ICSE6 Paris - August 27-31, 2012 - Jau-Yau Lu, Chih-Chiang Su, Jian-Hao Hong, En-Juan Chen

ICSE6-23

Prediction of Maximum General Scour Depth during a Flood for

Intermittent Rivers

Jau-Yau LU1, Chih-Chiang SU2, Jian-Hao HONG3, En-Juan CHEN4

1
Professor, Dept. of Civil Eng., Natl. Chung Hsing Univ.
250 Kuo-Kuang Rd., Taichung, Taiwan 402, R.O.C. - E-mail: jylu@mail.nchu.edu.tw
2
Postdoctoral research fellow, Dept. of Civil Eng., Natl. Chung Hsing Univ.
250 Kuo-Kuang Rd., Taichung, Taiwan 402, R.O.C. - E-mail: ccsu@nchu.edu.tw
3
Postdoctoral research fellow, School of Civil and Environmental Eng.
Nangyang Technological Univ., Singapore - E-mail: jhhong@ntu.edu.sg
4
Master, Dept. of Civil Eng., Natl. Chung Hsing Univ.
250 Kuo-Kuang Rd., Taichung, Taiwan 402, R.O.C. - E-mail: yellow70053@hotmail.com

Abstract
Most rivers in Taiwan are intermittent rivers. A short-term general scour usually occurs in the river during the rising
limb of a flood due to the transitory sediment imbalance and the steep slope gradient (Lu et al., 2008). However, during
the recession limb the entrained sediment particles tend to deposit on the riverbed to nearly the initial mean bed level
before the flood if the river is in quasi-equilibrium condition over the long term. This type of short-term general scour
has caused many casualties in Taiwan such as bridge and embankment failures, e.g. the Houfeng Bridge failure in
Dajia River (Typhoon Sinlaku, peak flow Qp= 4,225 m3/s, 2008) and the Shuangyuan Bridge failure in Gaoping River
(Typhoon Morakot, Qp= 27,360 m3/s, 2009). In this study, the maximum general scour depths of the floods were
measured using the numbered brick column technique for both gravel-bed and sand-bed rivers. A maximum general
scour depth of 4.5 m was measured near the river reach of Highway Bridge hydrological station (D50 = 96 mm, S0 =
1/90) in Dajia River, which was caused by a flood induced by Typhoon Sinlaku with a corresponding peak scoured flow
depth of 7.73 m. The experimental results indicate that the short-term general scour for the gravel-bed intermittent
rivers is usually very significant and cannot be neglected as it is one of the most important factors causing the bridge
failure (Lu et al., 2007). Based on the data collected in the Dajia River and Choshui River, a general scour formula was
tentatively developed with consideration of both the flow and sediment characteristics to predict the peak scoured flow
depth. It was found that the proposed equation gave better predictions as compared with the Blenchs (1969) formula.
In conjunction with the real-time ultrasonic water level gauge, the formula can be used as an efficient tool for the
bridge closure or emergency evacuation purposes.
Key words
Short-term general scour, scoured flow depth, local scour, gravel-bed, sand-bed, intermittent river, non-
equilibrium sediment transport

I INTRODUCTION
The characteristics of scour and fill in the riverbed during a flood have significant relation to the riverbed
stability (Galay, 1983). In hydraulic engineering, scour in a riverbed can be divided into general scour and
localized scour (Melville and Coleman, 2000). Localized scour, occurring within a short distance in a river
course, can be further categorized into contraction scour and local scour. Contraction scour usually occurs
when the waterway is restricted due to the bridge embankment and bridge foundation and/or pier in the local
river course, or the bridge built in a natural narrow reach. Local scour refers to riverbed erosion caused by
the local restriction or interference of water flow, attributable to the human-imposed structures in the river
course, such as bridge pier, cross-dyke, dam or ground sill etc. General scour is the long distance
longitudinal riverbed changes. Based on the time scale, it can be further divided into long-term general scour
and short-term general scour.
Long-term general scour is resulted from scours occurring for a relatively long time scale or even for
several years. It occurs most obviously in the river reach where man-made structures, such as dam, weir, or
ground sill are set, and also in the watershed where the base lowering of the riverbed occurred due to the
collapse or landslide induced by an earthquake. The causes of long-term general scour are complex. A river

