Sådhanå (2024)49:219  Indian Academy of Sciences

https://doi.org/10.1007/s12046-024-02456-5 Sadhana(0123456789().,-volV)FT3](012345
6789().,-volV)

Empirical model for predicting erosion on slope covered

by unconsolidated tephra
F TATA YUNITA1 , INDRATMO SOEKARNO1,*, JOKO NUGROHO1 and
UNTUNG BUDI SANTOSA2
1
Faculty of Civil and Environmental Engineering, Bandung Institute of Technology, Bandung, Indonesia
2
Association of Hydraulics Engineers, Jakarta, Indonesia
e-mail: [email protected]

MS received 16 March 2023; revised 8 December 2023; accepted 28 December 2023

Abstract. Volcanic eruption is known as multi-hazards to the surrounding environment and society causing the
formation of lahar as the frequent hazards that shortly occurred due to airborne tephra after an eruption. Erosion
triggered by rainfall on unconsolidated tephra material, such as volcanic ash, is the primary lahar initiation
mechanism. The time and scale of lahars vary based on eruptions and watershed conditions. The variability of the
erosion process is driven by a set of local factors including the grain size and spatial distribution of volcanic ash
thickness, slope, and rainfall intensity. Laboratory simulation experiments were conducted in a 3.00 m long, 0.75 m
wide, and 0.50 m deep flume to study the relationship of volcanic ash erosion rate to three driven parameters, namely
slope, rainfall intensity, and volcanic ash thickness. Three slope gradients were selected to represent gentle (14.1%),
mild (26.8%), and steep (46.6%) slopes. Meanwhile, the rainfall intensity ranged from 0.65 to 1.85 mm.min-1, and
the variations of volcanic ash layer thickness were 1.00 cm; 2.50 cm; and 5.00 cm. The erosion rate model was
generated from a dimensional analysis accommodating slope, rainfall intensity, flow discharge, and the ratio of
critical and applied boundary shear stress as independent variables. The variable coefficients were obtained by
parameter optimization of experiment data through nonlinear regression analysis. The erosion rate model per-
formance was tested using the Nash-Sutcliffe model Efficiency (NSE), Index of Agreement (IOA), and Root Mean
Square Error (RMSE). The performance of the volcanic ash erosion rate model was proven to be satisfactory with
the NSE[0.75, IOA[0.95, and RMSE values ranging from 0.005 to 0.009. This model has proven applicable better
for volcanic ash erosion than other erosion models which are sandy soil material-based experiments because of the
specific characteristics of volcanic material itself. The volcanic ash erosion model in this study can be implemented
to predict the sediment yield of tephra as part of volcanic disaster mitigation.

Keywords. Lahar; volcanic ash; empirical erosion model; laboratory experiment.

1. Introduction up to 600 km away [5], whilst Kelud eruption in 2014

ejected ash falling as far as 240 km from the volcano [6].
Volcanic eruption is known as multi-hazards to the sur- Although it is rare for large explosive eruptions to have an
rounding environment and society because of their potential instant impact on populated areas, the ejected tephra can
interaction with other phenomena and/or destructive trigger domino effects, whether directly or indirectly, for
mechanisms [1, 2]. Volcanic eruptions generally produce instance, lahars.
three types of material, namely gas, lava, and tephra. As the The lahar term is used to define a high-concentration
tephra fallout and dispersal are the most frequent and rapid-flowing mixture of volcanic material, debris, and
widespread effect that follows an eruption, their spread water [7–9]. The speed of lahar can reach more than
distance depends on several factors, such as the size and 100 km h-1 and flow through the river valley with the
density of particles, plume height, and wind speed and weight of concrete [10]. However, its size, speed, and
direction [3, 4]. St. Helens eruption in 1980 sent ash fallout concentration can vary as the water and sediment mixture
alter. Lahar generally starts small and grows as it erodes
and accumulates more material, and then multiplies in
Supplementary Information The online version contains volume. As it grows, its concentration, weight, and velocity
supplementary material available at https://doi.org/10.1007/s12046-
024-02456-5. rise as well and flow downstream by gravity. Therefore, this
flow transports a significant hazard to downstream
*For correspondence
219 Page 2 of 16 Sådhanå (2024)49:219

