Catena 70 (2007) 416 427

www.elsevier.com/locate/catena

Effectiveness and geomorphological impacts of check dams for soil

erosion control in a semiarid Mediterranean catchment:
El Crcavo (Murcia, Spain)
V.M. Castillo a,, W.M. Mosch a , C. Conesa Garca b , G.G. Barber a ,
J.A. Navarro Cano a , F. Lpez-Bermdez b
a
Departamento de Conservacin de Suelos y Agua, Centro de Edafologa y Biologa Aplicada del Segura. CSIC. Apdo.164, 30100 Espinardo Murcia, Spain
b
Departamento de Geografa, Universidad de Murcia, Campus La Merced, 30001 Murcia, Spain
Received 14 March 2005; received in revised form 6 November 2006; accepted 27 November 2006

Abstract

The construction of transverse structures (check dams) is a widely used method in Spanish Mediterranean areas to stabilise gullies and
ephemeral streams, to reduce channel incision and to prevent sediment deposition downstream. The effectiveness of check dams and their
effects on the morphology of ephemeral channels (ramblas) were investigated in a semiarid, highly degraded catchment. In the study area, 36
dams were surveyed, of which 29 were filled up with sediments, 2 had been destroyed and only 5 had still not completely filled with
sediments. The streams above the dams held sediments, which resulted in a decrease in the longitudinal gradient. Field observations of
changes in the cross-sectional shape of the stream channel, the composition of channel bed material, and bankfull stage measurements
indicated that the dams cause erosion downstream. The amount of sediments stored by the dams was found to be higher than the amount of
eroded material in the downstream reaches of the dam.
2006 Elsevier B.V. All rights reserved.

Keywords: Check dams; Ephemeral streams; Fluvial erosion; Sedimentation; Land degradation

