Catena 121 (2014) 112

Contents lists available at ScienceDirect

Catena
journal homepage: www.elsevier.com/locate/catena

Assessing channel response of a long river inuenced by

human disturbance
A. Latapie a, B. Camenen a, S. Rodrigues b, A. Paquier a, J.P. Bouchard c, F. Moatar b
a
b
c

Irstea Lyon, 5 rue de la Doua CS70077, 69626 Villeurbanne Cedex, France

Universit Franois Rabelais de Tours, E.A. 6293 GHCO, GoHydosystmes Continentaux, Facult des Sciences et Techniques, Parc de Grandmont, 37200 Tours, France
EDF-LNHE, 6 Quai Watier, 78400 Chatou, France

a r t i c l e

i n f o

Article history:
Received 20 September 2013
Received in revised form 25 April 2014
Accepted 27 April 2014
Available online xxxx
Keywords:
Morphological evolution
Hydraulic model
Reach delineation
Gravel mining
Middle Loire River

a b s t r a c t
This paper describes an approach to assess channel changes of a long anthropogenised river (the Middle Loire
River) over decadal timescales. Channel changes are evaluated along geomorphically homogeneous river
reaches. The classic geomorphic parameters (active channel width, bed slope, grain size) are complemented
with parameters extracted from a 1D hydraulic model: widthdepth ratio, effective bed shear stress and specic
stream power calculated for the biennial discharge assimilated to bankfull ow conditions. The delineation of
reaches is undertaken by combining visual inspection with the implementation of simple statistical tests to
corroborate discontinuities in ow and sediment transport. The 450 km long study area has been divided into
167 homogeneous reaches. The general trends observed over the last fty years are narrowing of the active channel
width and incision of the river bed. However, changes in bed level and active channel width are not consistently
correlated. Channel changes at the reach scale are mainly controlled by the presence of former sediment extraction
sites. Signicant incision is observed at the peak of the in-stream sediment mining period. This was followed by
channel recovery when extractions stopped. The 1D numerical model provides a more rigorous manner to derive
hydraulic parameters. The effective bed shear stress made dimensionless by its critical value and the width
depth ratio helps to explain observed channel responses more effectively by relating patterns to geological units
along the Middle Loire.
2014 Elsevier B.V. All rights reserved.

1. Introduction
Natural alluvial channels adjust their slope and shape in response to
water and/or sediment inputs, whether as a consequence of climate
change or anthropogenic inuences. Since channel changes can have
detrimental effects on ood extent, groundwater recharge and the
stability of infrastructure and ecology (Bravard et al., 1999), better understanding of channel adjustments is crucial to predict future evolution
and thus adapt river management strategies.
As river managers are working to understand and/or reverse channel
changes, geomorphic reach scale assessments are being developed to
support restoration interventions (e.g. Brierley and Fryirs, 2005;
Kellerhals et al., 1976; Montgomery and Bufngton, 1997; Rinaldi et al.,
2013; Rosgen, 1994). These approaches tend to perform poorly on
anthropogenised reaches as human inuences such as channel embankments and gravel extraction have signicantly altered some of the basic
variables that go into most classication schemes.
Often, designation of reach boundaries can be subjective and difcult
to reproduce (Miller and Ritter, 1996; Simon et al., 2007). In order to
improve the detection of threshold conditions along the river continuum,

E-mail address: [email protected] (B. Camenen).

http://dx.doi.org/10.1016/j.catena.2014.04.017
0341-8162/ 2014 Elsevier B.V. All rights reserved.

statistical algorithms have been implemented to enhance the robustness

of reach denitions (Leviandier et al., 2012; Orlowski et al., 1995). The use
of statistical algorithms for detecting homogeneous patches along a continuum has been widely applied in hydrology, ecology and geography
(e.g. Buishand, 1984; Clark et al., 2008; Clment and Pigay, 2003).
Whilst reach scale approaches are undeniably valuable, they generally contain limited analysis of vertical characteristics, with water depth
often estimated on site (Rosgen, 1994), or using hydraulic geometry
relationships (Leopold and Maddock, 1953), or a simple Manning
Strickler equation, or more recently using hydraulic models (Armas
et al., 2012). Site assessment is obviously not appropriate for long
river reaches. Although empirical equations can provide an acceptable
approximation, hydraulic models provide more robust information
(see Aggett and Wilson, 2009; Armas et al., 2012; Zilani and Surian,
2012).
This paper uses multi-disciplinary tools to analyse channel changes
among a relatively large river, the Middle Loire (France), that has been
characterized by different channel types and inuenced by human
disturbances. The specic objectives are: (1) to delineate homogeneous
river reaches using a visual distinction of changes approach and simple
statistical tests, (2) to analyse channel changes observed on the Middle
Loire River over the last 50 years and to assess if reaches delimited by
natural features behave differently from anthropogenized ones, (3)

A. Latapie et al. / Catena 121 (2014) 112

oods happened, but on a smaller scale. Dams built at the upstream end
of the catchment have only impacted upon the 1996, 2003 and 2008
oods.
At Gien, located 564 km downstream from source, ood events with
a return period of 2 years correspond to a discharge of 1600 m3/s. Fig. 2b
examines the ood frequency and magnitudes of ow recorded at Gien
between 1936 and 2011. The data were compiled from peak annual
discharges with a threshold of 1600 m3/s. Floods above 2700 m3/s
(equivalent to an event with a ten year return period) have been relatively rare throughout the period (5 occurrences).

to relate channel adjustments to hydraulic parameters computed

with a 1D hydraulic model, and (4) to assess the inuence of instream
sediment mining on the evolution of the Middle Loire River.
2. Study area
The Loire River is the longest river in France with a length of 1012 km.
Its drainage area covers 117,000 km2, one fth of France's area. Four
major tributaries feed the river: Allier, Cher, Vienne and Maine. Bedload
sediment inputs come mainly from the Allier and the Vienne. The reach
studied is the Middle Loire, which extends from the conuence with
the Allier River to the conuence with the Maine River. The downstream
boundary was extended to the gauging station of Montjean-sur-Loire to
facilitate hydraulic modelling, that is a distance of about 450 km
(Fig. 1). The Middle Loire River is characterized by a section with a multiple channel conguration downstream of the conuence with the Allier
River, a short meandering section upstream of Orlans and a multiple
channel system with the presence of numerous vegetated islands and
sand bars in the downstream section.

2.2. Geological settings

The Loire River basin is characterized by various lithological units
inuencing its layout and valley setting (Babonaux, 1970; Bross,
1982). Upstream of Gien (Fig. 1), the north/south alignment of the river
is associated with faults in the Hercynian socle (Alcayd and Gigout,
1976; Debrand-Passard et al., 1998; Nelhig, 2010). On the Middle Loire,
four geological units can be distinguished: Jurassic limestones from the
conuence with the Allier River to 90 km downstream, then Tertiary
formation (lacustrine limestones and alluvial formation) to downstream
of Blois (located at a distance of 696 km from source), Cretaceous chalks
to the conuence with the Maine River and Hercynian metamorphic
rocks on the downstream section (Fig. 3). The rst two geological units
are characterized by similar mean slope (sb1 sb2 0.4 m/km), whereas
units 3 and 4 have a lower mean slope value (sb3 = 0.29 m/km;
sb4 = 0.18 m/km).
The nature of sediment available is associated with crystalline and
volcanic rocks from the upstream Loire, locally enriched by int coming
from sedimentary formations of the Parisian Basin. Sediments are dominated by coarse sands, quartz, feldspar and gravels (Macaire et al.,
2013). Champion et al. (1971) investigated the thickness of alluvia in
the substratum and reported layers varying between 3 m at Amboise
(730 km from source) to 10 m at Cosne-Cours-sur-Loire (522 km from
source) with an average value of 5 m. Alluvia thickness is the largest
where the oodplain is wide and in conuences zones (Donsimoni
et al., 2008).

