See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/319075141

The equilibrium alluvial river under variable flow and its channel-forming
discharge

Article in Journal of Geophysical Research: Earth Surface · August 2017

DOI: 10.1002/2017JF004213

CITATIONS READS

15 292

4 authors:

Astrid Blom Liselot Arkesteijn

Delft University of Technology Delft University of Technology
50 PUBLICATIONS 479 CITATIONS 8 PUBLICATIONS 39 CITATIONS

SEE PROFILE SEE PROFILE

Victor Chavarrias Enrica Viparelli

Delft University of Technology University of South Carolina
11 PUBLICATIONS 51 CITATIONS 42 PUBLICATIONS 812 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Bed degradation in rivers View project

Bed surface texture in laboratory experiments View project

All content following this page was uploaded by Astrid Blom on 21 August 2017.

The user has requested enhancement of the downloaded file.

 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. ???, XXXX, DOI:10.1002/,

1 The equilibrium alluvial river under variable flow,

2 and its channel-forming discharge
1 1 1 2
Astrid Blom , Liselot Arkesteijn , Vı́ctor Chavarrı́as , Enrica Viparelli

1
Faculty of Civil Engineering and

Geosciences, Delft University of Technology,

Delft, Netherlands.

2
Department of Civil and Environmental

Engineering, University of South Carolina

at Columbia, USA.

D R A F T July 7, 2017, 11:40pm D R A F T

 X-2 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

3 Abstract. When the water discharge, sediment supply, and base level vary

4 around stable values, an alluvial river evolves toward a mean equilibrium or

5 graded state with small fluctuations around this mean state (i.e. a dynamic

6 or statistical equilibrium state). Here we present analytical relations describ-

7 ing the mean equilibrium geometry of an alluvial river under variable flow

8 by linking channel slope, width, and bed surface texture. The solution holds

9 in river normal flow zones (or outside the hydrograph boundary layer and

10 the backwater zone) and accounts for grain size selective transport and par-

11 ticle abrasion. We consider the variable flow rate as a series of continuously

12 changing yet steady water discharges (here termed an alternating steady dis-

13 charge). The analysis also provides a solution to the channel-forming water

14 discharge, which is here defined as the steady water discharge that, given the

15 mean sediment supply, provides the same equilibrium channel slope as the

16 natural long-term hydrograph. The channel-forming water discharge for the

17 gravel load is larger than the one associated with the sand load. The anal-

18 ysis illustrates how the load is distributed over the range of water discharge

19 in the river normal flow zone, which we term the ‘normal flow load distri-

20 bution’. The fact that the distribution of the (imposed) sediment supply spa-

21 tially adapts to this normal flow load distribution is the origin of the hydro-

22 graph boundary layer. The results quantify the findings by Wolman and Miller

23 [1960] regarding the relevance of both magnitude and frequency of the flow

24 rate with respect to channel geometry.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X-3

Key points

25 • We derive two relations linking the equilibrium channel slope, width, and surface

26 texture under variable flow in river normal flow zones.

27 • We define the channel-forming discharge as the steady discharge providing the same

28 equilibrium channel slope as the long-term hydrograph.

29 • The hydrograph boundary layer originates from the adaptation of the sediment sup-

30 ply toward its ‘normal flow load distribution’.

31

1. Introduction

32 The flow rate and sediment load in natural streams vary widely in space and time, as

33 precipitation and sediment supply from hillslopes and tributaries vary spatially and tem-

34 porally within the basin. In general the downstream water surface base level also changes

35 in time due to tides, storm surge, and, on a larger time scale, sea level change.

36 Despite such temporal variation, the river continuously adapts through changes in plan-

37 form, bed elevation, and bed surface texture, and approaches its graded or equilibrium

38 longitudinal profile [Gilbert, 1877; Mackin, 1948; Lane, 1955; Ahnert, 1994; Buffington,

39 2012; Blom et al., 2016]. Following Blom et al. [2016] we define the graded or equilibrium

40 river profile as the mean profile that the river approaches when flow, sediment supply,

41 and base level vary around stable values for a long time in the absence of subsidence or

42 uplift. The fact that a graded state in terms of channel slope and bed surface texture

43 also develops when the flow rate, the sediment supply, and the downstream base level

44 vary around stable values [e.g., Pickup and Rieger , 1979] is confirmed through laboratory

45 experiments [Wilcock et al., 2001], indirect analysis of field data [Wilcock and DeTemple,

D R A F T July 7, 2017, 11:40pm D R A F T

 X-4 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

46 2005], and numerical runs [Parker , 2004a; Wong and Parker , 2006; Parker et al., 2008;

47 Viparelli et al., 2011; Bolla Pittaluga et al., 2014]. This graded state is characterized by

48 a mean state with small fluctuations around it, which is termed a statistical equilibrium

49 or steady state equilibrium [Chorley and Kennedy, 1971, p.202] or dynamic equilibrium

50 [De Vries, 1993; Ahnert, 1994; Zhou et al., 2017] and needs to be considered over a period

51 of years or decades.

52 Variability of the flow rate due to flood waves is essential for, for instance, floodplain-

53 related mechanisms [e.g., Lauer and Parker , 2008; Lauer et al., 2016; Viparelli et al., 2013;

54 Fleischmann et al., 2016] and plant colonization on exposed bars [Tal and Paola, 2010;

55 Vargas-Luna, 2016]. Although extreme events may induce changes to the river system

56 that moderate flows cannot [Buffington, 2012], for instance the formation of new gullies,

57 the creation of meander cut-offs, floodplain scour, and the movement of huge boulders

58 [Wolman and Miller , 1960; Pickup and Warner , 1976; Lenzi et al., 2006], flow events of

59 moderate magnitude, which recur relatively frequently, are more effective in transporting

60 sediment than rare events of unusual magnitude [Wolman and Miller , 1960].

61 Various approaches have been used to formulate a representative steady water discharge

62 that, given enough time, would produce the same channel geometry as the natural long-

63 term hydrograph [e.g., NEDECO, 1959; Prins and De Vries, 1971; Pickup and Warner ,

64 1976; Pickup and Rieger , 1979; Hey, 1996; Emmett and Wolman, 2001; Copeland et al.,

65 2005]. This natural long-term hydrograph is here defined as the hydrograph that covers

66 the flow rate statistics of several years or decades. The concept of a channel-forming

67 discharge was first introduced in the design of irrigation channels, in which the variation

68 of the flow rate is generally limited [e.g., Lacey, 1930]. Application of this concept to

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X-5

69 the design, maintenance, and restoration of natural streams, which are subject to a much

70 wider range of flow rates, initiated a search for a single value of the water discharge

71 representative in its effect on channel geometry [Pickup and Rieger , 1979; Copeland et al.,

72 2005; Gomez et al., 2007]. Although the terminology has not always been consistent, the

73 following representative discharges have been proposed:

74 1. the bankfull discharge, which is the flow rate at which the channel is just filled to

75 the top of its banks and which therefore agrees to the condition of incipient flooding [e.g.,

76 Wolman and Leopold , 1957; Williams, 1978; Parker , 1978a, b; Emmett and Wolman,

77 2001; Gomez et al., 2007; Parker et al., 2007; Phillips and Jerolmack , 2016]. The bank-

78 full discharge is considered morphodynamically relevant as it represents the breakpoint

79 between processes of channel formation and floodplain formation [Copeland et al., 2005]

80 and above which the shear stress almost stops increasing with increasing discharge;

81 2. the effective discharge, which is the water discharge at which the product of the

82 probability density of the water discharge and the associated sediment load reaches a

83 maximum value [e.g., Wolman and Miller , 1960; Benson and Thomas, 1966; Andrews,

84 1980; Nash, 1994; Carling, 1988; Emmett and Wolman, 2001; Goodwin, 2004];

85 3. the discharge associated with a certain recurrence interval or a designated flood peak

86 frequency [e.g., Wolman and Leopold , 1957; Bray, 1975; Doyle et al., 2007];

87 4. the formative discharge for meanders, which is the steady water discharge that pro-

88 vides the same meander length as the long-term hydrograph [e.g., Benson and Thomas,

89 1966; Ackers and Charlton, 1970], or other morphodynamic units such as bars and islands

90 [Surian et al., 2009];

D R A F T July 7, 2017, 11:40pm D R A F T

 X-6 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

91 5. the slope-equivalent water discharge, which is the steady water discharge that, given

92 the mean sediment supply, results in the same equilibrium channel slope as the long-term

93 hydrograph [based on the approach by De Vries, 1974, 1993; Jansen et al., 1979; Howard ,

94 1980] or, equivalently, the steady water discharge that for a given slope provides the same

95 sediment flux as the long-term hydrograph [Doyle and Shields, 2008, their functional-

96 equivalent discharge];

97 6. the half-load discharge, which is the water discharge above and below which 50 %

98 of the total sediment load is transported [Van Bendegom, 1967; Prins, 1969; Vogel et al.,

99 2003; Klonsky and Vogel , 2011];

100 7. the channel-forming flood, which is the average of all water discharges at which gravel

101 is mobilized weighted by the probability of their occurrence [Phillips and Jerolmack , 2016].

102 Yet, despite the efforts to find an estimate of a single representative discharge, it has

103 also been realized that no single steady discharge can affect all characteristics of channel

104 geometry in a similar manner as the varying flow rate [NEDECO, 1959; Prins, 1969; Prins

105 and De Vries, 1971; De Vries, 1971, 1993; Biedenharn et al., 2008]. This calls for a precise

106 definition of the type of representative discharge one is considering [Prins and De Vries,

107 1971]: exactly what morphodynamic aspect should be reproduced by the representative

108 discharge?

109 Our objective is to determine relations linking the equilibrium channel slope, channel

110 width, and bed surface texture under variable flow, as well as to provide a definition

111 of the slope-equivalent or channel-forming discharge. To this end we follow the analysis

112 introduced by De Vries [1974, p.31-32], which is more accessible in Jansen et al. [1979,

113 p.119-121], De Vries [1993], Howard [1980], and Doyle and Shields [2008]. We extend

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X-7

114 their analysis to mixed-size sediment conditions. Our analysis also is an extension of the

115 Blom et al. [2016] equations for the equilibrium geometry of an alluvial river to a variable

116 flow rate.

117 In the next section we explain how we treat the variable flow rate in our analysis and

118 propose relations to determine the mean sediment load under a variable flow rate with

119 temporally constant statistics. In Section 3 we describe the equilibrium river geometry

120 under variable flow and unisize sediment conditions. To this end we consider a variable

121 flow rate that at each time is governed by steady flow. We extend this analysis to mixed

122 sediment by accounting for the mechanisms of grain size selective transport and particle

123 abrasion in Section 4. In Section 5 we validate the analytical model by comparing its

124 results for the equilibrium river geometry to those of two numerical time-marching mod-

125 els, one based on the backwater equation and one on the Saint-Venant [1871] equations.

126 We discuss channel response to changing boundary conditions through slope and width

127 adjustment in Section 6. In Section 7 we illustrate how the analytical model allows for

128 predicting the relative contribution of the long-term range of flow rate values to trans-

129 porting the (gravel and sand) load, which we term the ‘normal flow load distribution’.

130 In Section 8 we address the differences between the slope-equivalent discharge and the

131 effective discharge defined by Wolman and Miller [1960].

2. Hydrograph treatment and formulations for the mean load

132 As we focus on a dynamic or statistical steady state, we model the variability of the

133 water discharge, Qw [m3 /s], in terms of a changing flow rate of which the statistics are

134 temporally constant and expressed by the probability density function (PDF) of water

135 discharge, fw (Qw ) [s/m3 ].

D R A F T July 7, 2017, 11:40pm D R A F T

 X-8 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

136 Now let us consider the arbitrary probability density function of water discharge shown

137 in Figure 1a. The integral of the PDF by definition equals unity. The expected or mean

138 sediment load, E[Q] or Q̄, equals

∫ ∞
139 E[Q] ≡ Q̄ = Q(Qw )fw (Qw )dQw (1)
0

140 which is equivalent to Eq. (3.3-31) in Jansen et al. [1979], Eq. (2-31) in De Vries [1993],

141 and Eq. (4) in Doyle and Shields [2008]. The bar indicates the fact that we average over

142 a significant period and hence deal with the expected or mean value.

143 For illustration purposes we simplify the PDF of water discharge to a discrete distribu-

144 tion with two modes: a base flow rate, Qwbase , and a peak flow rate, Qwpeak (Figure 1b).

145 The probability that base flow occurs is denoted by α [-], and for a two modes case

146 the probability that peak flow occurs, by definition, equals 1 − α. For this two-modes

147 distribution of the flow rate Eq. (1) is simplified to

148 Q̄ = αQbase + (1 − α)Qpeak (2)

149 where the capacity-based sediment load under base and peak flow, Qbase and Qpeak , are

150 defined as Q(Qwbase ) and Q(Qwpeak ), respectively.

151 Similarly, under conditions with two sediment modes, gravel and sand, the mean gravel

152 and sand load, Q̄g and Q̄s , are equal to

∫ ∞
153 Q̄g = Qg (Qw )fw (Qw )dQw (3)
0
∫ ∞
154 Q̄s = Qs (Qw )fw (Qw )dQw (4)
0

155 where the subscripts g and s indicate gravel and sand, respectively. Eqs. (3) and (4) are

156 simplified to

157 Q̄g = αQgbase + (1 − α)Qgpeak (5)

D R A F T July 7, 2017, 11:40pm D R A F T
 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X-9

158 Q̄s = αQsbase + (1 − α)Qspeak (6)

159 for the two-modes water discharge regime of Figure 1b.

3. Equilibrium river geometry for unisize sediment

160 To describe the equilibrium river geometry under a variable flow rate, we first consider the

161 case of unisize sediment. For that purpose we combine Eq. (1) or (2) with a formulation

162 for the sediment load as a function of the flow rate. After listing our simplifications

163 in Section 3.1, we apply a power law load relation which leads to an explicit analytical

164 solution to the equilibrium river geometry (Section 3.2), as well as more complex load

165 relations leading to implicit analytical solutions (Section 3.3).

