upper bound to the flood discharge equal to about
twice the mean annual flood. While there may be an upper limit to flood
magnitude, it is certainly higher than twice the mean annual flood.
|
Compare floods and droughts from the standpoint of frequency analysis.
Both floods and droughts can be analyzed using frequency analysis. For
floods, however, only the peak flow is necessary; for droughts (low
flow), it is often necessary to determine not only the value of low flow
but also its duration, for instance, a 7-d average low flow.
|
What is the mean annual precipitation in the middle of the climatic spectrum?
The mean annual precipitation in the middle of the climatic spectrum is 800 mm.
|
What is the mean annual potential evaporation in the middle of the climatic spectrum?
The mean annual potential evaporation in the middle of the climatic spectrum is 1600 mm.
|
Why are the droughts in the Sahel likely to persist much longer than normal?
Long-lasting droughts such as that of the Sahel are likely to be driven by anthropogenic pressures such as deforestation or overgrazing.
|
PROBLEMS
Develop a spread sheet to calculate the mean, standard deviation, and skew coefficient of a series of annual maximum flows.
Test your work using the data of Example 6-1 in the text.
The annual maximum flows of a certain stream have been found to be normally distributed with mean 22,500 ft3/s and standard deviation 7500 ft3/s. Calculate the probability that a flow larger than 39,000 ft3/s will occur.
The standard unit corresponding to the value of 39,000 ft3/s is:
z = (39,000 - 22,500)/ (7500) = 2.2
Since the standard unit is positive, the value of 39,000 is located 2.2 standard deviations to the right of the mean. From Table A-5 (Appendix A), for z = 2.2: F (z) = 0.4861
Therefore, the cumulative probability is: 0.5 + 0.4861 = 0.9861
The probability that a flow larger than 39,000 ft3/s will occur is the complementary cumulative probability: 1.0 - 0.9861 = 0.0139; or 1.39 percent. ANSWER.
|
The 10-y and 25-y floods of a certain stream are 73 and 84 m3/s, respectively.
Assuming a normal distribution, calculate the 50-y and 100-y floods.
For return period T = 10 y, the complementary cumulative probability is: 1/T = 0.1. Therefore, the cumulative probability to the right of the mean is: F (z) = 0.5 - 0.1 = 0.4.
The corresponding standard unit (Table A-5, Appendix A) is z = 1.281. Likewise, for return period T = 25 y, the corresponding values are: 1/T = 0.04; F (z) = 0.46; and z = 1.751
Therefore: 73 = x̄ + 1.281 s; and 84 = x̄ + 1.751 s
Solving simultaneously: x̄ = 43.02; s = 23.4
For return period T = 50 y: 1/T = 0.02, F (z) = 0.48, z = 2.054
Therefore, the 50-y flood is:
Q50 = 43.02 + (2.054 × 23.4) = 91.1 m3/s. ANSWER.
For return period T = 100 y: 1/T = 0.01, F(z) = 0.49, z = 2.327
Therefore, the 100-y flood is:
Q100 = 43.02 + (2.327 × 23.4) = 97.5 m/s. ANSWER.
|
The low flows of a certain stream have been shown to follow a normal distribution.
The flows expected to be exceeded 95% and 90% of the time are 15 and 21 m3/s, respectively.
What flow can be expected to be exceeded 80% of the time?
For 95 percent exceedence probability, the cumulative probability to the left of the mean F(z) is: F(z) = 0.95 - 0.50 = 0.45.
The corresponding standard unit (Table A-5, Appendix A) is: z = 1.645. Likewise, for 90 percent exceedence probability, F(z) = 0.40, and z = 1.281
Therefore: 15 = x̄ - 1.645 s; and 21 = x̄ - 1.281 s
Solving simultaneously: x̄ = 42.11; s = 16.48
For 80 percent exceedence probability, F(z) = 0.30, and z = 0.842
Therefore, the flow expected to be exceeded 80% of the time is:
42.11 - (0.842 × 16.48) = 28.2 m3/s. ANSWER.
|
A temporary cofferdam for a 5-y dam construction period is designed to pass the 25-y flood.
What is the risk that the cofferdam may fail before the end of the construction period?
What design return period is needed to reduce the risk to less than 10%?
For a return period T = 25 y, and construction period n = 5 y, the risk
of failure of the cofferdam during the construction period (Eq. 6-24) is:
R = 1 - [1 - (1/T)]n = 1 - [1 - (1/25)]5 = 0.185
R = 18.5 percent. ANSWER.
