You are on page 1of 10

Journal of Hydraulic Research

ISSN: 0022-1686 (Print) 1814-2079 (Online) Journal homepage: http://www.tandfonline.com/loi/tjhr20

Longitudinal dispersion coefficient in ice-covered


rivers

Yongcan Chen, Zhigang Wang, Dejun Zhu & Zhaowei Liu

To cite this article: Yongcan Chen, Zhigang Wang, Dejun Zhu & Zhaowei Liu (2016):
Longitudinal dispersion coefficient in ice-covered rivers, Journal of Hydraulic Research, DOI:
10.1080/00221686.2016.1175519

To link to this article: http://dx.doi.org/10.1080/00221686.2016.1175519

Published online: 13 May 2016.

Submit your article to this journal

Article views: 26

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


http://www.tandfonline.com/action/journalInformation?journalCode=tjhr20

Download by: [University of Lethbridge] Date: 11 June 2016, At: 21:13


Journal of Hydraulic Research, 2016
http://dx.doi.org/10.1080/00221686.2016.1175519
© 2016 International Association for Hydro-Environment Engineering and Research

Research paper

Longitudinal dispersion coefficient in ice-covered rivers


YONGCAN CHEN (IAHR Member), Professor, Department of Hydraulic Engineering, Tsinghua University, Beijing 100084,
People’s Republic of China
Email: chenyc@tsinghua.edu.cn

ZHIGANG WANG, Engineer, Department of Hydraulics, China Institute of Water Resources and Hydropower Research,
Beijing 100038, People’s Republic of China
Email: bimonbird@163.com
Downloaded by [University of Lethbridge] at 21:13 11 June 2016

DEJUN ZHU, Associated Professor, Department of Hydraulic Engineering, Tsinghua University, Beijing 100084, People’s Republic
of China
Email: zhudejun@tsinghua.edu.cn

ZHAOWEI LIU (IAHR Member), Associated Professor, Department of Hydraulic Engineering, Tsinghua University,
Beijing 100084, People’s Republic of China
Email: liuzhw@tsinghua.edu.cn (author for correspondence)

ABSTRACT
The focus of this paper is on the ice-cover effects on the longitudinal dispersion coefficient in rivers. The dispersion coefficients due to either
the vertical or transverse velocity shear are calculated using Elder’s or Fisher’s triple integral equation, respectively, based on the assumptions of
a parabolic cross section and a vertical logarithmic velocity distribution in both the riverbed and the ice-cover zones. The results show that in ice-
covered natural rivers, where the aspect ratio is usually larger than 10, the dispersion coefficient due to the transverse velocity shear is 50 times greater
than that due to the vertical velocity shear. An analytical formula for longitudinal dispersion coefficient is further developed by the simplification of
Fischer’s triple integral equation, and it is validated against field tracer tests with discrepancy between observations and predictions less than 150%.
Comparing the proposed formula to the most widely-used Fischer’s formula demonstrates that for wide and shallow rivers the presence of ice-cover
will lead to about a three-fold increase in the magnitude of the longitudinal dispersion.

Keywords: Cross-sectional distribution of velocity; ice-covered rivers; longitudinal dispersion coefficient; parabolic cross section;
shear velocity

1 Introduction dispersion due to the velocity shear is caused by non-uniform


velocity distribution and turbulent diffusion and leads to a Fick-
Contaminated water, as soon as it is discharged to a river, will ian distribution of trace concentration. The concept of dispersion
undergo several stages of mixing as it is transported down- due to the mean velocity shear was first proposed by Taylor
stream by the flowing water. The effluent plume initially spreads (1954) and shortly after that applied to open channel flows and
vertically, transversely and longitudinally due to turbulent dif- natural rivers by Elder (1959) and Fischer, List, Koh, Imberger,
fusion. However, once it is vertically and transversely fully and Brooks (1979). The dispersion due to the effect of the stor-
mixed, the cloud of dissolved substance will spread primarily age zone denotes tracer exchange between the shear flow and
by longitudinal dispersion. the storage zones in which there is no (or very weak) longi-
Two distinct dispersion mechanisms are known to act in tudinal transport. The earliest contribution to the storage zone
rivers: dispersion due to velocity shear and dispersion due dispersion was made by Banks (1974) who proposed a cell-
to the storage zone effects (Atkinson & Davis, 2000; Davis in-series model to explain the deviation of measured cloud
& Atkinson, 2000; Davis, Atkinson, & Wigley, 2000). The from the Gaussian distribution predicted by Taylor’s theory.

Received 17 July 2014; accepted 4 April 2016/Currently open for discussion.