3.1. Simplifications and the ‘alternating steady flow’ assumption

166 We make a number of simplifying assumptions to obtain an analytical solution to the

167 equilibrium river geometry: (1) our model is one-dimensional; (2) we neglect subsidence

168 and uplift, base level change, and delta outbuilding; (3) we assume a rectangular cross-

169 section of width, B [m], and do not distinguish between flow through the main channel

170 and the floodplains; (4) we consider a relatively wide channel, which implies that we can

171 assume that the hydraulic radius equals the flow depth, H [m]; (5) we assume hydrostatic

172 pressure in the water column; (6) we assume the nondimensional friction coefficient, Cf

173 [-], to be independent of the local flow parameters and the surface texture; (7) we neglect

174 changes in sediment porosity; (8) we consider bed-material load only [i.e. bed load as

175 well as suspended bed-material load, Paola, 2001; Church, 2006], and (9) we neglect the

176 spatial lag in the adaptation to capacity-based transport rates [Phillips and Sutherland ,

177 1989].

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 10 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

178 A high precipitation rate and the associated temporally increased flow rate lead to a

179 flood wave that is advected downstream while diffusing. Consequently the water discharge

180 varies also spatially, which induces small temporal change in bed elevation. The fact that

181 bed elevation change under a varying flow rate is small can also be explained as follows.

182 In general the time scale of change in channel slope is much larger than the one of change

183 in the flow rate [Mackin, 1948; De Vries, 1974; Jansen et al., 1979; Howard , 1980, 1982;

184 De Vries, 1993]. In other words, there is insufficient time for the channel slope to adjust to

185 short-term changes in the flow rate. Furthermore, any change in slope due to the variable

186 flow rate starts at the domain boundaries. This reasoning is confirmed by the analyses of

187 Parker et al. [2008], Viparelli et al. [2011], and An et al. [2017].

188 In the lower course of a river the flood wave is typically longer than in its upper course,

189 as it has diffused while migrating downstream. Here we assume that the flood wave is

190 infinitely long and propagates infinitely fast, which implies that it does not dampen. Thus,

191 although we do consider the temporal variability of the flow rate, we do not consider the

192 dynamics of a flood wave. In other words, we assume the flow is steady for each value of

193 the flow rate, and term this an ‘alternating steady flow’.

194 We divide the fluvial reach in two types of zones: backwater zones that are controlled

195 by a downstream water surface base level and normal flow zones where the flow is not

196 affected by the base level (Figure 2). A ‘river normal flow zone’ is defined as a reach where

197 the flow can be reasonably approximated as uniform or normal, even under a variable flow

198 rate. This implies that for all flow rates the friction slope is more or less equal to the bed

199 slope and that, by definition, bed elevation changes are zero. Although Figure 2 shows one

200 normal flow zone and one backwater zone, a river typically consists of multiple backwater

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 11

201 and normal flow zones, yet backwater effects may prevail such that a strictly normal flow

202 zone is absent. The analysis presented in this paper is applicable to river normal flow

203 zones only and therefore we consider a static equilibrium state.

3.2. An explicit analytical model

204 We first apply a power law load relation without a threshold for significant transport

205 [e.g., Engelund and Hansen, 1967; Parker , 1990a, b; Syvitski et al., 2000; Emmett and

206 Wolman, 2001; Wilcock and Crowe, 2003; Goodwin, 2004; Barry et al., 2004, 2007; Doyle

207 and Shields, 2008; Houssais et al., 2015], which reflects Einstein’s [1950] probabilistic

208 philosophy regarding the nature of sediment transport.

209 We now summarize existing power law load relations into a simple generalized load

210 relation for unisize sediment:

( )r
∗ D
211 W =c τ ∗w (7)
Dref

212 where W ∗ [-] denotes the nondimensional sediment transport rate [W ∗ = q ∗ /τ ∗3/2 with

213 q ∗ denoting the nondimensional Einstein transport number, see Parker , 2004b, or Ap-

214 pendix A], τ ∗ [-] is the grain related Shields stress, c [-], w [-], and r [-] are constants,

215 and D [m] is the representative grain size of the sediment. We introduce the parameter

216 Dref [m] to nondimensionalize the right-hand term of Eq. (7) (Dref = 0.001 m). Eq. (7)

217 is simplified to the Engelund and Hansen [1967] load relation by setting w = 1, r = 0,
3/2 5/2
218 and c = 0.05Cf /Cf s where Cf s [-] denotes the nondimensional skin friction coefficient.

219 Appendix A provides more detailed information on the generalized load relation.

220 We apply Eq. (7) to estimate the sediment transport rates at base and peak flow rates,

221 Qbase and Qpeak , that are required in Eq. (2). To this end, we combine the formulation for

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 12 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

222 the sediment load, Eq. (7) or (A3), with the continuity equation for the flow (Qw = BU H

223 with U [m/s] denoting the mean flow velocity), and we use the normal flow equation,

224 H = (Cf Q2w /(SB 2 g))1/3 , to eliminate the flow depth H:

( )(2w+3)/3
G (D/Dref )r gSQw
225 Q= w (8)
D B 2w/3 Cf

226 where S [-] denotes channel slope, g [m/s2 ] the gravitational acceleration, and G [s4 /m2 ]

227 is a constant in the load relation (Eq. (A3b) in Appendix A).

228 We use Eq. (8) to replace Qbase and Qpeak in Eq. (2) and assume that the equilibrium

229 channel slope, S, does not vary with the varying flow rate (Section 3.1). We then find

230 the following equation for the equilibrium channel slope under variable flow in a reach

231 governed by normal flow:

( )3/(2w+3)
Cf B 2w/(2w+3) Dw Q̄
232 S= (9)
gQwdom G (D/Dref )r

233 where the slope-equivalent discharge, Qwdom , for the case of a two modes discharge is

234 given by

( )
(2w+3)/3 (2w+3)/3 3/(2w+3)
235 Qwdom = αQwbase + (1 − α)Qwpeak (10)

236 We generalize Eq. (10) to the case of the arbitrary PDF of water discharge in Figure 1a:
[∫ ∞ ]3/(2w+3)
237 Qwdom = Q(2w+3)/3
w fw (Qw ) dQw (11)
0

238 which is an elaborated form of the functional-equivalent discharge defined by Doyle and

239 Shields [2008, their Eq. (6)].

240 The current analysis conveniently provides a definition of the slope-equivalent or

241 channel-forming discharge, Qwdom , for conditions with varying flow. This representa-

242 tive water discharge is the steady water discharge that, given the mean sediment supply,

243 results in the same equilibrium channel slope as the natural long-term hydrograph.
D R A F T July 7, 2017, 11:40pm D R A F T
 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 13

244 Naturally, under variable flow the flow depth depends on the flow stage at time t or

245 discharge mode Qw , and, outside backwater zones (Figure 2), is found by combining Eq.

246 (9) with the normal flow equation:

1/3 ( )1/(2w+3)
Qw (t)2/3 Qwdom G (D/Dref )r
247 H(t) = (2w+2)/(2w+3) (12)
B Dw Q̄

248 The flow velocity at time t, U (t), is found by combining Qw = BU H with Eq. (12):
( )1/3 ( )1/(2w+3)
Qw (t) Dw Q̄
249 U (t) = (13)
Qwdom G B (D/Dref )r

250 The equations governing the equilibrium river geometry under unisize sediment condi-

251 tions are summarized in Block 1 of Table 1. Applying the Engelund and Hansen [1967]

252 load relation, these equations reduce to the ones listed in Block 2. The slope-equivalent

253 discharges in Eqs. (10) and (11) are simplified to the ones listed in the upper right-hand

254 corner of Table 1.

255 The equations for the equilibrium channel slope and flow depth under a variable flow

256 rate are similar to the Blom et al. [2016] equations obtained for the constant water dis-

257 charge case. Noteworthy is the resulting definition of a slope-equivalent or channel-forming

258 discharge. The slope-equivalent discharge is larger than the mean water discharge, which

259 is due to the nonlinear relation between water discharge and the sediment transport rate.

260 The equations provide an order of magnitude of the relative change that can be expected

261 on the long term if one of the controls changes with time. Let us consider an engineered

262 river with fixed banks in which the mean sediment supply decreases by a factor 2 (for

263 instance, due to a dam). According to the Engelund and Hansen [1967] load relation and

264 assuming the initial state is an equilibrium state, Eq. (9) tells us that the slope decreases

265 of the order of 30% eventually.

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 14 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

266 Eq. (9) also illustrates how in the equilibrium state the channel slope, S, scales with:

(Dw−r Q̄)3/(2w+3)
267 S∝ (14)
Qwdom

268 (w = 1 and r = 0 if applying the Engelund and Hansen [1967] load relation) which is very

269 similar to the Lane [1955] relation for channel slope:

DQ
270 S∝ (15)
Qw

271 where here Q and Qw are representative values of the sediment and water discharge. Yet

272 formulations somewhat different from Eq. (14) will be found when applying other load

273 relations.

274 Slope adjustment arises through reach scale aggradation/degradation or changes in

275 channel sinuosity [Buffington, 2012]. In cases where the river responds to changes in the

276 boundary conditions through width adjustment rather than slope adjustment [Howard ,

277 1980], one may impose channel slope, S, and use Eq. (9) to compute the equilibrium

278 channel width, B. Alternatively, in cases where the river responds to changes through

279 a combination of slope and width adjustment, one can combine the current formulation

280 for the equilibrium channel slope in Eq. (9) with a closure relation and solve for both

281 the equilibrium channel slope and width. We will address this topic in more detail in

282 Sections 4.2 and 6.

3.3. An implicit analytical model

283 After using the power law load relation that has led to an explicit solution to the equilib-

284 rium river geometry we here apply more complex load relations, for instance one with a

285 threshold value for significant transport. This leads to implicit analytical solutions to the

286 equilibrium river geometry. Following the same procedure as described in the previous

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 15

287 section, yet using the load relation by, for instance Meyer-Peter and Müller [1948] or

288 Fernandez-Luque and Van Beek [1976], Eq. (8) is replaced by

  ( )1/3 b
√ Q2w S 2 Cf s
289 Q = aB RgDD max 0, − τc∗  (16)
Cf2 B 2 g RD

290 where R [-] is the submerged specific gravity (R = (ρs − ρ)/ρ in which ρs [kg/m3 ] and ρ

291 [kg/m3 ] are the sediment and water density, respectively). According to Meyer-Peter and

292 Müller [1948] the values of the nondimensional constants in the load relation are a = 8,

293 b = 1.5, and τc∗ = 0.047, and according to Fernandez-Luque and Van Beek [1976] a = 5.7,

294 b = 1.5, and τc∗ = 0.037 ∼ 0.0455. The maximization function is included to indicate

295 that the excess Shields stress cannot be negative. As an example we apply Eq. (16) to

296 elaborate on the base and peak flow sediment transport rates in Eq. (2):
  ( )1/3 b
Q̄ Q2wbase S 2 Cf s
297 √ = α max 0, − τc∗  + ...
aB RgDD Cf2 B 2 g RD
  ( )1/3 b
Q2wpeak S 2 Cf s
298
 
(1 − α) max 0, − τc∗  (17)
Cf2 B 2 g RD

299 and solve Eq. (17) iteratively for the equilibrium channel slope, S, or channel width, B,

300 depending on the unknown variable in the equation, or add a closure relation and solve

301 for both channel slope and channel width. Eq. (17) is easily generalized to account for

302 any number of discharge values or an arbitrary PDF of water discharge.

303 The slope-equivalent discharge, Qwdom , which is the steady water discharge that yields

304 the equilibrium channel slope S or channel width B computed from Eq. (17), can be

305 determined from

  ( )1/3 b
Q̄ Q2wdom S 2 Cf s
306 √ = max 0, − τc∗  (18)
aB RgDD Cf2 B 2 g RD

307 The implicit character of the solution procedure here arises from the threshold for signif-

308 icant transport. This procedure can be applied to any existing load relation and will be
D R A F T July 7, 2017, 11:40pm D R A F T
 X - 16 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

309 validated in Section 5. Before doing so, we will extend the model to mixed-size sediment

310 conditions.

4. Equilibrium river geometry for mixed sediment

4.1. Simplifications

311 In this section we extend the model of the equilibrium river geometry under variable yet

312 normal flow to conditions with mixed-size sediment. Following Ferguson [2003], Gasparini

313 et al. [2004], and Blom et al. [2016], we limit our analysis to sediment mixtures character-

314 ized by two distinct modes, gravel and sand. Besides the assumptions listed in Section 3.1,

315 the mixed-size sediment model consists of a number of additional simplifying assumptions

316 [Blom et al., 2016]: (1) all gravel particles entering the reach at the upstream end have

317 the same grain size (and the same holds for sand); (2) all sediment arrives from upstream

318 through fluvial sediment transport; (3) only gravel particles abrade, which implies that

319 the gravel size gradually decreases with streamwise coordinate x [m], expressed by the

320 symbol Dgx [m], and the sand diameter Ds [m] is constant; and (4) the products of abra-

321 sion are sand and silt, and the fraction of abrasion product that is sand (versus silt), kss

322 [-], is imposed.

4.2. An explicit analytical model

323 Analogous to the unisize sediment case, we first apply a generalized power law load relation

324 to the mixed sediment case:

( )r
Di
325 Wi∗ =c τi∗wi (19)
Dref

326 where Wi∗ [-] denotes the nondimensional sediment transport rate of grain size fraction

327 i (here limited to either gravel or sand), Di [m] the representative grain size of size

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 17

328 fraction i, the coefficient wi [-] a size fraction dependent parameter, and τi∗ [-] is the grain

329 related Shields stress associated with size fraction i. Appendix B provides more detailed

330 information on this generalized load relation.