79
ICSE6 Paris - August 27-31, 2012 - Jau-Yau Lu, Chih-Chiang Su, Jian-Hao Hong, En-Juan Chen

reach with long-term general scour usually has a relatively low recovery ratio. To study the long-term trend
and evolution of a particular river reach, research is usually conducted through numerical simulation or
measurement of average riverbed elevation changes. Short-term general scour is the changes of riverbed
caused by one or several floods. Its riverbed elevation usually has restorative, i.e. the difference in riverbed
elevation before and after flood is insignificant.
Due to the effects of geography and typhoons in Taiwan, most rivers are usually intermittent rivers with
short-term general scours during high flows. Figure 1 shows a comparison of the average slopes (the ratio of
the elevation difference between the river origin and river mouth to the river length) for 10 rivers around the
world. The lengths of Dajia River and Choshui River in Taiwan are relatively short, and the corresponding
average slopes are therefore relatively high.
Bettess (2002) indicated that the factors which affect general scour include (1) change in cross-sectional
area of flow; (2) change in discharge; (3) sediment grading; (4) upstream sediment supply; and (5) duration
of flow. For a short duration flood the scour may not achieve an equilibrium condition. Lauchlan and May
(2002) summarized several general scour prediction equations. For example, Blenchs (1969) formula
depends on the regime theory can be expressed as follows
q 2/3
yms 1.20 1 / 6 for sand-bed of 0.06 D50 2 mm (1)
D50
q 2/3
y ms 1.23 1 / 12 for gravel-bed of D50 2 mm (2)
D50
in which q (m3/s/m) is the flow discharge per unit width of the main channel, D50 (mm) is the median size
of bed material, and yms is the mean scoured flow depth below the free surface. However, these empirical
formulas usually have certain limitations on their application due to the scale effect, and the effects of flow
characteristics, and bed material.
In summary, there was a lack of the field data measuring in intermittent rivers with relatively steep slopes
in the literature. For engineering practices, it is very important to develop a suitable short-term general scour
formula for the water resources management, flood mitigation planning, and the bridge closure or emergency
evacuation purpose. The main objectives of this study are: (1) to measure the scoured flow depth and the
maximum scour depth during the peak flow of a single flood on sand and gravel riverbeds, and (2) to derive
an empirical formula for predicting the short-term general scour for the high gradient intermittent rivers.
7,000

Dajia River (Taiwan) Colorado River (USA)
6,000 Choshui River (Taiwan) Yellow River (China)
Shinanogawa River (Japan) Amazon River (South America)
Mekong River (Souch Asia) Yangtze River (China)
5,000 Rhein River (Europe) Nile River (Africa)
Elevation (m)

4,000

3,000


2,000

1,000

0
0 1,000 2,000 3,000 4,000 5,000 6,000 7,000
Length of river (km)
Figure 1: Comparison of the average slopes for 10 rivers around the world (average slope is defined as the ratio
of the elevation difference between the river origin and river mouth to the river length).

II SITE DESCRIPTION
The main test sites in this study include Dajia River basin and Choshui River basin in the central region of
Taiwan. Both rivers originate from the Central Mountains, and flows into the Taiwan Strait, as shown in
Figure 2.
Dajia River is the third longest river in Taiwan with a total length of 124 km, average riverbed slope S0 of
1/60, and the drainage area of 1,236 km2. It is a gravel-bed stream with Shihkang Dam located 23.4 km
upstream from the river mouth. The average riverbed slope for the river reach downstream of Shihkang Dam

