Science of the Total Environment 816 (2022) 151660

Contents lists available at ScienceDirect

Science of the Total Environment

journal homepage: www.elsevier.com/locate/scitotenv

Research Paper

Simulating the effect of check dam collapse in a debris-flow channel

Tommaso Baggio ⁎, Vincenzo D'Agostino
Department of Land, Environment, Agriculture and Forestry, University of Padova, viale dell'Università 16, 35020 Legnaro, PD, Italy

H I G H L I G H T S G R A P H I C A L A B S T R A C T

• Erosion and deposition in a debris-flow

event with check dam collapses are
analysed.
• A new methodology is proposed to
model the effect of check dam collapse.
• When check dams break debris-flow
volume increases due to bed erosion at
the most.

a r t i c l e i n f o a b s t r a c t

Article history: Sequences of erosion control/consolidation check dams are the most widespread channel countermeasure in the
Received 5 August 2021 European Alps. Some of them were built in the past based on ancient technologies. Nowadays they may not be
Received in revised form 9 November 2021 fully adequate to mitigate the debris-flow/flood events that are becoming more frequent and intense. Conse-
Accepted 9 November 2021
quently, there is the remote possibility that they could fail with disastrous consequences as observed in some
Available online 14 November 2021
cases. A reliable methodology to reproduce the effect of check dam collapse has not yet proposed. Therefore
Editor: Fernando A.L. Pacheco the aim of this study is to define a procedure to simulate the effect of check dam collapse in a debris-flow event.
In this study we analysed the catastrophic debris flow occurred in the Rotian channel (Italian Alps) during which
a series of check dams collapsed magnifying the event and causing severe damages.
Keywords: With the aid of field data we reconstructed the event and used the simulation tool r.avaflow to reproduce the de-
Debris flow bris flow. We then defined three scenarios to simulate the event: (A) debris-flow propagation over an erodible
Check dam collapse channel; (B) propagation on a rigid channel bed combined with the release of impulsive masses to isolate the
Bed erosion analysis of the effect of check dam collapse; (C) a combination of the previous scenarios. The simulation perfor-
Propagation modelling
mance was assessed analysing the pre- and post-event LiDAR surveys.
Results showed that the C scenario accurately reproduced the observed debris-flow erosion pattern. In particular,
we found out that most of the entrained debris volume derived from bed erosion rather than the sediment
retained by check dams. The adopted method, which composes the contribution of bed erosion and check dam
collapse, could be of particular relevance for residual risk estimation when mitigation structures are old and
may fail with potential disastrous consequences.
© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://
creativecommons.org/licenses/by-nc-nd/4.0/).

⁎ Corresponding author.
E-mail address: tommaso.baggio@phd.unipd.it (T. Baggio).

https://doi.org/10.1016/j.scitotenv.2021.151660
0048-9697/© 2021 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

1. Introduction simulate structure collapse scenarios, the choice of an adequate simula-

tion tool is fundamental to reproduce the phenomena in the most reli-
Debris flows are rapid mass movements made up by a mixture of able way. For debris flow phenomena the tool should match different
water, fine particles, sediment and boulders (Hungr et al., 2014; Jakob characteristics. (i) It should consider the flow as a biphasic mass in
et al., 2005; Takahashi, 2007). They can develop high velocities (up to order to both accurately represent the behaviour of the debris flow phe-
10 ms−1) generating impressive impact forces (Rickenmann et al., nomenon (Kean et al., 2013) and to reproduce the released or entrained
2003) because of their high bulk density. When debris flows affect vil- mass since they are characterized by a higher debris content than the
lages they can completely destroy infrastructures and buildings causing flowing mass (Papa et al., 2004). (ii) The model should incorporate
severe damage and fatalities (Dowling and Santi, 2014; Larsen et al., the flow propagation over erodible bed, since erosion processes can
2001). For this reason, debris-flow mitigation structures have been be really strong after the structure collapse (Cucchiaro et al., 2019a).
built to reduce the associated risk. Such structures mitigate the risk in (iii) The simulation tool should consider (a) the possibility of introduc-
different ways (Huebl and Fiebiger, 2005; Mizuyama, 2008): they de- ing the volume into the computational domain through mass release to
crease the flow velocity and bed erosion rates (consolidation check represent the volume of debris released after the check dam collapse
dams), promote sediment deposition (flow breakers, open retention and (b) a solid-liquid hydrograph to introduce the debris-flow routing.
check dams, sediment traps, deposition areas, flood reservoir), prevent Different models for debris-flow propagation have been developed in
bed and bank erosion (channelization, bank reinforcements, bed sills) the last years and successfully tested with real case scenarios (Frank
and control the flow spread (deflection walls). Such countermeasures et al., 2015; Gregoretti et al., 2018; McDougall and Hungr, 2004;
are planned and located according to the morphology, sediment budget O'Brien et al., 1993; Rosatti and Begnudelli, 2013). Some of them incor-
and characteristics of the channel and basin (Johnson and McCuen, porate an erosion model and the possibility of releasing a mass. Among
1989; Osti and Egashira, 2008). All of these structures have the final the different tools, r.avaflow (Mergili and Pudasaini, 2021) resulted ad-
goal of reducing the potential consequences of debris-flow events. The equate for structures collapse scenarios thanks to its flexibility in intro-
hazard and risk map is calculated taking into account the mitigation ducing the mass in the computational domain, the presence of the
structures located in the watershed (Fuchs et al., 2007; Gentile et al., empirical erosion model (Mergili et al., 2017) and the multi-phase
2007; Rodríguez-Morata et al., 2019). A consolidation check dam is mass flow propagation model (Pudasaini and Mergili, 2019). Moreover,
the most common type of mitigation structure for debris flow within r.avaflow have already been used to back calculate geomorphologically
the channel bed (Piton and Recking, 2014). They are transversal struc- complex scenarios that include interaction between landslide, glacier,
tures often built in a staircase – like sequence in a way that decreases debris flow and lake (Baggio et al., 2018; Gylfadóttir et al., 2019;
bed erosion, sediment transport, front velocity, pulse behaviour and Mergili et al., 2020b, 2018). The empirical erosion model has been
bank destabilization (Zeng et al., 2009). Even if check dams have been analysed and calibrated for landslide (Mergili et al., 2020a) and debris
designed taking into account the geomorphological conditions after flow phenomena (Baggio et al., 2021).
analysing past events, they may not be representative of future events The need to strengthen our capacity to model the effect of the col-
(Hübl et al., 2005). In fact, in the last decades climate change has rapidly lapse of a series of check dams is still an open research question
triggered glacier melting, permafrost degradation and taken rainstorms (Zhang et al., 2019) and is motivated by the increasing attention of pub-
to extremes (Borga et al., 2014; Prein et al., 2016; Stoffel et al., 2014; lic authorities towards communication of the residual hazard and re-
Vagnon, 2020). The expected consequence is an increase of debris- lated risk (Hartmann et al., 2021). The aim of this study is to develop
flow severity and frequency in terms of water and sediment volume. and test a procedure to adequately simulate the effect of check dam col-
Moreover, check dams can be damaged by debris flow and debris lapse in case of a high-magnitude debris-flow event.
flood over decades, causing scouring erosion to the foundations, dam- For this purpose a recent extreme debris-flow event amplified by the
age to the crest and body and lateral erosion (Tacnet et al., 2012; failure of a series of check dams is presented. Thanks to pre- and post-
Victoriano et al., 2018). Furthermore the correct maintenance of mitiga- event surveys and a field campaign, the event is reconstructed. The
tion structures is increasingly lacking (Mazzorana et al., 2014). The con- debris-flow hydrograph is estimated and an operative procedure is pro-
sequences may lead to the collapse of check dams as observed in some posed to simulate the effect of check dam failure. Three simulation sce-
cases of extreme events (Benito et al., 1998; Chen et al., 2015; Cucchiaro narios are computed to reproduce the effects of check dam failure
et al., 2019a, 2019b; Wang, 2013). The volume of debris flow increases through mass release and channel erosion. The following discussion
both because of the released material during the collapse and the in- considers the reliability, implications and generalization of the results.
crease of local channel slope after the collapse (Piton et al., 2016). As The outcomes of the study highlight a replicable methodologic path
consequence of channel slope increase both entrainment (Gregoretti, that can be used to simulate the consequences of check dam failure in
2008; Lamb et al., 2008) and transport capacity (Armanini, 2015; debris-flow hazard scenarios for risk mapping.
Lanzoni et al., 2017; Takahashi, 2007) grow. Therefore the failure of a
check dam or a series of them can have incredibly severe consequences, 2. Material and methods
and debris flows can likely impact areas classified as having no or low
risk. Then, the concept that mitigation measures are commonly always In this section it is first described the characteristics of the Rio Rotian
considered adequate to decrease the event magnitude may not be to- catchment reporting the morphology of the channel, the pre-event mit-
tally confirmed (Piton et al., 2016). Scenarios of mitigation structures igation structures and fan description. Then, the debris-flow event oc-
collapse are seldom considered but they should at least be mentioned curred on 29th October 2018 is recounted, describing the methods to
and taken into account as a remote but possible scenario in the risk as- reconstruct the debris-flow characteristics and erosion pattern. Lastly,
sessment framework. the simulation model is depicted, as well as the input data to reproduce
Simulation tools can reproduce the dynamic of mass movements the event and evaluate the simulation performance.
(landslides, debris flows, floods) along a defined topography given a
set of input data (Scheidl et al., 2013). Therefore, the production of haz- 2.1. Study site: the rio Rotian channel
ard maps and then risk maps is commonly based on the output of hy-
draulic and mass flow propagation models (Mazzorana et al., 2009; The rio Rotian channel is a mountainous torrent located in the au-
Stecher et al., 2012). Simulation tools can be used to assess the effective- tonomous province of Trento, in the Eastern Italian Alps. The catchment
ness of check dams (Bernard et al., 2019; Osti and Egashira, 2008; has a total extent of 2.4 km2 and a mean slope of 26.4°, extending from
Remaître et al., 2008; Liu et al., 2012) and to predict the residual risk an altitude of 2042 to 824 m a.s.l. (Fig. 1). The shape of the basin is elon-
in the case of protection measure collapse (Tseng et al., 2012). To gated along the main channel path in a north – east direction. The