ISSN 0022-1686 print/ISSN 1814-2079 online


http://www.tandfonline.com
1
2 Y. Chen et al. Journal of Hydraulic Research (2016)

Subsequently, several models were developed to describe the (Ng, 2000), and water surface waves (Huang & Law, 2011) on
dispersion due to exchange process between the fluid masses the longitudinal dispersion were also studied. However, syste-
in storage zones and in main streams by Bencala and Walters matic research on the effects of ice cover has not yet been well
(1983), Haggerty and Gorelick (1995), Boano, Packman, Cortis, developed.
Revelli, and Ridolfi (2007), and Marion and Zaramella (2005). The existence of ice cover alters the flow characteristics
Although attention to the dispersion due to the storage zone greatly, with the maximum velocity shifted down and the
effects is growing, the velocity shear dispersion models are average velocity decreased (Lau & Krishnappan, 1980; Teal,
still widely used in engineering practice due to their simpli- Ettema, & Walker, 1994). This leads to changes in the shear
city (Sukhodolov, Nikora, Rowinski, & Czernuszenko, 1997; stress distribution (Meyer, 2009), vertical turbulent diffusivity
Sukhodolov, Sukhodolov, Kozerski, & Kohler, 2006). More- and transverse mixing capacity (Engmann & Kellerhals, 1974;
over, its corresponding longitudinal dispersion coefficient DL , Zhang & Zhu, 2011). The velocity distribution, vertical turbu-
which indicates the rate of cloud spreading, still serves as one lent diffusivity and transverse mixing capacity are all key factors
of the most important parameters in river water quality mod- contributing to the longitudinal dispersion, thus from Eqs (1)
els. To obtain DL , three types of method were proposed, namely and (2) one can expect a different longitudinal dispersion
the analytical approach, tracer experiment approach (Clark, coefficient in ice-covered rivers from that in ice-free streams.
Schlosser, Stute, & Simpson, 1996) and empirical formulae In this paper, we report a study of the effects of ice cover
Downloaded by [University of Lethbridge] at 21:13 11 June 2016

(Zeng & Huai, 2014), among which the analytical approach is on the longitudinal dispersion coefficient in rivers and we
more frequently used. propose an analytical expression to represent the longitudinal
Elder (1959) investigated the dispersion due to the verti- dispersion coefficient in ice-covered rivers. Firstly, the dis-
cal velocity shear in an infinitely wide channel and derived an persion coefficients DL1 and DL2 are computed using Eqs (1)
expression for the dispersion coefficient DL1 as: and (2), respectively, based on the assumptions of a parabolic
cross section and a vertical logarithmic velocity distribution in
 1  z  z
1 1 both the near riverbed and the near ice-cover zones. Then, an
DL1 =− û(z) û(z) dz dz dz (1)
H 0 0 Dt 0
analytical expression is derived from Eq. (2) to determine the
longitudinal dispersion coefficient in ice-covered rivers where
where H is the average water depth, Dt is the turbulent diffusi- the aspect ratio is so large that the dispersion due to the vertical
vity, û(z) is the deviation of the time-averaged velocity from velocity shear can be neglected. Finally, the analytical solution
the depth-averaged velocity at z, and z is the vertical distance is verified by the observed data obtained from field tracer tests
to the riverbed. Later, Fischer et al. (1979) derived a similar by Beltaos (1998) and the effects of ice cover on the longitudinal
expression for the dispersion coefficient DL2 due to transverse dispersion process in rivers are discussed.
velocity shear, which is:

 B  y  y
2 Parabolic cross sections in ice-covered rivers
1 1
DL2 =− h(y)û(y) h(y)û(y) dy dy dy
A 0 0 Dtt h(y) 0 In real rivers, the cross section is often composed of a deep part
(2)
and one or two adjacent shallow portions near the river banks;
where A is the cross-sectional area, B is the channel width, Dtt is
hence it is too irregular to be expressed in an analytical manner.
the transverse mixing coefficient, y is the transverse coordinate,
However, the deep part of a cross section can often be assumed
h(y) is the local water depth, and û(y) is the deviation of local
to be parabolic (Deng & Chu, 2001). Consequently, a parabolic
depth-averaged velocity at y from the cross-sectional average
curve can be used to represent the cross section in ice-covered
velocity.
rivers (Fig. 1) where the flow rate is usually small enough to
Elder (1959) and Fischer et al. (1979) studied the longi-
be restricted in the deep part of the cross section. According to
tudinal dispersion coefficients in natural streams. In order to
Fig. 1, the water depth distribution over the cross section can be
improve the accuracy of the longitudinal dispersion coefficient
obtained by:
in natural rivers, Deng, Singh, and Bengtsson (2001) calculated
the transverse mixing coefficient considering the processes of y y 
h(y) = −4Hm −1 (3)
both the transverse turbulent diffusion and the transverse dis- B B
persion. Seo and Baek (2004) used the beta probability density where Hm is the maximum water depth.
function to represent the transverse velocity distribution in a
natural stream. The function can be used to produce both sym- ice cover
metrical and skewed velocity profiles. Furthermore, Marion and B
Zaramella (2006) included sinuosity in the evaluation of the lon- O y
gitudinal dispersion coefficient. In recent years, the effects of Hm h(y)
cross-section shape (Chen & Zhu, 2007), riparian vegetation
(Perucca, Camporeale, & Ridolfi, 2009), suspended sediment Figure 1 Sketch of a parabolic cross-section in ice-covered rivers
Journal of Hydraulic Research (2016) Dispersion in ice-covered rivers 3

(a) (b)

