Review
Geomorphological and sedimentological analysis of flash-flood deposits
The case of the 1997 Rivillas flood (Spain)
Jose A. Ortega , Guillermina Garzón Heydt
Departamento Geodinamica, Facultad de Geología, Universidad Complutense de Madrid, C/Jose Antonio Novais, no. 2, 28040 Madrid, Spain
a b s t r a c ta r t i c l e i n f o
Article history:
Received 28 November 2008
Received in revised form 5 May 2009
Accepted 6 May 2009
Available online 14 May 2009
Keywords:
Flash-flood deposits
Sedimentary features classification
Hydraulic stability
Stream channel changes
Guadiana River
On the basis of the description of the 1997 Rivillas flood deposits, a morphosedimentary feature classification
is proposed. Mapping of the main morphosedimentary deposits in seven reaches along the basin has
provided abundant data for each defined typology and for a better adjustment of their stability fields.
Because of their unstable preservation environment, immediate post-flood field surveys with descriptions of
erosive and depositional features were undertaken. Up to 18 features were classified as either sedimentary or
erosive and mapped according to their genetic environments. Anthropic interference such as land use
changes produce modification of sediment supply and in channel and floodplain erosive processes causing
flash-floods to be more catastrophic. Erosive features are dominant over sedimentary ones, as the
sedimentary budget in the river is negative. By means of HEC-RAS (Hydrologic Engineering Center)
modelling, we were able to obtain mean values of the main variables limiting feature stability (velocity,
depth, stream powers and shear stress). These provide information regarding maximum stability threshold
and peak flood discharge. The ephemeral nature of riverine flash-flood deposits in this type of setting does
not mean that they are not significant, and their interpretation after recent floods can significantly improve
interpretation of the event dynamics and its flood hydrology and also be useful for flood risk mapping.
 2009 Elsevier B.V. All rights reserved.
1. Introduction
A flash-flood represents an abrupt and short duration rise in the
discharge of a stream with a dramatic contrast among the event and
the extended period between floods (Reid, 2004). The geomorphic
effects of flash-floods are significant especially if the event offers a
long lasting peak discharge (Costa and O'Connor, 1995). The term
flash-flood embraces several types of processes and geomorphic
environments, such as ephemeral streams (Greenbaum et al.,1998) or
alluvial rivers and fans, but the literature is more concerned with
bedload dynamics (Reid et al., 1998) or quantification of sedimenttransport
ratios (Cohen and Laronne, 2005). Flash-floods have also
been studied from various perspectives, such as their meteorological
control (Harnack et al., 2001), or as paleohydraulic approach for peak
discharge (Baker, 1973, 1978; Costa, 1983).
Flash-floods are usually related to ephemeral streams, and the Bijou
Creek paper on flood deposits (McKee et al., 1967) was taken as an
example for flashy, sheetflood sequence in a slightly modified model
by Miall (1977, 1996). Recent flash-floods deposits typically have
been described related to alluvial fans and pediments in semiarid
environments (Lucchitta and Suneson, 1981, Sneh, 1983; Stear, 1985)
consisting of a complex flow pattern of channel and sheet-like deposits
related to open, distributary (spreading out) systems, although a
special case of confined flash-flood deposits has been also described
by Sneh (1983).
As Stear (1985) pointed out, an important aspect that needs to be
stressed is the erosional effects of these catastrophic floods. In the
present study, we are dealing with a narrow alluvial plain confined by
valley slopes where erosional processes dominate over the depositional
ones. Although the flood deposits in these ephemeral streams
environments have low preservation potential, their interpretation
can notably improve the understanding of the flood dynamics and
hydrology.
In November 7th 1997, one of the most destructive floods occurred
in the Iberian Peninsula during the 20th century in the Rivillas stream,
affecting the town of Badajoz (Fig. 1). The flood caused 23 deaths and
material losses estimated at $150 M USD. These figures were
unexpected for a 34 km long tributary of the Guadiana River, with a
low relief drainage basin of 314 km2
. The disastrous character of the
event was assumed to be related to the flashy character of the event
and the land use on the floodplain. In fact, a shift toward intensive
farming at the watershed have been made, and the active channel
running through a built up area had been channelized, thus favouring
human occupation of the banks (Ortega et al., 1998).
Geomorphology 112 (2009) 1­14
 Corresponding author. Tel.: +34 913944850; fax: +34 913944845.
E-mail addresses: jaortega@geo.ucm.es (J.A. Ortega), minigar@geo.ucm.es
(G. Garzón Heydt).
0169-555X/$ ­ see front matter  2009 Elsevier B.V. All rights reserved.
doi:10.1016/j.geomorph.2009.05.004
Contents lists available at ScienceDirect
Geomorphology
journal homepage: www.elsevier.com/locate/geomorph
Because of the low preservation potential of the flash-flood
deposits, field surveys were conducted immediately after the event.
Detailed descriptions of the erosional and depositional features were
carried out for comparison with the hydraulic flood analysis. A
morphosedimentary classification of scours and deposits is proposed
here on the basis of the descriptions of the 1997 Rivillas flood
deposits.
The first objective of this paper is to establish a classification of
geomorphological and depositional features considering their spatial
distribution that might be useful to analysis of other flash-flood events
in similar environments. It is not our intention to present an
exhaustive literature review, but to provide a systematic description
and analysis of the 1977 Rivillas flood that could be use as a reference
for further studies on the flash-floods.
Another aim is to identify features generated during the
maximum flood discharge in order to achieve more information of
overbank sediment deposits, like Walling et al. (1997) suggest.
Several authors have dealt with the subject of trying to relate
sedimentary features to flow parameters, among them Leeder (1982)
and Miall (1996). Shouthard (1975) and Costello and Southard (1981)
tried to establish relations between water depth and bedforms.
Dalrymple et al. (1978) and Shouthard (1975) related features to flow
velocity, and Ashley (1990) related them to mean sediment size.
Simons et al. (1961) related features to water depth and velocity and
Froude number, and Shields (1936) and Leopold et al. (1964) to shear
stress. Baker and Costa (1987) established relationships between
stream power and incision or degradation, and recent studies
have addressed geomorphic changes and effectiveness (Costa and
O'Connor, 1995).