2
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

The fan apex is at an altitude of 824 m a.s.l. and the fan has an area of
0.38 km2. The channel crosses the fan centrally and joins with the Noce
river at an elevation of 782 m a.s.l. The village of Dimaro is located on the
fan, mainly consisting of a campsite on the left hand side of the channel
and settlements on the right hand side.
In the past, the Rotian channel produced three recorded debris-flow
events. The 1776 event was described as a catastrophic phenomenon:
the channel transported a large amount of sediment to the confluence
with the Noce river stopping the flow and deviating its original path.
In 1882 two debris-flow events of the same magnitude hit the village
destroying some settlements and depositing boulders in the fan area.
More recently, in September 2000, the Rotian channel produced a
small debris-flood event where most of the transported sediment was
deposited upstream of the fan apex and the flow remained confined
to the channel path.

2.2. The 27-30th October 2018 event

Between the 27th and 30th October 2018 the North-Eastern Italian
Alps were hit by an extreme storm, named “Vaia”. The event was char-
acterized by intense rainfall associated with high wind speeds, up to
200 km h−1 (Chirici et al., 2019). Focusing on the rio Rotian basin, the
debris flow occurred at the end of the Vaia storm on the 29th October.
The event was characterized by three different surges occurring at
19:02, 19:48 and 23:35 CET (Borga and Zaramella, 2020). An analysis
of the rainstorm was carried out by Borga and Zaramella (2020) using
the radar data of the Macaion station (located 32 km in the NE direction
with respect to the fan apex). They calculated a total precipitation
amount of 359 mm for the rio Rotian basin in the period 27 – 29th
October, corresponding to a return period (three days maximum) of
300 years. Furthermore, the most intense precipitation occurred at the
end of the event. A further rainfall analysis on the duration of the 3 h be-
fore the event resulted in a return period of 100 years (Borga and
Zaramella, 2020). The basin, already stressed by two consecutive days
of heavy rainfall, generated a catastrophic debris flow. The flow was
characterized by major bed destabilization and erosion, which took
place starting from the upper part of the basin. Propagating downward
the flow increased its sediment component and destroyed the series of
15 check dams. The collapsed structures released debris that was
engulfed by the flow causing a further bed instability and triggering a
severe erosion process. The open check dam situated at 989 m a.s.l.
retained a large quantity of the transported debris (32,000 m3) but its
Fig. 1. Overview of the rio Rotian catchment and fan, showing the position of mitigation storage capacity was not sufficient to effectively reduce the debris vol-
measures and the surveyed sections. In background the slope map. (contour lines ume of such an extreme event. Consequently, the debris flow continued
altitude difference: 25 m).
propagating downwards and flooded part of the fan, depositing a large
volume of debris along the roads of the village (160,000 m3). The flow
affected several settlements and the campsite, causing severe damage
geology of the basin is composed by colluvial, alluvial and glacial de- and one fatality.
posits in the lower part (below 1300 m a.s.l.), while the upper part is The debris flow was mainly formed by a fine sediment matrix (prin-
composed by stratified layers of limestone and clay. Due to these geo- cipally silt and sand) together with large rocks and boulders. According
logical characteristics, in the last centuries the channel progressively to field evidence, the debris flow can be classified as a muddy debris
deepened, generating a canyon-like incision of around 10–50 m of ele- flow (Jakob et al., 2005). The front of the debris flow transported big
vation difference compared with the top of the banks. The banks are boulders (diameters larger than 1 m and up to 6 m) depositing some
very steep (range 30–45°) and they are very probably the source of sed- of them in the fan, near the campsite (Fig. 2A). A really big boulder
iment supply to the stream in the case of intense runoff. (estimated volume of 18 m3) was transported probably by the front of
The channel is 4.8 km long and has a mean slope of 11.9°. It has been the flow and completely closed the open check dam (Fig. 2B). Parts of
consolidated with a series of 15 check dams built in 1977 (reach in the the collapsed check dams were entrained in the flow and pieces were
range from 1233 to 1030 m a.s.l.) and made of reinforced concrete. The found in the retention basins (Fig. 2C). In addition, the channel part up-
check dam mean height is 5.3 m (range 2.9–7.8 m) for a cumulative sum stream of the 15 consolidation check dams exhibited erosion and bank
of 78.9 m and a mean channel slope of 13°. Immediately downstream of destabilization (Fig. 2D).
the last check dam the channel bed has been enlarged in order to create Two days after the event (31st October 2018) the Mountain Basin
a depositional area formed by a series of 5 retention basins connected Service of the Trento province performed a drone survey involving the
each other by a sloped reach (total length 250 m, mean width 25 m). fan area. Seven days later (7th November 2018) the Service surveyed
In 2014, a 7.4 m high open check dam was constructed at 989 m a.s.l., the channel in the zone of the sediment trap and retention basin
immediately downstream of the retention basins (Fig. 1), to increase (1050 m a.s.l.), covering a torrent length of 350 m upstream of the
the deposition of the flowing solid component. check dam. Fifteen ground control points (GCP) were located within

3
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

Fig. 2. Pictures of the post-event conditions. The debris deposit at the campsite (A), the sediment trap (B), the check dam channel reach (C) and the upper part of the channel (D); the white
and red lines represent the minimum and maximum debris-flow front height with respect to the channel bed.

the surveyed area and their positions recorded with a high resolution On the 12th and 13th November we collected field data walking
GPS (estimated error of 0.05 m). The acquired images were processed along the channel bed from the confluence with the Noce river to an el-
to generate a point dense cloud and then a high resolution (0.1 m) dig- evation of 1480 m a.s.l. The grain-size distribution of the debris for three
ital surface model (Basin Mountain Service, 2018). Regarding the upper different sampling points was collected (two in the channel path and
part of the channel that was not surveyed with the drone flight, we used one in the retention basin). Also three channel sections were acquired
the 2019 LiDAR data (acquired on 14-15th June 2019 with a 0.5 m res- in places where the estimated channel erosion was weak and where
olution) as representative of the post-event conditions (no debris flood the maximum flow level was recognizable on both banks. For every sec-
and flow events occurred between the October 2018 event and the tion the upward thalweg profile was surveyed for a minimum distance
LiDAR survey). Thanks to the pre-event LiDAR survey (acquired in of 25 m. The profiles and sections were surveyed measuring the hori-
2014, 0.5 m resolution, error 0.1 m) we then performed a Dem of Differ- zontal and vertical distance with a TruePulse laser distance meter (accu-
ence (DoD) analysis, calculating the eroded and deposited volume pat- racy 0.1 m). The maximum flow depth (sign of the passage on the
tern for the whole channel path and fan area. We used the minimum banks), flow area and upward thalweg slope were derived. Empirical
level of detection method (Brasington et al., 2003) to estimate the asso- equations reported in the literature were used to back calculate the
ciated error. The DoD error is the result of the following equation peak flow velocity v (Hungr et al., 1984; Lo, 2000; Prochaska et al.,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2008; Rickenmann, 1999) and consequently the peak discharge. The
δuDoD ¼ δu2pre þ δu2post , where δupre and δupost are the pre- and post-
general form of velocity [m/s] empirical equation is v = a(HbSc) + d
event associated DTM errors, respectively. The DoD analysis was the where H is the flow depth [m], S the thalweg slope [m/m] and a, b, c, d
basis for the event reconstruction, input hydrograph, magnitude esti- are empirical coefficients reported in Table 1. The calculated peak dis-
mation and evaluation of the simulation performance. charge was compared with the empirical equation of Benini (2000),

Table 1
discharge peak reconstruction of the surveyed sections through empirical equations derived from field data. The suffix “mod” means that the surveyed max flow depth has been modified.
We also report the empirical equations to calculate the max flow velocity and the peak discharge.