(c) (d)
Downloaded by [University of Lethbridge] at 21:13 11 June 2016

Figure 2 Typical cross-sections of the Mudan River

The assumption of a parabolic cross-sectional shape in ice- suitable to represent the vertical velocity profile (Dolgopolova,
covered rivers was verified by the topography of the Mudan 2008; Meyer, 2009; Teal et al., 1994). A cross section is divided
River located in the north-east of China. The river is ice-covered into two zones by the maximum velocity line: the ice-cover zone
for approximately five months each year, on average. In 2010, and the riverbed zone (Fig. 3). In these two zones, the verti-
the Hydrological Bureau of Mudanjiang investigated the river cal velocity distributions follow the logarithmic law. Applying
topography along a 110 km reach nearby the Mudanjiang City. the logarithmic law at any transverse coordinate y will give the
They drilled holes in the ice-cover and placed the system of V60 cross-sectional velocity distribution as:
GNSS RTK (produced by Hi-Target Survey Instruments Com-

pany, Guangzhou, China) into the river flow to measure the ⎪ u∗i 30h(y) − 30z

⎪ hb (y) < z ≤ h(y)
horizontal position and bed elevation. The transverse intervals ⎨ κ ln Ki
of measured points were unequal and ranged from 10 m to 50 m. u(y, z) =
(4)

⎪ u∗b 30z
The longitudinal spans between measured cross sections var- ⎪
⎩ ln 0 < z ≤ hb (y)
κ Kb
ied between 1 km and 2 km and in total 63 cross sections were
measured. During the investigation, the flow was found to be
mostly restricted in the deep part of cross section. Using Eq. (3) where κ is the von Karman constant (typically 0.41),
to fit the measured cross sections it was found that the corre- Ki and Kb are the equivalent roughness heights of the
lating coefficients R2 were greater than 0.80, and the deep parts bottom of ice cover and the surface of riverbed; they
can be obtained from Manning’s roughness coefficients
of the cross sections were all of the parabolic type (four typi- √
using equations ni = κ/ g(ln [30(h − hb )/Ki ] − 1)−1 and
cal cross sections are shown in Fig. 2), although the overall √
forms were different from one another. The agreement between nb = κ/ g(ln [30hb /Kb ] − 1)−1 , respectively; hb (y) is the local
the parabolic curves and the measured data demonstrated that water column for the riverbed zone at y (i.e. distance from
the
ice-cover to the location of the velocity maximum), u∗i =
parabolic cross sections are a reasonable approximation for a √
significant proportion of ice-covered rivers. g(H − Hb )S and u∗b = gHb S are the local shear velocities
for the ice-covered zone and riverbed zone, respectively, S is
the hydraulic slope, g is acceleration due to gravity, and Hb is
the height of the local water column at the riverbed zone, where
3 Analytical formula for the longitudinal dispersion
h(y) = H.
coefficient
The vertical turbulent diffusivity, representing the effects
of flow turbulence and vertical advection, is one of the most
3.1 Velocity distribution and mixing characteristics
important parameters used to analyse the longitudinal dispersion
Although the ice cover makes the vertical velocity distribution due to the vertical velocity shear. Using the Reynolds analogy
different from that in ice-free rivers, the logarithmic law is still and assuming a linear distribution of shear stress in both the
4 Y. Chen et al. Journal of Hydraulic Research (2016)

Figure 3 Sketch of the vertical profile of the mean longitudinal velocity in ice-covered rivers

ice-cover and the riverbed zones (Lau & Krishnappan, 1980; u* and ζ as:
Meyer, 2009), the vertical turbulent diffusivity is determined.
Downloaded by [University of Lethbridge] at 21:13 11 June 2016

Then, the depth-averaged vertical turbulent diffusivity can be h∗ = h(y)/H (7)


expressed as: ∗
u = û(y)/U (8)
κu∗i [H − Hb ]2 + κu∗b Hb2 ξ = y/B (9)
Dt = (5)
6H
where U is the cross-sectional-averaged velocity. Based on the
The transverse mixing in natural rivers is a complex pro- channel geometry, one can get the cross section area as:
cess involving shear dispersion, turbulent diffusion, and maybe
advection by micro-spiral-motions for some special situations. A = BH (10)
In general, it is difficult to distinguish one from another. Thus
up till now, the only reasonable approach to study the process Substituting Eqs (7)–(10) into Eq. (2) results in the following
of transverse mixing in rivers is to perform tracer tests. Accord- simplified equation:
ing to the test results, it is advisable to calculate the transverse
 B  y  y
mixing coefficient Dtt of ice-covered rivers using: 1 1
DL2 =− h(y)û(y) h(y)û(y)dydydy
A 0 0 Dtt h(y) 0
Dtt = ku∗ R (6)  1
B3
=− H h∗ (ξ )Uu∗ (ξ )
BH 0
where k is a dimensionless coefficient, which is found to be  ξ  ξ
1
approximately 50% smaller than that for open water conditions × H h∗ (ξ )Uu∗ (ξ )dξ dξ dξ
√ ∗
0 ku∗ RH h (ξ ) 0
(Engmann & Kellerhals, 1974), and u∗ = gJR is the global 
shear velocity of the cross section. As k is always as assumed B2 U2 2 1 ∗
=− h (ξ )u∗ (ξ )
to be 0.15 in ice-free conditions, it is recommended to use the H u∗ k 0
value of 0.075 in the ice-covered rivers. Parameter R = A/χ is  ξ  ξ
1
the hydraulic radius with χ representing the wetted perimeter; × ∗ (ξ )
h∗ (ξ )u∗ (ξ )dξ dξ dξ
0 h 0
for wide and shallow ice-covered rivers R is approximated equal 2 2
2I B U
to H /2. = H u∗ (11)
k H u∗