The analysis of the hydrologic and hydraulic data of the Rivillas
flood was a prerequisite to interpret the factors controlling the spatial
distribution and characteristics of the observed erosional and
depositional features. In our study, HEC-RAS (Hydrologic Engineering
Center) hydraulic modelling was used in combination with highwater
marks measured after the flood, applying analogous techniques
as the ones used for palaeohydrological estimation (Patton et al.,
1979). Finally, we tried to elucidate what information can be inferred
from each feature for interpreting flood characteristics peak flood
parameters. We used the results obtained with HEC-RAS modelling to
estimate the most significant variables, such as depth, velocity, shear
stress and stream power in morphosedimentary features stability
fields under which bedforms are preserved. Hydraulic modelling has
been shown to be useful to determine floodplain flow characteristics
(Siggers et al., 1999), distribution of flow velocities across the
floodplain (Gee et al., 1990) or to predict net floodplain deposition
(Moody and Troutman, 2000).
Finally, there is an interesting issue from the applied perspective to
determine flood behaviour at particular sites including constrictions.
During high magnitude floods, in confined valley reaches, like the
studied one, there is a great variation in hydraulic conditions and flood
hazard severity along the floodplain; especially due to its complex
microtopography and morphology (Walling and He, 1998).
Fig. 1. Location of the study area and geological sketch of the Rivillas Stream.
2 J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
2. Regional setting and the 1997 flood event
The Guadiana River, with a drainage basin of 66,800 km2
and
semiarid conditions (rainfall 400­500 mm/y), is one of the largest in
the Iberian Pensinsula; it is the fifth in terms of discharge and length.
It crosses the Península from east to west through various geological
units before flowing into the Atlantic Ocean. The Rivillas stream is a
34 km long tributary of the Guadiana River with a basin of 314 km2
and an average gradient of 0.0075 m/m. The Rivillas main tributary is
the Calamon Stream. It has a similar size and length, so that the peak
discharge tends to occur at the same time at their confluence located
in Badajoz urban area, converging a high hazard and exposure.
From a geological point of view, most of the basin is excavated in
fine-grained Tertiary detritical sediments. Only the headwaters
present outcrops of metamorphosed limestones and shales (Fig. 1).
An important fact, that affects the sedimentary forms, is the
constrictions and modification of the river course, so that the flood
waters flow is different from the normal pattern of overbank flow and
sluggishness over the floodplain. In this case, the alluvial plain actually
works under flood conditions as a high velocity channel of flow in
semi confined environment.
On 5­6 November 1997, a highly active front crossed the Iberian
Peninsula in a SW­NE direction producing heavy rainfall that reached
historic maxima at almost every rain gauge station in the area and
equalling or exceeding the 500-year return period. A mesoscale
convective storm (MCS) released 120 mm of rain over the entire basin.
The precipitation event reached a high intensity since a significant
part of the rain fell in 1 h (Fig. 2). Moreover, all nearby rivers in Spain
and Portugal produced floods with catastrophic results in terms of
human and material damage: 23 deaths in Spain and 10 in Portugal.
The antecedent rainfall was also heavy, exceeding 84 mm in the
previous 3 days. This fact combined with the intense rainfall and the
confluence of the large Calamon tributary stream at the lower part of
the Rivillas basin within the built up area produced a very rapid rise in
the floodwaters. There was a lag of barely 2 h between rainfall and
flood peak in Badajoz.
The end result, in geomorphological terms, was a drastic transformation
in the floodplain, with the creation of numerous depositional and
erosive forms described and analyzed in this paper.
3. Methods
Fieldwork after a flash-flood needs to be as quick as possible in
order to describe the generated features before they are disturbed.
This is not only because of human activity, including recovering and
ploughing the floodplain, but also, because the large growth of
vegetation in the fertile floodplain soils that causes rapid alteration of
the flood features by bioturbation.
A reconnaissance survey of the river reaches where geomorphic
impacts were greatest was undertaken immediately after the flood
occurred. Seven reaches were chosen as representative of the upper,
middle, and lower basin sectors. Some other reaches were selected
because they presented high anthrophic occupation and transformation,
where peculiar features controlled by the influence of human
elements were formed. The length of the selected reaches ranged from
2 km to 500 m. Features in these areas were described and mapped,
and in some cases, a detailed topographic survey was undertaken in
order to perform HEC-RAS hydraulic modelling.
Morphosedimentary features were mapped in each reach on a
1:4500 scale topographic map, showing only features large enough
to be mappable at that scale. The river layout as it was in 1956
(the date of the oldest available aerial photographs) and the
present river, mapped after the flood, was also established. In
addition to the map of morphosedimentary features, maps of every
reach were carried out by interpreting aerial photographs taken in
1956 and 1982 (most recent available aerial photographs). Basin
changes were identified comparing both maps. This comparison
allowed us to determine, whether the morphosedimentary features
were present or not and to infer an aggradation­degradation
balance for the river.
Two methods were used to estimate the peak flows of the 1997
flood. The first was a unitary hydrograph using the HEC-1
programme, which produced an estimate of 653 m3
s-1
for peak
discharge; the hydrograph is shown in Fig. 2. The second one was an
hydraulic method using HEC-RAS programme yielded a peak
discharge of 799 m3
s-1
. We believe that the latter is more accurate
given that it was verified with flotsam data in three zones outside the
built up area. The peak discharge derived from the unitary
hydrograph was dependent on theoretical estimations of the number
of curves method and precipitation data that were not uniform all
over the basin.
Finally, we used HEC-RAS hydraulic modelling in order to estimate
some hydraulic parameters of the flood and relate them with the
depositional features. Flood peak discharge was estimated in three of
the analysed reaches. These were selected based on the variety of
features and because they are distributed representatively along the
river. The valley slope was used as a contour condition, on the
assumption that it was the same as the energy line slope; and a
Fig. 2. Rainfall distribution and hydrograph of the 1997 flood event for different sites at the Rivillas basin.
3J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
subcritical flow type was considered. Energy losses were calculated
from roughness using the table for cultivated areas presented in Chow
(1959), which provided better results for the three studied reaches
than the Cowan (1956) formula method. Contraction and expansion
losses were estimated at 0.1 and 0.3 for sections with no geometrical
changes and between 0.3 and 0.5 for sections with significant changes.