Velocity [m/s] Peak discharge [m3/s]

Section ID Max flow Wet area Slope Hungr et al. Lo (2000) Rickenmann Rickenmann Hungr Lo Rickenmann Rickenmann Mean
depth [m] [m2] [m m1] (1984) (1999) & Lo (1999) et al. (2000) (1999) & Lo (1999)
(2000) (1984) (2000)

v = 0.55H3/2 v = 3.32H2/3 v = 4.47H2/3 v = 8.90H0.30 Velocity x Wet area

S1/2 + 4.59 S1/5 + 0.70 S1/2 + 1.71 S1/2 + 1.06

SEZ. 1 3.3 25.4 0.41 6.7 5.4 8.0 8.6 169.3 136.7 203.6 218.3 187.0
SEZ. 1 1.8 15.8 0.41 5.4 3.8 5.9 6.6 85.2 59.7 92.5 103.5 78.3
mod
SEZ. 2 3.5 46.5 0.20 6.2 4.1 6.3 6.4 287.9 190.5 292.2 300.0 267.6
SEZ. 3 8.3 237.4 0.17 9.9 6.2 9.2 8.7 2354.4 1481.4 2176.8 2064.4 2019.3
SEZ. 3 5.8 82.6 0.17 7.7 5.1 7.6 7.5 638.8 419.8 628.6 616.2 575.9
mod

4
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

which considers the threshold for motion velocity of a sediment particle represented as a rigid wall since the opening had been closed by a big
transported by a debris flow (in this case we considered the maximum boulder, likely transported by the front of the debris flow. We located
particle diameter D that was moved): the input hydrograph 830 m upstream of the first check dam in corre-
spondence to the surveyed section 1 (Fig. 1).
 
ðγ − γÞ 0:5 0:5 To reproduce the Rotian debris flow three scenarios, representing
v ¼ 21:8 s D ð1Þ
γ three different ways to implement the check dam collapse, were tested:
A. Scenario representing the propagation of the debris flow over an
where γs and γ are the specific weights of the particle and flow,
erodible channel bed (check dams are considered erodible). CE is a
respectively.
function of a smoothed slope (SS [%], calculated as the mean value
Using the DoD map the erosion upstream section 1 (Fig. 1) was de-
within a circular moving window of 5 × 5 cells) as the most
rived to calculate the debris volume. For further verification, the ob-
performing function calibrated in (Baggio et al., 2021) for debris-
tained value is compared with empirical relationships (Mizuyama
flow entrainment over erodible channel.
et al., 1992).
Based on the peak discharge and total solid volume estimation, for a C E, exp 6 ¼ 10ð0:02Ss −6:75Þ ð4Þ
modelling aim an input hydrograph with a triangular shape was de-
rived, fixing the volumetric concentration of the solid component to
Moreover, the smoothed slope dependent function was varied in the
0.5 (Gregoretti et al., 2018). channel reach between the input hydrograph location and the check
dam 01, hereafter Area 1 (Fig. 4), in order to further improve the re-
2.3. Simulation model and input data liability of the erosion model. In Area 1 the simulated eroded vol-
umes adopting the function of Eq. (4) resulted considerably higher
To back calculate the debris flow, the simulation tool r.avaflow ver- than the observed one. Other two functions were tested (reported
sion 2.4 was adopted (Mergili and Pudasaini, 2021). The tool imple-
in Baggio et al., 2021) to simulate erosion in Area 1, always based
ments the multi-phase mass flow model described in Pudasaini and
on a smoothed slope:
Mergili (2019). The model can simulate the propagation of different
types of mass flows down a general topography. The mass can be intro- C Eexp2 ¼ 10ð0:025Ss −7Þ ð5Þ
duced in the calculation domain simultaneously or alternatively
through a hydrograph (discharge versus time relation) or a release C Eexp3 ¼ 10ð0:03Ss −7:5Þ ð6Þ
mass (raster map). An additional function of the model is the possibility
of releasing the mass from a certain cell at a given time.
Implementing r.avaflow for the Rotian debris-flow event, the multi-
phase model was reduced to a two-phase model, considering the solid B. Scenario of a debris-flow propagation over a non-erodible channel
component as debris material and the fluid one as water. The selected bed. The check dam collapses are reproduced by debris material re-
flow parameters are basically the same values reported in the user man- leased for every check dam. To calculate the release volume an up-
ual (Mergili and Pudasaini, 2021). We varied two flow parameters in ward sliding failure slope of 25° was supposed (angle of repose of
order to reproduce the rheology of a muddy debris flow. In particular, debris soaked material; assessed by means of local surveys of terrain
the basal friction angle was decreased to 16° and the fluid kinematic vis-
cosity was increased to 0.005 m2s−1. Furthermore, the tool provides an
empirical multi-phase erosion model, calculating for each time step (Δt)
the entrained volume in terms of erosive depths DE by means of the
following function of the flow momentum (M) (Eqs. (2) and (3)):
 
DE,s ¼ C E M s þ Mf α s,Emax Δt ð2Þ
 
DE,f ¼ C E M s þ Mf ð1−α s,Emax ÞΔt ð3Þ

where CE is the coefficient of erosion, αs,Emax the sediment volumetric

concentration of the eroded material and M is the flow momentum.
The subscripts s and f refer to the solid and fluid phase of the flow,
respectively.
The erosion model requires the assignment of CE, DE,max maximum
erosion depth, and αs,Emax. These parameters are defined by the user
alternatively as a single value or spatially distributed (raster map). The
maximum erosion depth was set to 10 m (maximum erosion
observed) but erosion was not permitted in correspondence to the
retention basin and sediment trap, since in those locations the channel
bed was reinforced with stones. Indeed, αs,Emax was set to 0.7,
representing the water saturated pre-event soil conditions. The value
is similar to that observed in the triggering area of two debris-flow ba-
sins of Eastern Alps (Gregoretti et al., 2019, 2018). Deposition is consid-
ered as the remaining mass at the end of the simulation time; in this
study it is fixed to 3000 s (50 min) for all simulations performed.
Regarding the DTM, the 2014 LiDAR derived terrain model was used,
representing the pre-event conditions. The original DTM was resampled
from a resolution of 0.5 m to 2 m for computational speed, adopting a
mean value method. The DTM was further modified to represent the Fig. 3. Profile sketch and plan view of the release depth calculation for the simulation of a
open check dam, since it was under construction in 2014. It has been check dam collapse.

5
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

collapses after the events) as the result of the lack of bed stabiliza- white and red lines). The modification of the bed bottom position
tion provided by the check dam Fig. 3. The depth of the released ma- were supported by information on the bed topography from the DoD
terial is then calculated as the difference between the original map.
topography (input DTM) and the assumed failure slope plane. The The calculation of the peak discharge of section 1 is particularly im-
calculated mass retained by the check dam is released at the passage portant, since it is the basis for the construction of the input hydrograph
of the simulated debris-flow peak discharge (estimated with a previ- for the simulation model. From Table 1 it is possible to identify the peak
ous simulation). The solid mass released for every check dam results discharge of section 1 that for the modified section resulted 78 m3 s−1,
variable in accordance with its height as highlighted in the sketch of calculated as the mean of the empirical equations. The estimated total
Fig. 3 and the solid ratio has been set to 0.8. Such mass was released solid volume from the DoD map resulted 33,600 m3. The value is com-
considering only the channel bed by means of a mask layer. An ex- puted as the erosion rate per channel length between section 1 and 2
ample of the final released mass height distribution of a check dam (equal to 32 m3m−1 according to the DoD) multiplied by the channel
collapse is represented in Fig. 3. The calculation was implemented length upstream of section 1 (1050 m). The estimation has been per-
as a pre-processing algorithm in the same script to launch the formed due to the impossibility to compute a reliable DoD map up-
mass flow propagation with r.avaflow. The input data required for stream of section 1 (presence of vegetation cover within the channel
every check dam is a polyline identifying the downstream border and steep slopes not well captured by the post-event survey). Using
of the check dam step and a mask outlining the upstream channel the equation of Mizuyama et al. (1992) the solid volume associated to
bed. It is also possible to adapt the algorithm to reproduce a partial the peak discharge of 78 m3 s-1 resulted 37,237 m3. The statistical anal-
collapse of a check dam in terms of height or width. In the first ysis of debris flows occurring in the Eastern Alps by Marchi and
case (height under collapse) it is just necessary to modify the eleva- D'Agostino (2004) corroborated the debris volume of such an event. Ac-
tion value of the lowest border of the computer generated slope. In cording to these authors, the maximum debris volume (Vs) related to
the second case (work width under collapse) the channel mask the upper catchment area A [km2] is given by the upper envelope
can be modified to simulate the check dam collapse for a particular equation: Vs = 70,000 A, calculating a value of 41,300 m3 for section 1
area of interest. (upper catchment area of 0.59 km2). This value is in line with the
debris volume derived from the DoD map upstream of the section 1.
Furthermore, the peak discharge in section 1 was confirmed by means
C. Scenario representing a combination of the previous two scenarios. of the limiting equation (Eq. (1)) for boulder entrainment (max grain
The mass representing the check dam collapse is calculated and re- diameter 1.0 m), obtaining a value of 81.6 m3s−1. The comparison
leased at the peak discharge passage as explained in B. Moreover, with equations reported in the literature and the DoD analysis
erosion is allowed in the computational domain and CE has been confirmed that the calculated peak discharge and associated debris
set to a unique value for the whole computational domain. Six volume can be consistently representative of the investigated event.
simulations were performed where CE was progressively varied We then used these values to build up an input hydrograph with a
from 10−7 to 10–6.2 kg−1 in order to calibrate the erosion triangular shape in accordance with field observations (Berti et al.,
coefficient in accordance with the observed erosion volume of the 2000; Gregoretti et al., 2019; Kean et al., 2012). We fixed the peak
check dam series in the torrent reach under analysis. discharge at 100 s after the hydrograph starts. After the peak, the dis-
charge progressively decreased to zero in a total time of 956 s. The
solid volumetric concentration was set to 0.5 since the event was
To evaluate the best scenario the observed net balance (difference mainly triggered by the movement of the soaked bed deposits within
between deposited and eroded volumes) was compared with those the whole channel path and eye witnesses reported the passage of a
simulated for two selected channel reaches. Area 1 represents the chan- dense debris flow.
nel reach between the input hydrograph and a section 50 m upstream of Regarding the section 2 the estimated peak discharge is equal to
check dam 1, while Area 2 encloses the channel part of the check dam 267 m3/s (mean of all equations). However, this value is probably
series (Fig. 4). In particular, a simulation was considered accurate overestimated since the tail of the debris flow eroded the channel
when it reproduces the balance volume within an error of 20% with re- bed increasing the apparent maximum flow depth, that was mea-
spect to the observed value. sured in the field. Besides, the estimated peak discharge of the sec-
tion 3 was assessed in 575 m 3/s (mean of all equations, Table 1).
3. Results Also in this case, the discharge assessment is uncertain since a depo-
sition followed by an intense bed erosion influenced maximum flow
3.1. Hydrograph reconstruction depth measure.