3.2 Longitudinal dispersion due to velocity shear where


 1  ξ  ξ
The longitudinal dispersion coefficient due to vertical veloc- ∗ ∗ 1
I =− u h u∗ h∗ dξ dξ dξ (12)
ity shear in ice-covered rivers, DL1 , can be estimated using 0 0 h∗ 0
Eq. (1), if the velocity distribution and the vertical turbulent dif-
To calculate the parameter I, the distributions of h* and u* are
fusivity Dt are available. With the velocity distribution and the
needed. If the cross section is parabolic, the transverse distribu-
transverse mixing coefficient Dtt known, the longitudinal dis-
tion of flow depth h(y) can be expressed using Eq. (3) and H can
persion coefficient due to the transverse velocity shear in an
be expressed as H = 2Hm /3, therefore:
ice-covered river can be estimated using Eq. (2). However, a
simple analytical solution can be obtained assuming a parabolic
cross-sectional shape. Let us define dimensionless variables h*, h∗ = −6ξ(ξ − 1) (13)
Journal of Hydraulic Research (2016) Dispersion in ice-covered rivers 5

As is known, U can also be calculated using the Manning 1


formula, expressed for ice-covered rivers as: S = 0.0009, H = 5m, n = 0.03m-1/3s
b
0.8
S = 0.0012, H = 5m, n = 0.03m-1/3s
b
1 H 2/3 1/2 0.6
S = 0.0015, H = 5m, n = 0.03m-1/3s
U= S (14) b
n 2 0.4 -1/3
S = 0.0009, H = 3m, nb = 0.03m s
-1/3
The channel sides can be expected to have little influence on the 0.2 S = 0.0009, H = 10m, nb = 0.03m s

e
velocity profile in the central region when the width-to-depth 0 S = 0.0009, H = 5m, n = 0.02m
b
-1/3
s
ratio is sufficiently large. Therefore, it is reasonable to assume S = 0.0009, H = 5m, nb = 0.04m
-1/3
s
-0.2
that the Manning formula is valid for the local depth-averaged
velocity Uv (y) over the cross section, thus: -0.4


-0.6
1 h(y) 2/3 1/2
Uv (y) = S (15) 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8
n 2
r
D −D
As û(y) = Uv (y) − U, Eq. (8) combined with Eqs (14) and (15) Figure 4 Relation curves of effect coefficient (ε = L1DL1−ice−free
L1−ice−free
) of
ice covers on the longitudinal dispersion due to vertical velocity shear
Downloaded by [University of Lethbridge] at 21:13 11 June 2016

can be transformed to:


versus the ratio (r = ni /nb ) of ice roughness to bed roughness
u∗ = [−6ξ(ξ − 1)]2/3 − 1 (16)

Substituting Eqs (13) and (16) into Eq. (12), the value of I can water depth and bed roughness. Such an effect is indicated by
be shown by numerical integration to be 0.0014. Combined the negative values of ε in Fig. 4, and could be explained by a
with k = 0.075, an analytical expression for the longitudinal much slower flow due to an additional drag of ice-covers. Fur-
dispersion coefficient due to the transverse velocity shear in thermore, the magnitude of ε reached the minimum when the
ice-covered rivers can be derived as: roughness ratio r = 1, where also the difference between disper-
2 2 sion coefficients for ice-covered rivers and ice-free rivers is the
B U largest. However, if the ice roughness was much smaller than
DL2 = 0.037 H u∗ (17)
H u∗ the bed roughness, the longitudinal dispersion coefficient in ice-
covered rivers might be larger than that in ice-free rivers. In
For ice-free rivers, Fischer (1975) proposed a similar analyti- such case, the velocity profile in ice-covered rivers was simi-
cal expression by simplifying Eq. (2) using data obtained from lar to that in ice-free rivers except in the region near the water
several flume and field experiments. The expression is given as: surface, and as a result, the non-uniformity of velocity distribu-
2 2 tion would become dominant. Besides, according to Fig. 4 and
B U the definition of ε, it could be concluded that the longitudinal
DL = DL2 = 0.011 H u∗ (18)
H u∗ dispersion coefficient due to the vertical velocity shear in ice-
covered rivers would increase with water depth and hydraulic
Although derived differently, Eqs (18) and (17) share the same slope, which was the same as in ice-free rivers.
appearance except that the numerical coefficient takes the value However, the longitudinal dispersion coefficient due to the
of 0.011 in Eq. (18) while it is 0.037 in Eq. (17). transverse velocity shear in ice-covered rivers followed a dif-
ferent pattern. To clearly illustrate it, eight hypothetical cases
4 Application and discussion were designed with different hydraulic parameters that could
influence the flow structure greatly, namely the water depth,
Based on Eq. (1), Wang, Chen, Zhu, and Liu (2013) investi- hydraulic slope, and Manning’s roughness coefficient. The cal-
gated the ice-cover effects on the longitudinal dispersion due culation results using Eq. (2) are listed in column (11) of
to the vertical velocity shear and presented an interesting rela- Table 1. It is shown that the longitudinal dispersion coefficient
DL1 − DL1−ice−free due to the transverse velocity shear increased with the hydraulic
tionship shown in Fig. 4, where ε = is the
DL1−ice−free slope in ice-covered rivers. In contrast, the dispersion coeffi-
coefficient of ice-cover effect on the dispersion coefficient, cient decreased with either the water depth or the roughness
DL1−ice−free is the longitudinal dispersion coefficient due to the coefficient of ice cover.
vertical velocity shear in ice-free flows, calculated using Elder’s Furthermore, Eq. (17) was also used to evaluate the lon-
(1959) formula and r = ni /nb is the ratio of ice-cover rough- gitudinal dispersion coefficient due to the transverse velocity
ness coefficient to the bed roughness coefficient. For most cases, shear and the corresponding results are shown in column (12)
the longitudinal dispersion coefficient in ice-covered rivers was of Table 1. The numbers in column (12) were approximately
smaller than that in ice-free rivers with the same hydraulic slope, the same as those in column (11) with the discrepancy less than
6 Y. Chen et al. Journal of Hydraulic Research (2016)

