Morphodynamics of a pseudomeandering gravel bar reach
J. Bartholdy a,*, P. Billi b,1
a
Institute of Geography, University of Copenhagen, ster Voldgade 10, DK 1350 Copenhagen K, Denmark
b
Department of Earth Sciences, University of Ferrara, Corso Ercole I d'Este, n.32, 44100 Ferrara, Italy
Received 27 June 1999; received in revised form 22 April 2001; accepted 9 May 2001
Abstract
A large number of rivers in Tuscany have channel planforms, which are neither straight nor what is usually understood as
meandering. In the typical case, they consist of an almost straight, slightly incised main channel fringed with large lateral
bars and lunate-shaped embayments eroded into the former flood plain. In the past, these rivers have not been recognised as
an individual category and have often been considered to be either braided or meandering. It is suggested here that this type
of river planform be termed pseudomeandering.
A typical pseudomeandering river (the Cecina River) is described and analysed to investigate the main factors responsible
for producing this channel pattern. A study reach (100 Â 300 m) was surveyed in detail and related to data on discharge,
channel changes after floods and grain-size distribution of bed sediments. During 18 months of topographic monitoring, the
inner lateral bar in the study reach expanded and migrated towards the concave outer bank which, concurrently, retreated by
as much as 25 m. A sediment balance was constructed to analyse bar growth and bank retreat in relation to sediment supply
and channel morphology. The conditions necessary to maintain the pseudomeandering morphology of these rivers by
preventing them from developing a meandering planform, are discussed and interpreted as a combination of a few main
factors such as the flashy character of floods, sediment supply (influenced by both natural processes and human impact), the
morphological effects of discharges with contrasting return intervals and the short duration of flood events. Finally, the
channel response to floods with variable sediment transport capacity (represented by bed shear stress) is analysed using a
simple model. It is demonstrated that bend migration is associated with moderate floods while major floods are responsible
for the development of chute channels, which act to suppress bend growth and maintain the low sinuosity configuration of
the river. D 2002 Elsevier Science B.V. All rights reserved.
Keywords: Pseudomeandering; Formative discharge; Lateral bar; Gravel-bed river; Cecina River; Northern Apennines
1. Introduction
Straight or low-sinuosity gravel-bed rivers are
common on active continental margins, where they
control a significant part of landscape development,
and act as important environmental indicators in the
geological record (Ibbeken and Schleyer, 1991). This
type of river commonly shows a distinct, alternate
bar morphology, often referred to as a bedform
pattern scaled with channel width (e.g. Jackson,
1975; Ashley, 1990). Some authors (e.g. Ackers and
Charlton, 1970; Keller, 1972; Lewin, 1976) regard
these large-scale bedforms as incipient meanders,
while Yalin (1992, p. 171) describes them as ``the `ca-
0169-555X/02/$ - see front matter D 2002 Elsevier Science B.V. All rights reserved.
PII: S0169-555X(01)00092-7
*
Corresponding author. Fax: +45-35322501.
E-mail addresses: JB@geogr.ku.dk (J. Bartholdy),
PBILLI@dicea.unifi.it (P. Billi).
1
Fax: +39-55495333.
www.elsevier.com/locate/geomorph
Geomorphology 42 (2002) 293­310
talysts' which accelerate the formation of meanders''.
In contrast, alternate lateral bars are commonly
observed as an equilibrium form (e.g. Bridge et al.,
1986; Laronne and Duncan, 1992). This makes low-
sinuosity rivers of special interest since they may
represent channels at the threshold of meandering that
are unable, for some reason, to develop ``true'' mean-
der bends.
The development of a straight, meandering or
braided channel pattern has been documented to
depend mainly on discharge, bed gradient (Leopold
and Wolman, 1957; Schumm, 1977; Church, 1996)
and bed material size (Parker, 1976), although other
factors, such as the amount and type of sediment
supply and the flow regime, may be relevant in
determining channel pattern. Our understanding of
the underlying conditions for planform development
can be significantly improved by field investigations
on the sedimentological, morphological and hydraulic
characteristics of low sinuosity channels and by the
detailed description of their response to specific run-
off events.
Several rivers in Tuscany have a channel plan-
form that is neither straight nor meandering. In the
typical case, they consist of an almost straight
macro-channel within which a sinuous base-flow
channel curves around large lateral bars that force
the water towards the opposite bank to erode large,
lunate embayments in the alluvial plain. Chute chan-
nels commonly develop at the inner side of the
lateral bar and interrupt the channel's tendency to
meander. For a short time following chute channel
formation, two channels may be active and the river
morphology bears a strong resemblance to that of a
braided stream, though the bar width is much larger
than the baseflow channel.
In the past, rivers with these characteristics were
not recognised as an individual category and have
been indiscriminately classified simply as braided or
meandering. It is suggested here that this type of
river planform be termed pseudomeandering, as first
suggested by Hickin (1972). An overview of sed-
imentological and geomorphological characteristics
of pseudomeandering rivers of Tuscany and the
description of their channel dynamics according to
a five-stage model is reported in the work by Teruggi
and Billi (1997).
The primary objectives of this paper are: (i) to
describe the channel pattern of a pseudomeandering
river in the northern Apennines; (ii) to analyse the
Fig. 1. Location map of the Cecina River valley, situated in the western part of the northern Apennines, southwest of Florence. The study reach
is marked with a triangle. The location of the outlet is marked with an arrow on the map of Italy (lower left).