Hydrograph reconstruction was based on the calculation of the solid 3.2. Erosional and depositional pattern analysis
volume (DoD map) and on the peak discharge reconstruction. Adopting
empirical relations, the mean velocity was derived for the debris-flow The DoD was computed as the difference between the 2019 and
front. Then, the bulked peak discharge was calculated through the prod- 2014 DTM. For our objective, involving the check dam collapse simula-
uct of the wet area surveyed in the field and the flow velocity of the tion, the focus was on the channel reach between the input hydrograph
front. location (section 1, Fig. 1) and the last downstream check dam (right
A total of three channel sections and the related upstream channel upstream of the retention basin). Two areas of interest were defined
profile were surveyed, collecting their horizontal and vertical coordi- for the calculation of the eroded and deposited volumes (Fig. 4). The
nates with respect to a reference point (locations are reported in error associated with the DoD map resulted 0.14 m and, adopting the
Fig. 1). The maximum flow depth that occurred during the 29th October minimum level of detection method, the eroded and deposited volumes
2018 event, the upward thalweg slope and wet area were estimated. within the two areas were calculated (Table 2).
Through the empirical equations, the peak discharges were derived, In the investigated areas the erosion processes resulted predom-
which are listed in Table 1. For section 1 the maximum flow depth inant with respect to the depositions one. For Area 1 (comprised be-
was modified, decreasing the value of 1.5 m. In fact, it was assumed tween 1435 and 1238 m a.s.l for a length of 817 m) intense erosion
that the channel deepening to the bedrock was caused by the tail of was observed, with an associated volume balance per channel length
the debris flow and not by the front surge under estimation. A sign of of 31 m3m−1. In Area 2 (covering the torrent reach between 1238
such front passage can be identified in the channel banks, (Fig. 2D, and 1029 m a.s.l for a length of 852 m) the erosion process became

6
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

Fig. 4. Maps representing the observed erosion and deposition depths (DoD) and simulated elevation changes respect the basal topography for scenarios A, B and C.

more severe. In this channel segment the erosion resulted 119 investigated in this study, the volume balance was computed for the
m3 m −1 . The net balance between eroded and deposited volumes zone of the retention basins (from just downstream of Area 2 to the
consisted in −25,630 m 3 (± 1855) and 102,393 m3 (± 4110) for open check, altitude range 1029–989 m a.s.l., length 290 m) resulting
Area 1 and 2 respectively. The erosion depth per channel length in in a net balance in favour of deposition of 20,048 m3 (± 2831). The
Area 1 is typical of debris-flow triggering reaches, while the high reach downstream of the open check dam up to the municipal road
values of Area 2 is the combined result of the check dam collapses (altitude range 1029–928 m a.s.l., length 531 m) exhibited erosion
and the successive bed erosion process. As to those areas not directly processes characterized by a net balance of −18,447 m3.

7
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

Table 2
analysis of the volume balance (deposition – erosion) based on the DoD map and comparison with the simulated scenarios. * Represent the balance error of the DoD for the investigated
area.

Area 1 Area 2

Scenario Volume balance Error (simulated Volume Volume balance Error (simulated Volume/channel
(deposited – eroded) /observed) /channel (deposited – eroded) /observed) length [m3/m]
[m3] length [m3]
[m3/m]

Observed −25,630 1855⁎ −31.4 −102,393 4110⁎ −119.2

Scenario A (CE: Area 1 - > CE,exp2; Area 2 −28,732 1.12 −33.72 −99,492 0.98 −116.8
- > CE,exp6)
Scenario B (mass release – without 0 – 0.0 −22,304 0.22 −26.2
erosion)
Scenario C (mass release and erosion, −30,864 1.20 −36.2 −109,096 1.07 −128.1
CE = 10–‐6.6.4)

3.3. Scenario results process representation. The operative procedure developed for
scenario C resulted more reliable to simulate the effect of check
Three scenarios A, B, C were computed, releasing the selected dam collapse than scenario A, since scenario C released the mass
input hydrograph in section 1. Scenario A involved the flow propaga- retained by the check dams at the passage of the peak discharge.
tion over an erodible channel, scenario B the release of a mass for The hydrograph in Fig. 5B of scenario C captured this aspect of im-
every check dam (sketch in Fig. 3) and scenario C was a combination pulsive mass release as highlighted by a greater peak discharge
of bed erosion and mass release. The total volume released by the than scenario A.
collapse of the check dam series at the passage of the peak discharge The output hydrographs reported in Fig. 5 show the discharge (de-
is equal to 22,304 m3 (scenario B and C). The erosion and deposition bris and fluid component) and solid concentration variation over time
patterns of the performed simulations are illustrated in Fig. 4. For for location A and B highlighted in Fig. 4. In location A scenarios A and
scenarios A and C were showed only the simulations best matching C predicted similar maximum discharges, equal to 107 and 121 m3s−1,
the observed erosion rates. For scenario A, the best performing sim- respectively. Instead, scenario B produced the same discharge as the
ulation is the result of the combination of two CE functions (CE,exp2 for input hydrograph (78 m3s−1), since entrainment was not permitted.
Area 1 and CE,exp6 for Area 2, Eqs. (5) and (4) respectively). Regarding Regarding the volumetric solid concentration, scenario A and C showed
scenario C the chosen C E coefficient is 10 –6.4 kg−1. In Table 2 the a stable value around 0.58, while for scenario B it resulted lower, equal
comparison between the observed and simulated erosion and to the input hydrograph (0.5, constant for the whole passage of the de-
deposition volumes is reported. For Area 1, representing the bris flow). Regarding the output hydrograph in location B, scenario C
channel reach between the input hydrograph and the first check predicted the highest discharge equal to 99 m3s−1, while scenario A
dam, scenario A performed best, resulting in a relative error of 12% and B resulted 75 and 43 m3s−1, respectively. Scenario A and C exhib-
of additional erosion with respect to the observed value. Scenario C ited a first surge in the front, followed by a higher peak. The discharge
predicted a slightly greater volume of eroded sediment volume re- of the front surge of scenario C resulted greater than scenario A as con-
spect to scenario A with an overestimated error for erosion of sequence of the mass released in the proximity of output location B. The
about 20%. The simulated erosion patterns of scenarios A and C re- solid concentration pattern showed a higher value in the front (range
sulted slightly different in Area 1. In particular, in the scenario A ero- 0.6–0.65) and it immediately decreased around 0.55 until 17 min of
sion was more variable (more intense for steeper slopes) than in the simulation time. Afterward the solid concentration constantly increased
scenario C for which erosion resulted steadier with a more intense to reach values around 1 at the end of the simulation time. This behav-
process upstream than check dam 01. Anyway, both scenarios accu- iour is the results of the deposition process caused by the open check
rately represented the erosion pattern observed in Area 1, since the dam. The biphasic representation of the flow simulated aggradation of
relative error resulted equal to or lower than 20%. Obviously, in the the solid component in this area, while the fluid component slowly sep-
scenario B no erosion occurred since it has not been permitted in arated, flowing downstream. The phenomenon is captured by the in-
the computational domain. Regarding Area 2, representing the crease of the solid concentration towards 1.
check dam series, both scenarios A and C represented the observed
eroded volume satisfactorily and provided a relative error of −2 4. Discussion
and 7% respectively. Instead scenario B predicted just a small amount
of eroded/released material (22,304 m3 ) corresponding to the re- In this study the Rio Rotian debris-flow event that occurred on the
leased debris of the check dam collapses only. 29th October 2018 was reconstructed. The peak discharge was first es-
Regarding the erosion pattern in Area 2, we can notice an ob- timated in the triggering area and then three different scenarios were
served intense erosion process upstream of check dam 01 and be- performed using the simulation model r.avaflow, aiming to reproduce
tween check dam 08 and 11 with associated erosion depths the effects of a check dam collapses during the event. The simulation
reaching a maximum of 11 m. Moreover, for check dam 06 erosion performance were analysed and evaluated using the observed erosion
occurred both upstream and downstream of its position. For check and deposition volumes derived from the DoD map.
dams 02, 04 and 05 erosion mainly occurred downstream, while for The event peak discharge and total volume at section 1 were evalu-
check dams 01 and 03 it occurred mostly upstream. Regarding Area ated with the DoD analysis and confirmed by the equations of Benini
2 (check dam series), scenario A and C exhibited a similar pattern. (2000) and Marchi and D'Agostino (2004) . Some uncertainties still re-
Scenario A eroded a large amount of material in correspondence main in the estimation of the hydrograph shape due to the impossibility
and downstream of the check dams while scenario C generated ero- of the discharge time pattern evaluation after the peak passage. How-
sion both upstream and downstream. Even if scenarios A and C are ever, the choice of a linearly shaped hydrograph is also supported by
both considered adequate to reproduce the observed eroded pat- the observations of two high magnitude debris flows recorded in the
tern, the two simulations have a different approximation in the triggering area (Simoni et al., 2020). The same pattern has been