infrastructure with its impacts and inundation effects. The process of lahar generation. However, the study of volcanic
damages range from the collapse of buildings and bridges to material movement on a slope is still inadequate because
the complete burial of structures. There are three major cat- observing these phenomena on the field after an eruption is
egories of lahars by origin: (1) lahars that are formed by the dangerous and highly challenging. Leavesley et al (1989)
direct and immediate result of eruptions through crater lakes conducted a field erosion experiment on a 20 cm tephra
breaking, snow or ice melting, and heavy rains falling during layer at Mount St. Helens slope four months after the 1980
or immediately after an eruption on unstable loose volcanic great eruption. Teramoto et al (2006) examined for seven
material; (2) lahars that commonly occur shortly after an months the effect of 5 cm volcanic ash on the rainfall-
eruption due to the collapse of a natural lake which is created runoff process at the Sakurajima Volcano slope. Even
in river that is blocked by new volcanic products; and (3) though those field experiments can depict the actual char-
lahars that are not related to concurrent volcanic activity, acteristics of topography, soil layer, and rainfall, the vari-
occurring by mobilization of loose tephra during heavy rain ation of slope, tephra layer, and rainfall highly depends on
or meltwater on steep slopes of volcanoes [10]. In tropical the nature of the site condition.
regions, lahars are primarily triggered by rain falling on Some scholars overcome these problems by using the
unconsolidated volcanic material, typically pyroclastic den- experimental model test in the laboratory [11, 30–33].
sity current (PDC) and/or tephra fallout deposits [11], usually Jones et al (2017) simulated the effects of the 10 cm vol-
during a short period after an eruption occurred. canic ash layer from Kelud and Chaitén tephra on rainfall-
In the early months of the post-eruption period, the lahars runoff properties within an area of 0.30 9 0.30 m2 with a
frequently occur due to airborne tephra, such as volcanic slope of 28.2 under several antecedent rainfall scenarios.
ash. The existence of volcanic material on the slope can Williams et al (2021) observed the influence of
drastically change hydrological conditions and sediment 0.58–1.17 mm.min-1 rainfall on tephra fall loading 10 cm
transport in watersheds near the eruption source [12, 13]. thick, placed on a 0.41 9 0.20 m2 container with a slope of
Therefore, after the eruption, the flood pattern will change 15. Duhita et al (2021) studied the effects of a 10 cm
significantly, and lahars will easily occur, even with light Merapi volcanic ash deposit on infiltration using a 1.50 9
rainfall. However, the time and scale of lahars vary based 0.75 m2 plot with an adjusted slope of 20 to 30 under 1.94
on eruptions and watershed conditions. This variability is mm.min-1 rainfall. However, those studies still had some
driven by a set of local factors including the grain size limitations, in the variability of parameters and scale.
distribution and spatial distribution of volcanic material Although they revealed a strong correlation between the
thickness, topography, and rainfall intensity [11]. Several existence of tephra on the rise of sediment yield, the amount
studies have examined the post-eruption lahar events, of tephra erosion itself has not been formulated yet. Thus,
which mainly focused on the rainfall-runoff relationship this study aims to obtain an empirical formula for predicting
and sediment transport on the river, and conducted in var- erosion on a slope covered by unconsolidated tephra through
ious volcanoes, including Miyakejima [14, 15], Sakurajima an experimental model test with a more varied range of
[16–18], Montserrat [19], St. Helens [13], Kelud [11], and parameters. This study examined the relationship of vol-
Chaiten [11, 20]. According to those studies, the changes in canic ash erosion rate to three driven parameters, namely
flood patterns are due to sudden changes in land cover. As slope, rainfall intensity, and volcanic ash thickness. Three
the volcano region is covered by the new volcanic material, slope gradients were selected to represent gentle (14.1%),
the vegetation is lost and the infiltration capacity and the mild (26.8%), and steep (46.6%) slopes. Meanwhile, the
slope roughness are reduced [21]. Those changes led to the rainfall intensity ranged from 0.65 to 1.85 mm.min-1, and
increase of overland flow and its velocity so that the vol- the variations of volcanic ash layer thickness were 1.00 cm;
canic material could be easily transported downstream. 2.50 cm; and 5.00 cm. Hopefully, through this study, the
The initiation of lahars on a slope may occur through erosion rate of volcanic material can be predicted to mitigate
shallow landslide mechanisms [22–25] and/or rill erosion the possible damages due to the lahar flood.
[26, 27]. The landslide blocks are generated when an
unstable volcanic material layer on a slope saturated due to
rainfall turns into a flow (fluidized) and moves together as a 2. Material and method
mixture of volcanic material and water. Meanwhile, the
erosion mechanism of volcanic material is preceded by the 2.1 Experiment set up
formation of gullies on slopes as the surface flow is con-
centrated. These gullies will gather and create a gully A laboratory-scale experimental model test with a rainfall
network that eventually enters the river valley. Based on simulator in the Lahar Laboratory of Sabo Experimental
several studies, the formation of lahar in the Merapi area is Station was utilized to examine the erosion mechanism of
generally dominated by the mechanism of rill erosion the unconsolidated volcanic material layer on a particular
[28, 29], especially in the earlier period of post-eruption. slope. Located in the Yogyakarta Special Region, this
Therefore, in this period it is important to understand the laboratory facility belongs to the Ministry of Public Works
erosion mechanism of volcanic material as an initiation and Housing of Indonesia. The experimental model test was
Sådhanå (2024)49:219 Page 3 of 16 219