2.1. Hydrologic characteristics

The Middle Loire River has a highly variable hydrologic regime: very
low discharge during the summer and high magnitude ows in winter
and spring. Cumulative departure from the mean annual discharge for
the period 18632011 is presented for Blois gauging station in Fig. 2a.
Blois is representative of the hydrologic regime of the Middle Loire; it
is located 250 km downstream of the conuence with the Allier River.
The curve ascends during wet periods and descends during drier
periods. Despite major ood events in 1846, 1856 and 1866, the end
of the nineteenth century presents a relatively stable ow regime.
Alternating wet (from 1905 to 1941 and from 1973 to 1984) and dry
periods (19411964 and 1984-today) are evident.
The last major ood events on the Loire River were recorded in the
second half of the nineteenth century. These extreme events had an estimated return period of 200 years and reached an estimated discharge
of 7 200 m3/s at Gien (Dacharry, 1996; Duband, 1996). Since then, other

la Mayenne

PARIS

e
rth

la

Sa

la Maine
MONTJEAN
SUR
LOIRE
(PK897)

ORLEANS
(PK635)

ir
le Lo

TOURS
(PK755)

SAUMUR
(PK823)

Loire

GIEN
(PK564)

LYON

BLOIS
(PK696)

l'Indr
e

la Creuse

le

Ch

er

NEVERS
(PK445)
la
Lo
ire
l'A
llie
r

la Vienne

Key
Dams
Power plants
Main cities

Fig. 1. Location map showing the study reach.

50

100 km

A. Latapie et al. / Catena 121 (2014) 112

(b)

Cumulative departure of Qannual from

mean value (m3/s)

1200
1000
800

drier

600
wetter

400
200
0
200

400
1860 1880 1900 1920 1940 1960 1980 2000 2020

Annual maximum daily discharge (m3/s)

(a)
1400

3200
3000
2800

Q10

2600
2400

Q5

2200
2000
1800
1600

1935

1945

1955

1965

1975

1985

1995

2005

Time (year)

Time (year)

Fig. 2. (a) Cumulative departure from the mean annual discharge for the period 18632011 at Blois gauging station, (b) Annual maximum daily discharge above Q2 (Q2 = 1 600 m3/s) for
the period 19362011 at Gien gauging station.

The valley of the Middle Loire River is naturally large, 3.5 km wide on
average. Lateral migration of the channel did not occur over the last
centuries because of anthropogenic inuences.
2.3. Anthropogenic inuences
The Loire River has been intensively modied over the last centuries.
Flood embankments were constructed between the 13th and 18th
century and have been raised following ood events (Dion, 1961). The
Middle Loire River is now constricted along 300 km of its length. Flood
embankments have conned the ow and prevented lateral migration
of the channel. Oblique groynes were erected in the 19th century to
maintain a single channel, to ensure sufcient depth conditions during
the low ow season and thus facilitate navigation in the main channel.
These structures have caused the deepening of the main channel as
intended but sediments have accumulated behind them, inducing the
development of vegetation, closing secondary channels and thus reducing
ood conveyance (Rodrigues et al., 2006).
The construction of bridges also inuences the river by conning the
ow and stabilizing the river bed: from 30 bridges in 1850, 50 bridges
can now be found on the Middle Loire. The presence of old bridge
piers also induces some ow disruption. As energy needs grew during
the 20th century, three dams were constructed at the upstream end of
180
b1

0.44 m/km

b2

160

Altitude above sea level (m)

3. Data and methods

Sb3 0.29 m/km

0.41 m/km

b4

0.18 m/km

140
120
100
80
60
40
UNIT 1:

20 JURASSIC
LIMESTONES

0
450

500

UNIT 2:
TERTIARY FORMATION
(sand, clay and limestones)

550

600

650

700

UNIT 3:
CRETACEOUS
CHALKS

750

800

3.1. Data and methodological framework

UNIT 4: HERCYNIAN METAMORPHIC ROCKS

the catchment, along with four nuclear power stations in the Middle
Loire (Fig. 1).
Sand and gravel extraction in the main channel started in an industrialized manner at the beginning of the 1950s. Dambre and Malaval
(1993) reported that nearly 83 106 m3 of sediment was extracted between 1949 and 1992 (i.e., about 2 106 m3/year in average). Given
that the mean annual total sediment transport (i.e., bedload and
suspended load) in the Loire River was estimated at 405,000 m3 for
the period 19531968 (Berthois, 1971), the sediment extraction rate
was higher than the replenishment rate. Extractions in the main channel
were signicantly reduced following a ministerial recommendation in
1981 before being formally prohibited in 1995. The decision to reduce
mining in the main channel was also dictated by the collapse of the
Wilson's Bridge in Tours in 1978, due to the scouring of a pier. Changes
in agricultural and human uses of the river banks should also be noted.
Islands and river banks used to be maintained as pasture lands
(Gautier et al., 2000). As this practice stopped, the vegetation developed,
affecting the connection of secondary channels at low ow and sediment
transport.
Since the end of granular extractions, local remedial actions have
been undertaken: protecting the toe of embankments, increasing ow
capacity by cutting down vegetation in secondary channels, removing
oblique groynes to restore ow conveyance etc. (Belleudy, 2000).

850

900

Distance from source (km)

Fig. 3. Longitudinal prole and the four main geological units of the Middle Loire River,
with their average slope.

The regional environment agency (DREAL Centre) regularly monitored two hundred sites to record water levels at low and high ows
since 1978. Aerial photographs taken during low ow conditions are
available for 1955, 1984, 1995, 2002 and 2010. A topographic survey
of the Middle Loire River was undertaken in 1995. Cross sections,
surveyed every 2 km on average, cover the main channel and can be
completed with oodplain data extracted from Lidar data collected in
2003. These data have allowed the determination of geomorphic characteristics and the construction of a hydraulic model.
3.2. Geomorphic parameters
3.2.1. Active channel width and change in river bed level
The longitudinal pattern of the active channel width was dened
using aerial photographs. The active channel width B corresponds to
the non-vegetated river width and thus includes secondary channels
and sand bars frequently altered during ood events. The channel active

A. Latapie et al. / Catena 121 (2014) 112

widths were measured on the 2002 and 2010 aerial photographs every
500 m in a GIS environment. This distance is consistent with the size of
the macroforms and complements previous studies undertaken by
Ginestet (1999) on the older photography dataset (1955, 1984, 1995).
The oodplain width Bf was extracted from the 2003 Lidar data. The resolution of the aerial photographs used to determine the active channel
width is relatively good but we cannot exclude distortion effects for
measurements away from the central photograph area. In particular,
the digitalization of river banks is challenging in areas covered by vegetation. The error is estimated at a maximum of 25 m for the dataset,
that is an error of about 7%.
Changes in river bed levels were calculated by assimilating the low
water prole to the bed prole. Comparisons were made for low
water levels recorded at similar discharges (19781986, 19861995,
19952002 and 20032011). These data were also used to estimate
the river bed slope. Errors associated with low water level measurements are estimated to be in the range of 3 cm.
3.2.2. Median grain size
The river bed material median grain size d50 was determined for all
granulometric sample analysis available in the main channel between
1970 and 2008 provided by the DREAL Centre. Despite the uncertainties
associated with the denition of a single river bed material grain size,
due to transverse variability in grain size between the main channel,
sand bars and islands (Babonaux, 1970), a grain diameter was associated
to variations in the longitudinal gradient so as to relate changes in median
grain size with changes in stream power and in shear stress (Fig. 4). The
sediments available on the Middle Loire originated mainly from the
upstream Loire, locally enriched by int from sedimentary formations
of the Parisian Basin. A downstream ning exponential law in the form
of the Sternberg (1875) formulation was adopted:
d50 d50x0 expax

with d50 as the river bed material median grain size at a distance x
from upstream, d 50(x = 0) is the river bed material median grain
size measured upstream (d50(x = 0) = 4.4 mm) and a is a coefcient
(a = 4.65 10 6 m 1 in our case). As observed in Fig. 4, there is a
signicant scatter and a general lack of data. The proposed downstream variation of the grain size is thus imprecise. Valverde et al.
(2013) made a recent extensive survey on the main channel grain
Cher