8
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

Fig. 5. Output hydrographs for sections A and B (locations are reported in Fig. 4) representing the total discharge (solid + fluid) and solid concentration over time of the simulation
scenarios. The input hydrograph is also reported for comparison.

observed in the routing area by Berti et al. (2000) and Kean et al. (2012). Regarding the simulations, scenarios A and C successfully
In addition, also varying the discharge pattern, the simulation results in reproduced the observed erosion volume of the check dam series (rela-
terms of erosion may not vary consistently, since the erosion function is tive error ≤ 20%). In particular, in scenario C a simulation representing
strictly dependent on the flow momentum and its highest value occurs the effect of check dam collapse was set up, calibrating an erosion coef-
in correspondence of the debris-flow front (Berger et al., 2011; McCoy ficient based on the observed eroded volume. A semiautomatic proce-
et al., 2012). Regarding section 2 and 3, their geomorphological condi- dure was developed to derive the spatially distributed volume
tions did not allow a reliable reconstruction of the peak discharge. In retained by the check dams. The effect of check dam collapse was then
section 2, the channel bed erosion process did not permit a reliable es- simulated through mass release in the computational domain
timation of the debris-flow front. Instead, for section 3 the peak dis- synchronised with the peak discharge passage. Furthermore, as ob-
charge evaluation resulted uncertain because of the large amount of served from scenario C, the calibrated CE coefficient was also
debris deposited. Therefore, the estimated peak discharge in section 2 appropriate to reproduce the erosion pattern of Area 1 (natural
and 3 (Table 1) could not be used for the evaluation of the investigated channel bed). Thanks to the DoD map the observed erosion pattern
scenarios. was reproduced as a combination of two smoothed slope dependent
Regarding the DoD we compared the eroded volumes per channel erosion functions reported in Baggio et al. (2021). For Area 1, where
unit with other observations reported in the literature in order to esti- the erosion process resulted less intense, (channel reach upstream of
mate its magnitude. Regarding Area 1 (natural channel bed) the erosion the check dam series) we validated the function CEexp2 (Eq. (5)).
resulted 31 m3m−1, which is a high value compared with other studies Instead for Area 2 (channel reach containing the collapsed check
in the Alps (Marchi and Cavalli, 2007; Marchi and D'Agostino, 2004). For dams) the function CEexp6 (Eq. (4)) represented better the erosion
Area 2 (check dam series), erosion resulted 119 m3 per channel unit process. Then, it is possible to conclude that the effect of check dam
length and can be considered as an extreme value for the eastern Alps collapse can be calibrated according to the reach erodibility by
(Marchi and Cavalli, 2007) and even worldwide (Hungr et al., 2005; simulating this intense erosion process similarly to a natural debris-
Rickenmann and Zimmermann, 1993). However such a high erosion flow channel.
value is the consequence of check dam collapse and is fully comparable, Based on all results, scenario C resulted more precise and then ap-
in terms of geomorphological estimation of debris-flow volumes, to that propriate to simulate check dam collapse than scenario A. However,
associated to a triggering area. Analysing erosion in Area 2, this occurred the simulation setup of scenario A could also be used for check dam col-
predominantly upstream and downstream of the check dam location lapse simulation, but disregarding real local erosion at the check dam lo-
(Fig. 4). Deposition process started just downstream of check dam 15 cations and aiming to focus only on the total amount of entrained debris
where the channel slope becomes gentler due to the presence of the re- from the channel bed. Accepting this approximation, the simulation
tention basins (mean channel slope of the first deposit equal to 8.4°). setup results easier since it does not require the input on the check
More importantly, we observed that: i) the spatially distributed bed dams position and a mask to extract the channel bed area.
erosion is the dominant mechanism in debris-flow entrainment in the The verifications through alternative modelling scenarios also con-
case of check dam failure; ii) this erosive pattern is not particularly dif- firmed the outcomes of other studies that highlighted the importance
ferent from that obtainable in the absence of check dams and it cannot of a careful simulation of debris-flow erosion (Armanini et al., 2009;
be expected in a particular reach of the torrent (e.g., always down- Gregoretti et al., 2019; Kean et al., 2013). This is always necessary in
stream of the collapsed work; Fig. 4). In fact, the total eroded volume case the travelling surge suffers a clear modification of its sediment con-
(102,393 m3) is around five time greater than the volume estimated centration at equilibrium conditions driven by the bed slope, according
and released by the collapse of the 15 check dams (22,304 m3) reported to Takahashi (2007), or due to the input of supplementary water dis-
in scenario B (a sketch for the calculation of the volume released is re- charge (e.g. inflow from a tributary branch) or due to the release of sup-
ported in Fig. 3). Therefore, the principal cause of debris-flow growth plementary debris (e.g. check dam failure as in our case). For such
is represented by bed erosion as consequence of check dams collapse. reasons simulation tools that include the erosion process have to be
A similar phenomenon is reported in Benito et al. (1998), where the col- adopted for the delineation of hazard and risk maps.
lapse of retention dams triggered an extreme debris-flow event in the In the last decades, attention towards debris-flow structure effec-
Biescas basin (Spain). Similarly to the Rotian channel, the Biescas debris tiveness and potential failure consequences has been rising due to
flow acquired debris from incision into the unconsolidated glacial de- their age and lack of maintenance, together with the increasing number
posits after the barrier collapse. A similar mechanism of entrainment of people living in high risk zones (Mazzorana et al., 2014). In particular,
is also reported in Chen et al. (2015) and Gong et al. (2020), where different studies pointed out that effectiveness assessment of mitigation
the collapsed check dams destabilized the channel bed, permitting an structures has to consider potential failures (Chen et al., 2015;
intense bed erosion process. Dell'Agnese et al., 2013; Piton et al., 2016). Thanks to our back-