80
ICSE6 Paris - August 27-31, 2012 - Jau-Yau Lu, Chih-Chiang Su, Jian-Hao Hong, En-Juan Chen

is approximately 1/90. Field experiments were conducted at Highway Bridge hydrological station (about 5.4
km upstream of the river mouth; S0 1/90, median size of bed material D50= 96 mm, geometric standard
deviation of the sediment g = D84 /D16 = 19.96), and the cross-section 27 (0.9 km downstream of
Houfeng Bridge, and about 17.7 km upstream of the river mouth; S0 1/90, D50= 136 mm, g =5.35).
Choshui River is the longest river in Taiwan with a total length of 186.4 km, average riverbed slope of
1/190 and the drainage area of 3,157 km2. The upper Choshui River belongs to gravel-bed reach, the
midstream belongs to gravel-bed or braided reach, and the downstream is a sand-bed reach. Our study was
performed in the sand-bed reach near the Silo Bridge (23.6 km upstream of the river mouth; S0 1/1,000,
D50= 2 mm, g = 9.86), and the gravel-bed reach at Mingchu Bridge (50.3 km upstream of the river mouth;
S0 1/100, D50= 35 mm, g = 10.49).
It is worth mentioning that the particle size distribution of the gravel-bed is often bimodal. Figure 3 shows
the particle size distributions of the Highway Bridge hydrological station in the Dajia River. As shown in the
figure, the percentages of sand (D< 2 mm) and gravel (D> 2 mm) are about 15% and 85%, respectively.
However, the particle size distribution of the sand-bed reach is unimodal.

E
119 120 121 122 123
26
1
ait

25
2
Str

ver
Ri
an

24
jia
iw

3
Da
ean
Ta

N
ic Oc

23
Pacif

22 Dajia River
Taiwan
21 1 Highway Bridge hydrological station
2 Houfeng Bridge
3 Shihkang Dam

5
4 6
Choshui River
Choshui River 4 Silo Bridge
5 Mingchu Bridge
6 Chichi Weir

Yu Mount (El.3952 m)
N 232812.05
E 1205726.86

Figure 2: Location map of Dajia River and Choshui River basins.

(a) (b)
99.99
99.95
99.9
Bed material
99.8 Highway Bridge hydrological station in Dajia River
Percentage of bed material (%)

99.5
99
98
95
90
80
70
%finer

60
50
40
30
20
10
5
2
1
0.5
0.2
0.1
0.05
0.01
0.001 0.01 0.1 1 10 100 1,000
D (mm)

Figure 3: (a) Cumulative particle size distribution; (b) Percentage of bed material in each size class (bimodal)
for the gravel-bed reach at Highway Bridge hydrological station in the Dajia River.

III METHOD
The main purpose of this study is to investigate the short-term general scour in the intermittent rivers.
Therefore, the measuring position has to be at least about 150 m upstream or downstream of a bridge. Our
study employs a numbered brick column to measure the general scour depth, which is a direct observation
method. A high efficient scour brick column laying technology was developed, which uses a 5 m long
supplementary hollow steel rod with openings at both ends and two excavators. The two excavators can
simultaneously place about 90 pieces of numbered bricks at a time, as shown in Figure 4. The laying work

81
ICSE6 Paris - August 27-31, 2012 - Jau-Yau Lu, Chih-Chiang Su, Jian-Hao Hong, En-Juan Chen

was performed near a side of the main channel during the low flow period. A transit was used to locate the
position of the brick column. Also, a flood stage gauge was installed on the embankment or flood-plain to
observe the corresponding peak flood level. After a flood, the riverbed was excavated to measure the short-
term maximum general scour depth (ds) in the river reach and the corresponding mean scoured peak flow
depth (yms).

Figure 4: High efficient brick column laying technology (Highway Bridge hydrological station, Dajia River).

IV RESULTS

IV.1 Typical flow and stage hydrographs for intermittent river

The number of occurrences of typhoon-induced flood events with short-term general scours is limited each
year for the intermittent rivers in Taiwan. Also, the data collections are very challenging during the rapidly
varied flows. Figures 5(a) and 5(b) depict the typical flow and stage hydrographs, respectively collected
during Typhoon Morakot for the Dajia River in 2009. One can see the rapid variation of the flow in this
intermittent river. A short-term general scour usually occurs during the rising limb and a deposition occurs
during the falling limb of the flood (Lu et al., 2008). For example, both the Houfeng Bridge failure in Dajia
River (Typhoon Sinlaku, peak flow Qp= 4,225 m3/s, which was close to the 5-yr flood Q5= 3,800 m3/s) and
the Shuangyuan Bridge failure in Gaoping River (Typhoon Morakot, Qp= 27,360 m3/s Q200) occurred near
the flood peaks in 2008 and 2009, respectively. Therefore, the maximum general scour depth (ds) is usually
measured near the flood peaks.
6,000
Shihkang Dam, Dajia River 2009 Typhoon Morakot
5,000

4,000
Q (m3/s)

3,000

2,000

1,000

0
58
Highway Bridge hydrological station, Dajia River

57
Stage (m)

56

55

54

53
Aug 10

Aug 11

Aug 12

Aug 13

Aug 14

Aug 15

Aug 16
Aug 5

Aug 6

Aug 7

Aug 8

Aug 9

Date

Figure 5: Typical hydrographs for the intermittent river during a typhoon: (a) flow hydrograph (b) stage
hydrograph (Typhoon Morakot, Dajia River, 2009).