The design return period needed to reduce the risk to less than 10%
(from Eq. 6-24) is:
10/100 = 1 - [1 - (1/T )]5
Solving for T:
T = 48 y. ANSWER.
|
Use the Weibull formula (Eq. 6-26) to calculate the plotting positions for the following series of annual maxima, in cubic feet per second:
1305, 3250, 4735, 5210, 4210, 2120, 2830, 3585, 7205, 1930, 2520, 3250, 5105, 4830, 2020, 2530, 3825, 3500, 2970, 1215.
The record length is: n = 20. The values are ranked in descending order, with rank m.
Using the Weibull formula, the probability of exceedence (percent) is: P = 100 [m / (n + 1)].
The return period is: T = (n + 1)/m. The calculations are shown in the following table.
Year
|
Annual Flood
(ft3/s) |
Ranked values (in descending order) |
Rank
m |
Probability of exceedence P (percent) |
Return period T (y) |
1 |
1305 |
7205 |
1 |
4.76 |
21.00 |
2 |
3250 |
5210 |
2 |
9.52 |
10.50 |
3 |
4735 |
5105 |
3 |
14.29 |
7.00 |
4 |
5210 |
4830 |
4 |
19.05 |
5.25 |
5 |
4210 |
4735 |
5 |
23.81 |
4.20 |
6 |
2120 |
4210 |
6 |
28.57 |
3.50 |
7 |
2830 |
3825 |
7 |
33.33 |
3.00 |
8 |
3585 |
3585 |
8 |
38.09 |
2.63 |
9 |
7205 |
3500 |
9 |
42.86 |
2.33 |
10 |
1930 |
3250 |
10 |
47.62 |
2.10 |
11 |
2520 |
3250 |
11 |
52.38 |
1.91 |
12 |
3250 |
2970 |
12 |
57.14 |
1.75 |
13 |
5105 |
2830 |
13 |
61.90 |
1.62 |
14 |
4830 |
2530 |
14 |
66.66 |
1.50 |
15 |
2020 |
2520 |
15 |
71.43 |
1.40 |
16 |
2530 |
2120 |
16 |
76.19 |
1.31 |
17 |
3825 |
2020 |
17 |
80.95 |
1.24 |
18 |
3500 |
1930 |
18 |
85.71 |
1.17 |
19 |
2970 |
1305 |
19 |
90.48 |
1.11 |
20 |
1215 |
1215 |
20 |
95.24 |
1.05 |
|
|
Use the Gringorten formula to calculate the plotting positions for the following series of annual maxima,
in cubic meters per second: 160, 350, 275, 482, 530, 390, 283, 195, 408, 307, 625, 513.
The record length is: n = 12. The values are ranked in descending order, with rank m.
Using the Gringorten formula (Eq . 6-27 with a = 0.44), the probability of exceedence (percent) is: P = 100 [(m - 0.44) / (n + 0.12)].
The return period is: T = (n + 0.12) / (m - 0.44). The calculations are shown in the following table.
Year
|
Annual Flood
(ft3/s) |
Ranked values (in descending order) |
Rank
m |
Probability of exceedence P (percent) |
Return period T (y) |
1 |
160 |
625 |
1 |
4.62 |
21.64 |
2 |
350 |
530 |
2 |
12.87 |
7.77 |
3 |
275 |
513 |
3 |
21.12 |
4.73 |
4 |
482 |
482 |
4 |
29.37 |
3.40 |
5 |
530 |
408 |
5 |
37.62 |
2.66 |
6 |
390 |
390 |
6 |
45.87 |
2.18 |
7 |
283 |
350 |
7 |
54.13 |
1.85 |
8 |
195 |
307 |
8 |
62.38 |
1.60 |
9 |
408 |
283 |
9 |
70.63 |
1.42 |
10 |
307 |
275 |
10 |
78.88 |
1.27 |
11 |
625 |
195 |
11 |
87.13 |
1.15 |
12 |
513 |
160 |
12 |
95.38 |
1.05 |
|
|
Modify the spread sheet of Problem 6-1 to calculate the mean, standard deviation, and skew coefficients of the logarithms of a series of annual maximum flows.
Test your work using the results of Example 6-4 in the text.
Fit a log Pearson III curve to the data of Problem 6-6.
Plot the calculated distribution on log probability paper, along with the Weibull plotting positions calculated in Problem 6-6.
The statistics of the logarithms can be calculated with the spread sheet developed in Problem 6-8. The results are:
mean: x̄ = 3.491; standard deviation: sy = 0.201;
skew coefficient: Csy = -0.382.
The calculations are shown in the following table.