9
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

analysis two joint scenarios for the simulation of a check dam series col- DE, max Maximum erosion depth
lapse were defined and successfully performed. Other studies investi- f Stands for fluid-phase
gated the debris flow – check dam interaction. Cui and Chen (2021) H Flow depth
and Ferrari et al. (2010) analysed the dam break effect on laboratory ex- M Flow momentum
periments and successfully reproduced the observed behaviour with QP Peak discharge
numerical models. Chen et al. (2019) and Marchelli and De Biagi S Slope of the thalweg
(2019) examined the check dam failure condition and debris-flow SS Smoothed slope
surge impacting the check dam through simulation models. Sodnik s Stands for solid-phase
et al. (2013) proposed a qualitative approach for the evaluation of the V Volume
hazard level in case of check dam failure. In the mentioned studies v Flow velocity
some limitations result from the investigation on small scale experi- γ Flow specific weight
ments, the use of single – phase and rigid bed propagation models. γs Particle specific weight
The present study, thanks to the availability of a well documented cata- Δt Time step length
strophic event and a specific implementation of the r.avaflow model, δupre Pre-event DTM error
overtakes these limitations and it demonstrates a good performance δupost Post-event DTM error
for a complicate real case. The proposed procedure can be directly ap- δuDoD Propagated DoD error
plied to investigate and assess the potential effects of check dam failure αs, Emax Solid volumetric concentration of the eroded material
in those torrents where mitigation structures may partially or totally
fail. Such a simulation is also capable of driving our decision on main- CRediT authorship contribution statement
taining or not ancient check dams in torrents that have been heavily
managed in the last centuries (1800 and 1900) and where the erosion Tommaso Baggio: Conceptualization, Methodology, Validation, In-
control works are in remote wild valleys with difficult access for vehi- vestigation, Writing - Original Draft.
cles and manpower. The results could be used to update the actual Vincenzo D'Agostino: Conceptualization, Writing – Review and
debris-flow risk assessment maps (Mazzorana et al., 2009) and help Editing, Supervision.
civil authorities in raising residents' awareness of debris-flow events.
Declaration of competing interest
5. Conclusions
The authors declare that they have no known competing financial
The study described a catastrophic debris-flow event occurred in the interests or personal relationships that could have appeared to influ-
Eastern Italian Alps at the end of the Vaia storm (October 2018). The ence the work reported in this paper.
event was characterized by the collapse of 15 check dams, which
destabilized the channel bed. The entrained material increased the Acknowledgements
debris-flow volume causing sever damages in the fan area. Thanks to
the LiDAR and field data, the depositional-erosional pattern of the debris This research received funding as part of the project “Bridging the
flow was successfully reconstructed. Three complementary different mass-flow modelling with the reality”, from the CARIPARO foundation
scenarios were implemented to simulate the check dam collapse (2724/2018). The study was additionally funded by Provincia
through bed erosion and mass release. For two of the proposed scenar- Autonoma di Trento through Accordo di Programma GPR. Authors wish
ios, a semi-automatic procedure was developed to calculate the mass to thank the Mountain Basin Service of the Autonomous Province of
release in the case of check dam collapse. The adopted mass flow Trento (IT) for the post-event data and they are also grateful to the three
model r.avaflow resulted particularly suitable for this purpose due to reviewers for their useful suggestions and comments.
its flexibility in introducing the mass within the computational domain.
The observed debris-flow entrainment process in the check dam area References
mainly occurred through bed erosions rather than released masses. In
Armanini, A., 2015. Closure relations for mobile bed debris flows in a wide range of slopes
fact, the total eroded volume resulted five times greater than the esti- and concentrations. Adv. Water Resour. 81, 75–83. https://doi.org/10.1016/J.
mated volume released by the fifteen check dams (scenario B). The re- ADVWATRES.2014.11.003.
sults, together with other companion literature findings, suggest that Armanini, A., Fraccarollo, L., Rosatti, G., 2009. Two-dimensional simulation of debris flows
in erodible channels. Comput. Geosci. 35, 993–1006. https://doi.org/10.1016/j.cageo.
the effect of check dam failures is often over-estimated as to its power
2007.11.008.
in magnifying the damage. The coefficient of erosion calibrated in Baggio, T., Mergili, M., Pudasaini, S., Carter, S., Fischer, J., 2018. Simulating snow process
Baggio et al. (2021) also resulted suitable to predict the eroded volumes chains: avalanche-river interactions with r.avaflow. International Snow Science
as consequence of check dam failure. This scenario could be immediate Workshop. Innsbruck, pp. 792–796.
Baggio, T., Mergili, M., D’Agostino, V., 2021. Advances in the simulation of debris flow ero-
to implement for an approximate simulation of the effect of check dam sion: the case study of the Rio Gere (Italy) event of the 4th August 2017. Geomor-
failure. phology 381, 107664. https://doi.org/10.1016/j.geomorph.2021.107664.
Summarizing all our findings, we can state that, in the lack of a Basin Mountain Service, 2018. Post-event Photogrammetric Survey of the rio Rotian
(original title, in italian: Rilievo fotogrammetrico post evento del rio Rotian).
model capable to simulate check dam collapse and debris flow erosion
Benini, G., 2000. Sistemazioni idraulico forestali. UTET — Unione Tipografico-Editrice
with a unique simulation, the proposed procedure resulted to be suit- Torinese. IT, Torino.
able to update the hazard maps in debris-flow channels where mitiga- Benito, G., Grodek, T., Enzel, Y., 1998. The geomorphic and hydrologic impacts of the cat-
tion structures are getting old, and to reduce the possible astrophic failure of flood-control-dams during the 1996-Biescas flood (Central Pyre-
nees, Spain). Z. Geomorphol. 42, 417–437. https://doi.org/10.1127/zfg/42/1998/417.
consequences for citizens and infrastructures. Berger, C., McArdell, B.W., Schlunegger, F., 2011. Direct measurement of channel erosion
by debris flows, Illgraben, Switzerland. J. Geophys. Res. Earth Surf. 116, n/a-n/a.
Nomenclature https://doi.org/10.1029/2010JF001722.
Bernard, M., Boreggio, M., Degetto, M., Gregoretti, C., 2019. Model-based approach for de-
sign and performance evaluation of works controlling stony debris flows with an ap-
a, b, c, d Empirical coefficients for flow velocity estimation plication to a case study at Rovina di Cancia (Venetian Dolomites, Northeast Italy).
CE Coefficient of erosion Sci. Total Environ. 688, 1373–1388. https://doi.org/10.1016/j.scitotenv.2019.05.468.
CEexp Coefficient of erosion derived with exponential function Berti, M., Genevois, R., LaHusen, R., Simoni, A., Tecca, P.R., 2000. Debris flow monitoring in
the acquabona watershed on the Dolomites (Italian alps). Phys. Chem. Earth Part B
D Maximum particle diameter Hydrol. Oceans Atmos. 25, 707–715. https://doi.org/10.1016/S1464-1909(00)
DE Erosion depth 00090-3.