331 As the transport relation in Eq. (19) does not include a threshold for significant trans-

332 port, hiding effects cannot be accounted for through adjusting the critical Shields stress,

333 yet the exponents r and wi can be adjusted to modify the grain size selectivity of the

334 load relation. Grain size selectivity increases with increasing values of wi − r. Varying wi

335 between the grain size classes i also adjusts the grain size selectivity. The load relation is

336 independent of grain size if wi = r. It reduces to the fractional form of the Engelund and

337 Hansen [1967] load relation applied by Blom et al. [2016] through setting wi = 1, r = 0,
3/2 5/2
338 and c = 0.05Cf /Cf s .

339 We apply the Hirano [1971] active layer equation to describe the conservation of mass of

340 gravel and sand at the bed surface [Ribberink , 1987; Parker , 1991; Stecca et al., 2014, 2016].

341 Following Blom et al. [2016] we add abrasion-related terms to the Hirano [1971] active

342 layer equation. Under graded or equilibrium conditions we can simplify these conservation

343 equations for the gravel and sand mass to [Blom et al., 2016]:

∗
344 Q̄g = p̄g0 Q̄0 e−xκ (20)
( ∗
)
345 Q̄s = Q̄0 − p̄g0 Q̄0 + kss p̄g0 Q̄0 1 − e−xκ (21)

346 where x∗κ [-] denotes the nondimensional streamwise coordinate: x∗κ = κβ(x − x0 ) with

347 the constant β [1/m] denoting the abrasion coefficient, and κ [-] is defined as κ = 1 + F

348 where F [-] denotes the volume fraction content of gravel at the bed surface. This implies

349 1 ≤ κ ≤ 2. For simplicity Blom et al. [2016] assume κ to be constant, which implies

350 that κ (just as β) is independent of x. Subscript 0 indicates the upstream end of the

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 18 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

351 considered reach. p̄g [-] is the mean gravel content in the sediment supply (p̄g = Q̄g /Q̄

352 and Q̄ = Q̄g + Q̄s ).

353 Eqs. (20) and (21) are not just the steady version of the Hirano [1971] active layer

354 equation with added abrasion terms [Blom et al., 2016], but also the steady version of

355 vertically-continuous sediment conservation models [e.g., Parker et al., 2000; Blom, 2008;

356 Viparelli et al., 2017] with added abrasion terms. The current analysis is therefore not

357 restricted to the application of the Hirano [1971] active layer equation.

358 If we neglect gravel particle abrasion, Eqs. (20) and (21) simplify to Q̄g = p̄g0 Q̄0 = Q̄g0

359 and Q̄s = Q̄s0 , which implies that the local gravel and sand load are equal to the gravel

360 and sand supply at the upstream end of the reach. We retain the effects of abrasion in our

361 subsequent analysis. Under conditions with abrasion the equilibrium river longitudinal

362 profile is concave upward and, as such, not governed by strictly normal flow [Blom et al.,

363 2016]. Yet the associated streamwise change is small enough to assume normal flow

364 conditions to hold.

365 Analogous to the procedure used to derive Eq. (8) or (16), we first apply Eq. (19)

366 or (B3) to predicting the gravel and sand load, we then combine the resulting equations

367 with the continuity equation for the flow and the normal flow equation, and we derive the

368 following relations for the gravel and sand load:

( )(2wg +3)/3
F (Dgx /Dref )r Gg gSQw
369 Qg = 2wg /3 w (22)
B Dgxg Cf
( )(2ws +3)/3
(1 − F ) (Ds /Dref )r Gs gSQw
370 Qs = (23)
B 2ws /3 Dsws Cf

371 where Gg and Gs are constants in the load relation defined in Eq. (B4). Besides assuming

372 changes in bed elevation due to the varying flow to be small, we also assume temporal

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 19

373 changes in surface texture to be small. This assumption is justified by the findings of

374 Parker and Klingeman [1982]; Parker et al. [1982], Wilcock et al. [2001], Parker [2004a],

375 Wilcock and DeTemple [2005], Parker et al. [2008], Ferrer-Boix and Hassan [2015], and

376 Litty and Schlunegger [2016].

377 We apply Eqs. (22) and (23) to compute the gravel and sand load during peak flow,

378 Qgpeak and Qspeak , and base flow, Qgbase and Qsbase . For reaches unaffected by backwater

379 effects, Eqs. (5) and (6) yield

( )(2wg +3)/3
F (Dgx /Dref )r Gg gSQwdom
380 Q̄g = 2wg /3 w (24)
B Dgxg Cf
( )(2ws +3)/3
1 − F (Ds /Dref )r Gs gSQwdom
381 Q̄s = 2ws /3 (25)
B Dsws Cf

382 where Qwdom is given by Eq. (10) and, for the arbitrary PDF of flow rate in Figure 1a, by

383 Eq. (11).

384 The combination of Eqs. (24) and (25) with Eqs. (20) and (21) while assuming the

385 coefficient wi to be independent of grain size (wg = ws = w and so Gg = Gs = G) yields

386 two equations describing the relation between the equilibrium channel slope, S, channel

387 width, B, and surface gravel fraction, F :

( )3/(2w+3)
Cf B 2w/(2w+3) Dsw µ Q̄0
388 S = (26)
gQwdom G (Dgx /Dref )r
( )w
1 Dgx ∗
389 F = p̄g0 e−xκ (27)
µ Ds

390 where
( )r ( ) ( )w
Dgx −x∗κ Dgx ∗
391 µ= 1 − p̄g0 + kss p̄g0 (1 − e ) + p̄g0 e−xκ (28)
Ds Ds

392 Combining the equation for the channel slope with the normal flow equation and the

393 continuity equation for the flow yields the formulations for the flow depth and flow velocity

394 at time t or discharge mode Qw . We have summarized the resulting formulations in Block
D R A F T July 7, 2017, 11:40pm D R A F T
 X - 20 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

395 5 in Table 1, and, for the Engelund and Hansen load relation, in Block 6 of Table 1. The

396 resulting equations without the effects of gravel particle abrasion (β = 0) are listed in

397 Blocks 3 and 4. For the case of a single steady discharge, the equations in Block 6 of

398 Table 1 reduce to the equations derived by Blom et al. [2016].

399 There is a drawback to assuming coefficient wi to be independent of grain size (wg =

400 ws = w), which also holds for the fractional form of the Engelund and Hansen load

401 relation. The assumption implies that, although sand is more mobile than gravel and the

402 load is finer than the bed surface (i.e. the sediment transport is grain size selective), the

403 predicted load does not coarsen with increasing Shields stress. The fact that the sediment

404 load coarsens with increasing Shields stress and that its grain size distribution becomes

405 closer to the one of the bed surface is shown in field cases [Kuhnle, 1992; Kuhnle and

406 Willis, 1992; Wathen et al., 1995] and in laboratory experiments [Van der Scheer et al.,

407 2002, Figure 7.1]. Despite this simplification, the resulting explicit equations describing

408 the equilibrium river geometry provide useful insight on the relation among the equilibrium

409 channel slope, surface texture, and channel width.

410 Another yet related consequence of the assumption that wi is independent of grain size

411 is the fact that the resulting slope-equivalent discharge, Qwdom , is not affected by the

412 grain size class. In the next section we will illustrate that generally the slope-equivalent

413 or channel-forming discharge does depend on the grain size fraction.

414 Just as in the analysis for a constant flow rate [Blom et al., 2016], we find one equilibrium

415 state for engineered rivers where the channel width is fixed. Such an engineered river

416 responds to changes in boundary conditions through changes in channel slope and bed

417 surface texture. Yet natural streams respond to changes in boundary conditions through

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 21

418 changes in channel slope, surface texture, and channel width [Howard , 1980], and in that

419 case there is a range of equilibrium states for which the channel is able to transport the

420 load supplied from above.

421 In engineered rivers where the channel width cannot adjust, it is relatively straightfor-

422 ward to apply the presented formulations. When applying them to natural streams two

423 strategies can be followed to formulate a closure relation: (1) one can introduce a closure

424 relation linking, for instance, channel width to other parameters, or (2) depending on the

425 available data and the particular field conditions one can solve equations (26) and (27)

426 for two unknown parameters. For instance, we can impose the channel slope rather than

427 the channel width, which may be a suitable option in those cases where the time scale

428 of slope adjustment is much larger than the time scale of width adjustment [Howard ,

429 1980; Mosselman, 2009]. For such cases we rewrite Eq. (9) to find a formulation for the

430 equilibrium channel width, B:

( )3/2w ( )(2w+3)/2w
G (Dgx /Dref )r gSQwdom
431 B= (29)
Dsw µ Q̄0 Cf

432 The fact that channel width increases with increasing flow rate is confirmed by many

433 authors [e.g., Leopold and Maddock , 1953; Dade et al., 2011]. It is, however, not straight-

434 forward to compare Eq. (29) with empirical relations for the channel width [e.g., Parker

435 et al., 2007; Biedenharn et al., 2008; Li et al., 2014], as the list of included variables dif-

436 fers largely. We will address the topic of channel response through width adjustment in

437 further detail in Section 6.

4.3. An implicit analytical model

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 22 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

438 For a sediment transport relation that (a) contains a threshold below which transport is

439 negligible, (b) contains a hiding coefficient, or (c) is a conditional function [e.g., Wilcock

440 and Crowe, 2003], we can only implicitly express channel slope or width and bed surface

441 texture and need to iteratively find their solutions. We here provide the implicit formula-

442 tions for a load relation with a threshold for significant transport, yet for clarity without

443 a hiding coefficient. For two discharge modes, the mean gravel load, Q̄g , and the mean

444 sand load, Q̄s , in Eqs. (5) and (6) yield

  ( )1/3 b
Q̄g Q2wbase S 2 Cf s
445 √ = α max 0, − τc∗  + ...
aB RgDg Dg F Cf2 B 2 g RDg
  ( )1/3 b
Q2wpeak S 2 Cf s
446 (1 − α) max 0, − τc∗  (30)
Cf2 B 2 g RDg
  ( )1/3 b
Q̄s Q2wbase S 2 Cf s
447 √ = α max 0, − τc∗  + ...
aB RgDs Ds (1 − F ) Cf2 B 2 g RDs
  ( )1/3 b
Q2wpeak S 2 Cf s
448 (1 − α) max 0, − τc∗  (31)
Cf2 B 2 g RDs

449 where the mean gravel and sand loads, Q̄g and Q̄s , at streamwise coordinate x under

450 conditions with gravel abrasion are predicted using Eqs. (20) and (21), provided that the

451 mean sediment supply rate Q̄0 and the mean gravel content in the supplied sediment, p̄g0

452 are known. Eqs. (30) and (31) are easily generalized to account for the arbitrary PDF of

453 water discharge in Figure 1a. This then leads to two relations that allow for solving for

454 two unknowns, provided that the remaining parameters are known.

455 Analogous to the derivation of the slope-equivalent or channel-forming discharge for

456 unisize sediment in Eq. (18), we define the channel-forming discharge as that steady

457 water discharge that, given the gravel load or the sand load, provides the same equilibrium

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 23

458 channel geometry as the natural long-term hydrograph:

  ( )1/3 b
Q̄g Q2wdomg S 2 Cf s
459 √ = max 0, − τc∗  (32)
aB RgDg Dg F Cf2 B 2 g RDg
  ( )1/3 b
Q̄s Q2wdoms S 2 Cf s
460 √ = max 0, − τc∗  (33)
aB RgDs Ds (1 − F ) Cf2 B 2 g RDs

461 This implies that the steady water discharge that provides the same equilibrium channel

462 slope as the natural long-term hydrograph and transports the gravel load downstream

463 (Qwdomg ) is generally larger than the water discharge that suffices to transport the sand

464 load (Qwdoms ). This difference increases with increasing ratio of gravel size to sand size.

465 Thus, there generally exists no uniquely defined slope-equivalent water discharge.

466 We also define a total load related channel-forming discharge, Qwdom , which can be

467 solved from

  ( )1/3 b
Q̄g Q̄s Q2wdom S 2 Cf s
468 √ + √ = max 0, − τc∗  + ..
aB RgDg Dg F aB RgDs Ds (1 − F ) Cf2 B 2 g RDg
  ( )1/3 b
Q2wdom S 2 Cf s
469
max 0, − τc∗  (34)
Cf2 B 2 g RDs

470 This total load related slope-equivalent discharge, Qwdom , is a weighted average of the

471 slope-equivalent water discharges for gravel and sand, and can be considered the channel-

472 forming discharge in the mixed sediment case.

5. Validation of the analytical model

473 We here apply two numerical time-marching models to validate the analytical model for

474 the equilibrium river geometry under variable flow. One time-marching model solves the

475 Saint-Venant [1871] equations and the other one the backwater equation [e.g., Parker ,

476 2004b]. In the latter model the flow rate varies yet the flow is assumed steady at each

477 time, whereas the first model accounts for the dynamics of flood waves.

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 24 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

478 The time-marching model runs are made using the one-dimensional numerical research

479 code Elv, which solves for (1) the flow, (2) bed elevation, and (3) surface texture in a

480 decoupled manner. The Saint-Venant equations are solved using an implicit Preismann

481 scheme. Specific settings comprise the time step, ∆t = 600 s, and the streamwise length

482 of a grid cell, ∆x = 2000 m. The morphodynamic state regarding bed elevation and bed

483 surface texture is updated after each hydrodynamic update. The backwater equation is

484 solved in a space-marching manner, starting from the downstream end and computing

485 the solution in upstream direction (as the flow is subcritical) using a first-order explicit

486 Euler scheme. The streamwise grid size is again ∆x = 2000 m and the flow solver does

487 not require a time step. Bed elevation and bed texture are updated after each time step

488 (∆t = 1 day) using the Exner [1920] and Hirano [1971] equations.

489 The streamwise gradients in the gravel and sand loads are computed using a second

490 order centered finite-difference stencil. In case of aggradation the grain size distribution

491 of the depositional flux to the substrate is set equal to the one of the active layer [Hirano,

492 1971], yet application of the Hoey and Ferguson [1994] formulation for the depositional

493 flux does not change the results. We keep track of changes to the substrate sediment

494 associated with changes in mean bed elevation. The active layer thickness, the vertical

495 size of the grid cells for keeping track of the bed stratification, porosity, and the initial

496 conditions of the time-marching runs (bed elevation, surface and substrate texture) do

497 not affect the equilibrium channel slope and bed surface texture.