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310294
Fig. 2. Below: Longitudinal profile of the Cecina River with the location of the study site. Above: Typical grain-size (D50) of the riffle sediments
in the river (from Billi and Paris, 1992).
Fig. 3. Discharge records for the study period at Ponte di Monterufoli 2.25 km downstream of the study site. The discharge values are based
on a gauging station run by Servizio Idrografico di Pisa (Regional Hydrographic Service). Values smaller than 5 m3
/s are not reported. The
dates of the four surveys are marked with numbers on top of the diagram.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310 295
morphology and texture of a riffle-bar/pool se-
quence and (iii) to relate the observed morphology
and sedimentology to different discharge conditions.
In the study, a reach of the Cecina River was
surveyed four times over a period of 18 months, from
April 1996 to October 1997.
Fig. 4. Vertical aerial photos at the same scale from August 1954; May 1986 and March 1994 of the river reach between the bridge Ponte di
Monterufoli (to the left) and the study site, 2250 m upstream of the bridge (last full side bar to the right in the 1986 and 1994 pictures). The scale
is indicated by the longitudinal profile underneath. It marks the distance upstream of the bridge. Flow is from right to left.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310296
2. Area of study
The Cecina River is located in central Tuscany,
west of Siena. Its catchment area is 900 km2
and its
main stem is about 80-km long (Fig. 1). It is located
between the Arno River to the North and the
Ombrone River to the South and outflows into the
Tyrrhenian Sea at the town of Cecina, approximately
50 km south of Pisa.
The geology of the study area consists mainly of
marine sediments which have been subjected to ex-
tensive block faulting oriented NW­SE, i.e. perpen-
dicular to the course of the Cecina River (Billi et al.,
1991; Carmignani et al., 1995), resulting in a few
discontinuities of the otherwise typical concave lon-
gitudinal profile of the river (Fig. 2). The study reach,
located approximately 55 km from the source and 25
km from the outlet, has an elevation of 37 m a.s.l. and
a bed gradient of 1.85 Â 10  3
.
Bed material D50 (Fig. 2) exhibits a general,
though irregular, downstream decrease. Median sizes
close to  6 phi (64 mm), 40 km from the source, are
replaced by values around  4.5 phi (23 mm), 70 km
from the source (10 km from the outlet) (Billi and
Paris, 1992).
The average annual precipitation is 900 mm (Cog-
nigni, 1996), but the river is highly ephemeral, with
a flashy character, since flood flows are induced
mainly by intense cloudbursts, most commonly dur-
ing autumn and early spring. This is illustrated by the
hydrograph covering the study period (Fig. 3). Flow
records are obtained from the gauging station at
Ponte di Monterufoli, 2250 m downstream of the
study site. The maximum discharge during the study
Fig. 5. Vertical aerial photos from August 1954; May 1986 and March 1994 of the study reach marked with a white arrow. Flow is from top to
bottom. On the 1994 picture, the surveyed area is marked with a black frame. On the 1986 and 1994 pictures, the black arrow points to a brick-
built water well on the property line between two fields. The well fell into the river as a result of bank erosion. In 1994, it was located close to
the cut bank. It is now in the process of being buried by the growing side bar (see Fig. 7B).
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310 297
Fig. 6. Maps of the study reach: (A) April 1996, (B) September 1996, (C) May 1997 and (D) October 1997. The flow is from left to right marked with an arrow in (A), (B) and (C),
sample locations are indicated in (D). The X/Y-plane is the same in all four maps referring to 0.0 as the observation point and the y-direction as magnetic north. The equidistance of the
contour lines (m above sea level) is 0.20 m.
J.Bartholdy,P.Billi/Geomorphology42(2002)293­310298
period was 227 m3
/s and discharge exceeded 200 and
100 m3
/s on one and four occasions with a duration
of 2 and 42 h, respectively. Discharges usually
accepted as descriptive for a river, i.e. Q1.58 and
Q2.33, are 275 and 419 m3
/s, respectively
The general morphology and recent changes in
the river valley at the study site are illustrated in Fig.
4. Three vertical aerial photographs at the same scale
(August, 1954; May, 1986 and March, 1994) are
shown together with a diagram of the riverbed pro-
file derived from levelling between the flow gauge
and the study site. The main valley floor is 800­
1000-m wide and somewhat less inclined (1.3 Â
10  3
) than the riverbed. The elevations of the valley
at the flow gauge and the study site are 36.5 and
40 m a.s.l., respectively (Carta Tecnica Regionale
1:5000, 1979).
The change in land use between the dates of the
two first areal photographs is striking. In 1954, the
valley was tapestried with many small fields, while
in 1986 (as well as at present) larger cultivated plots
are seen. This change took place in the valley bottom
and the surrounding hillsides. By contrast, further
inland, farm abandonment during the 1950s and
1960s has left former upland fields uncultivated,
causing a general decrease in the sediment supply
to the river (Billi and Rinaldi, 1997). During the
same period, the sediment supply deficit had been
exacerbated by extensive bed material mining that
was ultimately forbidden in the whole of Tuscany in
1978 (Billi and Rinaldi, 1997). The resulting reduc-
tion in sediment supply resulted in channel incision
and the river cut down to form a relatively straight,
deep channel with alternate bars. As gravel extrac-
tion ceased after 1978, a consistent supply of bed
material sediment load was re-established and aug-
mented also by bank erosion associated with channel
widening.