1. Introduction construction of check dams in ephemeral channels (ramblas),

Forest Hydrological Restoration Projects are frequently
applied by the Spanish Forest Administration to reduce soil
erosion and to combat desertification. These projects consist
of the reforestation of degraded hillslopes, and the
Research sponsors: The research was carried out as part of the projects
Seguimiento y Evaluacin de los efectos sobre el Medio Natural de la
Sequa y los Procesos Erosivos. 2: Seguimiento de las actuaciones de lucha
contra la desertificacin, funded by the Comunidad Autnoma de la Regin
de Murcia e INTERREG IIC Programme and Respuesta Hidrolgica y
Geomorfolgica de los Sistemas Fluvio-Torrenciales afectados por proyec-
tos de Restauracin Hidrolgico-Forestal en cuencas semiridas del Sureste
Peninsular funded by the Fundacin Instituto Euromediterrneo de
Hidrotecnia.
Corresponding author. Tel.: +34 968396349; fax: +34 968396213.
E-mail address: victor@cebas.csic.es (V.M. Castillo).
0341-8162/$ - see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.catena.2006.11.009
usually constructed in low order catchments in the upper part
of the drainage network. The small dams (in general less than
5 m high) are supposed to stabilise the channels and reduce
erosion. However, their actual impact is not well known.
Most research on the impact of dams has focused on the
influence of large dams and reservoirs, but less attention has
been paid to the effects and efficiency of small check dams.
As far as we know, there is little information on the
effectiveness of check dams in controlling gullying and soil
erosion. Marston and Dolan (1999) concluded that engi-
neering structures to control sediment export in an arid
watershed in Wyoming (U.S.A) were not always effective or
even necessary. In a study on the effectiveness of loose rock
check dams for gully control in northern Ethiopia, Nyssen
et al. (2004) found that 39% of the dams surveyed had
been destroyed during the first 2 years. Their collapse was
 V.M. Castillo et al. / Catena 70 (2007) 416427
strongly associated with the runoff energy, expressed as the
product of drainage area (A) by the slope gradient (S).
The upstream geomorphological effects of dams differ
from the downstream effects. Channel aggradation gradually
takes place upstream, because of the increase in base level
(e.g. Garca Ruiz and Puigdefabregas, 1984). As a
consequence the dams are rapidly filled up with sediment,
especially in semiarid environments, where sediment yield is
high (Poesen and Hooke, 1997).
Downstream effects seem to be more complex. Changes
in discharge and sediment load can cause an alteration of the
cross-sectional shape, channel form, slope gradient and grain
size of the bed material (Brandt, 2000). Erosion can increase
because of an increase in the erosion capacity of the clear
water, after deposition of sediment behind the dam.
The study of how fluvial processes are affected by dam
construction in ephemeral channels is complicated by the
absence of fluvial processes most of the time. The mor-
phological regime of ephemeral channels is unsteady, be-
cause the responses to irregular, torrential rainfalls are rapid
and cause extreme morphological dynamics in the ephem-
eral channels: channels that are dry for several months may
suddenly carry high discharges and sediment load (Conesa
Garcia, 1995). For this type of environment, large floods
that transform channels and produce large sediment move-
ments have return periods of 26 years (Conesa Garcia,
1995). These high magnitude, low frequency floods ap-
pear to govern channel development (Knighton and Nanson,
1997).
Fig. 1. Location and drainage network of the Crcavo catchment, and the position of the check dams. The check dams studied in detail are indicated in black.
417
The variability of torrential rainfalls both in space and
time, coupled with the effect of transmission losses in the
channels makes it difficult to obtain good data about the
discharge and sediment load of ephemeral channels. As
processes are difficult to measure, this study has concen-
trated on changes in channel morphology.
The objectives of this paper were to evaluate the
effectiveness of check dams for gully control as implemented
in the study area, to identify morphological effects of check
dams in ephemeral streams based on field observations, and
to analyse the impacts on erosiondeposition dynamics
caused by the establishment of the check dams.
2. Study area
The study was carried out in the Crcavo catchment
(3813 N; 131 W) in the NW of the province of Murcia.
This area was subject to a Forest Hydrological Restoration
Project that started in 1969 and its revision began in 1977.
During these projects about 36 check dams were constructed
in the catchment (Fig. 1), and in 1988 a large dam was built
at the outlet of the main channel for flood control purposes.
The Crcavo catchment has an area of about 2732 ha. It
mainly consists of limestone, marls and quaternary deposits.
Gypsum is present locally. The relief is formed of limestone
and dolomite ridges with occasional outcrops of Keuper
marls in the northwest. The lower parts of the terrain consist
of extended pediment surfaces that have been developed in
the Miocene marls and Quaternary deposits. Ephemeral
418 V.M. Castillo et al. / Catena 70 (2007) 416427
channels and gullies dissect the pediment surfaces, which are
largely used for agriculture, or have been subject to refor-
estation with pines (Pinus halepensis Mill.). Most other
slopes are covered with seminatural pines or recent pine
reforestation (Fig. 2).
The climate is semiarid. Mean annual rainfall is about
279 mm and evapotransporation reaches 848 mm (measured
at the nearby Alfonso XIII reservoir). Both the intensity and
volume of any rainfall are concentrated in a few days,
particularly in autumn, and rainfalls of 100, 200 and even
300 mm in a few hours have been registered (Lopez-
Bermudez and Romero Diaz, 1992).
3. Methodology
An inventory of all check dams in the catchment was
made and the volume of sediment stored behind the dams
was estimated. Then, morphological effects in the upstream
and downstream reach of four selected check dams were
studied by a geomorphological field survey. Finally, the
impact of the structures on the erosiondeposition dynamics
in the channels was assessed by empirical methods.
3.1. Geomorphological field survey and erosion measurements
All dams in the study area were georeferenced by GPS
(Geoexplorer3C Trimble). Raw data were differentially
corrected with the data supplied by a station base in Murcia
City. The basic properties of the dams were assessed (height,
width, structure type, year of construction), and the surface
area of the sediment wedge behind the dam was mapped by
GPS. This sedimentation area can be distinguished in the field
Fig. 2. General view of Crcavo catchment.
by the changes in grain size of the sediments or the appearance
of bedrock, changes in vegetation cover or a change in the
longitudinal gradient. An estimation of the amount of
sediments stored by the dam can be made from the dimension
of the dam and the surface area of the sedimentation area
mapped by GPS. The capacity of the dam for storing sediment
depends on the dimension of the dam, the slope of longitudinal
profile, and the cross-sectional form of the channel.
The big streams channels (ramblas) usually show a U-
form in cross section. For these channels, it is assumed that
the width of the channel at the surface is more or less the
same as the width of the former channel in depth. The
volume of sediment stored behind the dams was computed as
a prismatic channel with rectangular section (Lien, 2003;
May and Gresswell 2003):
V 1=2ws ls h 1=2S h
where, V is the volume of the sediments (m3), ls is the
longitudinal length of the surface area of sedimentation (m),
ws is the average width of sediment wedge (m), defined as
S / ls, with S the surface area of sediment wedge (m2),
mapped with GPS, and h is the height of the sediments
measured from the basis of the dam (m).
Most of the small streams in the upper part of the
catchment, and small tributary streams show a V-form in
cross section. In this case, the mean width of the channel
filled in with sediment w was estimated as:
w wb ws =2
where wb, is the width of the channel at the basis of the
dam, and ws the mean width of the surface area of
 V.M. Castillo et al. / Catena 70 (2007) 416427
sedimentation. The volume of the sediment stored behind
the dams was estimated as:
V 1=2 w ls h
where h is the height of the sediments measured from the
basis of the dam, ls is the longitudinal length of the surface
area of sedimentation, and w is the mean width of the
channel filled in with sediments.
Four check dams (numbers 1, 7, 8 and 33) were selected
for a detailed study of their effects in the ephemeral channels
(Fig. 1). Geomorphological maps of the channel bed
upstream and downstream of the dams were made. On the
maps, the type of sediments, occurrence of bedrock and
alluvial forms are indicated.
Downstream cross sections were established about 30 to
50 m from the dams where the local effects on hydraulic flow
characteristics were negligible. The cross sections measured
upstream of the dams were located within the sedimentation
area. At each cross section the grain size distribution of the
main geomorphological units (channel bed, alluvial bars and
floodplains) were analysed by taking volumetric samples. In
downstream cross sections, in which an armor layer exists, a
stratified sampling scheme was applied (Bunte and Abt,
2001) Surface material was removed to a depth of about
largest surface particle size and analysed separately. Later,
subsurface material was sampled until the largest particle
represents less than 5% of the total sample weight (Reid and
Dunne, 1996).
Furthermore, depth and width at bankfull stage were
measured in the upstream and downstream reaches. The
bankfull stage was determined as the water level that just fills
the available cross section (Dunne and Leopold, 1978).
During the runoff event on 2324th October 2000,
modifications of the channel bed were measured with the use
of 40 cm erosion pins. The pins were placed on alluvial bars
and in the channels, upstream and downstream of dam
number 1 and number 7. In the field, cross sections were
surveyed and flood marks observed in order to estimate
maximum discharges, using Manning's equation to calculate
the flow velocity v (m s 1)