498 The models are applied to a base case that is characterized by a cycled synthetic hydro-

499 graph governed by two peaks and a period of one year (Figure 3). The simple shape of the

500 hydrograph creates an unnatural shape of the PDF of water discharge yet suffices here.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 25

501 In the base case we impose channel width (B = 250 m) and solve for channel slope and

502 bed surface texture. We run the models for different values of the mean gravel content in

503 the supplied sediment. The domain length is 400 km. The friction coefficient, Cf , equals

504 0.008, and for simplicity we assume that friction is associated with skin friction only. The

505 mean sediment supply rate equals 0.03 m3 /s. The abrasion coefficient β here equals 0

506 m−1 , and the gravel and sand sizes are 20 mm and 1 mm, respectively.

507 We apply three load relations: the generalized power law load relation (GR, settings

508 r = 0.05, wg = ws = w = 0.4, and c = 2.3) in Eq. (19), the Wilcock and Crowe [2003]

509 load relation (WC), and the Ashida and Michiue [1972] load relation (AM).

510 The outcomes of the analytical model are nearly identical to the results of the two

511 numerical time-marching models (Figure 4). For the time-marching models we consider

512 a position 20 km downstream from the upstream end of the domain, which is sufficiently

513 upstream from the backwater zone. Figure 4 confirms the applicability of the analytical

514 model under normal flow conditions where the effects of the dynamics of flood waves

515 (wave flattening and lengthening) are negligible. If the dynamics of flood waves are not

516 negligible, the statistics of the local hydrograph need to be known from measured data,

517 modelling, or otherwise.

518 Figure 4 also shows how both channel slope and bed surface gravel content increase

519 with increasing mean gravel content in the sediment supply, which confirms the analysis

520 by Blom et al. [2016].

521 The PDF of the flow rate needs to cover the statistics of the flow over some years or

522 decades. The analytical model accounts for that long-term variability of the flow rate.

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 26 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

523 The order of flow events or history is here irrelevant since in the considered normal flow

524 zones the equilibrium state is a static one.

525 The analytical model enables us to rapidly find a solution to the equilibrium river geom-

526 etry under variable flow (numerical time-marching models take much longer to compute

527 this), and provides insight on the relation among the gravel and sand supply rates, the

528 PDF of water discharge, channel slope, channel width, and bed surface texture. In ad-

529 dition, due to the complexities of (modeling) bank erosion processes, currently available

530 time-marching models may not be capable of simulating channel response through width

531 adjustment.

6. Channel response through slope or width adjustment

532 The importance of the variability of the flow rate or the PDF of water discharge is illus-

533 trated by the following example of an engineered river with fixed banks. Let us compare

534 two cases (A and B) that are equal except for the larger standard deviation of the wa-

535 ter discharge in Case A. This difference, for a given channel width, results in a smaller

536 equilibrium channel slope in Case A. A larger standard deviation of the water discharge

537 results in a larger channel-forming discharge, and so a smaller channel slope suffices to

538 transport the same amount of sediment downstream.

539 Rather than response through slope adjustment (through reach scale aggrada-

540 tion/degradation or changes in channel sinuosity), river response to changes in the bound-

541 ary conditions may occur through width adjustment [e.g., Curtis et al., 2010; Dade et al.,

542 2011; Tealdi et al., 2011; Buffington, 2012]. For reaches where the time scale of slope

543 adjustment is much larger than the time scale of width adjustment [Howard , 1980], the

544 explicit solution for the channel slope in Eq. (26) was rewritten to a solution to the chan-

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 27

545 nel width in Eq. (29), which illustrates that for this type of channel response (i.e. through

546 width adjustment) the larger standard deviation of the water discharge in Case A yields

547 a larger equilibrium channel width. We expect that generally channel response (to chang-

548 ing boundary conditions) is of a mixed type, so through adjustment of a combination of

549 (1) channel width, (2) channel slope, and (3) surface texture [e.g., Buffington and Mont-

550 gomery, 1999; Dade et al., 2011]. In such a case one will need to add a closure relation

551 to the proposed formulations.

552 We here assess channel response to changes in the variability of the flow rate applying

553 a lognormal distribution of the flow rate with a mean value of the flow rate, µw , equal to

554 2000 m3 /s. Parameters of the base case include pg0 = 0.5, S = 2.4 · 10−4 , and B = 250 m.

555 For an engineered river with fixed banks that responds to changes in the controls through

556 adjusting its slope, Figure 5a confirms that the equilibrium channel slope decreases with

557 increasing variability of the flow rate.

558 The equilibrium surface texture only mildly responds to changes in the variability of

559 the flow rate (Figure 5b). This result may be surprising as field and flume studies have

560 shown that changes in sediment supply can be accommodated by small changes in sur-

561 face texture [Dietrich et al., 1989; Buffington and Montgomery, 1999; Ferguson et al.,

562 2015]. Adjustment of surface texture can indeed accommodate changes in sediment sup-

563 ply, yet a change in the characteristic flow rate likely does not affect the surface texture

564 [Litty and Schlunegger , 2016; Blom et al., 2016] and, under a constant width, is generally

565 accommodated through channel slope adjustment [Blom et al., 2016, Figure 2].

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 28 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

566 For a river that responds to changing boundary conditions through adjusting its channel

567 width, Figure 5c confirms that, for otherwise equal parameters, the equilibrium channel

568 width increases with increasing variability of the flow rate.

569 The equilibrium surface texture hardly responds to an increasing variability of the flow

570 rate: the bed surface becomes slightly finer with increasing variability of the flow rate

571 (Figures 5b and d). This holds for both types of channel response. In addition, the

572 similarity between these subfigures b and d illustrates that remarkably, for the considered

573 case, the type of channel response (through either width or slope adjustment) has almost

574 no effect on the adjustment of the surface texture to the changing variability of the flow

575 rate.

7. The ‘normal flow load distribution’ and the hydrograph boundary layer

576 The equilibrium channel slope, width, and surface texture adjust such that they enable

577 the stream to transport the long-term mean sediment supply downstream. In the river

578 normal flow zone the effect of the short-term variation of the sediment supply is not felt.

579 Worded differently, the river normal flow zone is located so far downstream that there is no

580 effect of the short-term variation of the sediment supply on reach-averaged parameters. As

581 such, in a river normal flow zone the local river geometry (channel slope, surface texture,

582 and channel width) is determined by the long-term mean value of the sediment supply

583 and not by its temporal variation.

584 The resulting equilibrium river geometry determines the distribution of the sediment

585 flux over the period of the cycled hydrograph (i.e. over the range of discharge values).

586 This distribution of the gravel and sand load over the range of water discharge in a river

587 normal flow zone is here termed the ‘normal flow load distribution’ of gravel and sand.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 29

588 The fact that the sedigraph (in the field or imposed as a boundary condition in a nu-

589 merical model) at the upstream end of a reach is generally different from the sedigraph in

590 the river normal flow zone (i.e. the normal flow load distribution) leads to the presence

591 of the ‘hydrograph boundary layer’ (HBL) [e.g., Parker , 2004a; Wong and Parker , 2006;

592 Parker et al., 2008; Viparelli et al., 2011; An et al., 2017], which is illustrated in Figure 6.

593 This HBL is the adaptation zone over which the distribution of the (gravel and sand) load

594 over the range of the water discharge spatially adjusts toward its ‘normal flow load distri-

595 bution’. A HBL is characterized by downstream migrating fluctuations in bed elevation,

596 slope, and bed surface texture that may dampen with streamwise position [e.g., Parker

597 et al., 2008].

598 The length of the river normal flow zone is influenced (i.e. reduced) by both the upstream

599 HBL and the downstream backwater zone. Such HBL and backwater effects not only occur

600 at, respectively, the upstream and downstream ends of a reach (Figure 6), but also arise

601 from a spatial change in channel width or friction, a bifurcation, or a tributary. The

602 normal flow zone is the reach that remains when HBL zones and backwater zones are

603 subtracted.

604 We represent the normal flow load distribution for gravel, sand, and total load using

605 the parameter, νX [s/m3 ]:

QX (Qw )fw (Qw )

606 νX (Qw ) = (35)
Q̄X

607 where X ∈ {g, s, t}. For X = g the parameter QX [m3 /s] indicates the gravel load, for

608 X = s the sand load, and for X = t the total load. The parameter Q̄X indicates the

609 expected value of the parameter QX , which normalizes νX such that the integral of νX

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 30 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

610 over water discharge equals unity. As an example, νg (Qw ) is the probability density of

611 the gravel load transported at a water discharge Qw .

612 Given the mean gravel and sand supply, the analytical model allows for predicting the

613 normal flow load distribution for each size fraction. We illustrate this for a simplified

614 Rhine River case, which is a typical example of an engineered river that responds to

615 changes in the controls through slope adjustment. Figure 7 shows the measured water

616 discharge in the Rhine at Lobith in the German-Dutch border area in the period 1901-

617 2013, as well as the associated PDF of the water discharge. Now, for illustration purposes,

618 let us assume that normal flow prevails at Lobith. Furthermore, we impose the mean sand

619 load at Lobith equal to 0.56 Mt/a and the mean gravel load to 0.11 Mt/a [Frings et al.,

620 2014, 2015]. We set the channel width, B, equal to 300 m (width over which sediment

621 is transported), the nondimensional friction coefficients, Cf and Cf s , to 0.007, the sand

622 size, Ds , to 1 mm, and the gravel size, Dg , to 10 mm.

623 Table 2 lists values of the equilibrium channel slope and bed surface gravel content at

624 Lobith predicted based on the three load relations applied in Section 5. The predicted

625 values of the channel slope are similar. According to Frings [2011] the channel slope at

626 Lobith is about 11·10−5 (his Figure 1) and the surface gravel content equals more or less

627 0.6 (his Figure 6), which implies that the predicted value of the equilibrium channel slope

628 is somewhat smaller than the measured slope. This results in channel bed degradation,

629 which is indeed the case in the German and Dutch Rhine although the reach around

630 Lobith seems to have stabilized over the past few years [Blom, 2016]. Despite the model

631 simplifications the Lobith case illustrates how the analytical model can be applied as a

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 31

632 tool for rapid assessment of the equilibrium river geometry and can aid in the assessment

633 of the need for a more detailed study using a time-marching numerical model.

634 Figure 8 shows the predicted normal flow load distributions for the gravel, sand, and

635 the total load for the three load relations. The results obtained with the threshold-

636 based Ashida and Michiue load relation (Figure 8c) are significantly different from those

637 obtained with the other two relations, which do not account for a threshold for significant

638 transport. A threshold-based load relation results in gravel immobility for small values

639 of the flow rate or the gravel transport being limited to relatively high flow rates. For

640 the generalized load relation (GR) and the Wilcock and Crowe load relation (WC), the

641 distribution of the gravel load, νg , is closer to the distribution of the water discharge, fw .

642 Figure 8 confirms the ideas by Wolman and Miller [1960] in that extreme flow events,

643 which are largely relevant to flood risk assessment, are of much less importance to the

644 equilibrium river geometry as these peak flow rates do not occur sufficiently frequently to

645 contribute significantly to transporting the supplied sediment load downstream.

8. The channel-forming versus the effective discharge

646 We here compare the slope-equivalent or channel-forming discharges for the gravel and

647 sand load, Qwdomg and Qwdoms , in Eqs. (32) and (33) and the one of the total load Qwdom

648 in Eq. (34) to the effective discharge first defined by Wolman and Miller [1960]. We

649 distinguish between gravel, sand, and total load effective discharges. The gravel-related

650 effective discharge, Qweff g , is the water discharge at which the product of the probability

651 density of the water discharge and the associated gravel load reaches a maximum value,

652 and the sand-related and total load-related effective discharges, Qweff s and Qweff , are

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 32 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

653 defined likewise:

654 Qweff g = mode(Qg (Qw )fw (Qw )) (36)

655 Qweff s = mode(Qs (Qw )fw (Qw )) (37)

656 Qweff = mode ((Qg (Qw ) + Qs (Qw ))fw (Qw )) (38)

657 As the current settings of the generalized load relation (wi = w, independent of grain

658 size) do not allow for a coarsening of the load with increasing Shields stress, the normal

659 flow load distributions for gravel, sand, and total load are equal, as well as the channel-

660 forming water discharges for gravel, sand, and the total load. For other settings or other

661 load relations the channel-forming water discharge associated with the gravel load is typ-

662 ically larger than the one associated with the sand load. This also holds for the effective

663 discharges, which supports findings by Lenzi et al. [2006] in that in a mountain stream the

664 bedload associated effective discharge is larger than the one associated with suspended

665 load.

666 For the generalized load relation (GR) and the Wilcock and Crowe load relation (WC),

667 the channel-forming discharges are similar to their effective counterparts (Figures 8a-b).

668 For threshold-based load relations (such as the Ashida and Michiue [1972] load relation,

669 Figure 8c), the discharge values contributing to transporting the gravel load are signif-

670 icantly larger than for load relations without such a threshold (GR and WC). As such,

671 the resulting larger channel-forming discharge (particularly for the gravel load) allows for

672 a smaller equilibrium slope and a larger channel width compared to non-threshold load

673 relations (Figure 5). However, threshold-based relations may overpredict to some extent

674 given that they erroneously predict zero transport at flows below the threshold [e.g., Barry

675 et al., 2004, 2007], which can cover a substantial range of the PDF of the flow rate.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 33

676 For the threshold-based relation (Figure 8c) the gravel-related effective discharge ap-

677 pears to be significantly larger than its channel-forming counterpart.

678 In addition, for the threshold-based load relation in Figure 8c, the gravel-related

679 channel-forming discharge is associated with a Shields stress (τg∗ = 0.033) that is only
∗
680 slightly larger than its critical value (τcg = 0.027). This is consistent with findings with

681 respect to the bankfull discharge in gravel-bed rivers [Parker , 1978b, 2004b; Parker et al.,

682 2007; Phillips and Jerolmack , 2016; Pfeiffer et al., 2017]. The sand-related channel-

683 forming discharge in Figure 8c is associated with a Shields stress (τs∗ = 0.29) that is
∗
684 significantly larger than its critical value (τcs = 0.14), which is consistent with the find-

685 ings by Parker [1978a, 2004b].