There is no available gauging station to calibrate the final results, but
flotsams were used as markers of the maximum flood level (Baker,
1978). Those field measurements considered less reliable owing to the
proximity of perturbations (such as buildings, fences, etc.) were
discarded.
4. Watershed and floodplain anthropogenic changes
From the comparison between the 1956 and 1982 aerial
photographs and from the basin situation after the flood, several
changes were identified. The most important land use changes and
damage occurred at Badajoz city, which achieved urban expansion
mainly by spreading out over the Rivillas and Calamon floodplains.
The scale of the disaster was also favoured by the false sense of
safety induced by the channelization of the streams. A remarkable
aspect to take into account, however, is the extraordinary scale of
the damage to the city, not only because of the presence of buildings
but also to the increase in bedload induced by anthropic changes
upstream.
The land use changes in the basin consists mainly of removal of the
original vegetation cover, changes of crop types, floodplain transformation,
and road construction. Comparison of aerial photographs has
revealed that changes were particularly dramatic in the previous
years, and their direct effect on the floodplain can be observed on
stream reaches showing direct anthropic influence.
Anthropic and land use changes may be differentiated as affecting
the basin or the floodplain. The principal alterations in the basin are a
consequence of the conversion of "dehesa-type" open forest pastureland
into vineyards and olive orchards, especially on hillslopes. This
fact favoured intensive soil erosion and removal of large amounts of
sediment into the floodplain leading to the development of small of
alluvial fans on the flanks of the floodplain in the first stages of the
flood event.
On the alluvial plain, the principal changes consisted of channel
realignment with meanders obliteration and removing of riparian
vegetation. Some of the results of these changes can be seen on the
Cansini and Romera reaches (Figs. 3 and 4). Sediment delivery
increased from valley side slope enhanced alluvial fan deposition on
the flanks of the floodplain. The river's sinuosity was reduced from
1.32 to 1.07, producing local floodplain erosion with the recovery of
earlier channels as stream velocity and power increased. The removal
of riparian vegetation diminished channel and alluvial plain flow
resistance, favouring erosion. However, where riparian vegetation had
grown on the stream banks without restraint from the abandonment
of pastures, channel constriction and reduction in flow capacity
occurred. In addition, anthropic constructions (such as causeways,
bridges, roads, and buildings) also limited the flow capacity of the
channel and floodplain.
5. Morphosedimentary features
The morphosedimentary features recorded on the field were used
to produce maps (Fig. 5) and to interpret and classify the identified
features. A total of 17 features either depositional or erosive were
classified (Table 1) and subdivided according to their spatial location.
Feature size was also considered on the following scale: macroform
(decametric), mesoform (metric), and microform (decimetric).
Depositional features were mapped and classified into three
geomorphic environments: channel bank, floodplain, and slope
deposits. Twelve groups of deposits are described according to their
architecture and location in relation to the river and their genetic
interpretation on the basis of texture, structure, size and depth, and
lithofacies (Miall, 1996). Five types of erosive features were found and
have been classified as scours, pools, and channel incisions. The origin
of most of them, however, is controlled by anthropic activities or
structures on the alluvial plain.
Fig. 3. Changes occurred at the Cansini reach between 1956 (A) and (C) when the channel
was sinuous and 1982 (A) and (B) after channel straightening and intensive farming.
4 J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
The first geomorphic environment considered, channel bank
deposits, are related to ordinary channel water outflow onto the
floodplain. They commonly build up in the early stages of the flood
and are subsequently reshaped. Under natural floodplain conditions,
the largest grain sizes are generally deposited on the channel banks,
and finer sediments spreads over the floodplain. In the case of the
1977 Rivillas flood complex morphologic adjustments occurred, due
to confinement of the channel by relatively high, artificial sandy levees
and anomalous riparian vegetation and to the narrowness of the
floodplain, which confined the flood instead of causing energy
dissipation. To our knowledge, some of the features developed by
the Rivillas flood have not been described in the literature before.
Three types of channel bank depositional features have been
identified: crevasse splays, linear levees and "digitated" levees.
5.1. Crevasse splays
In this group we included the crevasses splays that had not been
reworked. They present the characteristic crevasse splay morphology
(Miall, 1977) and coarse sand and even granules and pebbles showing
a progressive decrease in the grain size away from the outlet (Moya
et al., 1998a).
Numerous examples of this type are found on the banks, as the
channel's efficiency was only 2% of peak discharge. Consequently,
from the earliest stages of the flood, there was overbank flow and
generation of crevasse splays of sizes ranging up to 160 m long and
30 m wide, although some later evolved into longitudinal bars (Fig. 6).
They have been found in most cases associated with anthropic
changes in channel curves and direction, but they also occur under
natural conditions such as channel constrictions and expansions.
5.2. Linear levees
Linear levees are natural levees formed by immediate overbank
accretion of coarse sands granules and pebbles as classically described in
the literature (Singh, 1972). However, they present extensive continuity
along the artificially straightened channel, favoured by anthropogenic
Fig. 4. Anthropogenic changes in stream sinuosity index (s) and riparian vegetation
cover (v) in Romera reach between 1956 and 1997.
Fig. 5. Map of sedimentary features in the Cansini reach. Arrows show main flow direction during the flood describing a meandering trajectory.
5J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
accumulation of channel dredging deposits and agricultural dumps. They
reach 50 m in continuous length and 40 cm in height.
5.3. Digitated levees
Changes in riparian vegetation, cultivation practices and the
presence of abundant loose sediments have favoured the development
of a peculiar type of levee designated as "digitated levees" (Moya
et al., 1998a,b). These special sigmoid levees, derived from the linear
levees, present a discontinuous elongated and finger like morphology
as they are reshaped during advanced stages of flooding.
A former linear levee is split into discontinuous forms that
evolve as shown in Fig. 7. The initial continuous levee is spread out
into a special type of crevasse splay and reworked by floodplain
longitudinal flow that elongates the deposit downstream. When
flooding subsides, the reversed flow that returns into the stream
channel from the floodplain dissects the features, reshaping their
morphology into a digitated form. These deposits consist of
coarsening upwards and planar cross-bedded medium size sands,
with intercalated beds of fines. They reach up to 13 m in length and
50 cm in height.