In 1954, the reach between the flow gauge and
the study site was straight (sinuosity = 1.02), 50-m
wide and punctuated by alternating, elongated lateral
bars. In 1986, the overall sinuosity was roughly the
same (1.03), but, locally, the growth of bars had
deflected the baseflow channel against the bank, for-
ming lunate-shaped cut banks, in places cut off by
chute channels. In 1994, the sinuosity of the base-
flow channel was 1.18. The reach upstream of the
study site (the latter is marked with a white arrow on
Fig. 5) displays morphological evidence of a com-
plete cycle of change. In 1986, the growth of the
prominent bar had forced the left bank to retreat by
approximately 100 m with respect to the 1954 pic-
ture, and vegetation had covered the innermost,
upstream part of the bar, compensating for approx-
imately half of the opposite bank retreat. Between
1954 and 1986, the river widened, the alternate bar
configuration moved about 140-m downstream and
the sinuosity of the baseflow channel increased. This
development continued, and in 1994, an additional
erosion of about 120 m was related to a rotating
channel migration. It is characteristic that during this
massive cutbank retreat, the transition zone between
this and the downstream bend was fixed in its
original position. This downstream bend was chosen
as the study site because here the lateral bar has
constantly grown from 1986 to the present, altering
the local flow conditions and approaching the thresh-
old condition for a chute cut off, although a cut off
has not, as yet, occurred. Bar expansion was asso-
ciated with almost equivalent bank erosion opposite.
The black arrow in the 1986 picture (Fig. 5) marks a
water well, which fell into the river some time before
1994, and is now being buried by the growing bar on
the opposite bank (Fig. 7B).
3. Methods
The topography of the study site was surveyed
using a total station. Within the 300-m-long reach,
the survey established 15 detailed cross-sections.
Elevations were surveyed by levelling and referred
to a level of 33.00 m a.s.l., corresponding to the staff
0-reference of the Regional Hydrographic Service
gauging station at Ponte di Monterufoli, 2.25 km
downstream of the study site.
During the last field campaign (October 1997), 21
bed material samples were collected from the bar,
along a longitudinal transect, extending from the up-
stream riffle to the downstream pool, and along a
transverse transect, coinciding with the maximum bar
width (Fig. 6D). At each sampling point, surface (one
Dmax thick top layer) and subsurface samples were
shovelled into 10-l buckets. In places with sediment
finer than 20 mm, a single bulk sample of 0.5 l was
collected. The gravel samples were truncated in the
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310 299
field at 20 and 2 mm. The two sub-samples finer than
20 mm were weighed with an accuracy of F 5 g and
split samples were sent to laboratory for standard sieve
and Andreassen pipette analysis at 1/4 phi intervals.
The coarse fraction was spread out on a 1 Â1-m black
plate with white, 10-cm-long bars for scale. Vertical
photos were taken from a tripod and processed by
means of WinChips (1998) image analysing software.
The area of every particle projection was measured
with an accuracy of 1 mm2
, and particle diameter was
calculated as the diameter of the circle with the same
area as the particle projection. This diameter and a
particle density (rs) of 2650 kg/m3
were taken to
calculate the volume of the equivalent sphere, VS,
and the single particle frequency by weight of the
coarse fraction as: f (%) = VS rS(W/WS) Â 100, where W
is the weight of the coarse fraction and WS is the total
weight of the related equivalent spheres. The sub-
sample results were combined into a total grain-size
distribution, based on original fraction frequencies.
4. Sediment
Bed material D50 and the topography of the study
reach measured in April 1996 and October 1997, are
shown in Fig. 8. The sediment on top of the right bank
is very fine (D50 = 0.004 mm) and represents approx-
imately 2 m of overbank deposits resting on coarse-
grained bar deposits at the base of the cut bank. When
undermined by erosion, this massive, fine-grained
layer collapses to form large (as much as 2 m in
diameter) mud blocks, which come to rest the base of
the cut bank (Fig. 7A).
Fig. 7. Photos of the study reach taken in October 1997. (A) The concave cut bank (approximately 4-m high). Notice the old gravel bar at the
base overlain by the fine-grained overbank deposits. Parts of the eroded overbank deposits are left in form of giant ``mud blocks'' at the cliff-
foot. (B) The upstream part of the study reach lateral bar viewed from the cut bank. Notice the old water well to the left (see Fig. 5), now in
the process of being buried by the prograding bar. (C) The gravel bar during the October 1997 survey. In foreground, the theodolite total
station is installed over the observation point. In the distance, the gravel is being sieved and the coarse fraction photographed. (D) A close up
of the largest particles on the bar surface near the observation point. Notice the 30-cm-long ruler for scale.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310300
The grain size of bar surface sediment gradually
decreases from the low-flow channel margin to the
inner part of the bar, reflecting a transition to finer
overbank deposits. In the longitudinal profile (Fig. 8),
notwithstanding some scatter, a downstream fining is
also evident for both surface and subsurface samples.
A combination of the subsurface grain size distribu-
tions proportional to the deposition on the bar has a
mean grain size of  1.5 phi (2.9 mm) and a sorting
coefficient of 2.3 phi. Therefore, though the bar on the
surface seems to be dominated by coarse gravel, the
bar growth is based on material in the sand-gravel
transition.