(1) | (2) | (3) | (4) | (5) |
Return period T (y) |
Probability of exceedence P (percent) |
Frequency factor K (Csy = -0.382) |
y
|
Flood Discharge Q = x (ft3/s) |
1.05 |
95 |
-1.746 |
3.140 |
1380 |
1.11 |
90 |
-1.316 |
3.226 |
1683 |
1.25 |
80 |
-0.818 |
3.327 |
2123 |
2 |
50 |
0.064 |
3.504 |
3192 |
5 |
20 |
0.855 |
3.663 |
4603 |
10 |
10 |
1.234 |
3.739 |
5483 |
25 |
4 |
1.612 |
3.815 |
6531 |
50 |
2 |
1.844 |
3.862 |
7278 |
100 |
1 |
2.043 |
3.902 |
7980 |
200 |
0.5 |
2.218 |
3.937 |
8650 |
|
The log Pearson fit to the data is given in Cols. 1 (or 2) and 5. These results are plotted on log probability paper, as shown by the solid line of Fig. M-6-9. Also shown in this figure are the plotting positions calculated in Problem 6-6. ANSWER.
Fig. M-6-9 Flood frequency analysis by Log Pearson III method.
|
|
Fit a Gumbel curve to the data of Problem 6-6.
Plot the calculated distribution on Gumbel paper, along with the Weibull plotting positions calculated in Problem 6-6.
The statistics can be calculated with the spread sheet developed in Problem 1. The results are:
mean x̄ = 3407 ft3/s; standard deviation s = 1492 ft3/s.
From Table A-8 (Appendix A), for n = 20, the mean and standard deviation of the Gumbel variate are 0.5236 and 1.0628, respectively. The calculations are shown in the following table.
(1) | (2) | (3) | (4) | (5) |
Return period T (y) |
Probability of exceedence P (percent) |
Gumbel Variate
y |
Frequency factor
K |
Flood Discharge Q = x (ft3/s) |
1.05 |
95 |
-1.113 |
-1.540 |
1109 |
1.11 |
90 |
-0.838 |
-1.281 |
1496 |
1.25 |
80 |
-0.476 |
-0.941 |
2003 |
2 |
50 |
0.367 |
-0.147 |
3188 |
5 |
20 |
1.500 |
0.919 |
4778 |
10 |
10 |
2.250 |
1.624 |
5830 |
25 |
4 |
3.199 |
2.517 |
7162 |
50 |
2 |
3.902 |
3.179 |
8150 |
100 |
1 |
4.600 |
3.836 |
9130 |
200 |
0.5 |
5.296 |
4.490 |
10,106 |
|
The Gumbel fit to the data is given in Cols. 1 (or 2) and 5. These results are plotted in Gumbel (extreme value) paper, as shown by the solid line of Fig. M-6-10. Also shown in this figure are the plotting positions calculated in Problem 6-6. ANSWER.
Fig. M-6-10 Flood frequency analysis by GumbelI method.
|
|
Develop a spread sheet to read a series of annual maxima, sort the data in descending order,
and compute the corresponding plotting positions (percent chance and retum period) by the Weibull and Gringorten formulas.
Given the following statistics of annual maxima for stream X: number of years n = 35; mean = 3545 ft3/s; standard deviation = 1870 ft3/s. Compute the 100-y flood by the Gumbel method.
With return period T = 100 y, use Eq. 6-39 to calculate the Gumbel variate: y = 4.6. From Table A-8 (Appendix A), for record length n = 35 y, the mean and standard deviation of the Gumbel variate are:
ȳn = 0.5403; and σn = 1.1285.
Therefore, the frequency factor (from Eq. 6-40) is:
K = (y - ȳn)/ σn = 3.5974.
The 100-year flood is:
Q100 = x̄ + Ks = 3545 + (3.5974 × 1870)
Q100 = 10,270 ft3/s. ANSWER.
|
Given the following statistics of annual maxima for river Y: number of years n = 45; mean = 2700 m3/s; standard deviation 1300 m3/s; mean of the logarithms = 3.1; standard deviation of the logarithms = 0.4; skew coefficient of the logarithms = - 0.35. Compute the 100-y flood using the following probability distributions: (a) normal, (b) Gumbel, and (c) log Pearson III.
(a) For return period T = 100 y, the complementary cumulative probability is: 1/T = 0.01.
Therefore, the cumulative probability to the right of the mean is: F(z) = 0.50 - 0.01 = 0.49.
The corresponding standard unit (Table A-5, Appendix A) is: z = 2.327.
The 100-y flood is: Q100 = 2700 + (2.327 × 1,300) = 5,725 m3/s. ANSWER.
(b) For return period T = 100 y, the Gumbel variate (Eq. 6-39) is: y = 4.6. For record length n = 45, the mean and standard deviation of the Gumbel variate (Table A-8) are 0.5463 and 1.1519, respectively.
Therefore, the frequency factor (from Eq. 6-40) is: K = 3.519.
The 100-y flood is: Q100 = 2,700 + (3.519 × 1,300) = 7,275 m3/s. ANSWER.