6. Alluvial plain deposits
This group does not consist of ordinary overbank deposits related
to outflows from the channel and leading to vertical aggradation like
the regular bedsets described by Bridge (1984) as produced by
discrete overbank floods. In the Rivillas stream, the narrow alluvial
plain nearly functions as a channel during floods and the valley
slopes are sometimes acting as channel banks. Under these
conditions high energy forms develop on the floodplain. Within
these alluvial plain we have recognised seven types of deposits:
longitudinal, semilunate bars and sand flats (macroforms) and
mudball bars, shadow bars, longitudinal and transverse gravel bars
(mesoforms).
6.1. Longitudinal bars
Longitudinal bars are the most frequent depositional feature in all
the river reaches and consisted of sandy accumulations elongated in
the direction of flow but slightly curved. Most are built up by former
crevasse splays deposits removed at the high-water stage. In the
Cansini reach, several longitudinal bars have been formed as a
continuation of crevasse splays (Fig. 6). The preservation of this type
of depositional feature is favoured by the flashy character of the flood,
which allows features to be preserved from hydraulic conditions, as
proposed by Jones (1977) and suggested by Miall (1977) for the Bijou
Creek model (McKee et al., 1967).
These bars are composed of medium to fine sands, occasionally
presenting some coarser interbedded material from lateral supply.
They have an elongated and narrow geometry reaching 180 m in
length and 50 cm in height. Trough cross-bedding occurs and dune
and ripple bedforms appear on the surface. The proposed origin
involves the migration of material eroded from the levees that evolve
into crevasse splay lobes that continue growing further downstream
(Fig. 6). The stream-lined sand accumulation results in longitudinal
bars with an undulating surface as they are not located in maximum
flow velocity areas where erosion prevails.
Based on the location of the bars in relation to the channel and
floodplain, we propose a sinuous flow model when the floodplain is
entirely inundated. Flow sinuosity is possibly comparable to the model
described by Simons et al. (1961) for flumes with large width/depth
ratios, as we are looking at a similar situation of floodplain
confinement on the Rivillas stream during peak discharge.
6.2. Semilunate bars
Semilunate bars are lateral or marginal bars developed at the
floodplain margins. They are composed of fine sands and silts with
parallel or low-angle bedding and rectilinear or sinuous ripple structures
on the surface. They form narrow and elongated deposits up to 160 m in
long and 30 cm thick, parallel to the river. We have not found
individualised descriptions of similar bars in the literature, as these
cannot be considered as lateral channel bars with tractive structures and
relative high flow energy. Genetically, theycould be related to slackwater
deposits that result from flow dissipation in protected areas where
deposits accumulate and preserve. Because flotsams are usually
associated with them, they are good maximum water stage indicators,
being their top surface 20 cm below the maximum flotsam height.
6.3. Sand flats
Sand flats deposits imply that the entire alluvial plain is acting as a
flat, shallow, and homogeneous extraordinary channel while the
ordinary channel provides a continuous overbank flow from its
breached, confining levees. Sand flats were not very frequent; sandy
accumulation features are located marginally to the channel and
laterally to the floodplain. They are formed by composite bars and
combined composite bars as described by Allen (1980). Cant and
Walker (1978) suggested that they were produced by nucleation of
small sandy bars, and Miall (1978) considered that they contained
internal three-dimensional structures. Their main characteristics are
fine to medium sands and silts with cross-bedding. Deposits are up to
10 m in length and 30 cm in thickness. Curved megaripples 1.5 m long
appear on the surface (Fig. 8F).
In view of their marginal position, they are likely to reflect shallow
water. In the sand flats located closer to the channel, where the water
depth is higher during the peak flood, megaripples and composite and
combined forms were developed.
6.4. Longitudinal mudball bars
A special feature caused by the intense erosion undergone by the
floodplain is the formation of longitudinal mudball bars. These are
produced by the erosion of the upper argillic horizon in patches to a
depth of 30 cm. Although they are formed under natural conditions,
Table 1
Morphosedimentary features classification.
Process Location Architecture Grain
size
Feature Size
Sedimentation Floodplain Sand flats Sand Sand flats Macro and
mesoform
Bars Longitudinal bar Macroform
Semilunate bar
Shadow bar Mesoform
Gravel Longitudinal Macro-
mesoformTransverse
Mudball Longitudinal Mesoform
Channel
bank
Splays Sand Crevasse splay Macroform
Linear levees Mesoform
Digitated levees
Lateral Alluvial fans Sand
and
gravel
Symmetric fans Macroform
Asymmetric
fans
Erosion Floodplain Scours ­ Longitudinal
scour fields
Macroform
Lee hollows Meso and
microform
Channel and
floodplain
Pools Pools (kolks) Mesoform
Tributaries Channels Tributary
entrenchment
Macroform
Channel Former channel
entrenchment
Macro and
mesoform
6 J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
their origin has been favoured by changes in land use; like the
elimination of former open forest (dehesas) and increased plugging
depths achieved by modern equipment. The cohesive upper soil
horizon is literally stripped away, and the eroded material deposited
in the immediate proximity forms poorly developed bars. These bars
are shallow and up to 40 m long. The mud clast may initially present
imbrications, but they tend to acquire a subspherical shape in a short
distance and become degraded loosing their shape further downstream
(Fig. 8E).
6.5. Shadow or lee bars
The formation of these mesoforms is related to the presence of
anthropic obstacles in the floodplain, such as wells, houses, and any
structure that can cause flow diversion and the build up of a bar
downstream. The deposits consist of coarse sand with no internal
stratification or visible structures, and the length depends on the
obstacle size with a maximum length of 6 m and a maximum
thickness of 50 cm.
6.6. Longitudinal and transverse gravel bars
In spite of the lack of coarse material in the channel, some
longitudinal and transverse gravel bars related to antrophogenic
deposits have been found, including mill tailings from a factory and
other types of dumping material.
Particle size is variable; but tabular, marble blocks up to 40 cm long
have been observed showing imbrication fabrics. Bars may be up to
Fig. 6. Longitudinal bar development from a crevasse splay at the Cansini site. (A) General view of crevasse splay and longitudinal bar; (B) bar development stages; and (C) sinuous
flow model and longitudinal bars distribution on the floodplain.