v R2=3 S 1=2 =n
where R represents the hydraulic radius (m), S channel slope
(m m 1), and n the Manning's roughness coefficient.
The discharge Q (m3 s 1) is calculated as
QvA
with A representing the cross-sectional area (m2).
The roughness coefficient (n) was estimated from channel
characteristics, including channel bed material, type of
vegetation and vegetation cover, occurrence of obstacles,
and alteration of the section (Cowan, 1956). To increase the
accuracy and reproduction of data, the channel slope (S) of the
reach was used rather than the estimated water surface slope.
419
All the calculations were made using the WinXSPRO
software package (Hardy et al., 2005), which also supports
the estimation of shear stress for each water stage considered
at specific cross sections.
3.2. Calculations to estimate erosion
The scouring and deposition caused by four selected
discharges at the reaches upstream and downstream of dam
numbers 1, 7, 8 and 33 were calculated. The selected
discharges were representatives of low-frequent flows (1
1.5 m3 s 1), the estimated bankfull discharge (4 m3 s 1), and
over-bank flow (10 m3 s 1). Width, water depth, hydraulic
radius and velocity for these discharges at each upstream and
downstream cross section were calculated by WinXSPRO.
Values between 0.035 and 0.045 were taken for Manning's
n, estimated according to the method described in the
previous section. For each discharge, general scouring
during floods was estimated by using the incipient motion
criterion. The mean critical water velocity for which
sediment movement is initiated was calculated as follows
(Martin Vide et al., 1993):

y 1=6 r
qs Qq
vcr 1:414
0
gD
D q
where vcr is the critical velocity (m s 1), y0 is the water depth
(m), D is the characteristic diameter of the bed material (m),
the density of the water (kg m 3), s is the density of the
sediment (kg m 3), and g is the acceleration due to gravity
(m s 2).
We adopted a clean water approach, which implies a
permanent uniform flow that does not carry sediments.
Therefore, the method will overestimate erosion, since water
usually carries sediment. The formula predicts the potential
maximum erosion during an event, which can be compen-
sated, to a greater or lesser extent, by deposition after the
event. For each discharge and corresponding water depth the
critical velocity was calculated. D84 was taken for the
characteristic diameter, because coarse grained material
prevails on the surface of the channel bed (Martin Vide
et al., 1993). Since A = Q / v we can calculate a new cross-
sectional area A for the critical velocity. The calculated area
corresponds to a new water depth, which can be calculated
by y0 = (A / A)d y0. The difference between the water depths
(y0 y0) is the scour depth.
Local scouring due to turbulence of the water downstream
of the dam was estimated by Jaeger's method modified by
Novak (Martin Vide et al., 1993):
 1=3 !
y0
e 0:55 6H 0:25 0:5
q y0
D90
where e is the scour depth (m), H the vertical distance
between the water level downstream and upstream of the
dam (m), q the unit discharge per width (m2 s 1), y0, is the
420 V.M. Castillo et al. / Catena 70 (2007) 416427

