A grassed waterway and earthen dams to control muddy floods from a
cultivated catchment of the Belgian loess belt
Olivier Evrard a,d,, Karel Vandaele b
, Bas van Wesemael a
, Charles L. Bielders c
a
Département de Géographie, Université catholique de Louvain, Place Louis Pasteur, 3, B-1348 Louvain-la-Neuve, Belgium
b
Watering van Sint-Truiden, Interbestuurlijke samenwerking Land en Water, Minderbroedersstraat, 16, B-3800 Sint-Truiden, Belgium
c
Département des Sciences du Milieu et de l'Aménagement du Territoire, Université catholique de Louvain, Croix du Sud, 2,
Bte 2, B-1348 Louvain-la-Neuve, Belgium
d
Fonds pour la formation  la Recherche dans l'Industrie et l'Agriculture (F.R.I.A.), Belgium
Received 10 October 2007; received in revised form 10 January 2008; accepted 11 January 2008
Available online 31 January 2008
Abstract
Muddy floods, i.e. runoff from cultivated areas carrying large quantities of soil, are frequent and widespread in the European loess belt. They
are mainly generated in dry zero-order valleys and are nowadays considered as the most likely process transferring material eroded from cultivated
hillslopes during the Holocene to the flood plain. The huge costs of muddy flood damages justify the urgent installation of control measures. In the
framework of the `Soil Erosion Decree' of the Belgian Flemish region, a 12 ha-grassed waterway and three earthen dams have been installed
between 2002­2004 in the thalweg of a 300-ha cultivated dry valley in the Belgian loess belt. The measures served their purpose by preventing
any muddy flood in the downstream village, despite the occurrence of several extreme rainfall events (with a maximum return period of
150 years). The catchment has been intensively monitored from 2005­2007 and 39 runoff events were recorded in that period. Peak discharge (per
ha) was reduced by 69% between the upstream and the downstream extremities of the grassed waterway (GWW). Furthermore, runoff was
buffered for 5­12 h behind the dams, and the lag time at the outlet of the catchment was thereby increased by 75%. Reinfiltration was also
observed within the waterway, runoff coefficients decreasing by a mean of 50% between both extremities of the GWW. Sediment discharge was
also reduced by 93% between the GWW's inflow and the outlet. Before the installation of the control measures, specific sediment yield (SSY) of
the catchment reached 3.5 t ha- 1
yr- 1
and an ephemeral gully was observed nearly each year in the catchment. Since the control measures have
been installed, no (ephemeral) gully has developed and the SSY of the catchment dropped to a mean of 0.5 t ha- 1
yr-1
. Hence, sediment transfer
from the cultivated dry valley to the alluvial plain should dramatically decrease. Total cost of the control measures that are built for a 20 yearperiod
is very low (126  ha- 1
) compared to the mean damage cost associated with muddy floods in the study area (54  ha- 1
yr- 1
). Similar
measures should therefore be installed to protect other flooded villages of the Belgian loess belt and comparable environments.
 2008 Elsevier B.V. All rights reserved.
Keywords: Muddy floods; Grassed waterway; Earthen dams; Runoff control; Sediment delivery; Cost-efficiency
1. Introduction
Muddy floods consist of water flowing from agricultural
fields carrying large quantities of soil as suspended sediment or
bedload (Boardman et al., 2006). They are therefore considered
as a fluvial process rather than a mass movement one. Even
though they are frequent and widespread in the European loess
belt, they are mainly reported from central Belgium (Verstraeten
and Poesen, 1999; Evrard et al., 2007a), northern France
(Souchre et al., 2003) and southern England (Boardman et al.,
2003). Muddy floods cause numerous off-site impacts, such as
flooding of property, sedimentation and eutrophication in
watercourses.
About 90% of muddy floods observed in the Belgian loess
belt are generated on cultivated hillslopes (10­30 ha) and in dry
zero-order valleys (30­300 ha; Evrard et al., 2007a). Numerous
Available online at www.sciencedirect.com
Geomorphology 100 (2008) 419­428
www.elsevier.com/locate/geomorph
 Corresponding author. Département de Géographie, Université catholique
de Louvain, Place Louis Pasteur, 3, B-1348 Louvain-la-Neuve, Belgium.
Tel.: +32 10 47 29 91; fax: +32 10 47 28 77.
E-mail address: olivier.evrard@uclouvain.be (O. Evrard).
0169-555X/$ - see front matter  2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.geomorph.2008.01.010
studies carried out in cultivated catchments of the European
loess belt showed that most sediments produced during the
Holocene have been stored in the dry valley bottom near the
catchment outlet and have not been delivered to downstream
rivers (e.g. Bork et al., 1998; Lang et al., 2003; Rommens et al.,
2005; de Moor and Verstraeten, in press). Rommens et al.
(2006) also estimated the Holocene alluvial sediment storage in
a small (52 km2
) river catchment of the Belgian loess belt. They
showed that sediment supply towards the alluvial plain has
increased dramatically since Medieval times compared to the
rest of the Holocene period and occurred at a mean rate of 1.3 t
ha-1
yr-1
. Since 50% of sediment eroded from hillslopes was
stored in colluvial deposits, mainly located in dry zero-order
valley bottoms, muddy floods caused by severe erosion on agricultural
land are the mostly likely process transporting sediments
from the dry valleys to the alluvial plains. During heavy rainfall
in late spring and summer, ephemeral gullies form in these dry
valleys. These shallow (0.1 m) but wide (3 m) gullies act as
an important conveyor of sediment and may aggravate the offsite
damage produced by muddy floods (Nachtergaele and
Poesen, 2002; Verstraeten et al., 2006).
The huge costs associated with this damage, which appears to
have occurred more frequently during the last decade, justifies the
urgent installation of mitigation measures (Evrard et al., 2007a).
Two types of measures can be carried out to control muddy floods.
On the one hand, alternative farming practices implemented at the
field scale, such as sowing of cover crops during the intercropping
period, reduced tillage or double sowing in zones of concentrated
flow, limit runoff generation and erosion production (Gyssels
et al., 2002; Leys et al., 2007). However, the implementation
of these practices directly depends on the farmer's willingness.