7J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
40 m long and 20 m wide. Their presence implies that the flow had
enough energy to entrain and transport planar blocks. These particles
were transported several hundred meters, revealing high energy
conditions.
7. Hillslope features: alluvial fans
The entire basin is subject to intense slope erosion, favoured
chiefly by deforestation and crop transformation. The sediment
supply on the Rivillas floodplain increased considerably leading
to the development of alluvial fans on the valley margins, which
were later reworked by the flood acting as sediment sources. The
alluvial fans studied here have been classified based on their plan
view morphology into symmetric or asymmetric. Fans are
present at the mouth of low order streams flowing, and those
reworked by the longitudinal flow of the flood have attained
asymmetric geometry. Deposits consist of alternating fine and
coarse layers, 40 cm deep. Sinuous sand ripples are found on the
surface.
Alluvial fans in semiarid zones receive large amounts of sediments
during high energy flash-floods (Bull, 1977). In the Rivillas basin,
sediment is largely derived from piping in the slopes (Fig. 8A, B), and
stripping of argillic horizon stripping on cultivated crop fields or
infilled creeks. Sediment supplies are greatest in areas of changed land
use, such as vineyards and olive orchards with gradients exceeding 3%,
together with "dehesa" open forest.
8. Erosional features
Five groups of erosional features have been identified; most of
them are related to anthropic activities on the floodplain. These
features may be located on the floodplain (longitudinal scour fields
and lee hollows), on the floodplain and channel (pools, kolks), on
tributaries (tributary entrenchment), or directly in the channel
(former channel entrenchment).
8.1. Longitudinal scour fields
These stream-lined microforms occur as swarms and are designated
in this work as longitudinal scour fields (Fig. 8C). The maximum
depth of the depression is 60 cm, the length is 200 m, and water depth
ranges from 1 to 50 cm. The forms observed are either straight or
meandering and according to Allen (1969), this is because whenever
the critical stress is exceeded erosion gradually augments transforming
small longitudinal scours into meandering ones. They are
produced essentially by excess flow energy, which is further enhanced
Fig. 7. Digitated levees. (A) Development sequence; (B) return flow from the inundated floodplain into the channel; (C) digitated levees view.
8 J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
Fig. 8. (A) Asymmetrical alluvial fan developed from a lateral stream and elongated downstream by the main flow; (B) small alluvial fans related to piping processes on vineyards
slopes; (C) longitudinal scour fields from to ploughing parallel to the main direction of flow; (D) entrenchment in a tributary stream at Romera reach; (E) mudball bar development;
and (F) downstream edge of a sand flat.
9J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
by inappropriate farming practices such as deep ploughing parallel to
the main direction of flow.
We believe that this form was generated during the early stages of
the flood. Erosion intensified up until peak flow was reached; and then
as the waters subsided, many furrows were filled with sandy deposits
from longitudinal bars. The large extension attained by scour fields in
all the studied reaches indicates the high bedload transference
capability of the flood waters.
8.2. Lee hollows
Lee hollows form just downstream of obstacles related to elements
destroyed and swept away by the floodwaters, such as blocks of brick
fabric or large sections of fences. This suggests that the hollows were
formed in the later stages of the flood, subsequent to overflow onto
the plain and following the high energy peak flow that damaged
buildings. In some cases, sand bars are covering the displaced block,
which acquire a half-moon or horse-tail shape from the vortices
caused by the obstacle.
8.3. Pools
Unlike the previous hollows, the main vortex occurs at a fixed
point, and the pools deepen to as much as 60 cm reaching a maximum
length 10 m. Their occurrence is always associated with obstacles of
anthropic origin, such as roads, paths, and bridges. This form always
develops downstream of the obstacle, with the more abrupt face
upstream and gentle slope downstream. Finally, some of them may
expand into longitudinal scours.
They are initiated in the highest flow stages of the flood when
stream power is at a maximum. The erosional work, however,
continue after the peak flow have subsided as long as there is water
circulating over the obstacle because a waterfall is created enhancing
the erosional capability of the flow.
8.4. Tributary entrenchment
This process has occurred preferentially in the upper part of the
basin where tributary watersheds have undergone considerable
alteration from farming. In some cases, the tributary channels have
disappeared as a result of artificial earth remobilization. During this
exceptional rainfall event, many tributaries recovered their channels,
by incision to depths as high as 30­40 cm (Fig. 8D). The sediment
produced by tributary entrenchment was largely accumulated in fans
and bars.
8.5. Former channel entrenchment
On several reaches, the former channel (which can be seen in the
aerial photographs from 1956), has been reactivated and entrenched
showing hollows not associated with obstacles. Because of the recent
anthropic changes on the river sinuous stretches had been cut off by
channelization works and infilled. During the flood, entrenchment or
reoccupation in artificially abandoned channels occurred where the
substrate was least cohesive. The sequence of formation is indicated in
Fig. 9 for reach 3.
9. Stability conditions of the morphosedimentary features and
discussion
In order to produce some results on the most important variables
that condition hydrologic conditions under which the morphosedimentary
features form and remain preserved, HEC-RAS hydraulic
modelling results for flood peak discharge have been taken into
consideration. Three reaches (Fig. 1): Romera (site 5), Cansini (site
3) and Galache (site 1) were selected based on their landform
variety. Final results are shown in Table 2. Differences between flood
water level according to flotsam and to the calculated maximum
discharge were 34 and 44 cm using Chow's and Cowan's roughness
values respectively. If a reliability analysis is run on the flotsam
measurements, the difference may be reduced to 26 cm (Table 2;
Fig. 10).
The modelling provides quantitative data on the most important
hydraulic variables for the different morphosedimentary features
allowing us to derive a number of significant conclusions about the
flood. The considered variables were water depth, velocity, shear
stress, and stream power. Morphosedimentary maps of the three
studied reaches and HEC-RAS hydraulic output data were used to
calculate the range of hydraulic conditions within which features are
formed and preserved.
The results (Table 3) suggest that depth is not a limiting factor for
the preservation of features after the flood. Some present a wide
range of depths, up to 140 cm for longitudinal sandy bars and
150 cm for crevasse splays. In both cases, the maximum depth
threshold is high, implying that the resistance of the feature to
erosion by water declines. Alluvial fans on the floodplain remain
stable under a water flow b1 m deep. Erosive forms show the
highest formation depth values, between 100 and 150 cm for scours;
while former channel entrenchment attains maximum depths of
170 cm.