10
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

Borga, M., Zaramella, M., 2020. Evento di piena del 27–29 ottobre 2018 sul bacino del Rio Johnson, P.A., McCuen, R.H., 1989. Slit dam design for debris flow mitigation. J. Hydraul.
Rotian: stima della precipitazione e valutazione della sua severità. Eng. 115, 1293–1296. https://doi.org/10.1061/(ASCE)0733-9429(1989)115:9(1293).
Borga, M., Stoffel, M., Marchi, L., Marra, F., Jakob, M., 2014. Hydrogeomorphic response to Kean, J.W., Staley, D.M., Leeper, R.J., Schmidt, K.M., Gartner, J.E., 2012. A low-cost method
extreme rainfall in headwater systems: flash floods and debris flows. J. Hydrol. 518, to measure the timing of postfire flash floods and debris flows relative to rainfall.
194–205. https://doi.org/10.1016/j.jhydrol.2014.05.022. Water Resour. Res. 48. https://doi.org/10.1029/2011WR011460.
Brasington, J., Langham, J., Rumsby, B., 2003. Methodological sensitivity of morphometric Kean, J.W., McCoy, S.W., Tucker, G.E., Staley, D.M., Coe, J.A., 2013. Runoff-generated debris
estimates of coarse fluvial sediment transport. Geomorphology 53, 299–316. https:// flows: observations and modeling of surge initiation, magnitude, and frequency.
doi.org/10.1016/S0169-555X(02)00320-3. J. Geophys. Res Earth Surf. 118, 2190–2207. https://doi.org/10.1002/jgrf.20148.
Chen, X., Cui, P., You, Y., Chen, J., Li, D., 2015. Engineering measures for debris flow hazard Lamb, M.P., Dietrich, W.E., Venditti, J.G., 2008. Is the critical shields stress for incipient sed-
mitigation in the Wenchuan earthquake area. Eng. Geol. 194, 73–85. https://doi.org/ iment motion dependent on channel-bed slope? J. Geophys. Res. Earth Surf. 113,
10.1016/j.enggeo.2014.10.002. 2008. https://doi.org/10.1029/2007JF000831.
Chen, H.X., Li, J., Feng, S.J., Gao, H.Y., Zhang, D.M., 2019. Simulation of interactions between Lanzoni, S., Gregoretti, C., Stancanelli, L.M., 2017. Coarse-grained debris flow dynamics on
debris flow and check dams on three-dimensional terrain. Eng. Geol. 251, 48–62. erodible beds. J. Geophys. Res. Earth Surf. 122, 592–614. https://doi.org/10.1002/
https://doi.org/10.1016/j.enggeo.2019.02.001. 2016JF004046.
Chirici, G., Giannetti, F., Travaglini, D., Nocentini, S., Francini, S., D’Amico, G., Calvo, E., Larsen, M.C., Wieczorek, G.F., Eaton, L.S., Torres-Sierra, H., 2001. Natural hazards on aluvial
Fasolini, D., Broll, M., Maistrelli, F., Tonner, J., Pietrogiovanna, M., Oberlechner, K., fans: the debris flow and flash flood disaster of December 1999, Vargas State,
Andriolo, A., Comino, R., Faidiga, A., Pasutto, I., Carraro, G., Zen, S., Contarin, F., Venezuela. In: Sylva, W. (Ed.), Proceedings of the Sixth Caribbean Islands Water Re-
Alfonsi, L., Wolynski, A., Zanin, M., Gagliano, C., Tonolli, S., Zoanetti, R., Tonetti, R., sources Congress. Mayagüez, Puerto Rico, pp. 1–7.
Cavalli, R., Lingua, E., Pirotti, F., Grigolato, S., Bellingeri, D., Zini, E., Gianelle, D., Liu, J., Nakatani, K., Mizuyama, T., 2012. Effect assessment of debris flow mitigation works
Dalponte, M., Pompei, E., Stefani, A., Motta, R., Morresi, D., Garbarino, M., Alberti, G., based on numerical simulation by using Kanako 2D. Landslides 102 (10), 161–173.
Valdevit, F., Tomelleri, E., Torresani, M., Tonon, G., Marchi, M., Corona, P., Marchetti, https://doi.org/10.1007/S10346-012-0316-X 2012.
M., 2019. Forest damage inventory after the “Vaia” storm in Italy. For. - Riv. di Selvic. Lo, D.O.K., 2000. Review of Natural Terrain Landslide Debris-resisting Barrier Design.
ed Ecol. For. 16 (1), 3–9. https://doi.org/10.3832/efor3070-016. Marchelli, M., De Biagi, V., 2019. Dynamic effects induced by the impact of debris flows on
Cucchiaro, S., Cavalli, M., Vericat, D., Crema, S., Llena, M., Beinat, A., Marchi, L., Cazorzi, F., protection barriers. Int. J. Protective Struct. 10, 116–131. https://doi.org/10.1177/
2019a. Geomorphic effectiveness of check dams in a debris-flow catchment using 2041419618798378.
multi-temporal topographic surveys. Catena 174, 73–83. https://doi.org/10.1016/j.ca- Marchi, L., Cavalli, M., 2007. Procedures for the documentation of historical debris flows:
tena.2018.11.004. application to the Chieppena Torrent (Italian Alps). Environ. Manag. 40, 493–503.
Cucchiaro, S., Cazorzi, F., Marchi, L., Crema, S., Beinat, A., Cavalli, M., 2019b. Multi- https://doi.org/10.1007/s00267-006-0288-5.
temporal analysis of the role of check dams in a debris-flow channel: linking struc- Marchi, L., D’Agostino, V., 2004. Estimation of debris-flow magnitude in the Eastern Italian
tural and functional connectivity. Geomorphology 345, 106844. https://doi.org/10. Alps. Earth Surf. Process. Landf. 29, 207–220. https://doi.org/10.1002/esp.1027.
1016/j.geomorph.2019.106844. Mazzorana, B., Hübl, J., Fuchs, S., 2009. Improving risk assessment by defining consistent
Cui, P., Chen, J., 2021. Numerical simulation of dam-break mudflow based on the and reliable system scenarios. Nat. Hazards Earth Syst. Sci. 9, 145–159. https://doi.
Herschel- Bulkley model. EGU, p. 6979. org/10.5194/nhess-9-145-2009.
Dell’Agnese, A., Mazzorana, B., Comiti, F., Von Maravic, P., D’Agostino, V., 2013. Assessing Mazzorana, B., Trenkwalder-Platzer, H.J., Fuchs, S., Hübl, J., 2014. The susceptibility of con-
the physical vulnerability of check dams through an empirical damage index. J. Agric. solidation check dams as a key factor for maintenance planning. Österr. Wasser
Eng. 44, 9–16. https://doi.org/10.4081/jae.2013.e2. Abfallwirtschaft 66, 214–216. https://doi.org/10.1007/s00506-014-0160-4.
Dowling, C.A., Santi, P.M., 2014. Debris flows and their toll on human life: a global analysis McCoy, S.W., Kean, J.W., Coe, J.A., Tucker, G.E., Staley, D.M., Wasklewicz, T.A., 2012. Sedi-
of debris-flow fatalities from 1950 to 2011. Nat. Hazards 71, 203–227. https://doi.org/ ment entrainment by debris flows: in situ measurements from the headwaters of a
10.1007/s11069-013-0907-4. steep catchment. J. Geophys. Res. Earth Surf. 117, n/a–n/a. https://doi.org/10.1029/
Ferrari, A., Fraccarollo, L., Dumbser, M., Toro, E.F., Armanini, A., 2010. Three-dimensional 2011JF002278.
flow evolution after a dam break. J. Fluid Mech. 663, 456–477. https://doi.org/10. McDougall, S., Hungr, O., 2004. A model for the analysis of rapid landslide motion across
1017/S0022112010003599. three-dimensional terrain. Can. Geotech. J. 41, 1084–1097. https://doi.org/10.1139/
Frank, F., McArdell, B.W., Huggel, C., Vieli, A., 2015. The importance of entrainment and t04-052.
bulking on debris flow runout modeling: examples from the Swiss Alps. Nat. Hazards Mergili, M., Pudasaini, S.P., 2021. R.avaflow - the mass flow simulation tool. R.avaflow 2.3
Earth Syst. Sci. 15, 2569–2583. https://doi.org/10.5194/nhess-15-2569-2015. software. URL. https://www.avaflow.org. (Accessed 4 December 2021).
Fuchs, S., Heiss, K., Hübl, J., 2007. Towards an empirical vulnerability function for use in Mergili, M., Fischer, J.T., Krenn, J., Pudasaini, S.P., 2017. R.avaflow v1, an advanced open-
debris flow risk assessment. Nat. Hazards Earth Syst. Sci. 7, 495–506. https://doi. source computational framework for the propagation and interaction of two-phase
org/10.5194/nhess-7-495-2007. mass flows. Geosci. Model Dev. 10, 553–569. https://doi.org/10.5194/gmd-10-553-
Gentile, F., Tiziana, A.E., Ae, B., Liuzzi, G.T., 2007. Debris-flow Risk Analysis in South 2017.
Gargano Watersheds (Southern-Italy). https://doi.org/10.1007/s11069-007-9139-9. Mergili, M., Frank, B., Fischer, J.T., Huggel, C., Pudasaini, S.P., 2018. Computational experi-
Gong, X.-L., Chen, K.-T., Chen, X.-Q., You, Y., Chen, J.-G., Zhao, W.-Y., Lang, J., 2020. Charac- ments on the 1962 and 1970 landslide events at Huascarán (Peru) with r.avaflow:
teristics of a debris flow disaster and its mitigation countermeasures in Zechawa lessons learned for predictive mass flow simulations. Geomorphology 322, 15–28.
Gully, Jiuzhaigou Valley, China. Water 12, 25. https://doi.org/10.3390/w12051256. https://doi.org/10.1016/j.geomorph.2018.08.032.
Gregoretti, C., 2008. Inception sediment transport relationships at high slopes. J. Hydraul. Mergili, M., Mergili, M., Jaboyedoff, M., Pullarello, J., Pudasaini, S.P., 2020a. Back calculation
Eng. 134, 1620–1629. https://doi.org/10.1061/(ASCE)0733-9429(2008)134:11 of the 2017 Piz Cengalo-Bondo landslide cascade with r.avaflow: what we can do and
(1620). what we can learn. Nat. Hazards Earth Syst. Sci. 20, 505–520. https://doi.org/10.5194/
Gregoretti, C., Degetto, M., Bernard, M., Boreggio, M., 2018. The debris flow occurred at Ru nhess-20-505-2020.
Secco Creek, Venetian Dolomites, on 4 August 2015: analysis of the phenomenon, its Mergili, M., Pudasaini, S.P., Emmer, A., Fischer, J.T., Cochachin, A., Frey, H., 2020b. Recon-
characteristics and reproduction by models. Front. Earth Sci. 6, 80. https://doi.org/10. struction of the 1941 GLOF process chain at Lake Palcacocha (cordillera Blanca,
3389/feart.2018.00080. Peru). Hydrol. Earth Syst. Sci. 24, 93–114. https://doi.org/10.5194/hess-24-93-2020.
Gregoretti, C., Stancanelli, L.M., Bernard, M., Boreggio, M., Degetto, M., Lanzoni, S., 2019. Mizuyama, T., 2008. Structural countermeasures for debris flow disasters. Int. J. Erosion
Relevance of erosion processes when modelling in channel gravel debris flows for ef- Control Eng. 1, 38–43. https://doi.org/10.13101/ijece.1.38.
ficient hazard assessment. J. Hydrol. 568, 575–591. https://doi.org/10.1016/j.jhydrol. Mizuyama, T., Kobashi, S., Ou, G., 1992. Prediction of debris flow peak discharge.
2018.10.001. Interpraevent. Bern, Switzerland, pp. 99–108.
Gylfadóttir, S.S., Mergili, M., Jóhannesson, T., Helgason, J.K., Sæmundsson, Þ., Fischer, J.-T., O’Brien, J.S., Julien, P.Y., Fullerton, W.T., 1993. Two-dimensional water flood and mudflow
Pudasaini, S.P., 2019. A Three-Phase Mass Flow Model Applied for the Simulation of simulation. J. Hydraul. Eng. 119, 244–261. https://doi.org/10.1061/(asce)0733-9429
Complex Landslide–Glacier–Lake Interactions in Iceland. Geophysical Research Ab- (1993)119:2(244).
stracts, in. Osti, R., Egashira, S., 2008. Method to improve the mitigative effectiveness of a series of
Hartmann, S., Pedoth, L., Dalla Torre, C., Schneiderbauer, S., 2021. Beyond the expected- check dams against debris flows. Hydrol. Process. 22, 4986–4996. https://doi.org/
residual risk and cases of overload in the context of managing alpine natural Hazards. 10.1002/hyp.7118.
Int. J. Disaster Risk Sci. 12, 205–219. https://doi.org/10.1007/s13753-020-00325-3. Papa, M., Egashira, S., Itoh, T., 2004. Critical conditions of bed sediment entrainment due
Hübl, J., Strauss, A., Holub, M., Suda, J., 2005. Structural mitigation measures. 3rd Probabi- to debris flow. Nat. Hazards Earth Syst. Sci. 4, 469–474. https://doi.org/10.5194/
listic Workshop: Technical Systems + Natural Hazards. Wien, pp. 115–126. nhess-4-469-2004.
Huebl, J., Fiebiger, G., 2005. Debris-fow mitigation measures. Debris-Flow Hazards and Re- Piton, G., Recking, A., 2014. The dynamic of streams equipped with Check Dams. River
lated Phenomena. Springer, Berlin, Heidelberg, pp. 445–487. Flow 2014. CRC Press, pp. 1437–1445 https://doi.org/10.1201/b17133-192.
Hungr, O., Morgan, G.C., Kellerhals, R., 1984. Quantitative analysis of debris torrent haz- Piton, G., Carladous, S., Recking, A., Tacnet, J.M., Liébault, F., Kuss, D., Quefféléan, Y., Marco,
ards for design of remedial measures. Can. Geotech. J. 21, 663–677. https://doi.org/ O., 2016. Why do we build check dams in Alpine streams? An historical perspective
10.1139/t84-073. from the French experience. Earth Surf. Process. Landf. 42, 91–108. https://doi.org/
Hungr, O., McDougall, S., Bovis, M., 2005. Entrainment of material by debris flows. Debris- 10.1002/esp.3967.
Flow Hazards and Related Phenomena. Springer, Berlin, Heidelberg, pp. 135–158 Prein, A.F., Rasmussen, R.M., Ikeda, K., Liu, C., Clark, M.P., Holland, G.J., 2016. The future in-
https://doi.org/10.1007/3-540-27129-5_7. tensification of hourly precipitation extremes. Nat. Clim. Chang. 7, 48–52. https://doi.
Hungr, O., Leroueil, S., Picarelli, L., 2014. The Varnes classification of landslide types, an org/10.1038/nclimate3168.
update. Landslides 11, 167–194. https://doi.org/10.1007/s10346-013-0436-y. Prochaska, A.B., Santi, P.M., Higgins, J.D., Cannon, S.H., 2008. A study of methods to esti-
Jakob, M., Hungr, O., Jakob, D., 2005. Debris-flow Hazards and Related Phenomena. mate debris flow velocity. Landslides 5, 431–444. https://doi.org/10.1007/s10346-
Springer. 008-0137-0.