82
ICSE6 Paris - August 27-31, 2012 - Jau-Yau Lu, Chih-Chiang Su, Jian-Hao Hong, En-Juan Chen

IV.2 Short-term general scour during peak flow of Typhoon Sinlaku

A numbered brick column (90 pieces, 0.06 m thick each) was placed 155 m downstream of the Highway
Bridge at the main channel in Dajia River before Typhoon Sinlaku (2008) to measure the short-term general
scour depths. As shown in Table 1, the maximum general scour depth ds was 4.5 m (75 0.06 m), which
essentially occurred near the flood peak. Also, the recovery ratio of the river reach after the flood peak was
rather high. The bed level after the flood was 1.66 m higher than that before the flood. The results indicate
that the riverbed variation is very severe near the flood peak (which was close to Q5 in this case) for the
midstream and downstream of the gravel-bed reach tn the intermittent river in Taiwan. The measured scour
record was much greater than that in a perennial river with a mild slope or a mountain gravel-bed river in the
literature (Nawa and Frissell, 1993).
The water stage near peak flow measured by the flood stage gauge was about 55.32 m. Accordingly, the
scoured flow depth during the peak flow was approximately 7.73 m. Under the effect of high flood peak, the
short-term maximum general scour depth ds was more than half of the scoured flow depth (ds> 0.5yms). It
could cause serious impact on the safety of bridges or other river-crossing structures. For example, the
failure of Houfeng Bridge led to a casualty of 6 death and the crash of 3 vehicles in Dajia River caused by
the flood induced by Typhoon Sinlaku in 2008 (Hong et al., 2012). The Typhoon Morakot (Qp= 27,360 m3/s)
also caused the failure of Shuangyuan Bridge in Gaoping River in southern Taiwan in 2009. Even though
Shuangyuan Bridge was adjacent to the river mouth, it was collapsed due to the severe short-term general
scour during the flood peak, causing the falling of 12 people and 8 vehicles into the river. Therefore,
acquiring the maximum general scour depth and the scoured flow depth is of great importance to the
engineering and disaster prevention.

Distance (1) (2) (3)=(2)-(1) (4) (5)=(1)-(4) (6)=(2)-(5) (7) (8)=(7)-(5)

from Preflood bed level Bed level Difference of bed Maximum Bed level deposited height Water stage Scoured flow
Highway
(Top level of numbered after flood level between general scour during peak during recession of peak flow depth yms
Birdge before and after depth ds flow
brick column)
flood
155 m 52.09 m 53.75 m 1.66 m 4.5 m 47.59 m 6.16 m 55.32 m 7.73 m
(No. 90) (75 0.06 m) (No. 15)

Table 1: Short-term general scour measured by a numbered brick column during 2008 Typhoon Sinlaku (Qp=
4,225 m3/s) near Highway Bridge hydrological station downstream of Dajia River.

IV.3 Unusual rating curve of gravel bed intermittent river

Figure 6 shows the loop stage-discharge rating curve at Highway Bridge hydrological station for the flood
induced by Typhoon Morakot in the Dajia River in 2009. Due to the severe scouring of riverbed near the
flood peak, the direction of the loop-rating curve is clockwise, which is opposite to the direction of the
traditional rating curve in a perennial river or the ideal case in the text book (Henderson, 1966).

Figure 6: Loop stage-discharge rating curve for the flood induced by Typhoon Morakot at Highway Bridge
hydrological station in Dajia River.