Velocity is the most significant factor in the formation of
sedimentary or erosional features (Table 3; Fig. 11). The threshold
between the development of erosional and depositional features has
been established at around 1 m s-1
, although this limit is diffuse. In
longitudinal sandy bars that have not been destroyed after flooding,
the maximumvalue is 1.1 m s-1
, although in most cases this limit is less
than 0.9 m s-1
. In crevasse splays, thresholds are similar (b0.9 m s-1
),
but alluvial fans display lower range, between 0.7 and 1.1 m s1
.
Erosional features require the highest velocity, but that does not
indicate the moment of formation. Scours and former channel
entrenchment features from in high velocity areas (between 1 m s-1
and 1.67 m s-1
).
In relation to shear stress there is a clear threshold between
sedimentary and erosional features developed on the floodplain.
Longitudinal bars withstand a shear stress range of 9 to 47 N m-2
,
with values most commonly around 20 N m-2
; crevasse splay values
tend to be above 16 N m-2
, with a maximum of 31 N m-2
, and alluvial
fans present similar values between 15 and 39 N m2
. In the case of
erosional features, shear stress values can be as high as 93 N m-2
and
most commonly over 30 N m-2
; and in former channel entrenchments,
they are over 35 N m-2
.
The last of the variables considered, stream power, presents a similar
distribution for erosional features (22 and 152 W m-2
for scours, and
slightly lower for former channel entrenchment, 36­69 W m-2
) and
sedimentary features (10­38 W m-2
for fans, 0.8­29 W m-2
for
crevasse splays, and 15­55 W m-2
for longitudinal bars).
For the peak flood (at Calamon and Rivillas confluence), we have
estimated a maximum stream power of 923 W m-2
and mean values
of 348 W m-2
. The unit energy expended, however, was 22600×103
J,
which results high when compared to the values analysed by other
authors (Costa and O'Connor, 1995; Magilligan et al., 1998). Extreme
geomorphic impact or effectiveness may be achieved either by high
stream power and unit energy expenditure (Jarrett and Costa, 1986;
Costa, 1994) or by low stream power but a long lasting peak flow on
the hydrograph (Osterkamp and Costa, 1987). As a result, we may
conclude that the energy expenditure values for the Rivillas flood are
high compared to those of others with larger peak discharge and
longer duration.
Our conclusions should be seen as a first approximation to feature
characterization, relying on descriptive and interpretation aspects. In
fact, the original goal when we undertook the analysis of the
morphosedimentary features was to infer practical information that
10 J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
could help us to gain a better understanding of flood evolution and
hazards. The identification of so many peculiar, or at least hitherto
unreported, features prompted investigation of their genetic significance
and comparison with modelling results, (initially hydraulic
modelling was undertaken only to establish flood discharge
parameters).
These results are only preliminary as HEC-RAS is a one-dimensional
hydraulic model with which only three mean values can be
obtained for each section: in this cases channel, right bank, and left
bank. Therefore, the estimated conditions and stability fields cannot
be considered to be well constrained from our data. Also, the real scale
of the variables is too large for estimation of small features such as
linear and digitated levees, pools, etc. Moreover, due to scale
limitations, the actual value of variables for small features may differ
considerably from those obtained by means of the hydraulic
modelling.
An exhaustive initial data collection with modelling of specific
features would supply far more information and with a higher
accurateness than we obtained. More detailed mapping combined
with depth analysis would also make possible to establish
sedimentation­erosion rates for a more quantitative balance. A
better section knowledge, such as the one offered by threedimensional
models would be important for more precise variable
adjustment.
10. Conclusions
Human interference is considered to be one of the main factors
involved in the described development of the morphosedimentary
features. Land use changes and anthropic structures produce changes
in runoff coefficient, sediment supply and erosional processes in the
Table 2
Differences between flotsam height and the one obtained from both roughnesses
considered options.
Site Average difference (m) Discharge
(m3
s-1
)Cowan Chow
Site 5 0.58 0.53 (0.43) 156
Site 3 0.33 0.23 (0.21) 180
Site 1 0.42 0.26 (0.15) 300
Rivillas (average) 0.44 0.34 (0.26) Results
marked with an asterisk () are considered reliable.
Fig. 9. Former channel entrenchment evolution (Cansini site). (A) Development of a channel entrenchment from an obliterated sinuous channel; (B) 1956 sinuous channel; (C) 1982
straightened channel; and (D) entrenched channel after the 1997 flood event.
11J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
channel and floodplain enhancing the flashy character of the flood and
its severity (capability to cause damage). There is a negative balance in
basin sedimentary budget, and erosive features are dominant over
depositional ones.
The study of some features has yielded detailed information about
water depth, flow velocity and other variables that are necessary to
understand flood evolution. Sand flats form under low flow regime
and shallow water depth. Longitudinal bars are sediment accumulation
zones and adopt a sinusoidal shape on the floodplain, following
flow lines. The coarser material in these bars is related to human
activity, as are bars of mudclast, both occurring from excess energy
whenever local conditions are favourable. Semilunate bars define a
marginal position on the floodplain and imply flow diversion, as do
shadow bars. Crevasse splays indicate remobilization of material, but
in a short time span because shapes are preserved. Digitated and
linear levees yield information about the start and end of the flood.
Fans show that slopes are the main source of sediment and the
Table 3
Variables calculated for the three analysed reaches.
Feature Reach Water
depth (cm)
Velocity
(m s-1
)
Shear
stress
(N m- 2
)
Stream
power
(W m-2
)
Bars and
splays
Longitudinal
sand bars
Site 5 50­100 0.4­0.9 9­40 4­33
Site 3 100­140 0.8­1 21­33 18­34
Site 1 60­140 0.7­1.1 19­47 15­55
Range 60­140 0.4­1.1 9­47 4­55
Crevasse
splays
Site 5 b150 0.6 16.5 11
Site 3 30­100 0.2­0.9 4­31 0.8­29
Range 30­150 0.2­0.9 4­31 0.8­29
Alluvial fans Site 5 b100 0.7­1.01 14.9­37.3 10­38
Site 3 b30 0.9 39.2 37
Site 1 b30 0.95 29.1 28
Range b100 0.7­1.01 15­39 10­38
Erosional
features
Longitudinal
scour fields
Site 5 150 1.37­1.67 81­93 120­152
Site 3 100­145 1­1.42 31­76 32­108
Site 1 130 0.68­1.2 17­57 22­71
Range 100­150 0.68­1.67 17­93 22­152
Former
channel
entrenchment
Site 5 170 1.1 58.7 69
Site 3 120 1 35.4 36
Range 120­170 1­1.1 35­58 36­69 Fig. 11. Stability fields for velocity and water depth obtained for some features by
modeling with HEC-RAS.