Except for sowing of cover crops (e.g. in Belgium; Bielders et al.,
2003), the adoption of such practices remains rather limited in
Europe (Holland, 2004). It will probably still take several years or
even decades before reduced tillage and double sowing are
applied generally. On the other hand, `curative' measures aim to
reinfiltrate or buffer runoff once it is formed, as well as to trap
sediments and pollutants. Typically, grass buffer strips, grassed
waterways (GWW) and detention ponds (retaining runoff for a
certain time behind a small dam) serve this purpose (Fiener and
Auerswald, 2005). Such curative measures are most effective
when they are implemented in the framework of integrated
catchment management. Hence, a local water board should be
responsible for deciding in consultation with farmers where to
install these measures within the catchment and for ensuring their
maintenance.
From 2001 onwards, municipalities in the Belgian Flemish
region are eligible for subsidies to draw up an erosion mitigation
scheme (Verstraeten et al., 2003). Several small-scale measures
such as dams and GWW are being installed in the field but there
is a need to evaluate their efficiency before generalising their
installation in problem areas. Furthermore, since muddy floods
are generated on large surfaces (10­300 ha; Evrard et al.,
2007a), the effect of control measures should be investigated at
similar scales. However, previous research has focused on the
effect of grass buffer strips and has mostly been carried out on
experimental plots (typically 500 m2
, see e.g. Van Dijk et al.,
1996; Patty et al., 1997; Le Bissonnais et al., 2004). With
respect to the effect of GWW in the European context, it has
only been assessed at the micro-catchment scale (max. 8 ha;
Fiener and Auerswald, 2005; Fiener et al., 2005). Large quantities
of concentrated runoff leading to muddy floods cannot be
generated on such small surfaces and a specific study is hence
needed at the scale of the larger catchments, which are the
source areas of muddy floods.
This paper evaluates the effectiveness of a GWWand earthen
dams installed in a cultivated 300 ha-catchment in the Belgian
loess belt in mitigating muddy floods in the downstream village.
The cost-efficiency of the control measures is also discussed.
2. Materials and methods
2.1. General context
The Belgian loess belt (9000 km2
) is a plateau with a mean
altitude of 115 m gently sloping to the North (Fig. 1a). Annual
mean temperature varies between 9­10 °C, while annual
precipitation ranges from 700­900 mm (Hufty, 2001). Soils are
mainly loess-derived Haplic Luvisols (World Reference Base,
1998). Arable land covers 65% of the total surface (Statistics
Belgium, 2006). During the last three decades, the area covered
by summer crops (sugar beet -- Beta vulgaris L., maize -- Zea
Mays L., potatoes -- Solanum tuberosum L. and chicory --
Cichorium intybus L.) increased at the expense of winter
cereals (Evrard et al., 2007a). The summer crops provide little
cover to the soil during the thunderstorms that occur in late
spring or early summer, which leads to the formation of a soil
surface crust with a very low infiltration rate. High quantities of
runoff are then generated on these crusted soils during intense
precipitation (Evrard et al., in press).
The region of Sint-Truiden has been repeatedly affected by
muddy floods, and the local water agency (Melsterbeek Water
Board) decided to tackle the problem (Fig. 1a). In the framework
of the `Erosion Decree' adopted by the Flemish government in
2001, they drew up an erosion mitigation scheme at the catchment
scale (200 km2
). Between 2002­2005, 120 grass strips and GWW
have been installed, covering a surface of 25 ha (0.13% of total
area). Furthermore, 35 earthen dams have been built.
2.2. Study area
Velm has the local reputation of a `devastated village', since
it was flooded several times during the last two decades. Runoff
loaded with sediments is generated in cultivated dry zero-order
valleys covering a total area of 930 ha that drain to the village
(Fig. 1b).
A 300 ha-catchment draining into Velm, locally known as
`Heulen Gracht', was selected for detailed monitoring (Fig. 2).
Cropland covers 79% of the catchment surface. Orchards (17%)
and roads (3%) are the other main types of land use. A typical
topsoil sample in this catchment contains 80% silt, 10% clay
and 10% sand and the mean slope reaches 1.3%.
An earthen dam was built close to the catchment outlet in April
2002 to prevent muddy floods (dam # 3 in Fig. 2). A GWW was
420 O. Evrard et al. / Geomorphology 100 (2008) 419­428
also sown in the thalweg in 2003, covering 12 ha (4% of total
catchment surface; Fig. 2). Grass species consist of a mix of Lolium
multiflorum Lam., Lolium perenne L., Festuca rubra L. subsp.
Rubra and Dactylis glomerata L. Two additional dams were built
across the GWW in August 2004 (dams # 1 and 2 in Fig. 2).
2.3. Impact of control measures on runoff
Rainfall is measured at 0.1 mm resolution using two tipping
bucket rain gauges located at the catchment outlet and just
upstream of the GWW (Fig. 2). The catchment was equipped
with a discharge measurement station in April 2006. It consists
of a San Dimas flume connected with a flowmeter (Sitrans
Probe LU, Peterborough, Ontario, Canada). Finally, a water
level logger (Global Water-WL 15, Gold River, California,
USA) was installed in May 2005 behind each of the three
earthen dams built across the GWW. A topographic survey was
carried out in Spring 2005 to determine the volume-depth
curves of the detention ponds (Table 1). Water temporarily
Fig. 1. (a) Location of Melsterbeek Catchment and Velm village in the Belgian loess belt. (b) Network of dry valleys draining to Velm village. Dotted lines represent
historical gullies and rills observed in the area.
Fig. 2. Land use and location of muddy flood control measures within the
Heulen Gracht Catchment.
Table 1
Characteristics of the detention ponds of the Heulen Gracht Catchment
Pond Max. dam
Height (m)
Width of
overflow (m)
Max. storage
volume (m3
)
Diameter
orifice
plates (m)
Volume/catch.
area (mm)
1 2.1 10.5 3500 0.2/0.25 1.46
2 2.2 12.6 3500 0.2/0.2 2.38
3 0.95 3.0 2000 0.25/0.25 0.67
Location of the ponds is given in Fig. 2.