9. Discussion

686 The definition of Copeland et al. [2005] of the dominant or channel-forming discharge

687 is: that single steady discharge that given enough time would produce channel dimensions

688 equivalent to those shaped by the natural long-term hydrograph. Here we further narrow

689 down the definition of the channel-forming discharge as the steady discharge that, given

690 a certain mean sediment load, provides the same equilibrium channel slope as the natural

691 long-term hydrograph. This slope-equivalent discharge corresponds to the functional-

692 equivalent discharge introduced by Doyle and Shields [2008], although they do not ex-

693 plicitly consider the limitation to equilibrium or graded conditions and do not extend

694 their analysis to mixed-sediment conditions. In our analysis we have addressed the rela-

695 tion between the slope-equivalent discharge and the commonly applied effective discharge

696 [e.g., Wolman and Miller , 1960], yet we have not addressed the relation with the bankfull

697 discharge. We recommend further research on this topic.

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 34 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

698 The validity of the proposed formulations is limited to the normal flow zones of a graded

699 or equilibrium alluvial river and does not provide formulations for the zone of the HBL and

700 the backwater zone (Figure 6). The HBL has been studied by Parker [2004a], Wong and

701 Parker [2006], Parker et al. [2008], Viparelli et al. [2011], and An et al. [2017]. Current

702 research by the second author focusses on a formulation of the equilibrium channel slope

703 in backwater-dominated zones of an alluvial river.

704 Our analysis is limited to equilibrium river reaches under conditions without subsidence

705 and uplift. Under transient or ungraded conditions [e.g., Howard , 1982], in which the

706 river is adjusting to changes in the boundary conditions, river characteristics approach

707 those of the equilibrium or graded river geometry described in this paper. Such temporal

708 adjustment toward the graded river geometry is usually simulated or reproduced using

709 time-marching numerical models. Yet if the change of the boundary conditions happens

710 slowly (i.e. the time scale of change is large compared to the time scale of channel response

711 to the changing boundary conditions), reach characteristics keep pace with the changing

712 boundary conditions. Under such quasi-equilibrium conditions [Mackin, 1948; Chorley and

713 Kennedy, 1971; Howard , 1982] one may apply the formulations proposed in the present

714 paper. Blom et al. [2017] consider the case of an abrupt gravel-sand transition, where they

715 apply equilibrium formulations for channel slope and surface texture to quasi-equilibrium

716 conditions.

717 We have limited our analysis to cases with one or two sediment fractions (gravel and

718 sand), as rivers are often characterized by bimodal distributions of grain size. However,

719 also cases with more two grain size classes allow for an implicit solution to the equilibrium

720 river geometry.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 35

721 Simplifications such as the rectangular cross-section and the neglect of floodplain pro-

722 cesses need to be a point of attention when applying the presented analytical models to

723 a field case. Yet it is precisely these simplifications that allow us to arrive at the explicit

724 and implicit analytical solutions to the equilibrium river geometry. The formulations are

725 specifically meant to provide insight and can be used as a tool for rapid assessment of the

726 river response to changes in the controls, but application to a field case should always be

727 undertaken with care. Most of the simplifications made in the current analysis may be

728 relaxed by applying a time-marching numerical model to compute the equilibrium river

729 geometry, yet the required computational time may make this impossible.

730 Concerning the assumption of a rectangular cross-section, we expect that a field site

731 with a clear division between main channel and floodplains could be assessed with the

732 analytical model through, for instance, the assumption that the Shields stress does not

733 increase with flow rate once the flow rate exceeds a bankfull value (i.e. under overbank

734 flow conditions). A complication here is the fact that the bankfull flow rate depends on

735 the predicted equilibrium channel slope, surface texture, and channel width. Another

736 strategy could be to include a relation between the flow rate and the cross-sectional area.

737 Despite the drawbacks associated with the simple form of the power law load relation, we

738 believe that its explicit solutions provide useful insight on how channel slope, width, and

739 surface texture on the long term respond to changes in the boundary conditions. Although

740 the explicit solutions do not account for the fact that the mobility difference between fine

741 and coarse sediment becomes smaller with increasing flow rate, it does account for grain

742 size selectivity (i.e. for the mobility difference between fine and coarse sediment). When

743 applying it to a field case, its constants need to be carefully adjusted to the considered field

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 36 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

744 site and preferably their values need to be validated against data from the field site that

745 were not used in calibrating them. One should apply a more advanced (yet mathematically

746 less convenient) load relation when the power law load relation is insufficiently applicable

747 to the specific field case.

748 In our time-marching numerical runs we have neglected temporal change of the active

749 layer thickness and bed load layer thickness [Armanini and Di Silvio, 1988; Parker , 1991].

750 In reality the depth of reworking of bed sediment during peak flow is larger than during

751 base flow. We recommend further research on this matter.

752 The current analysis has implications for the schematization of the hydrograph and grain

753 size specific sedigraphs in long-term numerical runs. The imposed hydrograph should

754 cover the natural long-term hydrograph (i.e. the long-term statistics of the flow rate), yet

755 the tails of the distribution of the flow rate may be less important if one is interested in

756 modelling morphodynamic change in a river normal flow zone.

757 For field scale applications of the present formulation, the following field measurements

758 should be preferably available: (1) long-term time series of water discharge or its PDF

759 (at least covering a few years); (2) annual volume of sediment supply or long-term mean

760 sediment supply rate; (3) mean fraction content of gravel in the sediment supply; (4) a

761 representative gravel size; (5) a representative sand size; (6) friction; (7) channel slope;

762 (8) surface fraction content of gravel; and (9) channel width. Field cases for which this

763 data set, especially items 2 and 3, is available are rare. We here underline the importance

764 of acquiring field data on the mean annual gravel and sand flux.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 37

10. Conclusions

765 We provide analytical formulations for the geometry of an equilibrium alluvial channel

766 under variable flow. The analysis is limited to the normal flow reaches of a river, that is

767 to reaches outside the hydrograph boundary layer and the backwater zone. The analysis

768 applies to both unisize sediment and sediment mixtures composed of gravel and sand. It

769 accounts for grain-size selective transport and particle abrasion.

770 Application of a power law load relation results in explicit analytical solutions to the

771 relation among the equilibrium channel slope, channel width, and bed surface texture.

772 Application of other load relations (e.g., including a threshold for significant transport, a

773 hiding coefficient, or a load relation that is a conditional function such as the one proposed

774 by Wilcock and Crowe [2003]) provides implicit analytical solutions that can be solved

775 iteratively.

776 The resulting equations aid in the prediction of long-term channel response to changes

777 in the boundary conditions. A river responds to such changes through adjustment of (a)

778 channel slope, (b) channel width, (c) surface texture, or (d) a combination of these types

779 of adjustment. In engineered rivers where the channel width is fixed, given the sediment

780 supply and PDF of water discharge, there exists one solution to the equilibrium or graded

781 river geometry in a river normal flow zone [Buffington, 2012]. In natural streams besides

782 slope also channel planform adjusts to changing conditions through width changes, which

783 implies that there exists a range of equilibrium states for which the channel is able to

784 transport the load supplied from above.

785 We propose two analytical relations linking: (1) PDF of flow rate, (2,3) gravel and

786 sand supply rates, (4) channel width, (5) channel slope, (6) surface texture, and (7,8) the

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 38 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

787 representative gravel and sand sizes. The relations can be used to solve for two unknown

788 parameters. In engineered rivers, given the PDF of the flow rate and the sediment supply

789 rates, one can solve for the equilibrium slope and surface texture. In natural streams where

790 the time scale of slope adjustment is much larger than the one of width adjustment, one

791 may impose the channel slope and use the formulations to solve for the equilibrium channel

792 width. In natural streams in which channel response to changing boundary conditions is

793 of a mixed nature (i.e. through adjustment of a combination of slope, width, and surface

794 texture), one may add an empirical relation to the proposed formulations and solve for

795 the combined set of parameters.

796 The equilibrium surface texture hardly responds to changing variability of the flow rate.

797 In addition, in the considered case the type of channel response to the changing flow rate

798 variability (i.e. through adjustment of either channel slope or channel width) hardly affects

799 how the surface texture adjusts to the changing flow rate variability.

800 We here define the slope-equivalent or channel-forming discharge as the steady water

801 discharge that provides the same equilibrium channel slope as the natural long-term hy-

802 drograph, given the mean sediment supply rate. Application of the power law load relation

803 allows for an explicit analytical solution to the channel-forming discharge in the river nor-

804 mal flow zone. This slope-equivalent discharge is associated with a grain size class: the

805 gravel-related channel-forming discharge is generally larger than the one for sand.

806 Our definition of the slope-equivalent discharge confirms the ideas by Wolman and

807 Miller [1960] in that channel form and so the representative discharge are a function of

808 both the magnitude and frequency of the flow rate. The analysis illustrates that extreme

809 flow events, which naturally are important in flood risk assessment, are less important to

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 39

810 the equilibrium river geometry because of their rarity. Worded differently, the tails of the

811 PDF of the flow rate seem to be of limited importance to the equilibrium river geometry

812 in a river normal flow zone. In addition, this implies that, if the effects of climate change

813 are limited to mostly the tails of the PDF of the flow rate, effects of climate change on

814 river geometry may be negligible.

815 In the equilibrium state the reach-averaged parameters of a river normal flow zone are

816 not affected by the short-term variation of the sediment supply. As such, in a river normal

817 flow zone the local river geometry (channel slope, surface texture, and channel width) is

818 determined by the long-term mean value of the sediment supply and not by its short-term

819 variation.

820 The analytical model allows for predicting how the gravel and sand load vary over

821 the long-term range of flow rate values in the river normal flow zone, which we term

822 the ‘normal flow load distributions’ of gravel and sand. The fact that (in the field or

823 the numerically imposed) sedigraphs for gravel and sand at the upstream end of a reach

824 generally differ from their ‘normal flow load distributions’ leads to the presence of the

825 hydrograph boundary layer.

826 When applying a threshold-based load relation, we find that the gravel-related channel-

827 forming discharge is only slightly larger than the one associated with its critical Shields

828 number, whereas the difference for the two corresponding values for the sand load is sig-

829 nificantly larger. These findings are consistent with earlier findings regarding the bankfull

830 discharge.

831 The proposed formulations can easily be extended with empirical relations between, for

832 instance, drainage area and channel width or water discharge [e.g., Hack , 1957; Tucker

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 40 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

833 and Bras, 1998] to find drainage area based relations for the graded channel slope, width,

834 and surface texture.

835 The formulations for the equilibrium river geometry provide useful and rapid insight on

836 the state that the river approaches as well as the long-term consequences of foreseen river

837 measures or natural change of the controls [e.g., De Vriend , 2015].

838 The relevance of the statistical equilibrium state of streams seems to be broader than is

839 generally anticipated. It has often been argued that the river’s equilibrium state is irrele-

840 vant as it may never arrive at that state due to continuously changing boundary conditions

841 and uplift or subsidence. Yet if the controls change so slowly that the stream can keep

842 pace with them, the stream continuously finds itself in a state of quasi-equilibrium, in

843 which the proposed formulations can be applied. This is illustrated by Blom et al. [2017]

844 for the case of an abrupt gravel-sand transition.

Appendix A: Generalized load relation for unisize sediment

845 We summarize the wide range of existing power law load relations into the following

846 generalized relation:

( )r
∗ D
847 W =c τ ∗w (A1)
Dref

848 where c [-], w [-], and r [-] are constants (w > r for a decreasing sediment load with

849 increasing grain size). The nondimensional sediment transport rate, W ∗ [-], the nondi-

850 mensional Einstein transport number, q ∗ [-], and the grain related Shields stress, τ ∗ [-],

851 are defined as [e.g., Parker , 2004b]:

q∗ Q τbs
852 W∗ = , q∗ = √ , τ∗ = (A2)
τ ∗3/2 B RgDD ρgRD

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 41

853 where R [-] denotes the submerged sediment density (R = (ρs − ρ)/ρ in which ρs [kg/m3 ]

854 and ρ [kg/m3 ] are the sediment and water density, respectively) and g [m/s2 ] is the

855 acceleration due to gravity. The grain related shear stress, τbs [N/m2 ], is defined as

856 τbs = ρCf s U 2 , where Cf s [-] denotes the nondimensional skin friction coefficient and U

857 [m/s] is the depth-averaged flow velocity.

858 Combining Eqs. (A1) and (A2), we find the following relation for the sediment transport

859 rate per unit width, q [m2 /s] (Q = qB):

( )r w+ 32
D G 2w+3 c Cf s
860 q= U , G= (A3)
Dref Dw (Rg)w+1

861 Eq. (A3) is rearranged into the relation proposed by Engelund and Hansen [1967]
3/2 5/2
862 through setting w = 1, r = 0, and c = 0.05Cf /Cf s :
3/2
GEH 5 0.05Cf
863 qEH = U , GEH = (A4)
D (Rg)2

864 It is important to note that the friction coefficient, Cf , in the Engelund and Hansen [1967]

865 load relation contains form drag.