Table 1 channel bed in the reach between two check dams. In the
Grain size diameters used for erosion calculations reaches upstream of the check dams sedimentation takes
Check Cross section Location of D50 D84 D90 place. Due to a decrease in the longitudinal gradient, fine
dam no. sample (m) (m) (m) sediments predominate in the sedimentation area upstream of
D1 Downstream (DS) A (alluvial bar) 0.004 0.015 0.025 the dams and due to the higher water content in the sedi-
Upstream (US) A (flood plain) 0.007 0.042 ments, vegetation develops very well. In general, the
B (main channel 0.005 0.022
morphological pattern upstream shows that more than one
bed)
C (alluvial bar) 0.004 0.018 inner channel often exists and that these channels are
D7 Downstream (DS) A (channel bank) 0.004 0.017 0.025 shallow. Measurements of the channel width and depth at
Upstream (US) A (main channel 0.002 0.006 bankfull stage show that channels upstream of the dams are
bed) very shallow relative to their widths. (Table 3). This is also
B (alluvial bar) 0.0005 0.008
indicated by the alluvial bars, which are lower than the bars
D8 Downstream (DS) A (alluvial bar) 0.0004 0.020
B (main channel 0.002 0.005 0.0071 downstream (Fig. 3). The decrease in channel capacity
bed) upstream, is due to channel bed aggradation and may lead to
Upstream (US) A (alluvial bar) 0.007 0.023 lateral erosion or overflows during rainstorm events.
B (main channel 0.017 0.055
bed)
D33 Downstream (DS) A (channel bank) 0.003 0.021 0.030 Table 2
Upstream (US) A (main channel 0.006 0.017 Characteristics of the check dams
bed) Check Year of Width Height Sediment Capacity Sediment
dam construction (m) (m) height (m3 100) b stored
no. upstream upstream
water level (m), and D90 is the characteristic bed material dam (m) a dam
diameter, expressed in mm. (m3 100)
The grain size diameters used for the erosion calculations 1 1970 17.0 2.7 2.7 28.80 28.80
are shown in Table 1. 2 1970 18.0 3.7 3.7 27.10 27.10
3 1970 11.6 3.5 3.5 n.a. n.a
4 1970 13.4 3.2 3.2 n.a n.a
4. Results and discussion 5 1970 13.8 3.8 3.8 n.a n.a
6 1980 15.8 3.6 3.6 n.a n.a
4.1. Geomorphological survey of the stream channels and 7 1970 34.8 4.7 4.7 273.90 273.90
erosion measurements 8 1970 20.0 4.5 4.5 115.70 115.70
9 1970 14.0 1.5 1.5 20.20 20.20
10 1980 16.0 4.0 2.6 133.00 54.20
Of the 36 dams located in the study area (Fig. 1), 29 dams 11 1970 15.0 1.3 D D D
were completely filled up with sediments. Two dams had 12 1970 8.2 2.0 2.0 21.30 21.30
been destroyed and only five were not completely filled with 13 1980 13.6 4.0 3.0 13.70 10.30
sediments (Table 2). In general, the dams built in the 14 1970 23.8 5.0 2.9 67.70 33.10
15 1980 16.7 3.0 3.0 n.a. n.a.
narrowest stream channels (barrancos) in the upper part of
16 1980 14.5 3.5 2.3 68.30 44.90
the catchment or in small tributary streams store less 17 1980 15.5 2.8 2.8 22.00 22.00
sediments than the dams in the rambla due to their smaller 18 1980 18.0 3.2 3.2 14.90 14.90
dimensions, a steeper longitudinal gradient and the V-form 20 1970 23.0 2.6 D D D
of the cross section of the channels. Castillo et al. (2002) 21 1970 10.0 3.0 3.0 29.70 29.70
22 1970 14.0 2.9 2.9 19.30 19.30
showed that, in the Crcavo basin, the probability of a check
23 1970 15.5 3.0 3.0 42.90 42.90
dam being filled up increases as the drainage area increases 24 1980 6.0 3.0 3.0 6.70 6.70
and the agricultural land in the catchment decreases. This 25 1980 12.5 3.2 3.2 16.60 16.60
seemingly contradictory relationship between sedimentation 26 1980 11.0 3.2 3.2 16.90 16.90
upstream of the dam and the extent of agricultural land is 27 1980 8.0 2.6 2.6 8.90 8.90
28 1980 7.5 2.6 2.6 22.60 22.60
probably due to the fact that agricultural land is merely found
28a 1980 9.0 3.2 3.2 27.00 27.00
in flatter areas of the catchment. Furthermore, the frequent 29 1980 10.0 2.6 1.8 5.10 3.40
ploughing of these areas breaks up the soil crust, increasing 30 1970 14.0 1.9 1.9 20.30 20.30
infiltration and reducing direct runoff and erosion. The soil 31 1970 9.0 2.0 2.0 13.70 13.70
conservation measures (small terraces and earth embank- 32 1980 11.0 2.9 2.9 12.00 12.00
33 1980 10.0 3.2 3.2 14.10 14.10
ments, and stone walls) adopted in the study area may also
34 1970 8.0 1.9 1.9 1.50 1.50
have decreased soil losses in the agricultural area in contrast 35 1980 9.0 2.9 2.9 7.80 7.80
to more degraded slopes. 36 1980 6.5 1.9 1.9 3.61 3.61
Processes and forms in the reaches upstream of the check Capacity and amount of sediments stored in the surveyed dams.
dams were different from those in the downstream reaches. a
D indicates a destroyed check dam.
b
The map in Fig. 3 shows the typical geomorphology of the n.a. indicates that the capacity of the check dams is not available.
 V.M. Castillo et al. / Catena 70 (2007) 416427 421

Fig. 3. Geomorphological map of the stream channel in the reach between check dam numbers 7 and 8, with indication of cross sections. The height of the bars
with respect to the channel bed is indicated. The graphs (a, b, c,) are labelled in flow direction.

Downstream of the dams, only one channel exists, in which
bedrock is exposed frequently. In general these channels have
a lower width/depth ratio. Fine sediments and vegetation are
only present on the flood plains or bars. The presence of
bedrock and the low width/depth ratio indicate erosion
downstream of the dam due to the high erosive power of
water without sediment, which has been trapped by the dam
and by the erosive power of the water falling from the top of the
dam. The erosion downstream seems to continue downstream
until the next dam forces sedimentation to occur.

Table 3
Depth, width and width/depth ratio at bankfull stage
Check dam 1 Check dam 7 Check dam 8 Check dam 33
Distance Depth Width Width/ Distance Depth Width Width/ Distance Depth Width Width/ Distance Depth Width Width/
from the (m) (m) depth from the (m) (m) depth from the (m) (m) depth from the (m) (m) depth
dam (m) ratio dam (m) ratio dam (m) ratio dam (m) ratio
Upstream 200 1.1 14.0 12.7 200 0.6 22.4 35.6 200 0.9 15.5 17.6 200 0.9 3.2 3.4
100 1.4 14.4 10.7 100 0.9 29.9 34.0 100 1.0 16.6 17.1 100 0.9 5.5 6.5
50 1.1 17.7 15.7 50 0.9 26.6 30.9 50 0.8 19.9 25.4 50 0.9 6.8 7.6
Downstream 50 0.8 10.4 13.0 50 1.3 14.4 11.3 50 1.0 15.9 16.0 50 1.1 9.1 8.7
100 1.1 9.5 8.6 100 1.5 13.9 9.4 100 1.1 15.1 13.6 100 1.1 4.0 3.6
150 1.0 10.5 10.2 150 1.3 15.3 11.6 150 1.3 17.8 14.2 150 1.0 6.7 6.6
200 1.3 12.8 10.2 200 1.1 12.1 11.3 200 1.6 10.7 6.9 200 0.8 3.3 4.2
422 V.M. Castillo et al. / Catena 70 (2007) 416427