102

Vienne

Maine

size characterisation. Their rst results indicated that the coefcient

a may be slightly smaller. Thus, one should be aware of uncertainties
linked to the determination of d50 as well as related parameters.
3.3. Hydraulic parameters
The geomorphic characteristics are complemented by parameters
extracted from a hydraulic model. The one dimensional hydraulic
model RubarBE, developed by Irstea (El Kadi Abderrezzak and Paquier,
2009), which solves the Barr-de-St-Venant equations by an explicit
second order Godunov type numerical scheme was selected. The
model allows the consideration of two roughness coefcients: one for
the main channel and one for the oodplain. The main channel Strickler's
coefcient Kmc was calibrated for the 1996 ood event and varies between
28 and 32 m1/3/s along the studied reach. For the oodplain, Kfp was set at
15 m1/3/s. Adjustments to the resistance coefcient were locally made in
order to obtain a good reproduction of the measured water levels available. Once calibrated, the model can be run for a succession of steady
discharges from low ows to extreme events. Using the model outputs,
the river width to mean water depth ratio (W/D) (widthdepth ratio)
can be determined for different discharges. In the same way, the mean
bed shear stress , dened by the following equation can be calculated:

with as the density of water, g, the acceleration of gravity Rh as the

hydraulic radius, and S as the energy slope. As this study focused on
effective sediment transport, the biennial discharge Q2, assimilated to
the bankfull or effective discharge (Andrews, 1980) was used. Lateral
ow inputs from the three main tributaries were estimated to maintain
the estimated biennial discharge along the whole Middle Loire River.
The mean bed shear stress is the sum of the stresses acting due to
the friction attributable to the grains themselves or skin friction, sf, the
resistance created by the presence of bed forms or form drag fd, and the
adding resistance due to bank effect, river sinuosity, etc., or side drag sd
(i.e., = sf + fd + sd). Bed forms reduce the ability of the ow to
transport sediment and should ideally be included (Frings and
Kleinhans, 2008); however due to the scale of the present study and
the amount of data available, only skin friction is considered. Moreover,
according to Meyer-Peter and Mller (1948), the effective shear stress,
eff sf, is the part responsible for transporting sediment. eff was
calculated using the following relationship (Meyer-Peter and Mller,
1948):
eff

Median grain size (m)

gRh S

K
Kg

!3=2

with Kg as the Strickler coefcient due to skin friction, Kg = 21/(d50)1/6

(Strickler, 1923), and K as the total Strickler coefcient. In order to characterize the potential sediment transport, values obtained for the effective
shear stress eff are compared to the critical bed shear stress for inception
of transport cr, calculated using the equation of Soulsby and Whitehouse
(1997).
Specic stream power , has been determined with the following
relationship (Bagnold, 1980; Parker et al., 2011):

103

QS
W

10

450

500

550

600

650

700

750

800

850

900

Distance from source (km)

Fig. 4. Downstream variation in median grain size for the main channel (circles indicate
measured values, crosses are measured d50 which have been excluded from the analysis).
Solid line represents the exponential law in the form of the Sternberg formulation. The
vertical dotted lines indicate the four geological units and the blue segments the location
of major tributaries.

where the energy slope S and the width W are computed for the
assumed bankfull discharge, i.e., Qbf = Q2.
3.4. Delineation of homogeneous reaches along the uvial continuum
The delineation of reaches along the uvial continuum generally
consists in a visual distinction of changes (Orr et al., 2008). A rst

A. Latapie et al. / Catena 121 (2014) 112

division of the river into homogeneous geomorphic reaches was conducted in two steps (Latapie et al., 2009):

those areas; reaches corresponding to those locations have simply

been characterized as heavily inuenced by human disturbance.
The variogram analysis was performed on the active channel width,
oodplain width, changes in bed level and hydraulic parameters.
Results illustrated that the active channel width, oodplain width and
the hydraulic parameters data are not spatially correlated. On the
other hand, changes in bed levels between 1986 and 1995 and between
1995 and 2002 provided a range value equal to 4 and 5 km respectively.
Those lengths are consistent with the length of the reach identied with
the visual preliminary approach.
Finally, discontinuities identied in relation to geomorphic, hydraulic
parameters and temporal variation in active channel width B and river
bed level z were assessed using the Data Over Threshold (DOT) test
and the Pettitt's test (Table 1).
The number of inection points detected with the DOT approach is
highly dependent on the spatial resolution used for calculation; except
for the velocity, the number of discontinuities identied on the hydraulic
parameters is nearly twice the values calculated on the geomorphic parameters. If discontinuities in lateral and vertical changes are considered,
the results appear to be very sensitive to the period considered. However,
the multi-temporal analysis allows the distinction of permanent structures from transitory ones; those permanent structures identied with
the DOT approach were then combined with results from the Pettitt's
test to detect robust discontinuities.
If we assume a tolerance of 250 m to account for differences in the
data resolution, 170 reaches are identied. With the exception of
three boundaries, they all coincide with the boundaries of the 167
reaches previously dened. The additional discontinuities identied
correspond to structures located in the cities of Orlans and Tours;
and as previously mentioned, those reaches are merged.
Each reach dened is characterized by an average value calculated
for every geomorphic, hydraulic and temporal change derived. The
following discussion is based on these average values only.

1. Identifying discontinuities on low water proles (1978, 1986, 1995,

2002, 2011).
2. Assessing the presence of singularities affecting the ow, sediment
transport and thus river behaviour. The presence of articial structures (bridge, dam, groyne, ), bedrock outcrops, conuences and
transition from a single channel to a multiple channel conguration
(presence of vegetated islands) was checked on the 1995 aerial
photographs complemented by the database on bedrock outcrops
compiled by Rethoret (2001).
The 1995 aerial photography dataset was selected in step 2. It is an
intermediate value in the period analysed. It corresponds to the ofcial
end of the mining period on the Middle Loire. The hydraulic model is
based on topographic data surveyed in 1995.
This delineation is pertinent but location of reach boundaries could be
corroborated with the implementation of some statistical tests. First,
variograms were generated for all geomorphic and hydraulic data to assess the degree and range of spatial autocorrelation and how they change
over distance. The range, dened by tting an experimental variogram
gives an indication of the data spatial dependence and can thus be
regarded as the maximum length a reach can have. Second, detection of
changes on the longitudinal plots of the geomorphic and hydraulic
parameters was considered. In order to dismiss discontinuities linked to
uncertainties associated with the derivation of the parameters and limit
the detection to main discontinuities, threshold values have been dened
as 50 m for the active channel width B, 6 cm for the bed variations z
and 20% for the hydraulic parameters. Results of these tests are referred
to as Data Over Threshold or DOT. Finally, the test of Pettitt (1979) was
implemented on all the parameters available to identify discontinuities
along the stream network. This test was initially designed to detect a single point of change. However, Alber and Pigay (2011) implemented the
approach on the Rhne basin and ran the algorithm iteratively as long as
a change point was detected to dene homogeneous river reaches (with
an risk equal to 0.05). Combining these approaches to detect changes
on the longitudinal plots of the geomorphic and hydraulic parameters
allowed the identication of the most signicant discontinuities.