Appendix B: Generalized load relation for mixed sediment

866 For mixed sediment we propose the following generalized power law load relation:
( )r
Di
867 Wi∗ =c τi∗wi (B1)
Dref

868 where wi [-] is a constant generally slightly increasing with grain size Di [e.g., Kuhnle,

869 1992]. The nondimensional sediment transport rate of size fraction i, Wi∗ [-], the nondi-

870 mensional Einstein transport number of size fraction i, qi∗ [-], and the grain related Shields

871 stress associated with size fraction i, τi∗ [-], are defined as [e.g., Parker , 2004b]:

qi∗ Qi τbs
872 Wi∗ = ∗3/2
, qi∗ = √ , τi∗ = (B2)
τi B RgDi Di Fi ρgRDi

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 42 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

873 where Fi [-] denotes the volume fraction content of size fraction i at the bed surface, and

874 Di [m] is the representative grain size of size fraction i.

875 Eqs. (B1) and (B2) lead to the following formulation for the mixed sediment transport

876 rate per unit width, qi [m2 /s]:

( )r
Di Gi 2wi +3
877 qi = Fi U (B3)
Dref Diwi

878 where the constant Gi [s4 /m2 ] is defined as

wi + 23
cCf s
879 Gi = (B4)
(Rg)wi +1

880 Eq. (B3) is cast into the fractional form of the Engelund and Hansen [1967] relation by
3/2 5/2
881 setting wi = 1, r = 0, and c = 0.05Cf /Cf s :

GEH 5
882 qiEH = Fi U (B5)
Di

883 where GEH is defined in Eq. (A4b). The fractional form of the transport relation by

884 Engelund and Hansen [1967] in Eq. (B5) was proposed nor validated by the original

885 authors yet was applied by Van der Scheer et al. [2002] and Blom et al. [2016].

886 In Figure 9 we compare the nondimensional gravel and sand load predicted using the

887 generalized load relation to data predicted using the load relations by Engelund and

888 Hansen [1967] (EH), Fernandez-Luque and Van Beek [1976] (FLvB), Ashida and Michiue

889 [1972] (AM), and Wilcock and Crowe [2003] (WC). It shows how the power law load

890 relation covers the general trends fairly well. Figure 9 also indicates the range of reasonable

891 values for r, wi , and c.

892 Acknowledgments. The numerical time-marching research code Elv has been de-

893 veloped within the research group of the first author at the Water Lab of Delft Uni-

894 versity of Technology over the past few years. The following researchers and students

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 43

895 have contributed to the development of Elv and its earlier versions: Vı́ctor Chavarrı́as,

896 Liselot Arkesteijn, Guglielmo Stecca, Roy van Weerdenburg, Clara Orrú, Velia Fer-

897 rara, Lodewijk de Vet, and Astrid Blom. We thank Roger Kuhnle for assisting us

898 in gathering field data. We thank editor John Buffington, the associate editor, two

899 anonymous reviewers and Rob Ferguson for their helpful comments on the manuscript.

900 Links to digital copies of NEDECO [1959], Van Bendegom [1967], Prins [1969], Prins

901 and De Vries [1971], De Vries [1971, 1974], and Jansen et al. [1979] can be found

902 at http://www.citg.tudelft.nl/ablom. The code will be available upon request

903 ([email protected]). All data for this paper is cited and referred to in the refer-

904 ence list.

References

905 An, C., Y. Cui, X. Fu, and G. Parker (2017), Gravel-bed river evolution in earthquake-

906 prone regions subject to cycled hydrographs and repeated sediment pulses, Earth Surf.

907 Process. Landf., doi:10.1002/esp.4195.

908 Ackers, P., and F. G. Charlton (1970), Meander geometry arising from varying flows, J.

909 Hydrol., 11, 230–252, doi:10.1016/0022-1694(70)90064-8.

910 Ahnert, F. (1994), Equilibrium, scale and inheritance in geomorphology, Geomorphology,

911 11 (2), 125–140, doi:10.1016/0169-555X(94)90077-9.

912 Andrews, E. D. (1980), Effective and bankfull discharges of streams in the Yampa River

913 basin, Colorado and Wyoming, J. Hydrol., 46 (3), 311–330, doi:10.1016/0022-1694(80)

914 90084-0.

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 44 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

915 Armanini, A., and G. di Silvio (1988), A one-dimensional model for the transport of

916 a sediment mixture in non-equilibrium conditions, J. Hydraul. Res., 26 (3), 275–292,

917 doi:10.1080/00221688809499212.

918 Ashida, K., and M. Michiue (1972), Study on hydraulic resistance and bed-load transport

919 rate in alluvial streams, J. Soc. Civ. Eng. Trans., 206, 59–69.

920 Barry, J. J., J. M. Buffington, and J. G. King (2004), A general power equation for

921 predicting bed load transport rates in gravel bed rivers, Water Resour. Res., 40 (10),

922 W10401, doi:10.1029/2004WR003190.

923 Barry, J. J., J. M. Buffington, and J. G. King (2007), Correction to “A general power

924 equation for predicting bed load transport rates in gravel bed rivers”, Water Resour.

925 Res., 43 (8), doi:10.1029/2007WR006103.

926 Benson, M. A., and D. M. Thomas (1966), A definition of dominant discharge, Int. Assoc.

927 Sci. Hydrol. Bull., 11 (2), 76–80, doi:10.1080/02626666609493460.

928 Biedenharn, D. S., C. C. Watson, and C. R. Thorne (2008), Fundamentals of flu-

929 vial geomorphology, in Sedimentation Engineering: Processes, Measurements, Model-

930 ing, and Practice, M. H. Garcia (ed.), Amer. Soc. Civil Eng., 355–386, doi:10.1061/

931 9780784408148.ch06.

932 Blom, A. (2008), Different approaches to handling vertical and streamwise sorting in

933 modeling river morphodynamics, Water Resour. Res., 44 (3), W03415, doi:10.1029/

934 2006WR005474.

935 Blom, A. (2016), Bed degradation in the Rhine River, WaterViewer, Delft

936 University of Technology, Available at http://waterviewer.tudelft.nl/#/

937 bed-degradation-in-the-rhine-river-1479821439344____47.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 45

938 Blom, A., E. Viparelli, and V. Chavarrı́as (2016), The graded alluvial river: Profile con-

939 cavity and downstream fining, Geophys. Res. Lett., 43, 1–9, doi:10.1002/2016GL068898.

940 Blom, A., V. Chavarrı́as, R. I. Ferguson, and E. Viparelli (2017), Advance, retreat, and

941 halt of abrupt gravel-sand transitions in alluvial rivers, Geophys. Res. Lett., under

942 review.

943 Bolla Pittaluga, M., R. Luchi, and G. Seminara (2014), On the equilibrium profile of river

944 beds, J. Geophys. Res., Earth Surface, 119 (2), 317–332, doi:10.1002/2013JF002806.

945 Bray, D. (1975), Representative discharges for gravel-bed rivers in Alberta, Canada, J.

946 Hydrol., 27 (1), 143–153, doi:10.1016/0022-1694(75)90103-1.

947 Buffington, J. M. (2012), Changes in channel morphology over human time scales,

948 in Gravel-bed Rivers: Processes, Tools, Environments, M. Church, P. M. Biron,

949 and A. G. Roy (eds.), John Wiley and Sons, Ltd., Chichester, UK, 433–463 doi:

950 10.1002/9781119952497.ch32.

951 Buffington, J. M., and D. R. Montgomery (1999), Effects of sediment supply on

952 surface textures of gravel-bed rivers, Water Resour. Res., 35 (11), 3523–3530, doi:

953 10.1029/1999WR900232.

954 Carling, P. (1988), The concept of dominant discharge applied to two gravel-bed streams

955 in relation to channel stability thresholds, Earth Surf. Process. Landf., 13 (4), 355–367,

956 doi:10.1002/esp.3290130407.

957 Chorley, R. J., and B. A. Kennedy (1971), Physical geography: a systems approach, 370

958 pp., Prentice Hall.

959 Church, M. (2006), Bed material transport and the morphology of alluvial river chan-

960 nels, Annu. Rev. Earth Planet. Sci., 34, 325–354, doi:10.1146/annurev.earth.33.092203.

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 46 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

961 122721.

962 Copeland, R., P. Soar, and C. Thorne (2005), Channel-forming discharge and hydraulic

963 geometry width predictors in meandering sand-bed rivers, in Impacts of Global Climate

964 Change, 1–12, doi:10.1061/40792(173)568.

965 Curtis, K. E., C. E. Renshaw, F. J. Magilligan, and W. B. Dade (2010), Temporal

966 and spatial scales of geomorphic adjustments to reduced competency following flow

967 regulation in bedload-dominated systems, Geomorphology, 118 (1-2), 105 - 117, doi:

968 10.1016/j.geomorph.2009.12.012.

969 Dade, W. B., C. E. Renshaw, and F. J. Magilligan (2011), Sediment transport con-

970 straints on river response to regulation, Geomorphology, 126 (1-2), 245–251, doi:

971 10.1016/j.geomorph.2010.11.007.

972 De Vriend, H. J. (2015), The long-term response of rivers to engineering works and climate

973 change, Proc. Inst. Civil Eng., 168(CE3), 139–145, doi:10.1680/cien.14.00068.

974 De Vries, M. (1971), Aspecten van zandtransport in open waterlopen, Tech. Rep., 71 pp.,

975 Afd. der Weg- en Waterbouwkunde, Technische Hogeschool Delft, Delft, Netherlands

976 (in Dutch).

977 De Vries, M. (1974), Sedimenttransport, Lecture Notes F10, 65 pp., Delft University of

978 Technology, Netherlands (in Dutch).

979 De Vries, M. (1993), Use of models for river problems, 85 pp., UNESCO.

980 Doyle, M. W., and C. A. Shields (2008), An alternative measure of discharge effectiveness,

981 Earth Surf. Process. Landf., 33 (2), 308–316, doi:10.1002/esp.1543.

982 Dietrich, W. E., J. W. Kirchner, H. Ikeda, and F. Iseya (1989), Sediment supply and the

983 development of the coarse surface layer in gravel-bedded rivers, Nature, 340, 215–217,

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 47

984 doi:10.1038/340215a0.

985 Doyle, M. W., D. Shields, K. F. Boyd, P. B. Skidmore, and D. Dominick (2007), Channel-

986 forming discharge selection in river restoration design, J. Hydraul. Eng., 133 (7), 831–

987 837, doi:10.1061/(ASCE)0733-9429(2007)133:7(831).

988 Einstein, H. A. (1950), The bed-load function for sediment transportation in open channel

989 flows, Tech. Rep. 1026, 71 pp., US Department of Agriculture.

990 Emmett, W. W., and M. G. Wolman (2001), Effective discharge and gravel-bed rivers,

991 Earth Surf. Process. Landf., 26 (13), 1369–1380, doi:10.1002/esp.303.

992 Engelund, F., and E. Hansen (1967), Monograph on sediment transport in alluvial

993 streams, Tech. Rep., 63 pp., Hydraul. Lab., Tech. Univ. of Denmark, Copenhagen,

994 Denmark.

995 Exner, F. M. (1920), Zur Physik der Dünen, Akad. Wiss. Wien Math. Naturwiss, 129(2a),

996 929–952 (in German).

997 Ferguson, R. I. (2003), Emergence of abrupt gravel to sand transitions along rivers through

998 sorting processes, Geology, 31 (2), 159–162, doi:10.1130/0091-7613(2003)031⟨0159:

999 EOAGTS⟩2.0.CO;2.

1000 Ferguson, R. I., M. Church, C. D. Rennie, and J. G. Venditti (2015), Reconstructing

1001 a sediment pulse: Modeling the effect of placer mining on Fraser River, Canada, J.

1002 Geophys. Res. Earth Surf., 120, 1436–1454, doi:10.1002/2015JF003491.

1003 Fernandez-Luque, R., and R. Van Beek (1976), Erosion and transport of bed-load sedi-

1004 ment, J. Hydraul. Res., 14 (2), 127–144, doi:10.1080/00221687609499677.

1005 Ferrer-Boix, C., and M. A. Hassan (2015), Channel adjustments to a succession of water

1006 pulses in gravel bed rivers, Water Resources Research, 51 (11), 8773–8790, doi:10.1002/

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 48 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

1007 2015WR017664.

1008 Fleischmann, A. S., R. C. D. Paiva, W. Collischonn, M. V. Sorribas, and P. R. M. Pontes

1009 (2016), On river-floodplain interaction and hydrograph skewness, Water Resour. Res.,

1010 52 (10), 7615–7630, doi:10.1002/2016WR019233.

1011 Frings, R. M. (2011), Sedimentary characteristics of the gravel-sand transition in the River

1012 Rhine, J. Sediment. Res., 81 (1), 52–63, doi:10.2110/jsr.2011.2.

1013 Frings, R. M., R. Düring, C. Beckhausen, H. Schüttrumpf, and S. Vollmer (2014), Fluvial

1014 sediment budget of a modern, restrained river: The lower reach of the Rhine in Germany,

1015 Catena, 122 (0), 91–102, doi:10.1016/j.catena.2014.06.007.

1016 Frings, R., K. Banhold, and I. Evers (2015), Sedimentbilanz des Oberen Rheindeltas

1017 für den Zeitraum 1991-2010, Tech. Rep. 2015.019, 43 pp., Institut für Wasserbau und

1018 Wasserwirtschaft, RWTH Aachen, Germany (in German).

1019 Gasparini, N. M., G. E. Tucker, and R. L. Bras (2004), Network-scale dynamics of grain-

1020 size sorting: implications for downstream fining, stream-profile concavity, and drainage

1021 basin morphology, Earth Surf. Process. Landf., 29 (4), 401–421, doi:10.1002/esp.1031.

1022 Gilbert, G. K. (1877), Report on the Geology of the Henry Mountains, 160 pp., US Gov-

1023 ernment Printing Office.

1024 Gomez, B., S. E. Coleman, V. W. K. Sy, D. H. Peacock, and M. Kent (2007), Channel

1025 change, bankfull and effective discharges on a vertically accreting, meandering, gravel-

1026 bed river, Earth Surf. Process. Landf., 32 (5), 770–785, doi:10.1002/esp.1424.

1027 Goodwin, P. (2004), Analytical solutions for estimating effective discharge, J. Hydraul.

1028 Eng., 130 (8), 729–738, doi:10.1061/(ASCE)0733-9429(2004)130:8(729).

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 49

1029 Hack, J. (1957), Studies of longitudinal stream profiles in Virginia and Maryland, U.S.

1030 Geol. Surv. Prof. Pap. 294-B, 59 pp., U.S. Govt. Print. Off., Washington D.C.

1031 Hey, R. D. (1996), Channel response and channel forming discharge, Final Report R&D

1032 6871-EN-01, University of East Anglia, Norwich, UK.