421O. Evrard et al. / Geomorphology 100 (2008) 419­428
stored in the detention ponds drains through a pipe at the bottom
of the dam. Water levels are converted to outflow discharges
using Eq. (1) (Ilaco, 1985):
Q  A Â 2gh 0:5
1
where Q is discharge (m3
s- 1
); A is the cross-section of the
drain (m2
); g is gravity acceleration (9.81 m s-2
) and h is the
hydraulic head (m).
The impact of the control measures on runoff is estimated by
comparing peak discharges per unit area (l s-1
ha-1
), runoff
coefficients (%), duration of runoff flow (h) and lag time (h) for
each event measured at both the upstream and downstream
extremities of the GWW. Since the distribution of these parameters
is normal as determined by Kolmogorov­Smirnov tests, paired
Student's t-tests have been carried out using the SAS Enterprise
Guide statistical package (SAS Institute Inc., Cary, North Carolina,
USA) to detect any significant differences between both
extremities of the GWW at 95% confidence intervals.
2.4. Impact of control measures on erosion
The rills and gullies that were formed during the monitoring
period have been mapped, their length, depth and width
measured. The cross-sectional area of erosion features has
been computed for 67 transects within the catchment. The mass
of eroded soil is determined using the mean value of bulk density
measured for cropland in loess soils of central Belgium (1.43 g
cm- 3
; Goidts and van Wesemael, 2007). Sediment thickness in
the detention pond located behind dam # 3 (Fig. 2) was measured
with an estimated precision of 5 mm with a meter on a 5 m-grid
after each important rainfall event (with 10 mm of cumulative
rainfall). Data are interpolated to estimate sediment volume and
mass. The calculated erosion rates are compared to the output of
an empirical relationship between catchment area (A, ha) and
specific sediment yield (SSY, t ha- 1
yr-1
) obtained for 26
cultivated catchments of the Belgian loess belt over a period of
2­46 years during the 20th century (Eq. (2); Verstraeten and
Poesen, 2001).
SSY  26A0:35
2
A suspended sediment sampler (ISCO-6712, Lincoln,
Nebraska, USA) was installed in the San Dimas flume and
connected to the water level sensor in order to determine the
erosion rate after the installation of the control measures.
Since there is no permanent flow, sampling only occurs when
the height of water in the flume exceeds 5 cm. A runoff sample
is then taken at a 5 min-time step until the end of the event.
Suspended sediment concentration is determined by drying
the samples in an oven at 105 °C for 24 h. Runoff samples
have also been taken manually at the outlet of the dam pipes
Fig. 3. Monthly distribution of observed rainfall events during the period 2003­2007 with N15 mm of cumulative precipitation, and number of recorded runoff events.
Notes to Table 2:
A: Rainfall amount.
D: Duration of the event.
T: Return period according to Delbeke (2001).
Imax: Maximum rainfall intensity in 5 min.
Q: Peak discharge.
RC: Runoff coefficient.
NA: Not available.
a
Data for the period 2002­2004 are available from `crest stage recorder' measurements. Such a recorder consists of a plastic tube with a length of water-sensitive
tape which changes colour on contact with water.
422 O. Evrard et al. / Geomorphology 100 (2008) 419­428
during the heaviest storms to compare the sediment concentrations
and discharges with the ones measured in the San
Dimas flume. A Student t-test has been carried out to detect
significant changes in sediment discharge between both
extremities of the GWW.
2.5. Impact of control measures on muddy floods
The Sint­Truiden fire brigade classifies its interventions
according to their nature (road accident, fire, riverine or muddy
flood). Such data are available for Velm village since 1977.
Table 2
Results of runoff measurements in the San Dimas flume and the detention ponds of the Heulen Gracht Catchment
Rainfall data Flume Pond # 1 Pond # 2 Pond # 3a
Date (d/m/y) A (mm) D (h) T (yr) Imax (mm h-1
) Q (m3
s-1
) RC (%) Q (m3
s- 1
) D (h) Q (m3
s- 1
) D (h) Q (m3
s-1
) D (h) RC (%)
09/05/2002 20 1 15 0.48
20/06/2002 25 1 10 0.36
03/08/2002 25 0.5 20 0.11
20/08/2002 50 3 75 0.22
28/08/2002 70 1 N200 N0.5
06/02/2003 10 24 b2 0.05
24/05/2003 20 5 b2 0.05
03/07/2003 18 2 2 0.05
29/08/2003 40 24 5 0.06
08/07/2004 14.2 8 b2 0.05
17/07/2004 12.2 5.3 b2 0.05
21/07/2004 20 0.2 25 0.28
23/07/2004 23.2 4 2 0.28
08/08/2004 11 0.8 b2 0.05
13/08/2004 11.2 23 b2 0.05
14/08/2004 20.6 14 b2 0.26
01/07/2005 17.4 0.8 2 0.08 7.25 0.23 16.5 b0.03
14/08/2005 36.6 20.5 5 NA NA 0.07
23/10/2005 25.6 12 b2 0.04 7.5 0 b0.03
25/10/2005 14.8 10 b2 0.05 6.75 0 b0.03