4.2. Changes in active channel width

An example of the longitudinal analysis of the active channel width
between 1955 and 2010 is given in Fig. 5a. The observed changes in active
channel width can be linked with the four geological units described in
Part 2.2. In geological unit 1, the active channel width decreased between
1955 and 2010 mainly because the Loire River evolved from a multiple
channel system to a single (constrained) channel. In the other geological
units, corresponding to more urbanized areas, the active channel width
remained stable except for reaches located upstream of the conuence
with the Cher River and downstream of the conuence with the Maine
River, where channel narrowing occurred. The reduction in active channel
width corresponds to areas where sediments have accumulated behind
oblique groynes, favouring the development of vegetation.

4. Results
4.1. Reach denition
The rst delineation based on a visual detection of discontinuities on
low water long prole and aerial photographs differentiated 167
reaches with length ranging from 250 m to 4750 m and an average
length of 2500 m. Due to the presence of numerous structures in the
cities of Orlans and Tours, all singularities were not accounted for in

Table 1
Number of discontinuities identied on the geomorphic and hydraulic parameters and on temporal lateral and vertical changes with the Data Over Threshold test (noted DOT in the table)
and the Pettitt's test.
Hydraulic parameters (Q2)

Geomorphic parameters
B
Date
Nb. values
DOT
Pettitt

Bf

1955

1984

1995

347
20

370
20

343
17

2002

2010

2003

339
21

311
24

366
56

eff

W/D

265
106

41
107

1995
1827

858
655
70

748
72

Temporal changes
B
Date
Nb. values
DOT
Pettitt

5584

z
8495

9502

0210

121
13

76
9

858
177
13

95
5

7886
86
34
70

8695
145
74
62

9502
145
62
53

0311
184
70
47

A. Latapie et al. / Catena 121 (2014) 112

(a)

(b)

600

8
6
4

500

B / t (m/year)

Active channel width (B) in m

1955
2010

550

450
400
350

2
0
2
4
6

300
8
250
200
450

10
12
500

550

600

650

700

750

800

850

900

5584

8495

9502

0210

Distance from source in km

Fig. 5. (a) Measured active channel width B for 1955 and 2010 (b) Rate of change in active channel width for the four periods considered (the red line in the boxplot represents the median,
the limits of the box provide the 25th and 75th percentiles and the whiskers extend to the most extreme data points not considered outliers). (For interpretation of the references to color
in this gure legend, the reader is referred to the web version of this article.)

The rate of change in active channel width B during the period

considered is presented in Fig. 5b. A reduction in active channel width
is observed for all periods, with a slight increase in the rate of change
over the last two decades.
The river reach scale approach complements longitudinal analysis
by allowing the identication and comparison of reaches with similar
behaviour along the study area. Thus, the reach boundaries are considered to check whether the presence of a natural or articial features
could be linked to a specic channel response. In order to facilitate representation and discussion of the results, we distinguished 4 types of
boundaries:
Type 1 characterizes a reach with xed boundaries delimited by
articial features or bedrock outcrops at the upstream and downstream end,

B / t (m/year)

10

Type 2 represents mixed reaches (i.e., with a xed boundary

upstream or downstream and a movable boundary downstream
or upstream),
Type 3 characterizes a reach with movable boundaries associated
with the presence of islands at both ends,
Type 4 indicates the presence of tributaries.
Fig. 6 illustrates the rate of changes in active channel width for the
different periods and the various reach boundaries identied.
Reaches of types 1, 2 and 3 exhibit similar behaviour with relative lateral stability between 1955 and 1995 before showing a
tendency to narrow in the next two periods (19952002 and
20022010). The presence of a conuence in a reach (type 4) has
a signicant effect upon active channel width, especially over the last
decade.

Type 1

Type 2

Type 3

Type 4

5
0
5
10
15

B / t (m/year)

10
5
0
5
10
15

7886

9502
8695

7886
0311

9502
8695

0311

Fig. 6. Rate of change in active channel width for reaches presenting different boundaries (see legend of Fig. 5 for explanation of box-plots).

A. Latapie et al. / Catena 121 (2014) 112

4.3. Changes in longitudinal prole

Changes in the longitudinal prole as well as the reach-averaged
slope were assessed using low water level measurements. Fig. 7 presents
the rate of change in reach-averaged slope for the different periods
considered for reaches grouped by their types of boundaries. In general,
the slope remains stable whatever the type of reach. However, a slight
decrease of the slope occurs for reaches delimited by xed boundaries
(type 1). This may correspond to erosion and reduction of slope downstream of structures. On the other hand, a slight increase of the slope
occurs for type 4 reaches (conuence), which may have been inuenced
by sediment input from the conuences together with cessation of
instream sediment mining (1995). It is also interesting to note that the
rate of change in slope is more variable for reaches presenting at least a
xed boundary (types 1 and 2) since 1995. Inversely, reaches delimited
by islands (type 3) present a stable slope for all periods considered.
Fig. 8 presents the rate of change in bed level for several periods for
reaches grouped by their types of boundaries. The general trend for all
types of reaches is a decrease in the incision rate over the years to attain
stability or even aggradation in the last decade. Similar to the rate of
change in active channel width, type 1, 2 and 3 reaches behave in a comparable manner (with type 3 reaches, i.e., delineated by islands showing
a slightly more pronounced erosion and reduction of the active width
for the period 19952002). Reaches characterized by the presence of a
conuence present a high incision rate between 1978 and 1984 before
attaining stability or even aggradation. Apart from the presence of conuences, it is difcult to match reach boundaries with a specic channel
response. The slope variation seems however to be sensitive to xed
boundaries with various local variations. Unexpectedly, no correlation
was found between the changes in slope and changes in river bed levels.
In the same way, the changes in active channel width and changes in
river bed levels are not consistently correlated. Thus, the signicant reduction of the active channel width observed for many reaches did not
systematically lead to erosion nor reduction of the slope. Rinaldi (2003)
reported similar behaviour with channel narrowing and no/or limited
incision in 4 cases (out of 42 sites) of alluvial rivers in Tuscany. Similarly,
the presence of bedrock controlled reaches in the Middle Loire has

prevented the upstream migration of bed degradation and the consequent dynamic response in the uvial system. In the absence of bedrock
outcrops, the observed adjustments may reect a time lag between the
disturbance and the channel response (Knighton, 1998). Finally, the
drier hydrological period observed after 1984 might have enhanced
vegetation development and hence contributed to the stabilization of
in channel bars and a reduction in active channel width.
5. Use of a 1D model to assess channel changes
As the instream sediment mining period cannot be directly associated
to systematic adjustments of channel properties (narrowing and deepening), the hydraulic parameters are considered to assess whether the
broad-scale adjustments to river morphology vary in relation to different
combinations of widthdepth ratio (e.g. Doll et al., 2002).
5.1. Longitudinal evolution of hydraulic parameters
The implementation of a 1D hydraulic model to calculate parameters
provides major advantages: multiple modelling scenarios can be envisaged and the derivation is more rigorous; in particular, the specic
stream power is generally calculated using the energy slope S assimilated
to the bed (or low water) slope Sb (Bernot et al., 1996; Nanson and
Knighton, 1996). This assumption can be discussed as the slope should
be estimated for the bankfull discharge. Based on the results of the 1D
model for the biennial discharge, we found 0.17 b SQ2/Sb b 19. Signicant
scatter reects the presence of articial and natural singularities, which
have a high inuence at low to intermediate discharges (such as Q2).
The hydraulic parameters obtained for the biennial discharge, Q2,
which may be assimilated to the bankfull discharge, are reported in
Fig. 9. Q2 is nearly constant for the rst 300 km of the Middle Loire before being affected by the main tributaries which double the discharge
(Fig. 9a).
The widthdepth ratio computed for Q2 varies between 57 and 421.
It is higher in the upstream and downstream part of the studied area
where multiple channels predominate (Fig. 9b). As the mean water
depth does not vary signicantly from the upstream to the downstream

Type 1

Type 2

Type 3

Type 4

x 10

S / t (m/year)

6
4
2
0
2
4
6
8
8

x 10

S / t (m/year)

6
4
2
0
2
4
6
8

7886

9502
8695

7886
0311

9502
8695

0311

Fig. 7. Rate of change in reach-averaged slope for reaches presenting different boundaries (see legend of Fig. 5 for explanation of box-plots).