1033 Hirano, M. (1971), River bed degradation with armouring, Trans. Jpn. Soc. Civ. Eng,

1034 3 (2), 194–195.

1035 Hoey, T. B., and R. Ferguson (1994), Numerical simulation of downstream fining by

1036 selective transport in gravel bed rivers: Model development and illustration, Water

1037 Resour. Res., 30 (7), 2251–2260, doi:10.1029/94WR00556.

1038 Houssais, M., C. P. Ortiz, D. J. Durian, and D. J. Jerolmack (2015), Onset of sediment

1039 transport is a continuous transition driven by fluid shear and granular creep, Nat.

1040 Commun., 6, 6527, doi:10.1038/ncomms7527.

1041 Howard, A. D. (1980), Thresholds in river regimes, in Thresholds in Geomorphology,

1042 Coates, D. R. and J. D. Vitek (eds.), pp. 227–258, Allen and Unwin, Boston.

1043 Howard, A. D. (1982), Equilibrium and time scales in geomorphology: Application to

1044 sand-bed alluvial streams, Earth Surf. Process. Landf., 7 (4), 303–325, doi:10.1002/esp.

1045 3290070403.

1046 Jansen, P. Ph., L. van Bendegom, J. van den Berg, M. de Vries, and A. Zanen (1979),

1047 Principles of river engineering: the non-tidal alluvial river, 509 pp., Pitman London.

1048 Klonsky, L., and R. M. Vogel (2011), Effective measures of “effective” discharge, J. Geol.,

1049 119 (1), 1–14, doi:10.1086/657258.

1050 Kuhnle, R. A. (1992), Fractional transport rates of bedload on Goodwin Creek, in Dy-

1051 namics of gravel-bed rivers, P. Billi, R. D. Hey, C. R. Thorne, and P. Tacconi (eds.),

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 50 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

1052 John Wiley and Sons, Chichester, 141–155.

1053 Kuhnle, R. A., and J. C. Willis (1992), Mean size distribution of bed load on Goodwin

1054 Creek, J. Hydraul. Eng., 118 (10), 1443–1446, doi:10.1061/(ASCE)0733-9429(1992)118:

1055 10(1443).

1056 Lacey, G. (1930), Stable channels in alluvium, Min. Proc. Inst. Civ. Eng., 229, 259–292,

1057 doi:10.1680/imotp.1930.15592.

1058 Lane, E. W. (1955), The importance of fluvial morphology in hydraulic engineering, Proc.

1059 Am. Soc. Civ. Eng., 81 (754), 1–17.

1060 Lauer, J. W., and G. Parker (2008), Modeling framework for sediment deposition, storage,

1061 and evacuation in the floodplain of a meandering river: Theory, Water Resour. Res.,

1062 44 (4), W04425, doi:10.1029/2006WR005528.

1063 Lauer, J. W., E. Viparelli, and H. Piégay (2016), Morphodynamics and sediment tracers

1064 in 1-D (MAST-1D): 1-D sediment transport that includes exchange with an off-channel

1065 sediment reservoir, Advances in Water Resources, 93, Part A, 135–149, doi:10.1016/j.

1066 advwatres.2016.01.012.

1067 Lenzi, M., L. Mao, and F. Comiti (2006), Effective discharge for sediment transport in

1068 a mountain river: Computational approaches and geomorphic effectiveness, J. Hydrol.,

1069 326 (14), 257–276, doi:10.1016/j.jhydrol.2005.10.031.

1070 Leopold, L. B., and T. Maddock (1953), The hydraulic geometrv of stream channels and

1071 some physiographic implications, Prof. Paper 252, US Dept. of the Interior, Geological

1072 Survey.

1073 Li, C., M. J. Czapiga, E. C. Eke, E. Viparelli, and G. Parker (2014), Variable shields

1074 number model for river bankfull geometry: bankfull shear velocity is viscosity-dependent

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 51

1075 but grain size-independent, J. Hydraul. Res., 53 (1), 36–48, doi:10.1080/00221686.2014.

1076 939113.

1077 Litty, C. and F. Schlunegger (2016), Controls on pebbles’ size and shape in streams of the

1078 Swiss Alps, The Journal of Geology, 125, doi:10.1086/689183.

1079 Mackin, J. H. (1948), Concept of the graded river, Geol. Soc. Am. Bull., 59 (5), 463–512,

1080 doi:10.1130/0016-7606(1948)59.

1081 Meyer-Peter, E., and R. Müller (1948), Formulas for bed-load transport, in Proc. 2nd

1082 Meeting Int. Assoc. Hydraul. Struct. Res., pp. 39–64, Stockholm.

1083 Mosselman, E. (2009), Bank protection and river training along the braided Brahmaputra-

1084 Jamuna River, Bangladesh, in Braided Rivers: Process, Deposits, Ecology and Manage-

1085 ment, G. H. Sambrook Smith, J. L. Best, C. S. Bristow, and G. E. Petts (eds.), Blackwell,

1086 Oxford, UK, 277–287, doi:10.1002/9781444304374.ch13.

1087 Nash, D. B. (1994), Effective sediment-transporting discharge from magnitude-frequency

1088 analysis, J. Geol., 102 (1), 79–95.

1089 NEDECO (1959), River studies and recommendations of Niger and Benue, North-Holland

1090 Publ. Amsterdam, 1000p.

1091 Paola, C. (2001), Modelling stream braiding over a range of scales, in Gravel-bed Rivers

1092 V, Mosley, M.P. (ed.), pp. 11–46, Wellington, New Zealand Hydrological Society.

1093 Parker, G. (1978a), Self-formed straight rivers with equilibrium banks and mobile

1094 bed. part 1. the sand-silt river, J. Fluid Mech., 89 (1), 109–125, doi:10.1017/

1095 S0022112078002499.

1096 Parker, G. (1978b), Self-formed straight rivers with equilibrium banks and mobile bed.

1097 part 2. the gravel river, J. Fluid Mech., 89 (1), 127146, doi:10.1017/S0022112078002505.

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 52 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

1098 Parker, G., and P. C. Klingeman (1982), On why gravel bed streams are paved, Water

1099 Resour. Res., 18 (5), 1409–1423, doi:10.1029/WR018i005p01409.

1100 Parker, G., S. Dhamotharan, and H. Stefan (1982b), Model experiments on mobile, paved

1101 gravel bed streams, Water Resour. Res., 18 (5), 1395–1408.

1102 Parker, G. (1990a), Surface-based bedload transport relation for gravel rivers, J. Hydraul.

1103 Res., 28 (4), 417–436, doi:10.1080/00221689009499058.

1104 Parker, G. (1990b), The “Acronym” series of Pascal programs for computing bed load

1105 transport in gravel rivers., External Memorandum M-220, University of Minnesota, St.

1106 Anthony Falls Hydraulic Laboratory.

1107 Parker, G. (1991), Selective sorting and abrasion of river gravel. I: Theory, J. Hydraul.

1108 Eng., 117 (2), 131–147, doi:10.1061/(ASCE)0733-9429(1991)117:2(131).

1109 Parker, G. (2004a), Response of the gravel bed of a mountain river to a hydrograph, in

1110 Proc. Int. Conf. on Slopeland Disaster Mitigation, Taipei, Taiwan, October 5-6.

1111 Parker, G. (2004b), 1D Sediment transport morphodynamics with applications to

1112 rivers and turbidity currents, E-Book, Available at http://hydrolab.illinois.edu/

1113 people/parkerg/morphodynamics_e-book.htm.

1114 Parker, G., C. Paola, and S. Leclair (2000), Probabilistic Exner sediment continuity

1115 equation for mixtures with no active layer, J. Hydraul. Eng., 126 (11), 818–826, doi:

1116 10.1061/(ASCE)0733-9429(2000)126:11(818).

1117 Parker, G., P. R. Wilcock, C. Paola, W. E. Dietrich, and J. Pitlick (2007), Physical

1118 basis for quasi-universal relations describing bankfull hydraulic geometry of single-

1119 thread gravel bed rivers, J. Geophys. Res., Earth Surface, 112, F04,005, doi:10.1029/

1120 2006JF000549.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 53

1121 Parker, G., M. A. Hassan, and P. Wilcock (2008), Adjustment of the bed surface size

1122 distribution of gravel-bed rivers in response to cycled hydrograph, in Gravel-Bed Rivers

1123 VI: From Process Understanding to River Restoration, Habersack, H., H. Piegay and

1124 M. Rinaldi (eds.), 241–285, doi:10.1016/S0928-2025(07)11127-5.

1125 Pfeiffer, A. M., N. J. Finnegan, and J. K. Willenbring (2017), Sediment supply controls

1126 equilibrium channel geometry in gravel rivers, Proc. Nat. Academy of Sciences, 114 (13),

1127 3346–3351, doi:10.1073/pnas.1612907114.

1128 Phillips, B. C., and A. J. Sutherland (1989), Spatial lag effects in bed load sediment

1129 transport, J. Hydraul. Res., 27 (1), 115–133, doi:10.1080/00221688909499247.

1130 Phillips, C. B., and D. J. Jerolmack (2016), Self-organization of river channels as a critical

1131 filter on climate signals, Science, 352 (6286), 694–697, doi:10.1126/science.aad3348.

1132 Pickup, G., and W. A. Rieger (1979), A conceptual model of the relationship between

1133 channel characteristics and discharge, Earth Surf. Process. Landf., 4 (1), 37–42, doi:

1134 10.1002/esp.3290040104.

1135 Pickup, G., and R. Warner (1976), Effects of hydrologic regime on magnitude and fre-

1136 quency of dominant discharge, J. Hydrol., 29 (1), 51–75, doi:10.1016/0022-1694(76)

1137 90005-6.

1138 Prins, A. (1969), Dominant discharge, Tech. Rep. S 78-III, 134 pp., Waterloopkundig

1139 Laboratorium Delft, Netherlands.

1140 Prins, A., and M. de Vries (1971), On dominant discharge concepts for rivers, in Proc.

1141 XIVth IAHR Congr. Paris, France.

1142 Ribberink, J. S. (1987), Mathematical modelling of one-dimensional morphological

1143 changes in rivers with non-uniform sediment, Ph.D. thesis, 206 pp., Delft University

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 54 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

1144 of Technology, The Netherlands.

1145 Saint-Venant, A. J. C. B. (1871), Théorie du mouvement non permanent des eaux, avec

1146 application aux crues des rivières et à l’introduction des marées dans leur lit, Comptes

1147 Rendus des séances de l’Académie des Sciences, 73, 237–240 (in French).

1148 Stecca, G., A. Siviglia, and A. Blom (2014), Mathematical analysis of the Saint-Venant-

1149 Hirano model for mixed-sediment morphodynamics, Water Resour. Res., 50, 7563–7589,

1150 doi:10.1002/2014WR015251.

1151 Stecca, G., A. Siviglia, and A. Blom (2016), An accurate numerical solution to the Saint-

1152 Venant-Hirano model for mixed-sediment morphodynamics in rivers, Adv. Water Re-

1153 sour., 93, Part A, 39–61, doi:10.1016/j.advwatres.2015.05.022.

1154 Surian, N., L. Mao, M. Giacomin, and L. Ziliani (2009), Morphological effects of different

1155 channel-forming discharges in a gravel-bed river, Earth Surf. Process. Landf., 34 (8),

1156 1093–1107, doi:10.1002/esp.1798.

1157 Syvitski, J. P., M. D. Morehead, D. B. Bahr, and T. Mulder (2000), Estimating fluvial

1158 sediment transport: The rating parameters, Water Resour. Res., 36 (9), 2747–2760,

1159 doi:10.1029/2000WR900133.

1160 Tal, M. and C. Paola (2010), Effects of vegetation on channel morphodynamics: results

1161 and insights from laboratory experiments, Earth Surf. Process. Landf., 35 (9), 1014–

1162 1028, doi:10.1002/esp.1908.

1163 Tealdi, S., C. Camporeale, and L. Ridolfi (2011), Long-term morphological river response

1164 to hydrological changes, Advances in Water Resources, 34 (12), 1643–1655, doi:10.1016/

1165 j.advwatres.2011.08.011.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 55

1166 Tucker, G. E., and R. L. Bras (1998), Hillslope processes, drainage density, and landscape

1167 morphology, Water Resour. Res., 34 (10), 2751–2764.

1168 Van Bendegom, L. (1967), Algemene waterbouwkunde. deel i: De natuur, Tech. rep., Afd.

1169 der Weg- en Waterbouwkunde, Technische Hogeschool Delft (in Dutch).

1170 Van der Scheer, P., J. S. Ribberink, and A. Blom (2002), Transport formulas for graded

1171 sediment: behaviour of transport formulas and verification with data, Tech. Rep. 2002R-

1172 002, 124pp, Civil Engineering, University of Twente, Netherlands.

1173 Vargas-Luna, A. (2016), Role of vegetation in river bank accretion, Ph.D. thesis, 241 pp.,

1174 Delft University of Technology, Netherlands.

1175 Viparelli, E., D. Gaeuman, P. Wilcock, and G. Parker (2011), A model to predict the

1176 evolution of a gravel bed river under an imposed cyclic hydrograph and its application

1177 to the Trinity River, Water Resour. Res., 47 (2), W02533, doi:10.1029/2010WR009164.

1178 Viparelli, E., J. W. Lauer, P. Belmont, and G. Parker (2013), A numerical model to

1179 develop long-term sediment budgets using isotopic sediment fingerprints, Computers &

1180 Geosciences, 53 (0), 114–122, doi:10.1016/j.cageo.2011.10.003.

1181 Viparelli, E., R. R. H. Moreira, and A. Blom (2017), Modelling stratigraphy-based gravel-

1182 bed river morphodynamics, in Gravel Bed Rivers: Process and Disasters, D. Tsutsumi

1183 and J. Laronne (eds.), Wiley Blackwell, 609–637.