83
ICSE6 Paris - August 27-31, 2012 - Jau-Yau Lu, Chih-Chiang Su, Jian-Hao Hong, En-Juan Chen

The process of sediment transport in the gravel-bed reach of the intermittent river in Taiwan is rather
complex. As mention previously, the river reach is characterized by (1) a great disparity in river discharge
before and after the flood (as shown in Figure 5), and (2) a bimodal and wide gradation of riverbed material
(as indicated in Figure 3). The critical velocities (threshold velocities) or discharges are quite different for
different particles of the bed material. In Figure 6, when the discharge rises from the base flow, only sand
and fine gravel particles are movable. Thus, the riverbed does not change significantly and the water level
rises rapidly. When the discharge is greater than about 700 m3/s, the riverbed scour (short-term general scour)
starts and the rising rate of water level begins to slow down. Note that the flow was always remained in the
main channel during the flood. After the discharge exceeds the critical discharge of the mode size of the
gravels (DG), most gravels start moving vastly. However, the flood is still hungry. The sediment transport
is non-equilibrium and the riverbed lowers down quickly. From Figure 6, it can be seen that the short-term
general scour (maximum ds value near the flood peak) during the rising limb is amazing (also shown in Table
1). Nevertheless, due to the gradual deposition of cobble and gravel particles on the riverbed during the
recession limb, the riverbed gradually approaches the bed level before the flood.
It has to be mentioned that if one only measures the cross-sectional changes before and after the flood, the
information regarding the scour and fill process cannot be fully grasped. At present, under the circumstances
of global warming and climate changes, the probability of the occurrences of highly intensified concentrating
rainfalls increases. Reasonably estimating the flood-peak related information in the intermittent rivers
[scoured flow depth (yms) and/or maximum general scour depth (ds)] are very helpful to the disaster
prevention and the bridge safety inspections.

IV.4 Scoured flow depth prediction formula

Table 2 presents the short-term general scour field data and the basic parameters collected during the
typhoon-induced floods, at the field experimental sites for both Choshui River and Dajia River. From Table
2, it can be observed that the sediment load during a flood in a sand-bed river reach is mainly dominated by
suspended load, and it is relatively closer to the sediment transport capacity. As a result, the scour potential
for the sand-bed reach is relatively small, and the Blench (1969) formula, which was developed for
canals/rivers in regime, is more applicable. In contrast, the gravel-bed river reaches in intermittent rivers
apparently are under non-equilibrium sediment transport conditions, and the applicability of the formula is
relatively low.

River Field experimental site, Qp B q(=Qp/B) S0 Bed material ds yms yms(m) by

basin flood event (m3/s) (m) (m3/s/m) (%) D50 (mm) g Riverbed (m) (m) Blench (1969)
Silo Bridge,
2268 506 4.46 0.1 2 9.86 1.2 3.29 3.04
2003 Typhoon Dujuan sand-bed
Silo Bridge, (unimodal)
8050 758 10.62 0.1 2 9.86 1.65 5.68 5.61
Choshui 2004 Typhoon Mindulle
River Mingchu Bridge,
2146 247 8.69 1 35 10.49 2.1 4.86 3.87
2003 Typhoon Dujuan gravel-bed
Mingchu Bridge, (bimodal)
7250 275 21.54 1 35 10.49 6 10.5 7.08
2004 Typhoon Mindulle
Highway Bridge, gravel-bed
4225 400 10.56 1.11 96 19.96 4.5 7.73 4.05
Dajia 2008 Typhoon Sinlaku (bimodal)
River Houfeng Bridge, gravel-bed
5410 230 23.52 1.11 136 5.35 1.56 5.12 6.71
2009 Typhoon Morakot (bimodal)
Note: Qp= peak discharge of a flood event; B= flow width; q= unit peak discharge for the main channel; S0= bed slope;
D50= median size of bed material; g= geometric standard deviation of the sediment; ds= maximum short-term general
scour depth; and yms= scoured flow depth
Table 2: Short-term general scour data in current study.