Fig. 4. Cross sections established in the upstream and downstream reach of check dam numbers 1, 7, 8 and 33 (DS = downstream, US = upstream), with
indication of bankfull stage.
 V.M. Castillo et al. / Catena 70 (2007) 416427
Fig. 5. View of check dam number 7 after flood event on October 2000.
At the foot of the dam number 7 (D7) a pool had
developed due to local erosion, which had trapped fine
sediment and a lobe of deposition occurred downstream of
dam number 8 (D8) due to channel meandering. Most cross
sections (Fig. 4) reflected the incision of the channel
downstream (DS) compared to the upstream (US) sections.
The processes of erosion and sedimentation in the stream
channel were measured in the field during a storm event that
affected the study area in October 2000. Precipitation of
50 mm in 19 h with a maximum intensity of 8 mm h 1 was
registered in a nearby meteorological station being operated
by the Agricultural Information System. Peak discharge and
average values of hydraulic variables were calculated for two
representative cross sections upstream of dam numbers 1
(D1) and 7 (D7). D1, (17 m wide and 2.7 m high) is located
in the lower part of the Barranco del Lobo, and D7, (35 m
wide and 4.7 m high) in the lower part of the Rambla del
Crcavo (Fig. 5).
The estimated peak discharges were 1.6 and 2.5 m3 s 1 at
D1 and D7, respectively. All the other hydraulic variables
estimated (mean velocity, shear stress and unit stream power)
were higher at D1 than at D7 (Table 4), because of the steeper
slope of the channel reach at Barranco del Lobo (D1). These
hydraulic variables indicate a medium energy flow. The values
of shear stress clearly exceed of the entrainment threshold
Table 4
Hydraulic variables for monitored check dams resulting from the floods on October 2000
Location Check dam no. Width Cross section area Hydraulic radius
(m) (m2) (m)
Bco. El Lobo 1 6.5 1.23 0.19
El Crcavo 7 20.5 3.46 0.17
423
value (77.5 N m 2) found by Reid et al. (1998) in a desert
gravel-bed river in the Negev (Israel) and by Batalla and
Martn-Vide (2001) in a sandy gravel-bed river in northeastern
Spain. Thus, despite the peak flow values being low, the flood
event could be classed as Class 3, according to Conesa's
(1995) classification, i.e. discharges sufficient to cover the
whole main channel bed and produce bedload movement.
The measurements taken at the erosion pins underlined
the distribution and magnitude of erosion and deposition.
Net deposition prevailed in the cross sections located in the
upstream sediment wedge close to both dams (D7: 0.01
0.1 m; D1: 0.010.05 m) (Figs. 6 and 7), while at the edges
of the sediment wedge and far from the dam, net scouring
was observed. The net scouring observed in the central
alluvial bars and the thalweg of the Barranco del Lobo (D1,
Fig. 7) can be attributed to the partial failure of the check
dam. This failure caused a reactivation of the erosion within
the sediment wedge, carving an inner channel with active
headward erosion. The net scour depth (0.050.08 m)
measured upstream of the dam for a unit stream discharge of
0.25 m2 s 1 matched well that predicted by the relationships
proposed by Leopold et al. (1966).
In contrast to the upstream effects, changes in down-
stream fluvial morphology during the flood event exhibited
no clear pattern. Both slight (less than 0.04 m) erosion and
Bed slope Flow velocity Discharge Shear stress Unit stream power
(m m 1) (m s 1) (m3s 1) (Nm 2) (W m 2)
0.0262 1.3 1.6 47.5 61.7
0.0087 0.7 2.5 14.3 10.0
424 V.M. Castillo et al. / Catena 70 (2007) 416427
Fig. 6. Erosion (white symbols) and sedimentation (black symbols)
measured with the use of pins in the check dam number 7 during the
rainstorm event of October 2000. Pins located in the channel are represented
as (). Pins located on bars are represented as (). The dimension of the
dam is not at scale. The grey area corresponds to the sediment wedge.
sedimentation took place on the alluvial bars downstream of
D7 (Fig. 6), and, at the same time, local scouring of as much
as 0.13 m was measured in the channel. No erosion could be
measured downstream of D1, because the channels mainly
consisted of bedrock and only one case of deposition was
observed on the alluvial-bed channel (Fig. 7).
4.2. Effects on erosiondeposition dynamic
The effects of check dams on the erosiondeposition
dynamic were analysed by estimating either general or local
scouring related to the passing of selected flood discharges
downstream and upstream of four check dams. Theoretical
scouring is defined as the potential erosion predicted, if the
channel bed only consists of alluvial sediments. In reality, the
vertical erosion will only continue at a very slow rate once
bedrock is reached. For D7 and D33, the channel bed
consists mainly of bedrock and so the calculations were
made using the sediment samples taken in the channel banks.
For D1, the material of the alluvial bar was used. For D8, the
real characteristics of the sediments in the channel bed were
used, assuming that they covered the whole channel,
including the parts where bedrock appeared.
Real scouring is the erosion that is possible, assuming that
the appearance of bedrock will stop vertical erosion. The
following assumptions, based on field observations and
cross section, were made:
1. The alluvial bar in the downstream (DS) cross section of
D1 has a maximal thickness of 0.5 m and bedrock will
appear at this depth.
2. The alluvial bar in the upstream (US) cross section of D7
has a thickness of 0.5 m and below this, the coarse gravel
of the channel bed will appear.
3. The alluvial bar in the downstream (DS) cross section of
D8 has a maximal thickness of 2 m and at that depth
bedrock will appear.
The calculations were made for the different zones of the
fluvial system (main channel, alluvial bars, and flood plains).
The average bed scouring for the whole cross section is
shown in Table 5.
Fig. 7. Erosion (white symbols) and sedimentation (black symbols)
measured with the use of pins in the check dam number 1 during the
rainstorm event of October 2000. Pins located in the channel are represented
as (). Pins located on bars are represented as (). The dimension of the
dam is not at scale. The grey area corresponds to the sediment wedge.
 V.M. Castillo et al. / Catena 70 (2007) 416427 425