A. Latapie et al. / Catena 121 (2014) 112

75

Type 1

Type 2

Type 3

Type 4

z / t (mm/year)

50
25
0
25
50
75
100
125
75

z / t (mm/year)

50
25
0
25
50
75
100
125

7886

9502
8695

7886
0311

9502
8695

0311

Fig. 8. Rate of change in bed level for reaches presenting different boundaries (see legend of Fig. 5 for explanation of box-plots).

(b)

(a)
500

3500

450

350

W/Dratio

Discharge Q2 in m3/s

400
3000

2500

300
250
200
150

2000

100
50

1500
450

500

550

600

650

700

750

800

850

0
450

900

500

Distance from source in km

550

600

45

40

35

6
5
4

10

5
650

850

900

850

900

20
15

600

800

25

550

750

30

500

700

(d)
50

(W/m2)

eff (N/m2)

(c)
10

0
450

650

Distance from source in km

700

750

800

Distance from source in km

850

900

0
450

500

550

600

650

700

750

800

Distance from source in km

Fig. 9. Longitudinal variation of (a) modeled bankfull discharge (Qbf Q2), (b) widthdepth ratio, (c) effective bed shear stress (cr in plain line), (d) specic stream power (cr in plain
line). The vertical dotted lines delineate the four geological units.

A. Latapie et al. / Catena 121 (2014) 112

5.2. Interest of the effective bed shear stress and widthdepth ratio
The effective bed shear stress made dimensionless by its critical
value for inception of transport and the widthdepth ratio computed
for the biennial discharge is presented in Fig. 10. These two parameters
allow a clear distinction between the four geological units. For units 1
and 4, the large W/D values are directly connected to the presence of

10
Unit 1
Unit 2
Unit 3
Unit 4
model

9
8
7
6

eff / cr

part of the Middle Loire River (from DQ2 3 m to DQ2 4 m), it appears logical that the W/D ratio behaves similarly to the active channel


width B W Q 2 =B1:5 . In geological unit 1, W/D decreases as the Middle
Loire changes from a multiple channel system to a single (constrained)
channel. In the middle part of the studied area (geological unit 2), the
widthdepth ratio is nearly constant and W/D 100. In geological
units 3 and 4, W/D increases simultaneously with discharge and the
presence of multiple channels. A larger dispersion is observed in the
downstream part of the model (second part of unit 3 and unit 4). This
is mainly due to the presence of multiple channels; the difference in
water levels between the two channels cannot be rigorously integrated
in our model inducing a larger dispersion in the hydraulic results.
The calculated effective bed shear stress eff is within the range of 0.9
to 8.2 N/m2 (Fig. 9c). In terms of sediment transport magnitude, the
results may be affected by the high entrainment threshold. The increasing
eff in unit 1 indicates a potential for erosion. A maximum is observed for
eff at a distance of 550 km from source. This forms the boundary
between unit 1 and unit 2. Finally, the decreasing eff in units 2, 3 and 4
indicates a potential for sediment deposition.
The calculated specic stream power is within the range of 4 to
47 W/m2. Downstream patterns of change in are highly correlated
to downstream changes in channel slope and width. The general trend
of the curve is directly the opposite of the widthdepth ratio plot. As
observed by Knighton (1999), a maximum of specic stream power is
evident, in this instance 650 km from source. Then, decreases to
values of less than 10 (except at conuences where a slight increase is
noticed). Channel changes have often been associated with specic
stream power (e.g. Bledsoe et al., 2002; Brookes, 1987; Orr et al., 2008).
Since threshold values are specic to a catchment, the analysis of downstream changes in stream power should be considered as an indicator of
zones of potential sediment transport. Large changes in stream power
from one reach to the next are generally associated to erosion (increase
of ) or sediment accumulation (decrease of ), whereas high values
alone are not necessary indicative of changes (Reinfelds et al., 2004;
Vocal-Ferencevic and Ashmore, 2012). It should be noted that, based on
the Strickler (1923) concept, we can derive 1.5, which explains the
similar behaviour of and albeit the larger dispersion in the specic
stream power.
To be more accurate in characterizing the potential sediment transport, the bed shear stress or specic stream power was compared to its
critical value for inception of sediment transport. The critical bed shear
stress (cr) was obtained from the Shields curve based on the grain size
estimates. The critical specic stream power (cr) was estimated using
the development of Bagnold (1980). As the river bed median grain size
decreases (from d50 4 mm to d50 0.6 mm), the critical bed shear
stress decreases from cr = 36 to 0.3 N/m2; and the critical specic
stream power from cr = 15 to 0.4 W/m2 (see plain line in Fig. 9(c)
and (d), respectively). Despite uncertainties in the estimation of the
critical bed shear stress and specic stream power, not only due to
some sensitivity to the slope (Camenen, 2012) but also because of the
scatter in the grain size measurements, Fig. 9(c) and (d) indicates that
critical values are generally exceeded all along the river. Results indicate
that sediment transport occurs at ows lower than the biennial discharge, which is consistent with measurements and analysis realized
around the site of Brhmont (776 km from source; see Claude et al.,
2012). This behaviour is typical in sandgravel bed rivers where sediment transport is inuenced not only by ood events but also by
lower discharges occurring over a longer period.

5
4
3
2
1
0

100

200

300

400

500

W/D
Fig. 10. Effective bed shear stress made dimensionless by its critical value for inception of
transport versus the widthdepth ratio computed for Q2.

multiple channels and islands. Since d50 is larger in unit 1 than in unit
4, eff/cr differs in those two units.
Unit 1: The eff/cr ratio is relatively small (eff/cr 1.5) and slightly
decreases with the W/D ratio (80 b W/D b 280).
Unit 2: The eff/cr ratio is intermediate (2 b eff/cr b 6.5) but
decreases rapidly with the W/D ratio (60 b W/D b 130).
Unit 3: The eff/cr ratio is relatively high (3 b eff/cr b 7.5), increasing
with the W/D ratio (70 b W/D b 330).
Unit 4: The eff/cr ratio is high (5 b eff/cr b 9) and slightly decreases
with the W/D ratio (120 b W/D b 420).
Using a continuous t for each parameter (eff, D, and W) in a similar
way as for the median grain size (Eq. (1)), it is possible to obtain a curve
for the eff/cr W/D relationship that is function of the longitudinal
distance (solid line in Fig. 10). It appears from this analysis that the hydraulic and geomorphic parameters allows the distinction of different
trends of channel response but the latter changes need to be associated
with human disturbance (in particular instream sediment mining).
Fig. 11 illustrates the rate of change in river bed level considering the
effective shear stress made dimensionless by its critical value and the
widthdepth ratio for the periods 19861995 and 19952002 (where
1995 is the year corresponding to the cross section data used to construct the hydraulic model and ofcial termination of instream mining).
Different trends can be distinguished for the two periods. Between 1984
and 1995, the rate of change in bed level decreases as the ratio /cr
increases. During the following period (19952002), the contrary is
observed (i.e., the rate of change in bed level increases as the ratio /cr
decreases). The widthdepth ratio equally illustrates different behaviours
for the two periods but with larger dispersion. Those trends indicate river
recovery since extraction stopped (Fryirs and Brierley, 2000).
6. Impact of sediment mining upon channel changes
Instream mining has directly altered the channel geometry and bed
elevation in the Middle Loire. One hundred and two ofcial sites have
been identied from archive data between Nevers and Saumur (located
respectively 445 km and 823 km downstream from source, Fig. 1).
Unfortunately, the amount of sediment extracted is not always quantied in the historical database. Based on the data available, we distinguished reaches with low average annual extraction rate (i.e., less
than 1000 T/year) from reaches with moderate extraction rate (i.e.,
less than 100,000 T/year) and intensive extraction rate (i.e., more than
100,000 T/year). The annual rate of change in bed level considering