(c) For return period T = 100 y and skew coefficient of the logarithms -0.35, the frequency factor is obtained from Table A-6 by linear interpolation: K = 2.066.
Therefore: y = 3.1 + (2.066 × 0.4) = 3.9264.
The 100-y flood is: Q100 = log-1 (3.9264) = 8,441 m3/s. ANSWER.
|
A station near Denver, Colorado, has flood records for 48 y, with station skew Csy = - 0.18.
Calculate a weighted skew coefficient.
The station skew is : Csy = -0.18.
The quantity G is the absolute value of the station skew: G = 0.18.
Using Eqs. 6-34a and 6-34c: A = -0.3156; B = 0.8932. With record length n = 48,
the mean square error of the station skew is (Eq. 6-34): (MSE)sy = 0.119.
The regional skew near Denver, Colorado, is obtained from Fig. 6-5: Csr = -0.1.
When Fig. 6-5 is used to obtain the regional skew, the mean square error of the regional skew is: (MSE)sr = 0.302.
Therefore, the weighted skew (Eq. 6-33) is:
Csw = [(0.302 × -0.18) + (0.119 × -0.1)] / (0.302 + 0.119) = -0.157.
Use Csw = -0.16. ANSWER.
|
Determine if the value Q = 13,800 ft3/s is a high outlier in a 45-y flood series with the following statistics: mean of the logarithms = 3.572; standard deviation of the logarithms = 0.215.
From Table A-7, the outlier frequency factor for record length n = 45 is: Kn = 2.727.
Using Eq. 6-35, the high outlier threshold (in log units) is:
yH = 3.572 + (2.727 × 0.215) = 4.158.
Therefore, the high outlier threshold is: QH = log-1 yH = 14,398 ft3/s.
Since the discharge Q = 13,800 ft3/s is less than the high outlier threshold, it is not considered to be a high outlier. ANSWER.
|
Using the Lettenmaier and Burges modification to the Gumbel method, fit a Gumbel curve to the data of Example 6-6 in the text. Plot the calculated distribution on Gumbel paper, along with plotting positions calculated by the Gringorten formula.
The statistics of the flood series are:
mean x̄ = 1704 m3/s; and standard deviation s = 795 m3/s.
Using Eq. 6-45: x = 1704 - [0.78 ln ln (T / (T - 1)) + 0.45] × 795.
With record length n = 16 and m = rank (in descending order), the Gringorten formula (Eq. 6-27 with a = 0.44) leads to:
P (percent) = 100 [(m - 0.44) / 16.12]; and T = 16.12 / (m - 0.44).
The calculations are summarized in the following table.
Return period T (y) |
Flood discharge Q = x (m3/s) |
1.05 |
656 |
1.11 |
827 |
1.25 |
1051 |
2 |
1574 |
5 |
2276 |
10 |
2742 |
25 |
3330 |
50 |
3766 |
100 |
4199 |
200 |
4630 |
|
Year
|
Annual Flood
(m3/s) |
Ranked values
(m3/s) |
Rank
m |
Probability of exceedence P (percent) |
Return period T (y) |
1972 |
2520 |
3320 |
1 |
3.47 |
28.79 |
1973 |
1850 |
3170 |
2 |
9.68 |
10.33 |
1974 |
750 |
2520 |
3 |
15.88 |
6.30 |
1975 |
1100 |
2160 |
4 |
22.08 |
4.53 |
1976 |
1380 |
3320 |
5 |
28.29 |
3.54 |
1977 |
1910 |
1910 |
6 |
34.49 |
2.90 |
1978 |
3170 |
1850 |
7 |
40.69 |
2.46 |
1979 |
1200 |
1730 |
8 |
46.90 |
2.13 |
1980 |
820 |
1480 |
9 |
53.10 |
1.88 |
1981 |
690 |
1380 |
10 |
59.31 |
1.69 |
1982 |
1240 |
1240 |
11 |
65.51 |
1.53 |
1983 |
1730 |
1200 |
12 |
71.71 |
1.39 |
1984 |
1950 |
1100 |
13 |
77.92 |
1.28 |
1985 |
2160 |
820 |
14 |
84.12 |
1.19 |
1986 |
3320 |
750 |
15 |
90.32 |
1.11 |
1987 |
1480 |
690 |
16 |
96.53 |
1.04 |
|
Using the first table, return period is plotted against flood discharge on Gumbel paper, as shown by the solid line of Fig. M-6-16. ANSWER.
Using the second table, return period is plotted against ranked values on Gumbel paper, as shown by the data points of Fig. M-6-16. ANSWER.
Fig. M-6-16 Gumbel method: Lettenmeir anf Burges modification.
|
|
http://enghydro.sdsu.edu |
|
150902 12:50 |
|