Table 1 Longitudinal dispersion coefficients for hypothetical cases

DL2 (m2 s−1 ) Error**

Case B (m) Hm (m) S (10−3 ) ni (m−1/3 s) nb (m−1/3 s) U (m s−1 ) Dt (m2 s−1 ) Dtt (m2 s−1 ) DL1 * (m2 s−1 ) Eq. (2) Eq. (17) (%)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) (13)

TC1 50 1 0.4 0.009 0.03 0.46 0.002 0.001 0.23 846.1 806.2 5
TC2 50 3 0.4 0.009 0.03 0.96 0.008 0.005 1.18 703.8 680.8 3
TC3 50 5 0.4 0.009 0.03 1.35 0.017 0.010 2.54 646.1 626.6 3
TC4 50 10 0.4 0.009 0.03 2.14 0.048 0.029 7.19 575.4 555.5 3
TC5 50 5 0.4 0.030 0.03 0.94 0.009 0.010 0.91 313.4 303.8 3
TC6 50 5 0.4 0.070 0.03 0.54 0.013 0.010 2.06 105.3 100.3 5
TC7 50 5 0.1 0.009 0.03 0.67 0.009 0.005 1.27 323.1 308.7 4
TC8 50 5 0.7 0.009 0.03 1.78 0.023 0.013 3.35 854.7 823.5 4

*DL1 is calculated using Eq. (1)


**Error = |Eq. (17) – Eq. (2)|/Eq. (2), is the absolute value of the difference between the dispersion coefficient by Eq. (2) and that by Eq. (17)
divided by the dispersion coefficient by Eq. (2).
Downloaded by [University of Lethbridge] at 21:13 11 June 2016

5%, which were listed in column (13). Differences in rough- section, velocity distribution, and turbulent diffusion should be
ness characteristics, which changed the cross-sectional distri- validated before the comparison of Eq. (17) with field tracer
bution of velocity, were responsible for these discrepancies. tests. For environmental fluid mechanics studies, a systematic
The equivalent roughness height was assumed to be constant survey was conducted on the lower Spree River near Berlin.
over the cross section in Eq. (4) that was used in Eq. (2). In The riverbed of the investigated reach was modelled by a
contrast, Manning’s roughness coefficient was assumed to be parabolic function (Sukhodolov & Uijttewaal, 2010). During
constant over the cross section in Eq. (14) that was used in the ice-covered periods, the vertical profile of streamwise velocity
derivation of Eq. (17). However, these discrepancies were so followed the logarithmic law and the shear stress over the water
small that Eq. (17) could be regarded as a good approximation depth was linearly distributed (Sukhodolov, Thiele, Bungartz, &
to Eq. (2). Engelhardt, 1999).
For wide and shallow ice-free rivers, the transverse velo- Beltaos (1998) reported several sets of field tracer dye tests
city shear dominated the process of shear dispersion, and as in ice-covered rivers which were conducted by the Alberta
a result, the dispersion coefficient due to the transverse veloc- Research Council. As not all test reaches extend far enough to
ity shear was much larger than that due to the vertical velocity achieve a Fickian dispersion, Beltaos (1998) proposed a method
shear (Fischer et al., 1979). To compare the dispersion due to to estimate the longitudinal dispersion coefficient using a cor-
either vertical velocity shear or transverse velocity shear in ice- rection coefficient l, which was used to estimate characteristic
covered rivers, the longitudinal dispersion coefficients due to length and assumed values from 0.48 to 4.5 (Beltaos, 1979).
the vertical velocity shear for the eight hypothetical cases were Consequently, the top five tests were chosen for the validation
calculated using Eq. (1) and listed in column (10) of Table 1, of Eq. (17) based on the coefficient l, since their values were
with h(y) being the average water depth H. For all of the cases large enough to represent typical river lengths as well as the
in Table 1, the aspect ratios were greater than 5 and the ratios longitudinal dispersion coefficients. The selected test reaches
of DL2 to DL1 were greater than 50. Furthermore, the ratios had large lengths L ranging from 45 km to 140 km as well as
of DL2 to DL1 increased with increasing the width-to-depth large widths B from 60 m to 142 m, while the water depths H
ratio (Case 1–4). The results indicate that the transverse velo- were normally below 1 m (Table 2). Thus the width-to-depth
city shear played a much more important role than the vertical ratios were about or even above 100, suggesting that the tested
velocity shear in the process of longitudinal dispersion in ice- reaches were all wide and shallow, and thus Eq. (17) could be
covered rivers where the width-to-depth ratio is usually greater used to evaluate the longitudinal dispersion coefficients. The
than 10. Consequently, the longitudinal dispersion coefficient in experimental longitudinal dispersion coefficients and the dis-
ice-covered rivers could be predicted using Eq. (17), in which persion coefficient predicted by Eq. (17) are listed in columns
only the transverse velocity shear was taken into account and (9) and (10) of Table 2, respectively. The discrepancies between
the vertical velocity shear was ignored. the values calculated by Eq. (17) and the experimental results
It is difficult to validate an analytical formula of the lon- are smaller than 150%, column (12) of Table 2. The agreement
gitudinal dispersion coefficient against a tracer dye test in is fairly good, bearing in mind that Eq. (17) was derived from
natural rivers, because such tracer tests could only be car- Eq. (2), which is based on the assumption of uniform straight
ried out during ice-cover periods, not to mention the effects flow while real rivers are always sinuous. Besides, the main
of potential storage zones. The assumption of parabolic cross channels of cross sections were not exactly parabolic, and the
Journal of Hydraulic Research (2016) Dispersion in ice-covered rivers 7