Based on the preliminary analysis of our field data, it was found that besides the discharge per unit width
for the main channel (q) and the median sediment size of bed material (D50), the short-term general scoured
flow depth also should be related to the bed slope (S0) and geometric standard deviation of the bed material
( g ), i.e.

yms fn q, S 0 , D50 , g (3)
Processing the field data as shown in Table 2 with regression analysis, the following relationship is
obtained:

84
ICSE6 Paris - August 27-31, 2012 - Jau-Yau Lu, Chih-Chiang Su, Jian-Hao Hong, En-Juan Chen

q 0.80 S 00.27 g0.74

yms 1.26 0.23 (4)
D50

in which yms, q, and D50 are in m, m3/s/m, and mm, respectively, and the coefficient of determination r2=
0.99.
The comparison of values estimated by Eq. (4) and Blenchs (1969) formula is presented in Figure 7,
indicating the formula derived in our research has a better performance. However, the field data are still
insufficient at present, Eq. (4) can be further improved when more data are available in the future.
14
q2/3
yms=1.23 D 1/12 , Blench (1969)
50
12 q0.80 S00.27 0.74
g
yms=1.26 D 50 0.23 , Current study

10
Computed, yms (m)

0 2 4 6 8 10 12 14
Measured, yms (m)

Figure 7: Comparison of scoured flow depth yms values predicted by Blench (1969) and current study, and the
measured values.

V CONCLUSIONS
Using a high efficient numbered brick column laying method, some short-term general scour data were
successfully measured during typhoon-induced floods for both gravel-bed and sand-bed reaches in the steep
intermittent rivers in Taiwan. The flow usually varies rapidly leading to a severe scour and fill during the
peak flow of a typhoon-induced flood, particularly for the bimodal gravel-bed reach in which the maximum
general scour depth can be as high as half of the scoured flow depth. From the experimental results, it can be
concluded that the scour potential is rather high in the steep gravel-bed river reach during floods with high
flows because the gradation of bed material is wide, and significant amount of sediment moves irregularly as
bed load with non-equilibrium sediment transport. For the sand-bed river reach with a milder slope, the scour
potential is lower because the sediment load is dominated by the near-saturated suspended load due to the
abundant sediment supply from the riverbed. In general, the scoured flow depth formula developed in the
current study gave better predictions as compared with the Blenchs (1969) formula. Due to the global
warming, more extreme typhoon-induced floods may occur. With slight modification when more data are
available, the proposed equation can be used as a powerful tool for the design of the cross-river hydraulic
structures and the emergency evacuations of the bridge failures.

VI ACKNOWLEGMENTS
This research was supported by the Water Resource Agency, Ministry of Economic Affairs of R.O.C. The
writers would like to acknowledge Dr. Zhong-Zhi Shi for his assistance during the field experiments.

VII REFERENCES
Bettess, R. (2002) A review of predictive methods for general scour. Proc. of First International
Conference on Scour of Foundations, Eds: H-C Chen and J-L Briaud, Texas, USA, Nov. 17-20, 162-174.
Blench, T. (1969) Mobile-bed fluviology. University of Alberta Press, Edmonton, Canada.

85
ICSE6 Paris - August 27-31, 2012 - Jau-Yau Lu, Chih-Chiang Su, Jian-Hao Hong, En-Juan Chen

Henderson, F.M.. (1966). Open Channel Flow. Macmillan, New York.

Galay, V.J. (1983). Causes of river bed degradation. Water Resour. Res., 19(5): 10571090.
Hong, J.H., Chiew, Y.M., Lu, J.Y., Lai, J.S., and Lin, Y.B. (2012). Case study: Houfeng bridge failure in
Taiwan, J. of Hydraul. Engrg., ASCE 138(2): 186199.
Lauchlan, C., and May, R. (2002). Comparison of general scour prediction equations for river crossings.
Proc. of First International Conference on Scour of Foundations, Eds: H-C Chen and J-L Briaud, Texas,
USA, Nov. 17-20, 189-197.
Lu, J.Y., Hong, J.H., Su, C.C., and Wu, I.Y. (2007) Field measurements of bridge scour in a steep gravel
river reach, In: River sediment in the environment. Session 5, Proc. of 10th International Symposium on
River Sedimentation, Moscow, Russia., Aug. 1-4, 5: 209-216.
Lu, J.Y., Hong, J.H, Su, C.C., Wang, C.Y., and Lai, J.S. (2008). Field measurements and simulation of
bridge scour-depth variations during floods. J. of Hydraul. Engrg., ASCE, 134(6): 810821.
Nawa, R.K., and Frissell, C.A. (1993). Measuring scour and fill of gravel streambeds with scour chains
and sliding-bead monitors. North American Journal Fisheries Management,13: 634-639.
Melville, B.W., and Coleman, SE. (2000). Bridge scour. Water Resources Publications, Highlands
Ranch, Colorado.

86