Fig. 10. Adjustment of roughness values in relation to flotsam data. Values that consider offering possible flow perturbation are marked with a circle.
12 J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
abundance of these reveal the impact of basin land use changes in the
basin. These sediments preserve the sequence of events in the flood
stages.
Erosional features were abundant and widely distributed although
not very diversified. In almost all cases they are related to human
activities or structures and reveal the high energy expenditure.
Tributary entrenchment reflects the landform recovery after human
induced obliteration. Similarly former channels entrenchment is
indicative of the capacity of fluvial systems for self-adjustment, as
they try to recover their original morphology channel pattern
favouring more effective energy dissipation.
The analysis of stability variables shows a clear differentiation
between erosional and sedimentary features in the floodplain. The
latter indicate the threshold of stability of a feature on the floodplain,
especially velocity, shear stress, and stream power. Depth, however, is
not a limiting factor for feature preservation.
The identification of some features can improve our knowledge
about attained conditions on the floodplain and has also an applied
interest. Flood hazard mapping should be based not only on the area
submerged by the flood and the water depth, but also on its velocity,
shear stress or stream power.
Mapping of flood prone areas is usually based on the delineation of
areas subject to inundation during flood events with a particular
return period. This strategy which is very dependent on statistical
approaches and on the quality of hydrological information, does not
consider the specific risk at a certain location.
Changes in channel pattern and straightening, together with
artificial fil1ing of channels that acted as subsidiary flood pathways,
are responsible for altering the hydraulic characteristics of
the floods and magnifying their potential for valley-floor erosion.
The lessons learnt from this flood have implications for the
evaluation of potential hazards and identification of risk locations.
These are the areas in which the application of mitigation
measures, either of preventive or corrective nature, should be of
top priority.
As Millar and Parkinson (1993) indicate it may be more useful the
information on actual floods and particular kinds of hazards, than
predicting the areas that might be affected by a theoretical flood with
a certain return period.
Acknowledgements
The authors wish to thank Angela Alonso and José Arribas for
kindly reading this article and for their constructive suggestions;
Julio Garrote for his support in the surveying; and Agustín Blanco for
figure drawing. We would also like to three anonymous reviewers for
their comments, which helped improve the original manuscript. This
work was funded by projects No. BTE-2003-045 and CGL2004-03049
of MYCIT.
References
Allen, J.R., 1969. Erosional current marks of weakly cohesive mud beds. J. Sediment.
Petrol. 39, 607­623.
Allen, J.R.L., 1980. Sand waves: a model of origin and internal structure. Sediment. Geol.
26, 281­328.
Ashley, G.M., 1990. Classification of large-scale subaqueous bedforms: a new look at an
old problem. J. Sediment. Petrol. 60, 160­172.
Baker, V.R., 1973. Paleohydrology and sedimentology of Lake Missoula flooding in
eastern Washington. Spec. Pap. - Geol. Soc. Am. 144.
Baker, V.R., 1978. Large-scale erosional and depositional features of the channeled
scabland. In: Baker, V.R., Nummedal, D. (Eds.), The Channeled Scabland. InNASA,
Washington D.C., pp. 81­116.
Baker, V.R., Costa, J.E., 1987. Flood power. In: Mayer, L., Nash, D. (Eds.), Catastrophic
Flooding. InAllen & Unwin, Boston, pp. 1­21.
Bridge, J.S., 1984. Large-scale facies sequences in alluvial overbank environments. J.
Sediment. Petrol. 54, 583­588.
Bull, W.B., 1977. The alluvial fan environment. Prog. Phys. Geogr. 1, 222­270.
Cant, D.J., Walker, R.G.,1978. Fluvial processes and facies sequences in the sandy braided
south Saskatchewan River, Canada. Sedimentology 25, 625­648.
Chow, V.T., 1959. Open Channel Hydraulics. Mc Graw-Hill, New York.
Cohen, H., Laronne, J.B., 2005. High rates of sediment transport by flashfloods in the
Southern Judean Desert, Israel. Hydrol. Process. 19 (8), 1687­1702.
Costa, J.E., 1983. Paleohydraulic reconstruction of flash-flood peaks from boulder
deposits in the Colorado Front Range. Geol. Soc. Am. Bull. 94, 986­1004.
Costa, J.E., 1994. Multiple flow processes accompanying a dam-break flood in a Small
Upland Watershed, Centralia, Washington: U.S. Geological Survey Water-Resources
Investigations Report 94-4026. 24 pp.
Costa, J.E., O'Connor, J.E., 1995. Geomorphically effective floods. In: Costa, J.E., Millar, A.,
Potter, K., Wilcock, P.R. (Eds.), Natural and Anthropogenic Influences in Fluvial
Geomorphology: AGU monograph, vol. 89, pp. 45­56.
Costello, W.R., Southard, J.B., 1981. Flume experiments on lower-flow-regime bed forms
in coarse sand. J. Sediment. Petrol. 51, 849­864.
Cowan, W.L., 1956. Estimating hydraulic roughness coefficients. Agr. Eng. 37 (7),
473­475.
Dalrymple, R.W., Knight, R.J., Lambiase, J.J., 1978. Bedforms and their hydraulic
stability relationships in a tidal environment, Bay of Fundy, Canada. Nature 275,
100­104.
Gee, D.M., Anderson, M.G., Baird, L., 1990. Large-scale floodplain modeling. Earth Surf.
Process. Landf. 15, 513­523.
Greenbaum, N., Margalit, A., Schick, A.P., Sharon, D., Baker, V.R., 1998. A high magnitude
storm and flood in a hyperarid catchment, Nahal Zin, Negev Desert, Israel. Hydrol.