Visual examination of composite distributions of
subsurface samples (Fig. 9) reveals the occurrence of
distinctive sub-populations, combined in variable pro-
portions. In the riffle and bar stoss sides, a coarse
sub-population with a mean grain size of about  5.5
phi (45 mm) is abundant and represents the concen-
tration of coarse particles associated with peak flood
phases when coarser particles are entrained and
transported. In gravel-bed rivers, selective transport
has been observed by few authors (e.g. Lenzi et al.,
1999) to prevail for discharges ranging from one to
two times critical discharge, while conditions favour-
able to equal mobility are postulated to occur above
Fig. 8. D50 and topographical change in the sampling transects. Above: Cross-section profiles of the study reach. Diamonds and triangles
represents surface and subsurface D50, respectively, referred to the October 1997 survey. Flow is away from reader. Below: Longitudinal
profile across the bar. Sediment samples symbols as above. Flow is toward the right. In places where a visual, field examination showed no
substantial difference between surface and subsurface material, only surface samples were collected. The numbers on the bottom of each
diagram refers to the sample numbers shown on the map of Fig. 6D.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310 301
this range. The finer mode, on the other hand,
represents bedload moving across the bar head that
was deposited during the receding flow. The second
and third diagrams in Fig. 9 represent material
deposited on the bar in the longitudinal and trans-
verse transects respectively (Fig. 8). Three sub-pop-
ulations are present: a coarse one, as discussed above
and two finer washed out from the poorly sorted,
fine sediment of the riffle and the stoss side. In the
longitudinal transect, deposited material is dominated
by the central intermediate population but all three
populations are clearly visible. The deposited mate-
rial, across the bar, is dominated by the fine pop-
ulation, with the other two forming a coarse tail.
5. Bar migration and bank retreat
Contour maps of the study reach (Fig. 6), show the
275-m-long bar, the pool and the cut bank associated
with it. During the monitoring period, the bar grew
transversely while both its upstream and downstream
parts were eroded. As the bar became wider and
shorter, the channel decreased its radius of curvature
from 170 to 110 m and migrated towards the retreat-
ing, concave, cut bank.
Bank erosion was most pronounced (up to 25 m,
equivalent to 17 m/year) in the area just downstream
of the point of maximum curvature (0 m on the x-
axis in Fig. 6), whereas, it was zero in the upstream
part and less than 5 m in the downstream part of the
embayment. Subtracting the October 1997 surface
from the April 1996 surface yields the map above
the block diagram in Fig. 10. It reveals a total bar
deposition of 3.9 Â 103
m3
. Most of this deposition
occurred in a belt (up to 40 m wide) along the
convex, outer side of the bar and represents lateral
accretion as the bar migrated towards the cut bank.
Erosion of the upstream and downstream parts of
the bar (0.5 Â 103
and 1.0 Â 103
m3
, respectively)
Fig. 9. Representative grain-size distributions of subsurface mate-
rial in three areas of the study reach: (I) The eroded area at the riffle
and stoss side of the bar; (II) The area of deposition in the longi-
tudinal transect; (III) The area of deposition in the cross-section of
the bar. The distributions are combined according to the magni-
tude of erosion/deposition between April 1996 and October 1997
surveys.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310302
reduced net bar deposition to 2.4 Â 103
m3
. Bar depo-
sition and bank erosion are reported as percentages of
the totals in Table 1. The temporal variation of de-
position followed the discharge pattern (Fig. 3). Bank
erosion also reflected the discharge pattern, but was
somewhat more sensitive to the duration of floods
with moderate to high discharge, as observed in other
pseudomeandering rivers of the northern Apennines
(Casagli eta l., 1999; Rinaldi and Casagli, 1999),
while bar deposition, was apparently more sensitive
to the magnitude of higher discharges. Average bank
retreat during the 18 months of study was 11 m, equi-
valent to 7 m/year. The total sediment deficit, that is
bank erosion (11.2 Â 103
m3
) minus net deposition on
the bar, is 8.8 Â 103
m3
or 79% of the bank-eroded
material. Hence, inner bar deposition replaced only
about 20% of the material eroded from the outer bank,
and the incoming supply of sediment was, therefore,
significantly smaller than the amount leaving the
study reach. The Cecina is an incised river and the
present, confined flood plain is about 2 m lower than
the old alluvial surface. Releases of relatively large
Fig. 10. Block diagram of the study reach based on the October 1997 survey. The flow is from lower right to upper left. In the foreground,
the densely vegetated part of the reach is not included in the map. The step up from the base level of the diagram (37-m above the sea)
shown here is artificially made by the drawing software. The map above shows the distribution of erosion and deposition between April
1996 and October 1997 surveys. The scale to the right is in metres and shows the legend chosen in order to improve the reading of the 0.2-m
equidistance map. Light grey indicates deposition, white is almost no change, dark grey and dashed indicate erosion.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310 303
amounts of bank material and a sediment deficit were,
therefore, to be expected. Even if the sediment per-
taining to the old alluvium is omitted, bank erosion is
still larger, being more than twice the value of bar
deposition. Thus, bar growth is not the cause of bank
retreat but is more likely the result of a vain attempt to
fill up the space left by the rapidly retreating bank.
This constraint to the vertical growth of the bar, paired
with wide variations of discharge, seems to prevent
the embryonic bends from developing into recog-
nizable meanders and emphasizes the relevance of
sediment supply to channel morphology in pseudo-
meandering rivers.
Erosion in the downstream part of the cut bank
(Fig. 10; x = 120­150 m) occurred during the first
three surveys, along with tightening of the bend
radius and with the channel moving away from the
cut bank. Erosion definitely ceased between the third
and the fourth survey and a secondary sediment
accumulation formed between the low-flow channel
and the concave bank in the downstream part of the
bend (Fig. 10; quadrant x = 125­150 and y = 25­50).