10

A. Latapie et al. / Catena 121 (2014) 112

100

19861995

19861995

19952002

19952002

z / t (mm/year)

50

50

100

z / t (mm/year)

100

50

50

100

eff / cr

10

100

200

300

400

500

W/D

Fig. 11. Rate of change in bed level related to the effective bed shear stress made dimensionless by its critical value and to the widthdepth ratio (legend for the distinction of the four
geological units is the same as for Fig. 10).

the amount of sediment extracted per reach is presented for different

periods between 1978 and 2010 in Fig. 12.
The general trend for all the reaches is river bed incision for the rst
three periods. For the latest date considered (20032011), incision has
stopped and some aggradation is visible independent of the former
rate of extraction sustained by the reach. Until the end of the instream
mining period (1995), reaches affected by high sediment mining tend
to have eroded more signicantly.

100

z / t (mm/year)

50

50

100

150

E1 E2 E3

E1 E2 E3

E1 E2 E3

E1 E2 E3

19781986

19861995

19952002

20032011

Fig. 12. Annual changes in bed level considering the amount of sediment extracted per
reach (E1: low extraction rate, E2: moderate extraction rate and E3: high extraction
rate; same explanation for each box as in Fig. 5).

As the data collected do not allow quantication of volume extracted

for each reach of the study area, ve reaches that were the most affected
by instream mining are considered. The authorized volume of sediment
mined (Vex) is compared to the volume calculated from changes in
low water level for the period 19782002. This volume of deposition
or erosion (Vd/e) was approximated by multiplying the change in low
water level by the active channel width and the length of the reach
(i.e., Vd/e = z B L t). As a result, Vd/e N 0 means aggradation
whereas Vd/e b 0 means erosion. Results shown in Table 2 provide an insight into the localized effects of instream sediment mining.
The volume of sediment ofcially extracted from the active channel
with permission from the water authorities is generally greater than the
volume calculated from the low water levels for the ve reaches considered (i.e., Vex N Vd/e). Prior to describing the above results in more detail,
the uncertainties associated with low water level records available for
1978 and with the location of extraction sites are explained.
The reach located 502.65 km from source has a bridge at its downstream boundary. Aggradation of the river bed occurred during the period
corresponding to the main extraction period. This behaviour is the opposite to what is expected and it is probably due to the presence of a localized mining pit which is not visible on the low water level long prole. A
period of incision is observed from 1984 to 1996 despite a reduction in the
amount of sediment removed. For this particular reach, it appears that the
channel response was poorly inuenced by the extensive extractions at
the reach scale.
The reach located 536.29 km from source has been subjected to intensive sediment mining to allow the construction of a nuclear power
plant. This reach is located just downstream of the transverse sill of
the nuclear power plant, which constrains sediment transport. The
volume of erosion calculated from available topographic data between
1978 and 1984 is 8.3 104 (m3/year). This differs from the aggradation observed from low water levels. The recalibration of the river prior

A. Latapie et al. / Catena 121 (2014) 112

11

Table 2
Comparison between the volume of sediment annually extracted (Vex) and the calculated volume of deposition or erosion (Vd/e) for ve reaches located by their distance from source (in km).
Distance from source

502.65
536.29
625.35
705.81
748.28

19962002

20032011

Vex (104 m3)

19781984
Vd/e (104 m3)

Vex (104 m3)

19841996
Vd/e (104 m3)

Vd/e (104 m3)

Vd/e (104 m3)

11.5
78.6
13.4
15.1
15.8

9.4
9.2
3.6
2.9
0.4

2.8
0.5
4.8
4.3
6.1

6.1
0.03
0.6
1.6
2.8

8.2
0.4
0.6
0.5
0.4

14.6
0.02
4.8
3.8
5.6

to the construction of the nuclear power plant has had a larger inuence
on the channel response than extractions. It is assumed that as records of
low water levels are not evenly distributed, gravel pits could be located in
between two measures, leading to an underestimation of changes in bed
levels at the reach scale. Another explanation could be that the degradation is very localized and calculated volumes at the reach scale are incorrectly spread over the whole length of the reach.
The three other reaches presented in Table 2 present an expected
behaviour indicating the inuence of extractions. Signicant erosion is
observed during the extraction period before the reach starts its recovery
at the end of the mining period (1995).
7. Conclusion
An understanding of channel response over recent decades is a prerequisite for assessing recovery potential and predicting how a reach
may respond to future disturbance (Fryirs and Brierley, 2000). Customary
analyses using longitudinal proles show salient features and spatial
interrelationships. These are enhanced by reach determinations,
which enable spatial distributions to be appraised for reach by reach
comparison.
The general trends observed on the Middle Loire River over the last
50 years are a narrowing of the active channel width and an incision of
the river bed (Gasowski, 1994). This has led to adjustments of island
widths (Dtrich et al., 2010) and secondary channels (Rodrigues
et al., 2006). These adjustments are similar to those reported on other
European rivers affected by human disturbances (Bravard et al., 1997;
Surian and Rinaldi, 2003). Changes in channel pattern have not
occurred on the Middle Loire River as it is highly constricted by ood
embankments. However lateral adjustments associated to bank erosion
occurred in the last 50 years. This lateral stability (i.e., no change in
channel pattern) differs from changes observed on the Upper Loire
River, upstream of the conuence with the Allier River, which is not
conned (Leteinturier et al., 2000).
The reach scale approach implemented on the Middle Loire River
complements observed longitudinal variation of geomorphic characteristics. Statistical tests corroborate the location of clear discontinuities
and so reach denitions. Given the Water Framework Directive context,
this approach could inform decision makers about further localized
studies to support decisions to take to achieve good conditions. Results
show that channel changes at the reach scale are largely inuenced by the
presence of former extraction sites. Sediment mining caused signicant
erosion but the river bed has been restored since gravel mining stopped
in the 1990s. Several reaches that were in severe erosion are now in
aggradation. Fryirs et al. (2009) refer to this as a response gradient. On
the other hand, the channel response for the active width appears slower
since no clear indication of recovery was observed.
Hydraulic parameters derived from a 1-D hydraulic model allow a
more precise denition of parameters, supporting modelling scenarios.
The effective bed shear stress made dimensionless by its critical value
for inception of transport eff/cr and the widthdepth ratio W/D computed for the biennial discharge help in the understanding of channel
responses. Although a 1D hydraulic model requires a signicant amount
of topographic and hydrologic data, it provides several additional indicators such as eff/cr and W/D for different discharges. With the