11
T. Baggio and V. D'Agostino Science of the Total Environment 816 (2022) 151660

Pudasaini, S.P., Mergili, M., 2019. A multi-phase mass flow model. J. Geophys. Res. Earth Stecher, M., Mazzorana, B., Hübl, J., 2012. Proposal of risk mitigation strategies based on a
Surf. 124, 1–23. https://doi.org/10.1029/2019jf005204. conceptual planning approach. A case study conducted in the Gadriabach study site,
Remaître, A., Van Asch T., W.J., Malet, J.P., Maquaire, O., 2008. Influence of check dams on Vinschgau valley, Italy. Interpraevent. Grenoble, pp. 811–820.
debris-flow run-out intensity. Nat. Hazards Earth Syst. Sci. 8, 1403–1416. https://doi. Stoffel, M., Tiranti, D., Huggel, C., 2014. Climate change impacts on mass movements -
org/10.5194/nhess-8-1403-2008. case studies from the European Alps. Sci. Total Environ. 493, 1255–1266. https://
Rickenmann, D., 1999. Empirical relationships for debris flow. Nat. Hazards 19, 47–77. doi.org/10.1016/j.scitotenv.2014.02.102.
Rickenmann, D., Zimmermann, M., 1993. The 1987 debris flows in Switzerland: docu- Tacnet, J.-M., Curt, C., Benjamin, R., Richard, D., 2012. Efficiency assessment for torrent
mentation and analysis. Geomorphology 8, 175–189. https://doi.org/10.1016/0169- protection works. Interpraevent. Grenoble, pp. 821–831.
555X(93)90036-2. Takahashi, T., 2007. Debris Flow: Mechanics. Taylor & Francis, Prediction and Counter-
measures.
Rickenmann, D., Weber, D., Stepanov, B., 2003. Erosion by debris flows in field and labo-
Tseng, W.H., Wang, H.W., Chou, S.C., Kao, Y.L., Shieh, C.L., 2012. Experiments on channel
ratory experiments. International Conference on Debris-Flow Hazards Mitigation:
evolution caused by check-dam failure. J. Mt. Sci. 9, 175–184. https://doi.org/10.
Mechanics, Prediction, and Assessment, Proceedings. 2, pp. 883–894.
1007/s11629-012-2252-6.
Rodríguez-Morata, C., Villacorta, S., Stoffel, M., Ballesteros-Cánovas, J.A., 2019. Assessing Vagnon, F., 2020. Design of active debris flow mitigation measures: a comprehensive
strategies to mitigate debris-flow risk in Abancay province, south-central Peruvian analysis of existing impact models. Landslides 17, 313–333. https://doi.org/10.1007/
Andes. Geomorphology 342, 127–139. https://doi.org/10.1016/j.geomorph.2019.06. s10346-019-01278-5.
012. Victoriano, A., Brasington, J., Guinau, M., Furdada, G., Cabré, M., Moysset, M., 2018. Geo-
Rosatti, G., Begnudelli, L., 2013. Two-dimensional simulation of debris flows over mobile morphic impact and assessment of flexible barriers using multi-temporal LiDAR
bed: enhancing the TRENT2D model by using a well-balanced generalized roe-type data: the Portainé mountain catchment (Pyrenees). Eng. Geol. 237, 168–180.
solver. Computers Fluids 71, 179–195. https://doi.org/10.1016/j.compfluid.2012.10. https://doi.org/10.1016/j.enggeo.2018.02.016.
006. Wang, G.L., 2013. Lessons learned from protective measures associated with the 2010
Scheidl, C., Rickenmann, D., McArdell, B.W., 2013. Runout prediction of debris flows and Zhouqu debris flow disaster in China. Nat. Hazards 69, 1835–1847. https://doi.org/
similar mass movements. Landslide Science and Practice. Springer, Berlin Heidelberg, 10.1007/s11069-013-0772-1.
pp. 221–229 https://doi.org/10.1007/978-3-642-31310-3_30. Zeng, Q.L., Yue, Z.Q., Yang, Z.F., Zhang, X.J., 2009. A case study of long-term field perfor-
Simoni, A., Bernard, M., Berti, M., Boreggio, M., Lanzoni, S., Stancanelli, L.M., Gregoretti, C., mance of check-dams in mitigation of soil erosion in Jiangjia stream, China. Environ.
2020. Runoff-generated debris flows: observation of initiation conditions and Geol. 58, 897–911. https://doi.org/10.1007/s00254-008-1570-z.
erosion–deposition dynamics along the channel at Cancia (eastern Italian Alps). Zhang, F., Yan, B., Feng, X., Lan, H., Kang, C., Lin, X., Zhu, X., Ma, W., 2019. A rapid loess
Earth Surf. Process. Landf. 45, 3556–3571. https://doi.org/10.1002/esp.4981. mudflow triggered by the check dam failure in a bulldoze mountain area, Lanzhou,
Sodnik, J., Kryžanowski, A., Martin, M., Mikoš, M., 2013. Torrential check dams as debris- China. Landslides 16, 1981–1992. https://doi.org/10.1007/s10346-019-01219-2.
flow sources. Landslide and Flood Hazard Assessment. Zagreb, pp. 6–9.

12