31/03/2006 26 21 b2 0.08 10.7 0.08 12.33 0.09 2
01/04/2006 7 14 b2 0.06 6.33 0.06 7 b0.03
05/05/2006 10.8 3 b2 38 0.008 0.07 0 0 b0.03
18/05/2006 6 1 b2 34 0.03 2 0 0 b0.03 1.6
21/05/2006 18 3 b2 38 0.25 4.4 0.10 9.33 0.10 10.4 0.08 4 5.3
26/05/2006 11.1 10 b2 8 0.04 3.4 0.07 11 0.07 11 b0.03 NA
29/05/2006 14.6 7 b2 18 0.14 6.7 0.10 12 0.10 14.4 0.08 7 4.5
14/06/2006 24.7 1 10 94 0.44 5.2 0.28 10.4 0.26 11.4 0.24 9.84 5.9
03/08/2006 12.9 4 b2 30 0.09 10.2 0 0 b0.03 8.1
04/08/2006 17.6 5 b2 36 0.11 9.6 0 0 b0.03 7.6
05/08/2006 8.2 2 b2 30 0.29 8.9 0.05 7.25 0.05 6.3 b0.03 7.1
14/08/2006 22.5 6 2 29 0.16 8.9 0.08 11.1 0.08 11.75 0.09 1.66 1.7
15/08/2006 10 3 b2 23 0.09 14 0.06 10 0.06 12.7 b0.03 11.2
16/08/2006 10.3 1.5 b2 32 0.33 20.9 0.09 10.4 0.09 12.5 0.07 3.9 4.5
19/08/2006 7.9 0.33 b2 43 0.12 2.5 0.08 13.25 0.07 14.66 b0.03 2
21/08/2006 23.4 4 2 37 0.37 9 0.17 14.33 0.13 15 0.11 8.75 3.9
17/11/2006 14.8 7 b2 17 NA NA 0.07 10.5 0.07 8 b0.03 NA
19/11/2006 8.4 5.5 b2 7 NA NA 0.05 17 0.05 21.9 b0.03 NA
07/12/2006 7 2.5 b2 6 0.11 4.6 0.05 14.5 0.05 18.5 b0.03 3.7
08/12/2006 12 3 b2 10 0.03 2.1 0.07 13.5 0.07 16.5 b0.03 1.6
12/12/2006 6.2 5 b2 6 0.03 6.1 0.06 NA 0.06 NA 0.04 NA 4.9
18/01/2007 16 17.5 b2 7 0.03 2.6 0.07 21.4 0.07 18.4 b0.03 2.1
19/01/2007 13 6 b2 11 0.12 4.1 0.10 17 0.10 18.4 0.08 4.5 3.3
26/02/2007 15.7 14.5 b2 17 0.14 3.6 0.09 15.5 0.09 18.5 0.07 3.5 2.4
28/02/2007 10.2 7 b2 12 0.09 4.5 0.06 9.5 0.06 11 b0.03 3.5
07/03/2007 10.6 22 b2 7 0.06 2.5 0.03 8 0.03 8.4 b0.03 2.0
25/05/2007 13.5 2 b2 74 0.19 3.3 0.04 4.5 0.04 3.25 b0.03 2.7
11/06/2007 43 1 150 110 1.47 22.7 0.44 14 0.39 15.15 0.37 16.25 16.4
18/06/2007 5 0.33 b2 19 0.10 19.6 0.04 NA 0.07 19 0.05 3 4.5
25/06/2007 10.8 10 b2 41 0.05 0.8 0.03 NA 0.04 7.7 b0.03 0.6
20/07/2007 14.5 6 b2 30 0.06 1.2 0.07 NA 0.03 7 b0.03 0.9
28/07/2007 15.9 9.5 b2 16 0.14 4.8 0.07 14 0.03 13 0.05 1.5 0.8
02/08/2007 19.4 4 2 20 0.31 9.5 0.16 17 0.13 16.5 0.10 8 3.9
09/08/2007 50 14 50 12 0.34 15.1 0.28 34 0.27 33 0.26 28.5 9.5
21/08/2007 20.8 17 b2 34 0.24 6.9 0.11 19.5 0.11 20 0.09 8 2.6
423O. Evrard et al. / Geomorphology 100 (2008) 419­428
Corresponding daily rainfall data are available for the Gorsem
(Sint­Truiden) station of the Belgian Royal Meteorological
Institute, located 5 km from the catchment. These muddy flood
reports allow a comparison of flood frequency in Velm village
before and after the installation of the control measures, taking
account of the rainfall return period.
3. Results
Between 2002­2007, 77% of events with 15 mm
precipitation occurred between May and September. Similarly,
70% of runoff events occurred during this period (Fig. 3).
3.1. Impact of control measures on runoff
The San Dimas flume recorded 39 runoff events in 2006 and
2007 (Table 2). Runoff coefficients of the catchment upstream of
the GWW calculated based on the San Dimas flume discharge
varied between 0.07% and 22.7%, with a mean of 7.4% (Table 2).
Usually, runoff coefficients were higher in spring and summer
(8.3%) than in autumn and winter (3.8%). The highest coefficients
were measured in August 2006, which was a very wet month, as
well as during the extreme event of June 11 2007. The heaviest
and most intense storms have led to the highest peak discharges
in the flume (0.44 m3
s-1
on June 14 2006 and 1.47 m3
s-1
on
June 11 2007; Table 2).
Some 53 rainfall-runoff events have been measured behind the
dams (Table 2). Before the installation of dams 2 and 3 in August
2004 (Fig. 2), runoff events with a discharge 0.03 m3
s-1
have
been observed at dam 3 during low-intensity precipitation (e.g. in
February, 2003; Table 2). After the installation of the two
additional dams, notable runoff has been measured behind the
dam at the outlet (dam 3; Fig. 2) during only 13 events. These
events correspond to: (i) prolonged periods of rain in winter
(30 mm in 48 h) or (ii) to heavy thunderstorms between May
and August (20 mm in a few hours). All runoff parameters are
significantly different between both extremities of the GWW at
the 95% confidence interval (Table 3).
An important and significant decrease of the peak discharge
per unit area (mean of 69%) was observed between the San
Dimas flume (just upslope of the GWW) and the catchment
outlet. Loss of runoff, probably due to infiltration in the GWW
and behind the dams, has also been observed. Runoff
coefficients decreased by a mean of 40% between both
extremities of the GWW (Table 3). The reduction was higher
during low-intensity rainfall (mean of 43% for events with an
Imax b40 mm h-1
) than during intense thunderstorms (mean of
20% for events with an Imax N40 mm h-1
; Table 2).