This type of side bar probably results from flow
expansion beyond the zone of streambed narrowing,
imposed by the downstream hook of the cut bank,
and it has been observed in several, similar reaches
in the area. Commonly, these bars are composed of
coarse sediment and they may play an important role
in protecting the downstream part of the cut bank
from erosion. This likely accounts for the lack of, or
the very slow, downstream progression of bends as
observed in other pseudomeandering rivers of the
northern Apennines (Brocchi, 1987).
In the central part of the study reach, where bar
sedimentation was largest, bank retreat was paralleled
by lateral bar growth. This is illustrated in the cross-
section shown in Fig. 11. The hydraulic geometry
parameters of this cross-section are calculated for ver-
tical increments of 0.10 m for each of the four surveys
(Fig. 12).
Cross-sectional area (Fig. 12A) constantly in-
creases from 0 at 36 m a.s.l. to a maximum of about
230 m2
at 39.8 m a.s.l. (i.e. the elevation of the old
alluvial plain margin), while wetted perimeter (Fig.
12B) and width (Fig. 12C) rise in a stepped pattern.
The minimum width/depth ratio occurs at an eleva-
tion of about 37.8 m a.s.l. This corresponds to a
discharge of 100 m3
/s, which is usually exceeded
several times a year (Fig. 3). The maximum width/
depth ratio (120) (Fig. 12D) occurs at the elevation of
the upper part of the bar (38.7 m a.s.l.). Field
Fig. 11. Cross profile of the study reach in the central part of the bend (x = 0; Fig. 6) from each of the four surveys: April 1996, September 1996,
May 1997 and October 1997. The succession of the surveys appears from the erosion in the cut bank.
Table 1
Bar deposition (2.4 Â 103
m3
) and bank erosion (11.2 Â 103
m3
)
expressed as percentages of the total change between the survey
periods
4/1996­
9/1996
9/1996­
5/1997
5/1997­
10/1997
Bar deposition 15 55 30
Bank erosion 14 71 15
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310304
evidence and hydraulic calculations (see Section 6)
indicate that this elevation corresponds to a discharge
close to Q1.5, which is usually regarded as the do-
minant or bankfull discharge. This finding contrasts
the concept of Williams (1978) who identified bank-
full conditions as associated with a minimum in the
width/depth ratio. Such a difference can be accounted
for by the strong asymmetry of the cross-section, with
width increasing at a faster rate than mean depth.
Between the first and the last survey, the maximum
width of the study reach increased by 10%, from 106
to 117 m, and the mean radius of curvature of the
baseflow channel decreased from 170 to 110 m. This
evolution is most likely a consequence of moderate
floods tending to form regular meanders. Field obser-
vations (Teruggi and Billi, 1997) indicate the ten-
dency to form regular meanders is interrupted by
larger floods. During these floods ( Q10­Q20), the
bar top is inundated but, given the small size and the
confinement of the present flood plain, no over bank
Fig. 12. Morphometric parameters as a function of water level from the four surveys of the reference cross-section at the point of maximum
curvature (x = 0; Fig. 6). (A) Cross-sectional area. (B) Wetted perimeter. (C) Channel width. (D) Width/depth-ratio. (E) Hydraulic radius. (F)
Mean depth.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310 305
expansion takes place since the flow is constrained
within the old alluvial plain banks. The morpholog-
ical effectiveness of larger floods is therefore fully
preserved and they are capable of generating chute
cutoffs. Since such chute cutoffs represent the typical
final stage in the cycle of adjustment of a pseudo-
meandering reach (Teruggi and Billi, 1997), two
distinct levels of morphologically effective floods
and an unbalanced sediment flux seem to be impor-
tant conditions favouring the development of a pseu-
domeandering channel pattern.
6. Dynamics
In order to investigate the morphological responses
to floods of different magnitude, a simple model has
been set up for the study reach. The model does not
include the forces arising from change in momentum
downstream and across the stream (Smith and Mc-
Lean, 1984). However, as stated by Dietrich (1987,
p. 200) simple calculations omitting these terms ``can
be used to explain the basic vertically averaged flow
pattern in bends''. This applies particularly well to
long reaches of nearly constant radius of curvature
like the one represented by the central cross-section
across the study reach.
At the centre of the bar, surface D84 (usually
selected to match hydraulic roughness, k) is 80 mm
(sample 9, Fig. 7), while the mean grain size of
deposited material across the bar is 40 mm. If these
two grain sizes, regarded as maximum and minimum
surface roughness in the profile, are used to calculate
Manning's n using n = k1/6
/25.4, (Engelund and
Petersen, 1974), an intermediate n of 0.024 is obtained
for the reference cross-section. This value is used in
the Manning formula
V  n1
d2=3
S1=2
1
where V = mean flow velocity, d = mean depth and
S = slope, to produce flow and channel dynamic simu-
lations (Fig. 13) by a simple model describing the
flow in the bend.
The longstream water slope is altered over the
bend as S = Sc(rc/r) (Allen, 1985), rc being the radius
of curvature at the centreline, r the local radius of
curvature and Sc the centreline slope. The local depth
is varied over the bend by subtracting the bed top-
ography from the local water level. The water level
is calculated across the channel starting from the
centreline by adding a local water level change on
the base of the transverse slope induced by the local
centrifugal acceleration, Str = V 2
/gr, where g is the
acceleration due to gravity. For a given discharge, the
water level over the bend was found by iteration,
calculating the local water level and discharge for
every metre over the cross-section. The calculations
were continued with an increment of 0.01 m of the
water level at the centreline, until the sum of the
local discharges corresponded to the discharge
chosen for the simulation. Bed shear stress was then
calculated from so = dSgq, where q is the density of
water.