Table 2 Characteristics and longitudinal dispersion coefficients of the test reaches in ice-covered conditions (data from Beltaos, 1998)

DL2 (m2 /s)

Case L (km) B (m) H (m) S (10−3 ) n (m−1/3 s) U (m s−1 ) Dtt (m2 s−1 ) Expt. Eq. (17) Eq. (18) Error* (%)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12)

EC1 102 105 0.65 1.26 0.035 0.48 0.0016 ≥ 982.0 2282.5 678.6 132
EC2 140 128 0.94 0.82 0.049 0.35 0.0022 ≥ 736.3 1285.5 382.1 75
EC3 45 60 0.90 0.81 0.070 0.24 0.0020 ≥ 162.8 142.6 42.4 12
EC4 121 116 1.44 0.79 0.073 0.31 0.0040 ≥ 828.9 445.0 132.3 46
EC5 71 142 1.00 0.93 0.060 0.32 0.0025 ≥ 1136.0 1131.7 336.5 0

*Error = |Eq. (17) – Expt.|/Expt., is the absolute value of the dispersion coefficient difference between prediction (Eq. (17)) and field test (Expt.)
divided by the measured dispersion coefficient.

experimental results from field tracer tests were not completely Thirdly, the prediction of the longitudinal dispersion coefficient
accurate since the dispersion coefficients cannot be measured was easier in ice-covered rivers. Compared with so many irregu-
Downloaded by [University of Lethbridge] at 21:13 11 June 2016

directly. larities in ice-free rivers, the cross section shapes are simpler and
Equation (18) was also used to predict the measured dis- a significant proportion of ice-covered rivers can be expected to
persion coefficients and the results are listed in column (11) be parabolic. Consequently, the longitudinal dispersion coeffi-
of Table 2. Clearly Eq. (18) underestimated the dispersion cient in ice-covered rivers could be calculated by an analytical
coefficients for all cases. Such underestimation implies that ice- solution with an improved precision. Nevertheless, new and
covered rivers had larger dispersion coefficients than ice-free accurate tracer tests in both real rivers and laboratory flumes
rivers, which was also concluded by Beltaos (1998). As men- are still needed.
tioned earlier, the dispersion coefficient due to the vertical velo-
city shear in ice-covered rivers was smaller than that in ice- 5 Conclusions
free rivers. In contrast, the dispersion coefficient due to the
transverse velocity shear in ice-covered rivers was larger than Under the assumptions that the cross section is parabolic and the
that in ice-free rivers. However, the transverse velocity shear velocity profiles in both the riverbed zone and the ice-cover zone
dominated the dispersion process in wide-and-shallow ice- follow the logarithmic law at each location of the cross section,
covered rivers. Recalling the derivation process of Eq. (17), the longitudinal dispersion coefficients due to the vertical veloc-
we can find that it is the transverse mixing, rather than the ity shear and the transverse velocity shear were calculated for
transverse velocity shear, that contributed to the difference ice-covered rivers. The calculation results showed that, for the
of the dispersion coefficients in ice-covered rivers and ice- longitudinal dispersion process in wide and shallow ice-covered
free rivers. Hence, smaller transverse mixing capacity resulted rivers, the transverse velocity shear was dominant and the effect
in a larger longitudinal dispersion coefficient in ice-covered of vertical velocity shear could be ignored. Then, an analytical
rivers. expression (Eq. (17)) was proposed to evaluate the longitudinal
Seo and Cheong (1998) stated that Eq. (18), compared with dispersion coefficient in wide and shallow ice-covered rivers,
the other theoretical and empirical equations available for real by simplifying the triple integral method developed by Fischer
ice-free rivers, predicts the dispersion coefficients better, within et al. (1979). The formula was verified by the measured data
400% error if compared to measured data. Therefore, Eq. (17) for the dispersion coefficients of several field tracer tests in
should be considered as a significant improvement in the current ice-covered rivers, with the discrepancy being less than 150%.
predictive capabilities as the discrepancy between the predicted A comparison between the present formula and Eq. (18), pro-
and the measured dispersion coefficients for ice-covered rivers posed by Fischer (1975) for ice-free rivers, showed that the
is smaller than 150%. The relatively good performance of the dispersion coefficient in ice-covered rivers was larger than that
equations proposed in the present study is due to three factors. in ice-free rivers under the same conditions. It is the trans-
Firstly, compared to hundreds of field tracer tests in ice-free verse mixing capacity, rather than the transverse velocity shear,
rivers, only a limited number of field tracer tests were performed that makes the dispersion in ice-covered rivers greater than that
investigating longitudinal dispersion in ice-covered rivers. Even in ice-free rivers. The prediction of the longitudinal disper-
worse, few theoretical or empirical works could be found. sion coefficient in ice-covered rivers was relatively accurate,
Secondly, the measurement of the longitudinal dispersion coef- since the flow was nearly uniform by being restricted in the
ficient was more accurate in ice-covered rivers. As mentioned main channels. Therefore, the proposed analytical expression
above, the flow was nearly uniform since it was small enough to could be a good tool to estimate the longitudinal dispersion in
be restricted in the main channel during the ice-covered period. ice-covered rivers.
8 Y. Chen et al. Journal of Hydraulic Research (2016)