Runoff was buffered during 5­12 h behind the three
successive dams. The mean duration of runoff was 38% longer
at the outlet than just upstream of the GWW. A long hydrograph
recession limb, corresponding to the progressive outflow of
runoff buffered behind the dams through the pipes, was
observed. The lag time increased by a mean of 75% after the
installation of the control measures.
Peak flow left the San Dimas flume and reached the outlet of
the first dam in a mean of 2 h 25 min (mean propagation velocity
of the peak discharge of 0.04 m s-1
). Peak outflow from the first
dam reached the outlet of the second pond in a mean of 32 min
(mean propagation velocity of 0.09 m s-1
). Some 64 min were
then needed for peak runoff to reach the outlet of the third dam
(mean propagation velocity of 0.14 m s-1
). The propagation of
the peak discharge was hence slowed down in the GWW, but the
decrease was not linear between the dams (Table 2).
3.2. Impact of control measures on specific sediment yield
According to Eq. (2), the specific sediment yield for a
catchment of the size of the Heulen Gracht should be 3.5 t ha- 1
Table 3
Summary of t-test results to detect significant differences in the flume (upstream of GWW) and at the outlet
Parameter Peak discharge per ha Runoff coefficient Flow duration Lag time Sediment discharge
Flume Outlet Flume Outlet Flume Outlet Flume Outlet Flume Outlet
Mean 0.8 0.2 7.5 4.4 9.6 15.5 1.2 5 3.2 0.2
SD 1.2 0.3 5.9 3.5 5 6 1.1 1.4 8 0.6
Observations 30 30 30 30 15 15 15 15 10 10
t stat 2.4 2.5 -2.8 -8.1 1.9
P (Tt) b0.01 b0.05 b0.01 b0.0001 0.064 (NS)
SD=standard deviation. NS=not significant.
Table 4
Muddy flood events requiring fire brigade interventions in Velm village between
1977­2002 and associated rainfall depth
Date Daily rainfall (mm) Duration (h) Return period
20/07/1980 38 b24 N5
10/08/1992 44.5 b24 N5
08/06/1996 21 b24
13/08/1996 47 b24 N10
30/05/1999 NA
08/05/2000 NA
03/06/2000 18.2 b24
14/07/2000 13.6 b24
25/07/2000 66.5 b24 N100
29/07/2000 23.2 b24
02/08/2001 40 b24 N5
09/05/2002 20 1 5
20/06/2002 25 1 10
20/07/2002 30 8 5
03/08/2002 25 0.5 20
20/08/2002 50 3 100
27/08/2002 70 1 N200
Return periods after Delbeke (2001).
424 O. Evrard et al. / Geomorphology 100 (2008) 419­428
yr-1
. The gullies observed since the 1940s and draining to Velm
are mapped in Fig. 1b.
Six major erosion events have been documented between
April­September 2002 (Table 4). The first erosion event
occurred in May 2002. The summer crops were already sown
at that time, and tillage erased erosion features remaining from
the winter period. Total volume of the rills and gullies reached
1500 m3
in September 2002, corresponding to a soil loss due
to rill and gully erosion of 2175 t. Rill and gully cross-section
was very variable (0.2­2.5 m2
), both between and within fields.
Erosion rate reached a mean value of 7.25 t ha-1
at the
catchment scale in 2002, without taking sheet erosion into
account.
Measurements of sediment concentrations in the flume as
well as in the outflow of the dams are available for 13 events
recorded in 2006 and 2007 (Table 5). Much of the sediment was
trapped behind the first dam. Sediment concentration in the first
dam's outflow is decreased by a mean of 86% compared to the
concentration measured in the San Dimas flume. It further
decreased by 16% due to trapping behind the second dam. In
contrast, an increase of sediment concentration (+38%) was
generally measured at the catchment outlet compared to the
second dam outflow. This is due to the inflow of runoff loaded
with sediments, flowing from row crop fields located along this
part of the GWW. However, sediment concentrations at the
outlet were reduced by a mean of 88% (0.9 g l-1
) compared to
the ones measured in the flume (mean of 5.4 g l-1
). Sediment
discharge was reduced by a mean of 93%, decreasing from
3.2 kg s-1
in the flume to 0.2 kg s- 1
at the outlet. However, this
difference was not statistically significant at 95% confidence
interval (Table 3). This is probably due to the rather low number
of events for which sediment data are available, as well as to the
important seasonal variation of runoff and sediment production
on cropland in the Belgian loess belt. Rainfall simulations have
shown that sediment production is much lower on crusted soils
in August (3 g l- 1
) than on fragmentary soils (40 g l- 1
) at the
end of spring (Evrard et al., in press). During the extreme event
of June 11 2007, 84 t of sediments were trapped behind the three
successive dams. It represents 72% of soil loss measured in the
San Dimas flume. No more rills have been observed in the
thalweg since the sowing of the GWW and the construction of
the dams despite the occurrence of several extreme events.
3.3. Impact of control measures on muddy floods
Soil erosion and flooding are ancient problems in the area.
Flooding of the nearby Gingelom village was already very
frequent during the 18th century (Aumann and Vandenghoer,
1989). Intense soil erosion was explicitly mentioned in the
1960s for the nearby Gingelom village (Fig. 1b; T'Jonck, 1967).
However, the off-site consequences have become more frequent
during the last two decades (Evrard et al., 2007a). Fire brigade
Table 5
Results of soil loss measurements in the San Dimas flume and sediment
concentrations in the outflow of the dams
San Dimas flume Dam 1 Dam 2 Dam 3
Soil loss Mean sediment N Sediment Sediment Sediment
Date
(d/m/yr)
(t) conc.(g l-1
) conc.
(g l- 1
)
conc.
(g l-1
)
conc.