In the first simulation (Fig. 13A), the largest
discharge occurred during the observation period
(227 m3
/s on 2/6/1997, close to Q1.5 = 275 m3
/s) is
simulated, based on the October 1997 river morphol-
ogy. The modelled water level reaches an elevation
slightly lower than that associated with the width/
depth ratio peak (Fig. 12D). During the flood, at the
location where sample 9 was collected, the bar was
accreting and bed shear stress reached 16 N/m2
. The
corresponding largest moveable particle (Dmax), cal-
culated using the method of Komar (1989) as dis-
cussed in Komar (1996):
Dmax  s00:045qs  qgD0:6
50 1
h i2:5
2
where qs is particle density (2650 kg/m3
), has a size
of 60 mm. The coarsest particles in sample 9 are 120
and 70 mm for surface and subsurface, respectively;
thus, Komar's equation gives a good prediction,
though it underestimates the maximum moveable
grain size at the surface. Other equations provide the
following values, intermediate between surface and
subsurface Dmax: Baker and Ritter (1975), 84 mm;
Williams (1983), 94 mm. A simple critical Shields
parameter of hc = 0.045 gives 22 mm, which is
between surface and subsurface D50, but closer to
the latter. Thus, the calculated maximum and median
particle sizes of subsurface material are of the right
order of magnitude, which supports the applicability
of the model.
The other three simulations indicate how a dis-
charge with a recurrence interval of 20 years (818
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310306
m3
/s) would affect the bend in three different mor-
phological situations: (i) the 1986 situation with a
radius of curvature of 400 m (estimated from the
1986 aerial photo of Fig. 4) (Fig. 13B). This is a
rough approximation, but the only one possible; (ii)
the topography and radius of curvature (170 m) from
the April 1996 survey (Fig. 13C); (iii) the topo-
graphy and radius of curvature (110 m) from the
October 1997 survey (Fig. 13D).
The patterns of Fig. 13B,C and D are, therefore,
controlled primarily by the radius of curvature, which
decreased from 400 to 170 and then to 110 m. In
simulation b, the water level is just beneath the old
floodplain level, while in simulations c and d, the
more tightly curved bend forces the inclined water
surface over the outer bank and flow depth becomes
smaller in the inner part of the bend. Here, bed shear
stress increases considerably as a result of the larger
longstream slope in the even more tightly curved
bend, while elsewhere on the cross profile, shear
stress varies according to variations in water depth.
The occurrence of a secondary shear stress maximum
in the inner part of the bar, around x =  50 m, is
relevant to channel pattern change. During the three
Fig. 13. Simulation of water level, velocity and bed shear stress in the cross profile of the central part of the study reach in four different
situations: (A) The highest measured discharge in the study period (June 1997) with topography and radius of curvature of October 1997
survey. (B) Discharge corresponding to a recurrence interval of 20 years with topography from the April 1996 survey and a radius of
curvature of 400 m. (C) Discharge corresponding to a recurrence interval of 20 years with topography from the April 1996 survey and a
radius of curvature of 170 m. (D) Discharge corresponding to a recurrence interval of 20 years with topography and radius of curvature (110
m) from the October 1997 survey.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310 307
simulations, this secondary shear stress maximum
increases from 34, to 41 and finally to 43 N/m2
. In
simulation d, the transition between lower and upper
flow regime is reached. In simulations c and d, shear
stress in the inner part of the bend exceeds the max-
imum bed shear stress in the centre of the channel
during a 1.5-year flood (simulation a). The simula-
tions show the present bend is so tightly curved that,
during a 20-year flood, the flow over the bar produ-
ces a bed shear stress, which is large enough to erode
a new main channel in the initial form of a chute cut
off. Lewis and Lewin (1983) observed that chute cut-
off frequency is maximum when the rc/w ratio (radius
of curvature to width) is in the range between one and
two. In the Cecina study site, the ratio is well within
this range, although no chute cutoff yet has formed,
as no large flood has occurred since the study bend
matured. Though the Cecina River has many geo-
morphological similarities to the Welsh rivers studied
by Lewis and Lewin (1983), the adjustment of bend
geometry alone is not enough to account for the de-
velopment of a chute cutoff and other factors, such as
a sediment imbalance and the above mentioned dual
set of formative discharge, need to be considered.
7. Discussion and conclusions
Channel pattern is controlled by the mutual inter-
action of many factors of which dominant discharge,
often taken as Q1.5 or Q2.33, is regarded as sufficient to
describe the hydrological part. In the case of Cecina
River, however, it seems necessary to consider the
effects of two different formative discharges to under-
stand its pseudomeandering planform. Frequent, mod-
erate floods are capable of shaping the river towards a
meandering pattern, but this development is interrup-
ted by large floods ( Q10­Q20), which cut through the
incipient point bars and force the main channel back
towards a straight planform. A plot of channel gra-
dient versus discharge (Leopold and Wolman, 1957)
places the Cecina River in the braided category
(though very close to the line that discriminates
braided from meandering rivers), whereas, a plot of
slope versus a combination of discharge and grain size
(Parker, 1979), using the mean grain size of deposited
material (2.9 mm), places the river just below the
braiding threshold with Q1.5 and very close to braid-
ing with Q20. As bank resistance plays a major role for
the planform development, this factor should not be
neglected. Ferguson (1973) suggested consideration
of the bank content of fine grained material, but in the
case of the study site, this is not possible since the cut
bank consists mainly of coarse gravel (slightly
cemented at the base) overlain by laminated sand
and massive silt and clay. Use of the slope/Froude
number ratio (S/Fr) in combination with depth/width
ratio (d/w), as suggested by Parker (1976), seems to
be an appropriate alternative since depth/width ratio is
implicitly related to bank resistance. Both Q1.5 and
Q20 plot in the meandering part of Parker's (1976)
diagram with Q20 close to the transition towards a
straight channel pattern.