Acknowledgements u* = dimensionless velocity (–)


u∗ = global shear velocity of the cross section,

The authors gratefully acknowledge the topography measure- u∗ = gRS, (m s−1 )
ment conducted by the Hydrological Bureau of Mudanjiang, u∗b = local shear velocity of riverbed zone,

China, during the ice-covered period. The authors also appre- u∗b = gHb S, (m s−1 )
ciate the helpful comments and suggestions of the anonymous u∗i = local shear
velocity of ice-cover zone,
reviewers. u∗i = g(H − Hb )S, (m s−1 )
û = deviation of local velocity from the cross-sectional
mean (m s−1 )
Funding
U = cross-sectional average velocity (m s−1 )
Uv = local depth-averaged velocity (m s−1 )
The work was partly supported by the National Natural Sci-
y = transverse coordinate (m)
ence Foundation of China [numbers 51039002 and 51279078],
z = vertical coordinate (m)
the Program for New Century Excellent Talents in Univer-
ε = effect coefficient of ice-cover (–)
sity [2012490811] and the Program for Changjiang Scholars
κ = the von Karman constant (–)
and Innovative Research Team in University of Ministry of
ξ = dimensionless transverse coordinate (–)
Education of China [IRI13025].
χ = wetted perimeter (m)
Downloaded by [University of Lethbridge] at 21:13 11 June 2016

Notation
References
A = cross-sectional area (m2 )
B = river width (m) Atkinson, T. C., & Davis, P. M. (2000). Longitudinal disper-
DL = longitudinal dispersion coefficient (m2 s−1 ) sion in natural channels: 1. Experimental results from the
DL1 = longitudinal dispersion coefficient due to vertical River Severn, UK. Hydrology and Earth System Sciences,
velocity shear (m2 s−1 ) 4(3), 345–353.
DL2 = longitudinal dispersion coefficient due to transverse Banks, R. (1974). A mixing cell model for longitudinal dis-
velocity shear (m2 s−1 ) persion in open channels. Water Resources Research, 10(2),
Dt = vertical turbulent diffusivity (m2 s−1 ) 357–358.
Dtt = transverse mixing coefficient (m2 s−1 ) Beltaos, S. (1979). Mixing characteristics of the Athabasca
g = acceleration due to gravity (m s−2 ) river below Fort McMurray winter conditions. Preparation
h(y) = local water depth at transverse coordinate y (m) for the Alberta Oil Sands Environmental Research Program
hb = local water column height of river bed zone (m) by Transportation and Surface Water Engineering Division,
h* = dimensionless water depth (–) Alberta Research Council.
H = mean water depth (m) Beltaos, S. (1998). Longitudinal dispersion in ice-covered
Hb = local water column height of riverbed zone where rivers. Journal of Cold Regions Engineering, 12(4), 184–201.
h(y) = H (m) Bencala, K. E., & Walters, R. A. (1983). Simulation of solute
Hm = maximum water depth (m) transport in a mountain pool-and riffle stream: A transient
I = integrated parameter (–) storage model. Water Resources Research, 19(3), 718–724.
k = dimensionless coefficient for transverse mixing Boano, F., Packman, A. I., Cortis, A., Revelli, R., & Ridolfi, L.
coefficient (–) (2007). A continuous time random walk approach to the
Kb = equivalent roughness height of the surface of river stream transport of solutes. Water Resources Research, 43,
bed (m) W10425.
Ki = equivalent roughness height of the bottom of ice Chen, Y. C., & Zhu, D. J. (2007). Study on longitudinal dis-
cover (m) persion coefficient in trapezoidal cross-section open channels.
n = composited Manning’s roughness coefficient (m−1/3 s) Proceedings of 32nd International Association of Hydraulic
nb = Manning’s roughness coefficient of the surface of river Engineering & Research (IAHR) congress, Venice, Italy,
bed (m−1/3 s) A1.a-010-O.
ni = Manning’s roughness coefficient of the bottom of ice Clark, J. F., Schlosser, P., Stute, M., & Simpson, H. J.
cover (m−1/3 s) (1996). SF6–3He tracer release experiment: A new method
r = ratio of ice roughness coefficient to river bed roughness of determining longitudinal dispersion coefficients in large
coefficient (–) rivers. Environmental Sciences & Technologies, 30(5),
R = hydraulic radius (m) 1527–1532.
S = hydraulic slope (–) Davis, P. M., & Atkinson, T. C. (2000). Longitudinal disper-
u = local mean velocity (m s−1 ) sion in natural channels: 3. An aggregated dead zone model
Journal of Hydraulic Research (2016) Dispersion in ice-covered rivers 9