(g l- 1
)
26/05/2006 1 0.9 2 0.4 0.2 NA
29/05/2006 20 10 10 0.2 0.1 NA
14/06/2006 120 30.9 24 1.7 1.3 1.9
21/08/2006 19 3.8 24 1.1 1.1 0.6
08/12/2006 5 0.7 2 0.4 0.4 0.3
26/02/2007 13 3.2 2 0.3 0.3 0.3
28/02/2007 4 1.2 2 0.3 NA NA
07/03/2007 2 1.1 2 0.9 0.3 NA
25/05/2007 2 2.1 2 0.7 NA NA
11/06/2007 117 5 24 1.6 1.9 2.2
18/06/2007 10 4.4 4 0.3 0.2 NA
02/08/2007 5 1.2 2 0.6 0.5 0.5
09/08/2007 16 0.9 3 0.9 0.2 0.3
N is the number of water samples taken in the flume. Two water samples have
systematically been taken behind the dams.
Fig. 4. Rainfall, inflow and outflow hydrographs measured during the thunderstorm of June 11 2007.
425O. Evrard et al. / Geomorphology 100 (2008) 419­428
interventions due to muddy floods in the Velm village
dramatically increased since 1980 (Table 4). All muddy floods
were triggered by heavy thunderstorms (between 14­70 mm of
rainfall, with a mean of 35.5 mm) and occurred between May
and August (Table 4). Six heavy storms (20­70 mm precipitation)
occurred in 2002, each leading to the flooding of the
village. The three events in August 2002 were rather extreme,
having a return period between 20- and N200 years (after
Delbeke, 2001).
Since the installation of the GWW and the two additional
earthen dams in 2004, no muddy flood has been recorded in
Velm village, despite the occurrence of several extreme events
(Table 2). The measures have particularly served their purpose
during the extreme event of June 11 2007 (having a return
period of 150 years, according to Delbeke, 2001), buffering
runoff during 17 h and preventing any flood in Velm village
(Fig. 4). Peak discharge per unit area decreased by 79% and the
lag time dramatically increased (from 10 min in the flume to 5 h
30 min at the outlet).
4. Discussion
4.1. Effectiveness of the grassed waterway and earthen dams
The propagation of the peak discharge was drastically
slowed down within the section with the GWW and the earthen
dams. However, there was no important reinfiltration in GWW
for moderate and extreme storms. This is due to a high soil
compaction (bulk density of 1.59 g cm- 3
in the GWW
compared to a mean of 1.43 g cm- 3
for cropland in the Belgian
loess belt according to Goidts and van Wesemael, 2007). This
confirms the results of rainfall simulations carried out in the
Belgian loess belt showing that grass strips and GWW have a
higher runoff coefficient (62­73%) than most cultivated soils
(13­58%; Evrard et al., in press).
Sediment trapping is very high and occurs mainly behind the
first dam, except during extreme events. These observations
confirm the main results of a former modelling exercise (Evrard
et al., 2007b). The model simulated that the GWW led to a 50%
decrease of peak discharge, which is consistent with field
observations. Our findings also agree with the results of a
similar study analysing the impact of a GWW on runoff and
erosion in a micro-catchment (8 ha) in southern Germany
(Fiener and Auerswald, 2005; Fiener et al., 2005). The German
ponds were very efficient in trapping sediments (between 50­
80% of sediments were trapped) and reducing peak runoff rates.
However, two main differences with our study can be outlined,
besides the different catchment sizes. In Germany, no event
having a return period of N5 years occurred during the 9-year
experiment, while we observed that the dams particularly served
their purpose during extreme events. Furthermore, an intensive
soil and water conservation scheme was implemented in the
German catchments draining to the ponds, limiting sediment
and runoff inputs (Auerswald et al., 2000). Our study shows that
even without widespread implementation of alternative farming
practices, the measures are effective in controlling muddy
floods. They offer, therefore, a solution that can be implemented
in the short term to protect the most endangered villages against
muddy floods.
4.2. Evaluation of erosion rates and sediment delivery
Based on field measurements in 2002, rill and gully erosion
rates reached 7.25 t ha-1
for that specific year. This figure does
not take sheet erosion into account. Often, interrill erosion has
been estimated as a fraction of total soil loss. This fraction
ranges between 10­20% of the total soil loss in the Belgian
loess belt (Govers and Poesen, 1988; Takken et al., 1999;
Steegen et al., 2000). Total erosion was hence underestimated in
our study and should be close to 8.3 t ha-1
. This figure is
consistent with the range of annual erosion rates measured in
central Belgium (6.5­12.3 t ha- 1
yr- 1
; Verstraeten et al., 2006).
Nachtergaele and Poesen (1999) calculated a mean ephemeral
gully erosion rate of 2.33 t ha-1
yr- 1
(over a six months
period during which summer ephemeral gullies remain active).
The ephemeral gully in the thalweg of the Heulen Gracht
Catchment was observed on all aerial photographs available for
the study area (between 1947­1996), always appearing at the
same location (Fig. 1b). Hence, no increase of gully erosion
throughout the study period was found. The highest erosion rate
was even observed in 1947 (3.43 t ha-1
yr-1
), which shows that
erosion is not a recent phenomenon in the study area.
Steegen (2001) showed that summer extreme events have a
particularly important effect on long-term landscape evolution.
For instance, a summer rainfall with a 10 year-return period that
occurred in a 250 ha-catchment in the Belgian loess belt
exported several times the mean long-term erosion rate (7 t ha- 1
yr- 1
for the extreme event, after Steegen et al., 2000; vs. 2.6 t
ha-1
yr-1
for the long-term mean, after Vandaele, 1997).
During the extreme event of June 2007, the three ponds
trapped sediments (84 t in total). Since the control measures
prevent the formation of rills and gullies in the catchment,
erosion rates are dramatically reduced. Only interrill erosion is
still observed at a mean rate of 0.5 t ha-1
yr- 1
, thereby
drastically decreasing sediment delivery to the alluvial plain.
4.3. Cost-efficiency of control measures
Immediately after thunderstorms, people need assistance
from the fire brigade and municipal services to pump water
from cellars and clean up the roads. Fire brigade interventions
after the thunderstorms of August 2002 in the Melsterbeek
Catchment (Fig. 1a) cost  25,000 (i.e. 125  km-2
). Muddy
floods also led to numerous cases of damage to private property.