It is suggested here that the Cecina River's pseu-
domeandering morphology results from the combina-
tion of the following main factors.
(1) Incision--the present overbank (inner bar top)
deposits are approximately 2 m lower than the old
(probably Middle Age, Benvenuti, personal commu-
nication) flood plain. This in itself is not a distinctive
factor for the development of pseudomeandering riv-
ers, but nor are the resulting implications that follow.
(2) Cohesive bank deposits--the channel is pre-
vented from braiding because of raised bank resist-
ance: banks consists mainly of coarse gravel (in
places slightly cemented at the base) overlain by a
massive bench of fine grained, overbank deposits.
(3) Sediment supply--though in the study reach,
the balance between incoming and outcoming sedi-
ment is negative, the sediment supply is sufficient for
side bars to grow and deflect the flow towards the
opposite banks during moderate floods. Therefore, bar
growth is no longer the cause but merely the result of
a vain attempt to fill up the space left by the rapidly
retreating bank.
(4) Flow regime--During frequent, moderate
floods, the morphology of the embryonic bends
develops towards a meandering pattern through a
gradual decrease of the radius of curvature. Infre-
quent, large floods ( Q20), being confined within an
incised channel, have gradient and power enough to
cut through the lateral bars and leave the incipient
meander bends as inactive, lunate embayments in the
elevated cut bank.
(5) Flash floods--the short duration of even large
floods, in combination with the enhanced bank resist-
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310308
ance, prevents the river from channel-width-adjust-
ments towards equilibrium during the largest floods.
On the existing basis, it is not possible to determine
whether pseodomeandering channel morphology is a
previously unrecognised type of stable river pattern or
merely represents a transitional, unstable phase, fol-
lowing incision in the present setting. Nevertheless,
the pseudomeandering river planform is very common
in Tuscany, and so far the modern morphology of the
Cecina River seems stable, in the long term, since it
results from the balance of opposite tendencies to form
meanders during frequent floods and to assume a
straight course during infrequent, large floods. It could
be argued that the development since 1954 (Figs. 4 and
5) represents the change from a straight towards a
meandering planform in the new lower level of the
incised river. Whether or not this development will be
interrupted by the next very large flood forming a new
straight planform remains to be seen. However, the Q20
simulations in the studied river reach, and the fact that
the first formed bends were actually cut off by a flood
between 1954 and 1986 (Fig. 4 at about 1000 m),
speaks in favour of the suggested cyclic development.
Acknowledgements
This research benefited from funds provided by the
two institutions involved. We are pleased to express
our gratitude to Ditte, Mette and Anders for their
assistance in the field, to Bjarne Fogh for his help with
analysing the grain-size photos, to Kent Prkesen for
the layout assistance, Ulf Thomas and Paul Christian-
sen for the logistic, and to Kirsten Simonsen at the
Skalling laboratory for the grain-size analysis. Gordon
Grant reviewed an earlier version of the manuscript.
We are pleased to express our gratitude for his good
comments and linguistic corrections. Likewise, two
anonymous referees are thanked for their valuable
suggestions.
References
Ackers, P., Charlton, F.G., 1970. The geometry of small meandering
streams. Proc. - Inst. Civ. Eng. S12, 289­317.
Allen, J.R.L., 1985. Principles of Physical Sedimentology. Allen &
Unwin, London, 272 pp.
Ashley, G., 1990. Classification of large-scale subaqueous bed-
forms: a new look at an old problem. J. Sediment. Petrol. 60,
160­172.
Baker, V.R., Ritter, D.F., 1975. Competence of rivers to transport
coarse bedload materials. Geol. Soc. Am. Bull. 86, 975­978.
Billi, P., Paris, E., 1992. Bed sediment characterisation in river
engineering problems. In: Bogen, J., Walling, D.E., Day, T.J.
(Eds.), Erosion and Sediment Transport Monitoring Pro-
grammes in River Basins, vol. 210. IAHS Press, Wallingford,
pp. 11­20.
Billi, P., Rinaldi, M., 1997. Human impact on sediment yield and
channel dynamics in the Arno River Basin (central Italy). IAHS
Publ. 245, 301­311.
Billi, P., Losito, G., Muschietti, M., 1991. Geological setting, tec-
tonic evolution and related seismic activity. In: Vannucchi, G.
(Ed.), Seismic Hazard and Site Effects in the Florence Area.
Dipartimento di Ingegneria Civile, Universita` di Firenze, Fire-
nze, pp. 1­9.
Bridge, J.S., Smith, N.D., Trent, F., Gabel, S.L., Bernstein, P., 1986.
Sedimentology and morphology of a low-sinuosity river: Calmus
River, Nebraska Sand Hills. Sedimentology 33, 851­870.
Brocchi, L., 1987. Dinamica morfologica del Fiume Arno dalla
confluenza con la Chiana alla Gonfolina. Caratteri morfologici.