applied to the River Severn, UK. Hydrology and Earth System Ng, C. O. (2000). Dispersion in sediment-laden stream flow.
Sciences, 4(3), 373–381. Journal of Engineering Mechanics, 126(8), 779–786.
Davis, P. M., Atkinson, T. C., & Wigley, T. M. L. (2000). Lon- Perucca, E., Camporeale, C., & Ridolfi, L. (2009). Estimation
gitudinal dispersion in natural channels: 2. The roles of shear of the dispersion coefficient in rivers with riparian vegetation.
flow dispersion and dead zones in the River Severn, UK. Advances in Water Resources, 32(1), 78–87.
Hydrology and Earth System Sciences, 4(3), 355–371. Seo, I. W., & Baek, K. O. (2004). Estimation of the longi-
Deng, Z. Q., & Chu, J. D. (2001). Longitudinal dispersion coef- tudinal dispersion coefficient using the velocity profile in
ficient in natural rivers. Advances in Water Science, 12(2), natural streams. Journal of Hydraulic Engineering, 130(3),
137–142 (in Chinese). 227–236.
Deng, Z. Q., Singh, V. P., & Bengtsson, L. (2001). Longitudinal Seo, I. W. & Cheong, T. S. (1998). Predicting longitudinal dis-
dispersion coefficient in straight rivers. Journal of Hydraulic persion coefficient in natural streams. Journal of Hydraulic
Engineering, 127(11), 919–927. Engineering, 124(1), 25–32.
Dolgopolova, E. N. (2008). Vertical transfer coefficient in natu- Sukhodolov, A. N., Nikora, V. I., Rowinski, P. M., &
ral streams. Water Resources, 35(4), 408–416. Czernuszenko, W. (1997). A case study of longitudinal
Elder, J. W. (1959). The dispersion of a marked fluid in turbulent dispersion in small lowland rivers. Water Environment
shear flows. Journal of Fluid Mechanics, 5(4), 544–560. Research, 69(7), 1246–1253.
Downloaded by [University of Lethbridge] at 21:13 11 June 2016

Engmann, J. E. O., & Kellerhals, R. (1974). Transverse mix- Sukhodolov, A. N., Thiele, M., Bungartz, H., & Engelhardt,
ing in an ice-covered river. Water Resources Research, 10(4), C. (1999). Turbulence structure in an ice-covered, sand bed
775–784. river. Water Resources Research, 35(3), 889–894.
Fischer, H. B. (1975). Discussion of “simple method for pre- Sukhodolov, A. N., & Uijttewaal, W. S. J. (2010). Assessment of
dicting dispersion in streams”. Journal of the Environmental river reach for environmental fluid dynamics studies. Journal
Engineering Division-ASCE, 101(3), 453–455. of Hydraulic Engineering, 136(11), 880–888.
Fischer, H. B., List, E. J., Koh, R. C. Y., Imberger, J., & Sukhodolov, T., Sukhodolov, A., Kozerski, H-P., & Kohler,
Brooks, N. H. (1979). Mixing in inland and coastal waters. J. (2006). Longitudinal dispersion in a lowland river
New York: Academic Press, 104–138. with submersed vegetation. Proceeding of 3rd Interna-
Haggerty, R., & Gorelick, S. M. (1995). Multiple-rate mass tional conference on fluvial hydraulics. Lisbon, Portugal,
transfer for modeling diffusion and surface reactions in media 631–637.
with pore-scale heterogeneity. Water Resources Research, Taylor, G. I. (1954). The dispersion of matter in turbulent flow
31(10), 2383–2400. through a pipe. Proceedings of the Royal Society of London,
Huang, G., & Law, A. W. K. (2011). Taylor dispersion of A, 223, 446–468.
contaminants by random waves. Journal of Engineering Teal, M. J., Ettema, R., & Walker, J. F. (1994). Estimation
Mathematics, 70(4), 389–397. of mean flow velocity in ice-covered channels. Journal of
Lau, Y. L., & Krishnappan, B. G. (1980). Ice cover effects on Hydraulic Engineering, 120(12), 1385–1400.
stream flows and mixing. Journal of Hydraulics Division, Wang, Z. G., Chen, Y. C., Zhu, D. J., & Liu, Z. W. (2013).
107(10), 1225–1242. Study on longitudinal dispersion caused by vertical shear in
Marion, A., & Zaramella, M. (2005). A residence time model ice-covered rivers. Proceedings of 2012 SREE conference
for stream-subsurface exchange of contaminants. Acta Geo- on hydraulic engineering (CHE 2012), SREE, Hong Kong,
physica Polonica, 53(4), 527–538. China, 23–28.
Marion, A., & Zaramella, M. (2006). Effects of velocity gra- Zeng, Y. H., & Huai, W. X. (2014). Estimation of longi-
dients and secondary flow on the dispersion of solutes in tudinal dispersion coefficient in rivers. Journal of Hydro-
a meandering channel. Journal of Hydraulic Engineering, environment Research, 8(1), 2–8.
132(12), 1295–1302. Zhang, W., & Zhu, D. Z. (2011). Transverse mixing in an unreg-
Meyer, Z. (2009). An analysis of the mechanism of flow in ice- ulated northern river. Journal of Hydraulic Engineering,
covered rivers. Acta Geophysica, 58(2), 337–355. 137(11), 1426–1440.

You might also like