development of Lidar data acquisition, which hopefully will give access

to bathymetric data as well, the application of the approach outlined
here will possibly be easier over long reaches.
Acknowledgements
We are grateful to EdF, Irstea and the Mission Val de Loire who
sponsored the work. The preparation of this manuscript was also
partly supported by the EU FP7 project REFORM (contract no.
282656). This research was carried out within the Plan Loire Grandeur
Nature 3 Masterplan. The research beneted greatly from data made
available by the DREAL Centre, the University of Tours, the GIP Loire
Estuaire, the SMN of Nantes, the CETE of Blois, the DDE of Indre et
Loire, the DDE of Loiret and the DDE of Maine et Loire. Fabien Pasquet
and Stphane Braud are particularly acknowledged for their assistance
in data collection. We also thank J.N. Gautier and the late lamented J.J.
Peters for valuable discussions. Finally, we would like to thank the
two anonymous reviewers for their signicant input to improve the
quality of this paper.
References
Aggett, G.R., Wilson, J.P., 2009. Creating and coupling a high-resolution DTM with a 1-D
hydraulic model in a gis for scenario-based assessment of avulsion hazard in a
gravel-bed river. Geomorphology 113 (12), 2134.
Alber, A., Pigay, H., 2011. Spatial disaggregation and aggregation procedures for characterizing uvial features at the network scale: application to the Rhne basin (France).
Geomorphology 125, 343360.
Alcayd, G., Gigout, M., 1976. Guides gologiques rgionaux (Regional geological guide).
Val de Loire (Anjou, Touraine, Orlanais, Berry) (in French).
Andrews, E.D., 1980. Effective and bankfull discharges of streams in the Yampa River
basin, Colorado and Wyoming. J. Hydrol. 46, 311330.
Armas, I., Gogoase-Nistoran, D.E., Osaci-Costache, G., Brasoveanu, L., 2012. Morphodynamic evolution patterns of Subcarpathian Prahova river Romania. Catena 100,
8399.
Babonaux, Y., 1970. Le lit de la Loire, Etude hydrodynamique uviale (The Loire's river
bed, uvial hydrodynamic study). Mmoires de la section de gographie., Ministre
de l'ducation nationale. Comit des travaux historiques et scientiques (in French).
Bagnold, R.A., 1980. An empirical correlation of bedload transportrates in umes and natural
rivers. Proc. R. Soc. Lond. A 372 (1751), 453473.
Belleudy, P., 2000. Restoring ow capacity in the Loire River bed. Hydrol. Process. 14,
23312344.
Bernot, V., Calland, V., Bravard, J.P., Foussadier, R., 1996. La sectorisation longitudinale du
Vidourle (Gard-Hrault), une mthode applique la gestion environnementale
d'un espace uvial mditrranen (Longitudinal delineation of the Vidourle river
(Gard-Hrault), a methodology applied to the environmental management of a
Mediterranean uvial corridor). Rev. Gogr. Lyon 71 (4), 323339 (in French).
Berthois, L., 1971. Les transports sdimentaires et l'rosion dans le bassin de la Loire (Sediment transport and erosion in the Loire River basin). Etude Ligriennes 11 (in French).
Bledsoe, B.P., Watson, C.C., Biedenham, D.S., 2002. Quantication of incised channel evolution and equilibrium. J. Am. Water Resour. Assoc. 38, 861870.
Bravard, J.-P., Amoros, C., Pautou, G., Bornette, G., Bournaud, M., Creuz des Chtelliers, M.,
Gibert, J., Peiry, J.L., Perrin, J.F., Tachet, H., 1997. River incision in south east france:
morphological phenomena and ecological effects. Regul. Rivers Res. Manag. 13,
7590.
Bravard, J.-P., Kondolf, G.M., Piegay, H., 1999. Environmental and societal effects of channel
incision and remedial strategies. Incised River Channels: Processes, Forms, Engineering
and Management. Wiley, pp. 303341.
Brierley, G.J., Fryirs, K.A., 2005. Geomorphology and River Management. , Blackwell.
Brookes, A., 1987. River channel adjustments downstream from channelization works in
England and Wales. Earth Surf. Process. Landf. 12, 337351.
Bross, R., 1982. Les processus sdimentaires dans le euve Loire (Sedimentary Processus
in the Loire River). (PhD thesis) , University of Angers, Angers, France (in French).
Buishand, T., 1984. Tests for detecting a shift in the mean of hydrological time series. J.
Hydrol. 73, 5169.

12

A. Latapie et al. / Catena 121 (2014) 112

Camenen, B., 2012. Discussion of understanding the inuence of slope on the threshold
of coarse grain motion: revisiting critical stream power by C. Parker, N.J. Clifford, and
C.R. Thorne. Geomorphology 139140, 3438.
Champion, M., Maillard, P., Cario, P., 1971. Les alluvions de la Loire dans la rgion Centre.
Inventaire de la production et des gisements (The Loire alluvium in the region Centre.
Inventory of production and deposit sites). Technical Report, Bulletin de Liaison des
Laboratoires des Ponts et Chausses (in French).
Clark, J.S., Rizzo, D.M., Watzin, M.C., Hession, W.C., 2008. Spatial distribution and geomorphic condition of sh habitat in streams: an analysis using hydraulic modelling and
geostatistics. River Res. Appl. 24, 885899.
Claude, N., Rodrigues, S., Bustillo, V., Brhret, J.-G., Macaire, J.-J., Jug, P., 2012. Estimating
bedload transport in a large sand-gravel bed river from direct sampling, dune tracking and empirical formulas. Geomorphology 179, 4057.
Clment, P., Pigay, H., 2003. Statistics and uvial geomorphology. In: Kondolf, G.M.,
Pigay, H. (Eds.), Tools in Fluvial Geomorphology. Wiley. & Sons, Chichester, UK,
pp. 597630.
Dacharry, M., 1996. Les grandes crues historiques de la Loire (Historical ood events of
the Loire River). La Houille Blanche 6 (7), 4753 (in French).
Dambre, J.-L., Malaval, P., 1993. Evaluation des conditions de poursuite de la politique de
limitation des extractions de matriaux dans le lit de la Loire du Bec d'Allier Nantes
(Evaluating the conditions to pursue gravel extractions in the Loire River main channel
between the conuence with the Allier River and Nantes). Technical report, Loire
Estuaire, nantes, France (in French).
Debrand-Passard, S., Macaire, J.-J., Clozier, L., Fleury, R., 1998. Particularits de l'volution
du systme solognot dans le bassin de la Loire au Quaternaire. corrlations possibles
(Particularities in the evolution of the Sologne system in the Loire bassin). Gol. Fr. 2,
5568 (in French).
Dtrich, S., Rodrigues, S., Macaire, J.-J., Bont, P., Brhret, J.-B., Bakyono, J.-P., Jug, P.,
2010. Caesium-137 in sandy sediments of the River Loire (France): assessment of
an alluvial island evolving over the last 50 years. Geomorphology 115 (12), 1122.
Dion, R., 1961. Etude des leves de la Loire. The Loire River's dykes, Flammarion, Paris,
France (in French).
Doll, B.A., Wise-Frederick, D.E., Buckner, C.M., Wilkerson, S.D., Harman, W.A., Smith, R.E.,
Spooner, J., 2002. Hydraulic geometry relationships for urban streams throughout
the Piedmont of North Carolina. J. Am. Water Resour. Assoc. 38, 641651.
Donsimoni, M., Berthier, H., Martin, J.C., Nachbaur, A., 2008. Dtermination de la prsence
de karsts sous les leves domaniales du bassin de la Loire et rduction du risque
d'effondrement de la digue li la prsence de ces conduits souterrains naturels
(Tranche 1: analyse gologique et hydrogologique) (Determination of the presence
of karsts underneath the state-owned ood embankments of the Loire catchment
and reduction of the risk of collapse of the embankment associated with the presence
of those natural underground features (phase 1: geologic and hydrogeologic analysis)).
Technical Report, BRGM (in French).
Duband, H., 1996. La gense des crues dans le bassin de la Loire (Genesis of ood events in
the Loire catchment area). La Houille Blanche 67, 5462 (in French).
El Kadi Abderrezzak, K., Paquier, A., 2009. One-dimensional numerical modeling of sediment transport and bed deformation in open channels. Water Resour. Res. 45
(W05404) (20 pp.).
Frings, R.M., Kleinhans, M.G., 2008. Complex variations in sediment transport at three
large river bifurcations during discharge waves in the river Rhine. Sedimentology
55, 11451171.
Fryirs, K., Brierley, G.J., 2000. A geomorphic approach to identication of river recovery
potential. Phys. Geogr. 21, 244277.
Fryirs, K., Spink, A., Brierley, G.J., 2009. Post-European settlement response gradients of
river sensitivity and recovery across the upper Hunter catchment, Australia. Earth
Surf. Process. Landf. 34, 897918.
Gasowski, Z., 1994. L'enfoncement du lit de la Loire (The entrenchment of the Loire's river
bed). Gocarrefour 69, 4145 (in French).
Gautier, E., Pigay, H., Bertaina, P., 2000. A methodological approach of uvial dynamics
oriented towards hydrosystem management: case study of the Loire and Allier rivers.
Geodin. Acta 1, 2943.
Ginestet, P., 1999. Sectorisation gomorphologique de la Loire entre le Bec d'Allier et
Monjean sur Loire (Geomorphic sectorisation of the Loire River between the conuence with the Allier River and Montjean sur Loire). (59 pp. in French).
Kellerhals, R., Churchand, M., Bray, D.I., 1976. Classication and analysis of river processes.
J. Hydraul. Div. 102, 819829.
Knighton, A.D., 1998. Fluvial Forms and Processes: A New Perspective. , Oxford University
Press.
Knighton, A.D., 1999. Downstream variation in stream power. Geomorphology 29 (34),
293306.
Latapie, A., Camenen, B., Paquier, A., Rodrigues, S., Moatar, F., Bouchard, J.-P., 2009.
Morphological evolution of the middle Loire River (France): denition of geomorphic
homogeneous reaches. Proc. 6th IAHR symposium on River, Coastal and Estuarine
Morphodynamics. , vol. 1, pp. 4350 (Santa Fe, Argentina).