Table 5
Predicted scouring caused by specific discharges at downstream (DS) and upstream (US) of selected check dams
Check dam no. Cross section Distance from Bed slope H (m) Q (m3 s 1) Water stage Average scouring in the wet
the dam (m) (m m 1) (m) perimeter (m)
Local General during floods
1 DS 35 0.031 2.5 1.0 0.25 0.23 0.00
1.5 0.3 0.31 0.01
4.0 0.45 0.61 0.13
10.0 0.75 1.00 0.41
US 100 0.026 1.0 0.5 0.00
1.5 0.55 0.00
4.0 0.7 0.00
10.0 1 0.11
7 DS 40 0.020 4.7 1.0 0.25 0.27 0.00
1.5 0.3 0.31 0.00
4.0 0.43 0.49 0.04
10.0 0.62 0.88 0.28
US 100 0.008 1.0 0.2 0.00
1.5 0.28 0.00
4.0 0.44 0.00
10.0 0.62 0.21
8 DS 50 0.023 4.5 1.0 0.36 0.37 0.02
1.5 0.39 0.47 0.06
4.0 0.51 0.87 0.29
10.0 0.69 1.60 0.77
US 90 0.012 1.0 0.38 0.00
1.5 0.42 0.00
4.0 0.56 0.00
10.0 0.8 0.00
33 DS 50 0.045 3.2 1.0 0.42 0.51 0.15
1.5 0.5 0.67 0.25
4.0 0.81 1.28 0.68
10.0 1.26 1.65 1.27
US 120 0.011 1.0 0.31 0.00
1.5 0.37 0.00
4.0 0.6 0.08
10.0 0.98 0.43
H is the vertical distance between the water level downstream and upstream of the dam.