1184 Vogel, R. M., J. R. Stedinger, and R. P. Hooper (2003), Discharge indices for water quality

1185 loads, Water Resour. Res., 39 (10), 1–9, doi:10.1029/2002WR001872.

1186 Wathen, S. J., R. I. Ferguson, T. B. Hoey, and A. Werritty (1995), Unequal mobility of

1187 gravel and sand in weakly bimodal river sediments, Water Resour. Res., 31 (8), 2087–

1188 2096, doi:10.1029/95WR01229.

D R A F T July 7, 2017, 11:40pm D R A F T

 X - 56 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

1189 Wilcock, P. R., and J. C. Crowe (2003), Surface-based transport model for mixed-size

1190 sediment, J. Hydraul. Eng., 129 (2), 120–128, doi:10.1061/(ASCE)0733-9429(2003)129:

1191 2(120).

1192 Wilcock, P. R., and B. T. DeTemple (2005), Persistence of armor layers in gravel-bed

1193 streams, Geophys. Res. Lett., 32 (8), doi:10.1029/2004GL021772, l08402.

1194 Wilcock, P. R., S. T. Kenworthy, and J. C. Crowe (2001), Experimental study of the

1195 transport of mixed sand and gravel, Water Resour. Res., 37 (12), 3349–3358, doi:10.

1196 1029/2001WR000683.

1197 Williams, G. P. (1978), Bankfull discharge of rivers, Water Resour. Res., 14 (6), 1141–

1198 1154, doi:10.1029/WR014i006p01141.

1199 Wolman, M. G., and L. B. Leopold (1957), River flood plains: Some observations on

1200 their formation, U.S. Geol. Surv. Prof. Pap. 282-C, 30 pp., U.S. Govt. Print. Off.,

1201 Washington D.C.

1202 Wolman, M. G., and J. P. Miller (1960), Magnitude and frequency of forces in geomorphic

1203 processes, J. Geol., 68 (1), 54–74.

1204 Wong, M., and G. Parker (2006), One-dimensional modeling of bed evolution in a gravel

1205 bed river subject to a cycled flood hydrograph, J. Geophys. Res., Earth Surface,

1206 111 (F3), F03018, doi:10.1029/2006JF000478.

1207 Zhou, Z., G. Coco, I. Townend, M. Olabarrieta, M. van der Wegen, Z. Gong, A. DAlpaos,

1208 S. Gao, B. E. Jaffe, G. Gelfenbaum, Q. He, Y. Wang, S. Lanzoni, Z. Wang, H. Winterw-

1209 erp, and C. Zhang (2017), Is morphodynamic equilibrium an oxymoron?, Earth-Science

1210 Reviews, 165, 257–267, doi:10.1016/j.earscirev.2016.12.002.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 57

fw a Pw α b
1−α

Qw Qw
Qwbase Qwpeak
Figure 1. Statistics of the flow rate expressed by (a) a probability density function

(PDF) of water discharge for an arbitrary long-term hydrograph and (b) the probabilities

of a two-modes water discharge.

‘normal flow’ zone backwater zone

peak flow

dη base flow
S=− sea
dx

Figure 2. Schematic of variable flow under a constant base level in a mildly sloping river:

the ‘normal flow’ zone and the backwater zone. Although the figure shows one normal

flow zone and one backwater zone, a river typically consists of multiple backwater zones

and normal flow zones. The proposed formulations for the equilibrium river geometry

under variable flow are valid in river normal flow zones.

D R A F T July 7, 2017, 11:40pm D R A F T

X - 58 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

water discharge Q w [m /s]

3
a
10000

5000

0
0 100 200 300
time [days]
probability density f [s/m3 ]

0.01
b
w

0.005

0
0 5000 10000
3
water discharge Q w [m /s]

Figure 3. (a) Synthetic hydrograph at the upstream end of the reach and (b) the

associated PDF of water discharge.

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 59

-4
10
5
a GR

channel slope S [-]

4
WC
3 AM

0
1
b
gravel fraction F [-]

0.8
bed surface

0.6

0.4 analytical
backwater
0.2
Saint-Venant
0
0 0.2 0.4 0.6 0.8 1
gravel fraction in the load p [-]
g

Figure 4. Validation of the analytical model: (a) equilibrium channel slope, S, in

the river normal flow zone, and (b) equilibrium bed surface gravel content, F , predicted

using the analytical model and two numerical time-marching models as a function of

the mean gravel content in the load, p̄g . The data of the time-marching models are

associated with a location 20 km downstream from the upstream end of the domain.

Predictions are made based on the generalized power law load relation (GR, settings

r = 0.05, wg = ws = w = 0.4, and c = 2.3) in Eq. (19), the Wilcock and Crowe [2003]

load relation (WC), and the Ashida and Michiue [1972] load relation (AM).

D R A F T July 7, 2017, 11:40pm D R A F T

X - 60 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

slope adjustment
10 -4
1

gravel fraction F [-]

channel slope S [-]

bed surface
0.5 GR
2 WC
a AM b
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/
w w
[-] /
w w
[-]

width adjustment
1
channel width B [m]

gravel fraction F [-]

400

bed surface
0.5
200
c d
0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
/ [-] / [-]
w w w w

Figure 5. Predicted channel response to a changing boundary condition through ad-

justment of channel slope (subfigures a-b) versus adjustment of channel width (subfigures

c-d). The considered boundary condition is the variability of the flow rate, which is ex-

pressed by σw /µw with σw denoting the standard deviation of the flow rate and µw its

mean value. Subfigures show predicted equilibrium values in the river normal flow zone,

for the case of adjustment of channel slope, of (a) channel slope, S, (b) surface gravel

content, F , and, for the case of adjustment of channel width, (c) channel width, B, and

(d) surface gravel content, F . Predictions are made using the explicit and implicit ana-

lytical models based on the generalized power law load relation (GR, settings r = 0.05,

wg = ws = w = 0.4, and c = 2.3) in Eq. (19), the Wilcock and Crowe [2003] load relation

(WC), and the Ashida and Michiue [1972] load relation (AM).

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 61

HBL ‘normal flow’ zone backwater zone

peak flow

dη base flow
S=− sea
dx

Figure 6. Schematic of variable flow under a constant base level: the hydrograph

boundary layer (HBL), the ‘normal flow’ zone, and the backwater zone.
water discharge Q w [m /s]
3

a
10000

5000

0
1910 1930 1950 1970 1990 2010
time [years]
-4
probability density fw [s/m ]

×10
3

6
b
4

0
0 5000 10000
3
water discharge Q w [m /s]

Figure 7. Measured flow rate in the Rhine River at Lobith in the period 1901-

2013 (data courtesy: Rijkswaterstaat): (a) the water discharge and (b) the associated

probability density function.

D R A F T July 7, 2017, 11:40pm D R A F T

X - 62 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

-4 GR
10
6
water
[s/m ]

a
3

4 sand
gravel
x

2 total
fw ,

0
10 -4 WC
6
[s/m ]

b
3

4
x

2
fw ,

0
10 -4 AM
6
[s/m ]

c
3

4
x

2
fw ,

0
0 2000 4000 6000 8000 10000 12000
3
water discharge Q [m /s]
w

Figure 8. The predicted ‘normal flow load distribution’, νX , for gravel (in red), sand

(in blue), and the total load (in green), which expresses the contribution of the long-term

range of water discharge to transporting the gravel, sand, and total load, for the Lobith

case shown in Figure 7. Analytical predictions are made based on (a) the generalized

power law load relation (GR, settings r = 0.05, wg = ws = w = 0.4, and c = 2.3) in

Eq. (19), (b) the Wilcock and Crowe [2003] load relation (WC), and (c) the Ashida and

Michiue [1972] load relation (AM). Dashed vertical lines indicate the channel-forming

discharge and the vertical bars the effective discharge (sand in blue, gravel in red, and

total in green). The grey area is the PDF of water discharge, fw .

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 63

r [-] w [-] c [-]

-1 0 1 0 0.5 1 2 4

FLvB
EH
AM
WC

0
10
q*s [-]

10 -5 a b c

10 -1 10 0 10 -1 10 0 10 -1 10 0
* * *
[-] [-] [-]
s s s

0
10
qg [-]

-5
10
*

d e f
-10
10
10 -2 10 -1 10 -2 10 -1 10 -2 10 -1
* * *
[-] [-] [-]
g g g

Figure 9. Comparison of the generalized load relation in Eq. (B1) or (B3) to other load

relations for a gravel-bedded river with a bed surface consisting of 90% gravel. Upper plots

show the predicted nondimensional sand load (sand size Ds = 1 mm) and lower plots the

gravel load (gravel size Dg = 10 mm), for a range of values for r, w (wg = ws = w), and

c. In the base case r = 0.05, w = 0.4, and c = 2.3. Besides the generalized load relation,

predictions are made using the Engelund and Hansen [1967] (EH), Fernandez-Luque and

Van Beek [1976] (FLvB), Ashida and Michiue [1972] (AM), and Wilcock and Crowe [2003]

(WC) load relations.

D R A F T July 7, 2017, 11:40pm D R A F T

X - 64 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW

Table 1. Overview of equations for the equilibrium river geometry under variable flow in a

river normal flow zonea

GENERALIZED LOAD RELATION (GR) ENGELUND AND HANSEN (1967) (EH)

(provided that wg = ws = w)

channel-forming discharge
[∫ ]3/(2w+3) [∫ ]3/5
∞ (2w+3)/3 ∞ 5/3
Qwdom = 0
Qw fw (Qw ) dQw Qwdom = 0
Qw fw (Qw ) dQw
( )3/(2w+3) ( )3/5
(2w+3)/3 (2w+3)/3 5/3 5/3
Qwdom = αQwbase + (1 − α)Qwpeak Qwdom = αQwbase + (1 − α)Qwpeak

UNISIZE BLOCK 1 BLOCK 2

( )3/(2w+3) ( )3/5
Cf B 2w/(2w+3) Dw Q̄ Cf B 2/5 D
S= S= Q̄
gQwdom G (D/Dref )r gQwdom GEH

Qw (t)2/3 Q
1/3 ( r )1/(2w+3)
Qw (t)2/3 Q
1/3 ( )1/5
wdom G (D/Dref ) wdom GEH 1
H(t) = B (2w+2)/(2w+3) Dw Q̄
H(t) = B 4/5 D Q̄
( )1/3 ( )1/(2w+3) ( )1/3 ( )1/5
Qw (t) Dw Q̄ Qw (t) D Q̄
U (t) = Qwdom G B (D/Dref )r
U (t) = Qwdom GEH B

MIXED BLOCK 3 BLOCK 4

2w/(2w+3)
( w
)3/(2w+3) 2/5 ( Ds µ )3/5
Cf B Ds µ Q̄ Cf B
S= S= Q̄
gQwdom G (Dg /Dref )r gQwdom GEH
( )w
1 Dg 1 Dg
F = µ Ds
p̄g F = µ Ds
p̄g
1/3
2/3 ( r )1/(2w+3) Qw (t)2/3 Q
1/3 ( )1/5
Qw (t) Q
wdom G (Dg /Dref ) wdom GEH 1
H(t) = B (2w+2)/(2w+3)
wµ
Ds Q̄
H(t) = B 4/5 Ds µ Q̄
( )1/3 ( w
)1/(2w+3) ( )1/3 ( )1/5
Qw (t) Ds µ Q̄ Qw (t) Ds µ Q̄
U (t) = Qwdom G B (Dg /Dref )r
U (t) = Qwdom GEH B

( )r ( )w
Dg Dg Dg
µ= Ds
(1 − p̄g ) + Ds
p̄g µ = 1 − p̄g + Ds
p̄g

MIXED & BLOCK 5 BLOCK 6

2w/(2w+3)
( w
)3/(2w+3) 2/5 ( Ds µ )3/5
Cf B Ds µ Q̄0 Cf B
ABRASION S = S= Q̄0
gQwdom G (Dgx /Dref )r gQwdom GEH
( )w ∗ ∗
p̄g0 e−xκ p̄ e−xκ
1 Dgx 1 Dgx
F = µ Ds
F = µ Ds g0

Qw (t)2/3 Q
1/3 ( r )1/(2w+3) Qw (t)2/3 Q
1/3 ( )1/5
wdom G (Dgx /Dref ) wdom GEH 1
H(t) = B (2w+2)/(2w+3)
wµ
Ds Q̄0
H(t) = B 4/5 Ds µ Q̄0
( )1/3 ( w
)1/(2w+3) ( )1/3 ( )1/5
Qw (t) Ds µ Q̄0 Qw (t) Ds µ Q̄0
U (t) = Qwdom G B (Dgx /Dref )r
U (t) = Qwdom GEH B

( )r ( ∗ ) ( ∗ )
1 − p̄g0 + kss p̄g0 (1 − e−xκ ) + µ = 1 − p̄g0 + kss p̄g0 1 − e−xκ +
Dgx
µ= Ds
( )w ∗ ∗
p̄g0 e−xκ p̄ e−xκ
Dgx Dgx
Ds Ds g0
a
The set of equations in Block 5 is the fundamental set of equations: the equations in the
other blocks can be found by simplifying the ones in Block 5. For the case of a single steady
discharge the equations in Block 6 reduce to the ones derived by Blom et al. [2016].

D R A F T July 7, 2017, 11:40pm D R A F T

 BLOM ET AL.: EQUILIBRIUM RIVER UNDER VARIABLE FLOW X - 65

Table 2. Measured and predicted channel slope S and bed surface gravel content, F , at Lobithb

measured GR WC AM

S ·10−5 (-) 11 5.7 5.6 4.8

F (-) 0.6 0.31 0.47 0.53

Predictions are based on the generalized power law load relation (GR, settings r = 0.05,
b

wg = ws = w = 0.4, and c = 2.3) in Eq. (19), the Wilcock and Crowe [2003] load relation (WC),
and the Ashida and Michiue [1972] load relation (AM), under the assumption of normal flow and
the hydrograph shown in Figure 7. Measured data originate from Frings [2011].

D R A F T July 7, 2017, 11:40pm D R A F T

View publication stats