Unpublished M.Sc. Thesis, University of Florence, 92 pp.
Carmignani, L., Decandia, F.A., Disperati, L., Fantozzi, P.L., Laz-
zarotto, A., Liotta, D., Oggiano, G., 1995. Relationships be-
tween the Tertiary structural evolution of the Sardinia-Corsica-
Provencal Domain and the Northern Apennines. Terra Nova 4,
128­137.
Carta Tecnica Regionale 1:5000, 1979. Regione Toscana.
Casagli, N., Rinaldi, M., Gargini, A., Curini, A., 1999. Pore water
pressure and streambank stability: results from a monitoring site
in the Sieve River, Italy. Earth Surf. Processes Landforms 24,
1095­1114.
Church, M., 1996. Channel morphology and typology. In: Petts, G.,
Carlow, P. (Eds.), River Flows and Channel Forms. Blackwell,
Oxford, pp. 185­202.
Cognigni, G., 1996. Geomorfologia dell'alveo del Fiume Cecina.
Unpublished M.Sc. thesis. University of Florence, 149 pp.
Dietrich, W.E., 1987. Mechanics of flow and sediment transport in
river bends. In: Richards, K. (Ed.), River Channels. The Institute
of British Geographars Special Publication 17, Blackwell, Ox-
ford, pp. 179­227.
Engelund, F.A., Petersen, F.B., 1974. Hydraylik. Den private ingen-
irfond, Danmarks Tekniske Hjskole, 233 pp.
Ferguson, R.I., 1973. Channel pattern and sediment type. Area 5,
38­41.
Hickin, E.D., 1972. Pseudomeanders and point dunes--flume study.
Am. J. Sci. 272, 762­799.
Ibbeken, H., Schleyer, R., 1991. Source and Sediment. A Case
Study of Provenance and Mass Balance at an Active Plate Mar-
gin (Calabria Southern Italy). Springer, Berlin, 286 pp.
Jackson, R.G., 1975. Hierarchical attributes and a unifying model of
bedforms composed of cohesionsless material and produced by
shearing flow. Geol. Soc. Am. Bull. 86, 1523­1533.
Keller, E.A., 1972. Development of alluvial stream channels: a five-
stage model. Geol. Soc. Am. Bull. 83, 1531­1536.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310 309
Komar, P.D., 1989. Flow-competence evolutions of the hydraulic
parameters of floods: an assessment of the technique. In:
Beven, K., Carling, P. (Eds.), Floods: Hydrological, Sedimen-
tological and Geomorphological Implications. Wiley, Chichester,
pp. 107­134.
Komar, P., 1996. Entrainment of sediments from deposits of mixed
drain sizes and densities. In: Carling, P., Dawson, M.R. (Eds.),
Advances in Fluvial Dynamics and Stratigraphy. Wiley, Chi-
chester, pp. 127­181.
Laronne, J.B., Duncan, M.J., 1992. Bedload transport paths and
gravel bar formation. In: Billi, P., Hey, R.D., Thorne, C.R.,
Tacconi, P. (Eds.), Dynamics of Gravel-Bed Rivers. Wiley, Chi-
chester, pp. 177­202.
Lenzi, M., D'Agostino, V., Billi, P., 1999. Bedload transport in the
instrumented catchment of the Rio Cordon: Part I. Analysis of
bedload records, conditions and threshold of bedload entrain-
ment. Catena 36, 171­190.
Leopold, L.B., Wolman, M.G., 1957. River channel patterns--
braiding, meandering and straight. U.S. Geol. Surv. Prof. Pap.
282B, 39­85.
Lewin, J., 1976. Initiation of bed forms and meanders in coarse-
grained sediment. Geol. Soc. Am. Bull. 87, 281­285.
Lewis, G.W., Lewin, J., 1983. Alluvial cutoffs in Wales and the
borderlands. In: Collinson, J.D., Lewin, J. (Eds.), Modern and
Ancient Fluvial Systems. IAHS Spec. Publ. 6. Blackwell, Ox-
ford, pp. 145­154.
Parker, G., 1976. On the cause and characteristic scale of meander-
ing and braiding in rivers. J. Fluid Mech. 76, 459­480.
Parker, G., 1979. Hydraulic geometry of active gravel rivers. J.
Hydraul. Div., Am. Soc. Civ. Eng. 105, 1185­1201.
Rinaldi, M., Casagli, N., 1999. Stability of streambanks formed in
partially saturated soil and effects of negative pore water pres-
sures: the Sieve River (Italy). Geomorphology 26, 253­277.
Schumm, S.A., 1977. Fluvial System. Wiley, New York, 338 pp.
Smith, J.D., McLean, S.R., 1984. A model for flow in meandering
streams. Water Resour. Res. 20, 1301­1315.
Teruggi, L.B., Billi, P., 1997. Sedimentology of a pseudomeander-
ing river (Cecina R., central Italy). G. Geol. 59 (1­2), 267­272.
Williams, G.P., 1978. Bankfull discharge of rivers. Water Resource
Research 14, 1141­1158.
Williams, G.P., 1983. Paleohydrological methods and some exam-
ples from Swedish fluvial environments. Geogr. Ann. 65A,
227­244.
Winchips, 1998 Published by the Institute of Geography, University
of Copenhagen. http://www.lh@geogr.ku.dk.
Yalin, M.S., 1992. River Mechanics. Pergamon, Oxford, 220 pp.
J. Bartholdy, P. Billi / Geomorphology 42 (2002) 293­310310