For low-frequent discharges (11.5 m3 s 1) no general
bed scouring during floods is predicted. Only in downstream
reaches, with steep or fine-grained bed material channel is
net bed scouring predicted: 0.25 m downstream of D33, and
0.06 m upstream of D8, respectively. This range of
discharges corresponds more or less with the moderate
event of October 2000. The predictions fit fairly well the
erosion/deposition values measured by the pins. Thus, most
of the pins (more than 90%) located upstream of D7 showed
sediment deposition with no net erosion. Upstream of D1, no
net erosion was measured at 62.5% of the erosion pins, while
the rest of the pins showed a net bed scour depth of less than
0.08 m. The observed scouring can be explained by the
presence of secondary flow cells inside the channel
excavated in the sediment wedge as a consequence of the
partial failure of the check dam.
At a discharge of 4 m3 s 1, a value that may be considered
as representative of the bankfull discharge, general scouring
ranging from 0.04 to 0.68 m was predicted for all down-
stream cross sections. In contrast, for the same discharge no
erosion is predicted for upstream reaches except in the case
of D33, which may reflect the stabilisation effect of the
check dams on the upstream channels.
At discharge rates that overflow the bankfull cross-section
(10 m3 s 1) erosion was generalized both upstream and
downstream of the check dams. For this magnitude of flood,
the sediment wedges filling the valley upstream of the dams
are sensitive to transient erosion phenomena capable of
removing large amounts of sediments which drastically
change the fluvial morphology. Maximum scouring depths
(0.430.21 m) were predicted in the cross sections upstream
of D33, which are characterized by being narrow, and
upstream of D7, with its fine-grained sediments bed.
Local scouring was predicted for the whole range of the
discharges analysed (Table 5). Regardless of the discharge
considered, the greatest scour depths were estimated
downstream of the dams with the greatest drop height (D8)
or highest unit discharge (i.e. flow discharge per unit width)
(D33). The smaller scour depths predicted downstream of the
dam with the highest drop height, (D7) at bankfull and over-
bank discharges may be due to the reduction of unit dis-
charge as a consequence of the widening of the channel.
426 V.M. Castillo et al. / Catena 70 (2007) 416427
Table 6
Estimated erosion and sedimentation in the upstream and downstream
reaches of the check dams
Check Erosion Sedimentation Net sedimentation
dam downstream upstream
no.
m3 ta m3 ta ta
1 1400 1540 2880 3168 1628
7 8000 8800 27,390 30,129 21,329
8 2900 3190 11,670 12,837 9647
33 570 627 1410 1551 924
a
Mass (t) = volume (m3); average bulk density of sediment. (1.1 t m 3).
The relationships found between predicted local scouring
and geometrical and hydraulic variables are in accordance
with the field observations reported by Lenzi and Comiti
(2003), suggesting a complex influence of drop height, flow
depth and dam spacing on scouring dynamics of steep
channels with a series of check dams. Furthermore, the
results highlight the need for implementing dissipation
energy structures downstream of check dams to minimize the
risk of failure because of undermining.
We can make a rough estimation of the total amount of
erosion that has taken place after dam construction, if we
assume that channels continue to be trenched downstream of
a given dam until the sediment wedge of the next dam is
reached. However, this calculation will overestimate the
amount of erosion because downstream erosion will decrease
as a consequence of a decrease in gradient. The amount of
erosion estimated is shown in Table 6. The total amount of
sedimentation in the upstream reach is much higher than the
amount of vertical erosion predicted for the downstream
reach.
5. Conclusions
The study area is an active erosional environment because
of the presence of easily erodible material including marls
and quaternary deposits and sparse vegetation cover. The
check dams control erosion and deposition within the
channels. In general the effects of dam construction are to
encourage sedimentation in the upstream reach and erosion
in the downstream reach. During the few torrential rain-
storms big amounts of sediment become available and so
most dams have been completely filled up. The dams store
sediments which otherwise would have been transported to
the Crcavo reservoir. However, it should be mentioned that
the amount of sediment that is stored behind the check dams
seem to be rather small in comparison with the vast amounts
of material available for erosion in channel banks and slopes.
Within the catchment, there are zones highly affected by
bank gullies that mobilize a lot of sediment, which is later
available for fluvial transport. Bank gully erosion mobilizing
high amounts of sediment is recognized as one of the major
processes involved in land degradation (Oostwoud Wijdenes
et al., 2000). The sedimentation upstream of check dams
causes a decrease in the longitudinal gradient and reduces
water flow velocity. Thus, upstream of the check dams the
channel is stabilised. However also a decrease in channel
capacity was measured, which may favour lateral erosion or
overflows.
The morphological changes in the channels indicate that
erosion increases downstream following dam construction.
In general, dam construction provokes changes in discharge
and sediment transport downstream, which is reflected in a
different morphology of the channel. As expected from
earlier publications (e.g. Garca Ruiz and Puigdefabregas,
1984; Martinez Castroviejo et al., 1990; Hooke and Mant,
2000), an increase in channel erosion can be observed
downstream of a check dam. The morphological changes
that took place during the study can be seen from (1) the
measurements carried out during the rainstorm event in
October 2000, which show that sedimentation prevails
upstream and erosion downstream, (2) the occurrence of
bedrock downstream, (3) lower width/depth ratios of the
channel downstream, and (4) cross-section geometry, which
shows more incision downstream. These changes can be
explained by (1) the trapping of sediment by the dam, which
causes an increase in the erosion capacity, and (2) an increase
in the turbulence of the water passing the dam. Since most