According to 1601 records submitted by Belgian households to
the Disaster Fund, mean damage amount was  4436
(SD=3406 ; Evrard et al., 2007a). The villages of Velm and
Gingelom were particularly affected by the thunderstorms of
May and August 2002 (Table 4). Households from these two
villages submitted 268 records to the Belgian Disaster Fund.
They received  636,967 (mean of  2377 per record).
Overall, muddy floods lead to a damage cost of 54  ha- 1
yr- 1
in the region of Velm. Total cost of the control measures
installed in the area reached 126  ha-1
. The measures are built
426 O. Evrard et al. / Geomorphology 100 (2008) 419­428
for a 20 year-period according to the Soil Erosion Decree.
Farmers receive additional subsidies each year for the
maintenance of grass strips (21  ha- 1
yr-1
). Compared to
the damage cost of muddy floods (54  ha- 1
yr-1
), the
investments would be cost-efficient in 3 years if the measures
are effective and no muddy flood occurs. Our results prove that
the measures serve their purpose. In Velm village, total
investment ( 351,528) represents the damage cost to private
properties caused by the single August 2002 flood.
The Flemish authorities calculated that the construction of all
the control measures proposed in the municipal erosion mitigation
schemes that were approved by their administration would cost
between 7.7­9.6 million  yr-1
during the period 2006­2025,
which is not disproportionate compared with the total damage
cost associated with muddy floods in the Flemish municipalities
of the Belgian loess belt (between 8­86 million  yr-1
; Evrard
et al., 2007a).
5. Conclusions
A 12 ha-grassed waterway and three earthen dams were
installed in a 300 ha-cultivated catchment in central Belgium, in
order to prevent muddy floods in the downstream village. These
measures served their purpose by preventing muddy floods in
the village, even during extreme events (with a maximum return
period of 150 years). Peak discharge per unit area was reduced
by a mean of 69% between both extremities of the GWW.
Furthermore, runoff was buffered during 5­12 h, due to the
combined effect of the GWW and the earthen dams. The lag
time increased by 75% after the installation of the control
measures. Sediment discharge at the catchment outlet decreased
by a mean of 93% compared to the one measured in the GWW's
runoff inflow. The measures also prevented any gully formation
in the thalweg, thereby reducing erosion to an interrill
phenomenon which occurs at a mean rate of 0.5 t ha-1
yr-1
,
whereas the specific sediment yield of a catchment of similar
size without control measures in the Belgian loess belt should
reach 3.5 t ha-1
yr-1
. This would dramatically decrease
sediment transfer from the cultivated dry valley to the alluvial
plain. Given they prevent muddy floods and remain costefficient,
similar control measures can be installed to protect
other flood prone areas in the Belgian loess belt and comparable
environments. These measures could be combined with
alternative farming practices, such as reduced tillage. However,
there is a need to study the impact of these practices on runoff
and erosion at the catchment scale.
Acknowledgements
The authors thank Marco Bravin for his technical assistance in
installing the catchment loggers and measurement station and
Elisabeth Frot for her help with data collection in 2002. The
installation of the measurement devices was financially supported
by the Land & Soil Protection Division of the Flemish Ministry of
Environment, Nature and Energy. Local farmers (Jean Lejeune,
Jean Boonen and Roland Meys) are also gratefully thanked for
implementing the erosion control measures.
References
Aumann, F., Vandenghoer, C., 1989. Baron Surlet de Chokier, regent van
België: Mens en Mythe (1769­1839). (In Dutch). Gemeentebestuur
Gingelom.
Auerswald, K., Albrecht, H., Kainz, M., Pfadenhauer, J., 2000. Principles of
sustainable land-use systems developed and evaluated by the Munich
Research Alliance on Agro-Ecosystems (FAM). Petermanns Geographische
Mitteilungen 144 (2), 16­25.
Bielders, C.L., Ramelot, C., Persoons, E., 2003. Farmer perception of runoff and
erosion and extent of flooding in the silt-loam belt of the Belgian Walloon
Region. Environmental Science & Policy 6, 85­93.
Bork, H.-R., Bork, H., Dalchow, C., Faust, B., Piorr, H.-P., Schatz, T., 1998.
Landschaftsentwicklung in Mitteleuropa. Klett-Perthes, Gotha. (In German).
Boardman, J., Evans, R., Ford, J., 2003. Muddy floods on the South Downs,
southern England: problem and responses. Environmental Science & Policy
6, 69­83.
Boardman, J., Verstraeten, G., Bielders, C., 2006. Muddy floods. In: Boardman,
J., Poesen, J. (Eds.), Soil Erosion in Europe. Wiley, Chichester, pp. 743­755.
Delbeke, L., 2001. Extreme neerslag in Vlaanderen. (In Dutch). Technical
report. Aminal, Afdeling Water. Ministerie van de Vlaamse Gemeenschap.
de Moor, J.J.W., Verstraeten, G., in press. Alluvial and colluvial sediment
storage in the Geul River catchment (The Netherlands) -- combining
field and modelling data to construct a Late Holocene sediment budget.
Geomorphology.
Evrard, O., Bielders, C.L., Vandaele, K., van Wesemael, B., 2007a. Spatial and
temporal variation of muddy floods in central Belgium, off-site impacts and
potential control measures. Catena 70, 443­454.
Evrard, O., Persoons, E., Vandaele, K., van Wesemael, B., 2007b. Effectiveness
of erosion mitigation measures to prevent muddy floods: a case study in the
Belgian loam belt. Agriculture, Ecosystems & Environment 118, 149­158.
Evrard, O., Vandaele, K., Bielders, C.L., van Wesemael, B., in press. Seasonal
evolution of runoff generation on agricultural land in the Belgian loess belt
and implications for muddy flood triggering. Earth Surface Processes &
Landforms.
Fiener, P., Auerswald, K., 2005. Seasonal variation of grassed waterway
effectiveness in reducing runoff and sediment delivery from agricultural
watersheds in temperate Europe. Soil & Tillage Research 87 (1), 48­58.