Leopold, L.B., Maddock, T., 1953. The hydraulic geometry of stream channels and
some physiographic implications. Professional Paper. U. S. Geological Survey.,
252 (57 pp.).
Leteinturier, B., Engels, P., Petit, F., Chiffaut, A., Malaisse, F., 2000. Morphodynamisme d'un
tronon de Loire bourbonnaise depuis le 18me sicle (Changing dynamics of the Loire
River in Burgundy (France) during the last 200 years). Gomorphol. Relief Process.
Environ. 4, 239252 (in French).
Leviandier, T., Alber, A., Le Ber, F., Piegay, H., 2012. Comparison of statistical algorithms for
detecting homogeneous river reaches along a longitudinal continuum. Geomorphology
138, 130144.
Macaire, J.-J., Gay-Ovejero, I., Bacchi, M., Cocirta, C., Patryl, L., Rodrigues, S., 2013. Petrography of alluvial sands as a past and present environmental indicator: case of the
Loire River (France). Int. J. Sediment Res. 28, 285303.
Meyer-Peter, E., Mller, R., 1948. Formulas for bed-load transport. Proc. IInd IAHR Congress. IAHR, Stockholm, Sweden, pp. 3964.
Miller, J.R., Ritter, J.B., 1996. An examination of the Rosgen classication of natural rivers.
Catena 27, 295299.
Montgomery, D.R., Bufngton, J.M., 1997. Channel reach morphology in mountain drainage
basins. Geol. Soc. Am. Bull. 109, 596611.
Nanson, G.C., Knighton, A.D., 1996. Anabranching rivers: their cause, character and classication. Earth Surf. Process. Landf. 21, 217239.
Nelhig, P., 2010. Gologie du bassin de la Loire (Geology of the Loire River catchment).
Geosciences 12, 1223.
Orlowski, L.A., Schumm, S.A., Mielke, P.W.J., 1995. Reach classication of the lower
Mississipi River. Geomorphology 14, 221234.
Orr, H.G., Large, A.R.G., Newson, M.D., Walsh, C.L., 2008. A predictive typology for
characterising hydromorphology. Geomorphology 100, 3240.
Parker, C., Clifford, N.I., Thorne, C.R., 2011. Understanding the inuence of slope on the
threshold of coarse grain motion: revisiting critical stream power. Geomorphology
126, 5165.
Pettitt, A., 1979. A non-parametric approach to the change point problem. Appl. Stat. 28,
126135.
Reinfelds, I., Cohen, T., Batten, T., Brierley, G., 2004. Assessment of downstream trends in
channel gradient, total and specic stream power: a GIS approach. Geomorphology
60, 403416.
Rethoret, H., 2001. Etat des connaissances des seuils et points durs en Loire - Campagne de
localisation 2000 entre le Bec d'Allier et Nantes (Inventory of articial and natural
outcroups on the Loire River 2000 Location survey between the conuence with
the Allier River and Nantes). Technical report, Direction Rgionale de l'Environnement
Centre (in French).
Rinaldi, M., 2003. Recent channel adjustments in alluvial rivers of Tuscany, central Italy.
Earth Surf. Process. Landf. 28, 587608.
Rinaldi, M., Surian, N., Comiti, F., Bussettini, M., 2013. A method for the assessment and
analysis of the hydromorphological condition of Italian streams: the Morphological
Quality Index. Geomorphology 180, 96108.
Rodrigues, S., Brhret, J.G., Macaire, J.J., Moatar, F., Nistoran, D., Jug, P., 2006. Flow and
sediment dynamics in the vegetated secondary channels of an anabranching river:
the Loire River (France). Sediment. Geol. 186, 89109.
Rosgen, D.L., 1994. A classication of natural rivers. Catena 22 (3), 169199.
Simon, A., Doyle, M., Kondolf, M., Shields, F.D.J., Rhoads, B., McPhillips, M., 2007. Critical
evaluation of how the Rosgen classication and associated natural channel design
methods fail to integrate and quantify uvial processes and channel response. J.
Am. Water Resour. Assoc. 43 (5), 11171131.
Soulsby, R.L., Whitehouse, R.J.S.W., 1997. Threshold of sediment motion in coastal environment. Proc. Pacic Coasts and Ports'97 Conf. University of Canterbury, Christchurch, New Zealand, pp. 149154.
Sternberg, H., 1875. ber lngen- und querprol geschiebefhrender sse (On the longitudinal and cross-sectional proles of bed-load transporting rivers). Z. Bauwesen
25, 483506 (in German).
Strickler, A., 1923. Beitrge zur frage der geschwindigkeitsformel und der rauhigkeitszahlen
fr strme, kanle und geschlossene leitungen (Contributions to the questions of velocity formulations and roughness values for rivers, canals, and closed ducts). Mitteilung
16, Amt fr Wasserwirtschaft, Bern, Switzerland (in German).
Surian, N., Rinaldi, M., 2003. Morphological response to river engineering and management in alluvial channels in Italy. Geomorphology 50, 307326.
Valverde, L., Rodrigues, S., Jug, P., Desmet, M., Recking, A., Tal, M., 2013. Variations of
grain size in relation with cross-section dynamics and tributary conuences in a
large sand-gravel bed river: the Loire (France). Proc. 8th International SedNet Conference, Lisbon, Portugal.
Vocal-Ferencevic, M.V., Ashmore, P., 2012. Creating and evaluating digital elevation
model-based stream-power map as a stream assessment tool. River Res. Appl. 28,
13941416.
Zilani, L., Surian, N., 2012. Evolutionary trajectory of channel morphology and controlling
factors in a large gravel bed river. Geomorphology 173, 104117.