Process. 12, 1­23.
Harnack, R., Appfel, K., Georgescu, M., Baines, S., 2001. The determination of observed
atmospheric differences between heavy and light precipitation events in New
Jersey, USA. Int. J. Climatol. 21 (12), 1529­1560.
Jarrett, R.D., Costa, J.E., 1986. Hydrology, geomorphology, and dam-break. Modelling of
the July 15, 1982 Lawn Lake Dam and Cascade Lake Dam Failures, Larimer County,
Colorado, U.S. Geological Survey Professional Paper 1369. 78 pp.
Jones, C.M., 1977. Effects of varying discharge regimes on bedform sedimentary structures
in modern rivers. Geology 5, 567­570.
Leeder, M.R., 1982. Sedimentology. Process and Product. Allen and Unwin, London.
Leopold, L.B., Wolman, M.G., Miller, J.P., 1964. Fluvial Processes in Geomorphology.
Freeman, San Francisco, CA.
Lucchitta, I., Suneson, N., 1981. Flash flood in Arizona. Observation and their application
to the identification of flash-flood deposits in the geologic record. Geology 9,
414­418.
Magilligan, F.J., Phillips, J.D., James, J.A., Gomez, B., 1998. Geomorphic and sedimentological
controls on the effectiveness of an extreme flood. J. Geol. 106, 87­96.
McKee, E.D., Crosby, E.J., Berryhill, H.L., 1967. Flood deposits, Bijou Creek, Colorado, June
1965. J. Sediment. Petrol. 37, 829­851.
Miall, A.D., 1977. A review of the braided river depositional environment. Earth Sci. Rev.
13, 1­62.
Miall, A.D.,1978. Lithofacies types and vertical profile models in braided rivers deposits:
a summary. In: Miall, A.D. (Ed.), Fluvial sedimentology: Can Soc Petrol Mem, vol. 5,
pp. 597­604.
Miall, A.D., 1996. The Geology of Fluvial Deposits. Springer, Berlin.
Millar, A.J., Parkinson, D.J., 1993. Flood hydrology and geomorphic effects on river
channels and flood plains: the flood of November 4­5, 1985, in the South Branch
Potomac River Basin of West Virginia. In: Jacobson, R.B. (Ed.), Geomorphic Studies
of the Storm and Flood of November 3­5, 1985, in the Upper Potomac and Cheat
River Basins in West Virginia and Virginia: U.S. Geological Survey Bulletin, vol.1981,
pp. E1­E96.
Moody, J.A., Troutman, B.M., 2000. Quantitative model of the growth of floodplains by
vertical accretion. Earth Surf. Process. Landf. 25, 115­133.
Moya, M.E., Garzón, G., Ortega, J.A., Centeno, J.D., 1998a. Estructuras sedimentarias
resultado del desbordamiento del Arroyo Rivillas, Badajoz (Espaa), Noviembre de
1997. Fin. Proc. Cong. Nacional de Geología. Lisboa. Portugal (In Spanish).
Moya, M.E., Garzón, G., Ortega, J.A., 1998b. Depósitos de la avenida del Arroyo Rivillas,
Badajoz. Noviembre de 1997. V Reunión Nacional de Geomorfología, Granada, Spain
(In Spanish), pp. 33­36.
Ortega, J.A., Garzón, G., Moya, M.E., 1998. Parámetros climáticos y ambientales de la
avenida del río Rivillas. V Reunión Nacional de Geomorfología, Granada, Spain (In
Spanish), pp. 237­245.
Osterkamp, W.R., Costa, J.E., 1987. Changes accompanying an extraordinary flood on
sand-bed stream. In: Mayer, L., Nash, D. (Eds.), Catastrophic Flooding. InAllen &
Unwin, pp. 201­224.
Patton, P.C., Baker, V.R., Kochel, R.C., 1979. Slack-water deposits: a geomorphic
technique for the interpretation of fluvial paleohydrology. In: Rhodes, D.D.,
Williams, G.P. (Eds.), Adjustment of the Fluvial System, pp. 225­252.
Reid, I., 2004. Flash-flood. In: Goudie, A.S. (Ed.), Encyclopedia of Geomorphology.
InRoutledge, London. 1156 pp.
Reid, I., Laronne, J.B., Powell, D.M., 1998. Flash-flood and bedload dynamics of desert
gravel-bed streams. Hydrol. Process. 12 (4), 543­557.
Shields, A.,1936. Application of similarity principles and turbulence research to bedload
movement. Hydrodynamics Laboratory Publication No. 167. InInstitute of Technology,
Pasadena, CA.
Shouthard, J.B., 1975. Bed configurations. SEPM Short Course N°2, pp. 5­43.
Siggers, G.B., Bates, P.D., Anderson, M.G., Walling, D.E., He, Q., 1999. Earth Surf. Process.
Landf. 24, 211­231.
Simons, D.B., Richardson, E.V., Nordin, C.F., 1961. Sedimentary structures generated
by flow in alluvial channels. In: Middleton, G.V. (Ed.), Primery Sedimentary
Structures and their Hydrodynamic Interpretation: Soc Econ Paleontol Mineral
Spec Publ, vol. 12, pp. 34­52.
Singh, I.B.,1972. On the bedding in the natural-levee and point bar deposits of the Gomti
River, Uttar Pradesh, India. Sediment. Geol. 7, 309­317.
13J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14
Sneh, A., 1983. Desert stream sequences in the Sinai Peninsula. J. Sediment. Petrol. 53,
1271­1279.
Stear, W.M.,1985. Comparison of the bedform distribution and dynamics of modern and
ancient sandy ephemeral flood deposits in the Southwestern Karoo Region, South
Africa. Sediment. Geol. 45, 209­230.
Walling, D.E., He, Q., 1998. The spatial variability of overbank sedimentation on river
floodplains. Geomorphology 24, 209­223.
Walling, D.E., Owens, P.N., Leeks, G.J.L., 1997. The characteristics of overbank deposits
associated with major flood event in the catchment of the River Ouse, Yorkshire,
UK. Catena 31, 53­75.
14 J.A. Ortega, G. Garzón Heydt / Geomorphology 112 (2009) 1­14