channel beds consist of bedrock, real erosion only takes
place at a low rate. Our calculations show that the erosion
that has occurred downstream is less than the amount of
sedimentation upstream.
From this, we can conclude that besides their positive
upstream effects (storing sediments and decreasing vertical
erosion), check dams may also have some significant
negative effects. They cause erosion downstream, although
less than the amount of sediments stored in the upstream
reaches, while any failure of check dams from lack of
maintenance may reintroduce sediments into the channel.
These effects have major implications for ephemeral channel
management including the control of flood propagation and
sediment transfer and delivery into the main rivers and
reservoirs downstream.
Optimal placement and maintenance of the sediment
control structures is a crucial factor for improving their
effectiveness and to reduce sediment mobilization through
the drainage system.
Acknowledgements
The research was carried out as part of the projects
Seguimiento y Evaluacin de los efectos sobre el Medio
Natural de la Sequa y los Procesos Erosivos. 2: Seguimiento
de las actuaciones de lucha contra la desertificacin, funded
by Comunidad Autnoma de la Regin de Murcia e
INTERREG IIC Programme and Respuesta Hidrolgica y
Geomorfolgica de los Sistemas Fluvio-Torrenciales afecta-
dos por proyectos de Restauracin Hidrolgico-Forestal en
cuencas semiridas del Sureste Peninsular funded by
Fundacin Instituto Euromediterrneo de Hidrotecnia. The
authors thank Peter Sandercock for providing the WinXSPRO
 V.M. Castillo et al. / Catena 70 (2007) 416427
software and Paco Castillo and Raul Gallego for their help in
the field. S. A. Brandt and I. Reid are thanked for their com-
ments that allowed a significant strengthening of the paper.
References
Batalla, R.J., Martn-Vide, J.P., 2001. Thresholds of particle entrainment in a
poorly sorted sandy gravel-bed river. Catena 44, 223243.
Brandt, S.A., 2000. Classification of geomorphological effects downstream
of dams. Catena 40, 375401.
Bunte, K., Abt, S.R., 2001. Sampling surface and subsurface particle-size
distributions in wadable gravel-and cobble-bed streams for analyses in
sediment transport, hydraulics, hydraulics and streambed monitoring.
Gen. Tech. Rep.RMRS-GTR-74. U.S. Department of Agriculture, Forest
Service, Rocky Mountains Research Station, Fort Collins, CO.
Castillo, V., Barber, G.G., Mosch, W., Navarro-Cano, J.A., Conesa, C.,
Lpez-Bermdez, F., 2002. Seguimiento y evaluacin de los trabajos de
restauracin hidrolgico-forestal. In: Lpez Bermdez (Ed.) Segui-
miento y evaluacin de los efectos sobre el medio natural de la sequa y
los procesos erosivos en la Regin de Murcia. Consejera de Agricultura,
Agua y Medio Ambiente de la Regin de Murcia. Murcia. pp. 166233.
Conesa Garcia, C., 1995. Torrential flow frequency and morphological
adjustments of ephemeral channels in southeast Spain. In: Hickin,
E.J. (Ed.), River Geomorphology. John Wiley & Sons Ltd, Chichester,
pp. 169192.
Cowan, W.L., 1956. Estimating hydraulic roughness coefficients. Agricul-
tural Engineering 37, 473475.
Dunne, T., Leopold, L.B., 1978. Water in Environmental Planning. W.H
Freeman and Co., San Francisco, USA.
Garca Ruiz, J.M., Puigdefabregas, J., 1984. Efectos de la construccion de
pequeas presas en causes anastomosados del pirineo central. Cuadernos
de Investigacin Geogrfica Tomo XI, fascculos 1 y 2, Logroo.
Hardy, T., Panja, P., Dean, M., 2005. WinXSPRO, a channel cross-section
analyser. User's Manual Version 3.0. Gen.Tech.Rep. RMRS-GTR-147.
U.S. Department of Agriculture. Forest Service. Rocky Mountain
Research Station, Fort Collins, CO.
Hooke, J.M., Mant, J.M., 2000. Geomorphological impacts of a flood event
on ephemeral channels in SE Spain. Geomorphology 34, 163180.
Knighton, D., Nanson, G., 1997. Distinctiveness, diversity and uniqueness
in arid river systems, In: Thomas, D.S.G. (Ed.), Arid Zone Geomor-
phology: Process, Form and Change in Dryland, 2nd ed. John Wiley &
Sons Ltd., Chichester, pp. 169192.
427
Lenzi, M.A., Comiti, F., 2003. Local scouring and morphological
adjustments in steep channels with check-dams sequences. Geomor-
phology 55, 97109.
Leopold, L.B., Emmett, W.W., Myrick, R.M., 1966. Channel and Hillslope
Processes in a Semiarid Area of New Mexico. USGS Professional Paper.
U.S. Government Printing Office, Washington, p. 352G.
Lien, H.P., 2003. Design of slit dams for controlling stony debris flow.
International Journal of Sediment Research 18, 7487.
Lopez-Bermudez, F., Romero Diaz, M.A., 1992. Gnesis y consecuencias
erosivas de las lluvias de alta intensidad en la regin mediterrnea.
Cuadernos de Investigacin Geogrfica 1819, 728.
Marston, R.A., Dolan, L.S., 1999. Effectiveness of sediment control
structures relative to spatial pattern of upland soil loss in an arid
watershed, Wyoming. Geomorphology 31, 313323.
Martin Vide, J.P., Rosell Estelrich, R., Nierola Chifoni, D., Gmez
Navarro, L., 1993. La avenida del 9 de septiembre de 1992 en la riera de
las Arenas. Dep. Ingeniera Hidrulica, Martima y Ambiental,
Universidad Politcnica de Catalua, Barcelona.
Martinez Castroviejo, R., Inbar, M., Gomez Villar, A., Garcia Ruiz, J.M.,
1990. Cambio en el cauce aguas debajo de una presa de retencin de
sedimentos. 1a Reunin Nacional de Geomorfologa. Sociedad Espaola
de Geomorfologa, Teruel, pp. 457468.
May, C., Gresswell, R., 2003. Processes and rates of sediment and wood
accumulation in headwater streams of the Oregon coast range, USA.
Earth Surface Processes Landforms 28, 409424.
Nyssen, J., Veyret-Picot, M., Poesen, J., Moeyersons, J., Haile, M., Deckers,
J., Govers, G., 2004. The effectiveness of loose rock check dams for
gully control in Tigray, northern Ethiopia. Soil Use and Management 20,
110.
Oostwoud Wijdenes, D.J., Poesen, J., Vandekerckhove, L., Ghesquiere, M.,
2000. Spatial distribution of gully head activity and sediment supply
along an ephemeral channel in a Mediterranean environment. Catena 39,
147167.
Poesen, J.W., Hooke, J.M., 1997. Erosion, flooding and channel management
in Mediterranean environments of southern Europe. Progress in Physical
Geography 21 (2), 157199.
Reid, L.M., Dunne, T., 1996. Rapid Evaluation of Sediment Budgets Catena
Verl., Reiskirchen.
Reid, I., Laronne, J.B., Powell, D.M., 1998. Flash-flood and bedload
dynamics of desert gravel-bed streams. Hydrological Processes 12,
543557.