Fiener, P., Auerswald, K., Weigand, S., 2005. Managing erosion and water
quality in agricultural watersheds by small detention ponds. Agriculture,
Ecosystems & Environment 110 (3­4), 132­142.
Goidts, E., van Wesemael, B., 2007. Regional assessment of soil organic carbon
changes under agriculture in Southern Belgium (1955­2005). Geoderma
141, 341­354.
Govers, G., Poesen, J., 1988. Assessment of the interrill and rill contributions to
total soil loss from an upland field plot. Geomorphology 1, 343­354.
Gyssels, G., Poesen, J., Nachtergaele, J., Govers, G., 2002. The impact of
sowing density of small grains on rill and ephemeral gully erosion in
concentrated flow zones. Soil & Tillage Research 64 (3­4), 189­201.
Holland, J.M., 2004. The environmental consequences of adopting conservation
tillage in Europe: reviewing the evidence. Agriculture, Ecosystems &
Environment 103 (1), 1­25.
Hufty, A., 2001. Introduction  la climatologie. De Boeck Université, Brussels.
(In French), 542 pp.
Ilaco, 1985. Agricultural compendium for rural development. Elsevier,
Amsterdam.
Lang, A., Bork, H.-R., Mäkel, R., Preston, N., Wunderlich, J., Dikau, R., 2003.
Changes in sediment flux and storage within a fluvial system: some
examples from the Rhine catchment. Hydrological Processes 17,
3321­3334.
Le Bissonnais, Y., Lecomte, V., Cerdan, O., 2004. Grass strip effects on runoff
and soil loss. Agronomie 24, 129­136.
Leys, A., Govers, G., Gillijns, K., Poesen, J., 2007. Conservation tillage on
loamy soils: explaining the variability in interrill runoff and erosion
reduction. European Journal of Soil Science 58, 1425­1436.
Nachtergaele, J., Poesen, J., 1999. Assessment of soil losses by ephemeral gully
erosion using high-altitude (stereo) aerial photographs. Earth Surface
Processes and Landforms 24, 693­706.
427O. Evrard et al. / Geomorphology 100 (2008) 419­428
Nachtergaele, J., Poesen, J., 2002. Spatial and temporal variations in resistance
of loess-derived soils to ephemeral gully erosion. European Journal of Soil
Science 53, 449­463.
Patty, L., Réal, B., Gril, J.J., 1997. The use of grassed buffer strips to remove
pesticides, nitrate and soluble phosphorus compounds of water. Pesticide
Science 49, 243­251.
Rommens, T., Verstraeten, G., Poesen, J., Govers, G., Van Rompaey, A., Peeters,
I., Lang, A., 2005. Soil erosion and sediment deposition in the Belgian loess
belt during the Holocene: establishing a sediment budget for a small
agricultural catchment. The Holocene 15 (7), 1032­1043.
Rommens, T., Verstraeten, G., Bogman, P., Peeters, I., Poesen, J., Govers, G.,
Van Rompaey, A., Lang, A., 2006. Holocene alluvial sediment storage in a
small river catchment in the loess area of central Belgium. Geomorphology
77, 187­201.
Souchre, V., King, C., Dubreuil, N., Lecomte-Morel, V., Le Bissonnais, Y.,
Chalat, M., 2003. Grassland and crop trends: role of the European Union
Common Agricultural Policy and consequences for runoff and soil erosion.
Environmental Science & Policy 6, 7­16.
Statistics Belgium, 2006. http://www.statbel.fgov.be.
Steegen, A., 2001. Sediment deposition in and export from small agricultural
catchments. Unpublished PhD thesis, Faculty of Sciences, Geography, K.U.
Leuven.
Steegen, A., Govers, G., Nachtergaele, J., Takken, I., Beuselinck, L., Poesen, J.,
2000. Sediment export by water from an agricultural catchment in the Loam
Belt of central Belgium. Geomorphology 33, 25­36.
Takken, I., Beuselinck, L., Nachtergaele, J., Govers, G., Poesen, J., Degraer, G.,
1999. Spatial evaluation of a physically-based distributed erosion model
(LISEM). Catena 37, 431­447.
T'Jonck, G., 1967. La lutte contre l'érosion. Un cas  Gingelom. L'amélioration
de la structure agraire: problmes techniques particuliers. (In French).
Société Nationale Terrienne, pp. 9­15.
Vandaele, K., 1997. Temporele en ruimtelijke dynamiek van bodemerosieprocessen
in landelijke stroomgebieden (Midden-België): een terreinstudie.
(In Dutch). Unpublished PhD thesis, Faculty of Sciences, Geography, K.U.
Leuven.
Van Dijk, P.M., Kwaad, F.J.P.M., Klapwijk, M., 1996. Retention of water and
sediment by grass strips. Hydrological Processes 10, 1069­1080.
Verstraeten, G., Poesen, J., 1999. The nature of small-scale flooding, muddy
floods and retention pond sedimentation in central Belgium. Geomorphology
29, 275­292.
Verstraeten, G., Poesen, J., 2001. Factors controlling sediment yield from small
intensively cultivated catchments in a temperate humid climate. Geomorphology
40, 123­144.
Verstraeten, G., Poesen, J., Govers, G., Gillijns, K., Van Rompaey, A., Van Oost,
K., 2003. Integrating science, policy and farmers to reduce soil loss and
sediment delivery in Flanders, Belgium. Environmental Science & Policy 6,
95­103.
Verstraeten, G., Poesen, J., Goossens, D., Gillijns, K., Bielders, C., Gabriels, D.,
Ruysschaert, G., Van Den Eeckhaut, M., Vanwalleghem, T., Govers, G.,
2006. Belgium. In: Boardman, J., Poesen, J. (Eds.), Soil Erosion in Europe.
Wiley, Chichester, pp. 385­411.
World Reference Base, 1998. World reference base for soil resources. World
Resources Report, vol. 84. FAO, Rome, Italy.
428 O. Evrard et al. / Geomorphology 100 (2008) 419­428