conducted by constructing a rigid container box of 3.00 m demplot area. This uniformity (Cu ) was calculated using the
9 0.75 m 9 0.50 m to set up the erosion demonstration plot Christiansen formula (1943) [41, 42]:
(demplot). The demplot inclination could be adjusted using Pn 
a chain pulley at the top of the demplot section. The 1 jx xj
Cu ¼ 100 ð2Þ
demplot slope was selected in four variations, namely nx
8 (14.05%), 15 (26.79%), and 25 (46.63%). The range of
slope was chosen to ensure the sediment movement in the where Cu is the coefficient of uniformity (%); x is the
experiment occurred by erosion mechanism, since for sandy measured rainfall (mm hour-1); x is the average of
soil, for slopes of more than 30 the sediment movement is measured rainfall (mm hour-1); and n is the number of
dominated by slope failures, while on slopes of less than 3 measurements. Based on ASTM (2019) regarding the
sand particles tend to settle. The water level in the runoff rainfall simulator for land erosion models, the minimum
tank was measured every 5 or 10 minutes after the runoff value of the rainfall uniformity coefficient is 70% [43].
started. The sediment trap was replaced every 2 to 15 From the calibration results, it is found that the rainfall
minutes and dried oven before weighing. The rainfall produced by valve SV4 gave Cu lessthan70% for all rainfall
simulator provided rainfall coverage over the entire dem- intensity, whilst valve SV5 dan SV6 had C u more than 70%
plot area. The demplot layout and the measurement and for rainfall intensity ranging from 0.65 to 2.08 mm.min-1.
observation instruments are presented in figure 1. Thus, the rainfall in the model test would only be carried
out within this range of values. However, during the
experiment, it was found that the slope failure soon
2.2 Rainfall simulator calibration occurred at rainfall greater than 1.85 mm.min-1. Since this
study focused on the erosion of volcanic ash, the rainfall
Rainfall simulator calibration includes calibrating raindrop range counted in this experiment was only between 0.65 to
diameter and rainfall uniformity distribution produced by the 1.85 mm.min-1.
rainfall simulator. The rainfall intensity in each calibration
was constant through time, with several variations of rainfall
intensity tested ranging from 0.30 to 1.23 mm.min-1. This 2.3 Volcanic material properties
calibration was carried out in an area of 3.00 m 9 2.00 m
which would be used in the experiment. Rainfall for the The material bed of the demplot consisted of two layers to
experiments was generated by a set of nozzles 7.5 m high represent unconsolidated tephra as the top layer and com-
above the demplot base allowing raindrops to reach terminal pacted sandy material as the subsoil layer. For the experi-
velocity before impact. ments, the tephra layer thickness was varied, namely 1 cm;
The calibration method for raindrop diameter was con- 2.5 cm; and 5 cm, while the subsoil layer thickness was
ducted by the Flour Method [34–36] to ensure the raindrop 20 cm for all scenarios. The topsoil was taken from the
size similar to natural rainfall. Many scholars have for- upper reaches of the Woro River (78 35’ 47‘‘ S, 1108 28’
mulated the function of the relationship between rainfall 10’’ E), after a small-scale eruption of Mt. Merapi on March
intensity and its diameter in the form of a power function 3rd, 2020, while the material for the subsoil layer originated
[37–40] as follows: from the slope of Mt. Merapi in Gendol River Watershed
(78 35’ 1‘‘ S, 1108 26’ 59’’ E). The material properties were
D50 ¼ y:iz ð1Þ tested using laboratory standard test methods. The material
in the experiment resembles the natural conditions in Mt.
where D50 is the average raindrop diameter (mm); i is
Merapi. The mass and bulk density of volcanic ash is
rainfall intensity (mm hour-1); y is coefficient (hour); and z
2.62 kg m-3 and 1.65 kg m-3. The water content of volcanic
is the dimensionless power function constant. The values of
material in the experiment is between 6-8%. The volcanic
y and z are found diverse in different cases because of the
material grain size of the tephra and subsoil layer is shown
variability of rainfall characteristics in each research loca-
in table 1.
tion. This study used van Dijk formula developed in West
Java, Indonesia, with the value of y = 1.15 and z = 0.211
[38]. The process of calibration and comparison between
2.4 Data collecting and analysis
the natural raindrops size and the raindrops produced by the
rainfall simulator is shown in figure 2. The calibration Data collected from the experiment includes rainfall, run-
results that the raindrop diameters from valves SV4 and off, and erosion amount, which were the average of data
SV5 were between 0.81–0.86 mm, while the valve SV6 measured every 5 to 15 minutes depending on their inten-
produced raindrops with diameter of about 2.75–2.86 mm. sity and volume. The water temperature used in the
Hence, only valve SV6 could produce raindrops relatively experiment was between 26C to 28C, while the air
similar to natural rainfall. humidity during the experiment was recorded between 64%
Rainfall uniformity calibration was carried out by mea- to 87%. The data measured from the experiment are sum-
suring rainfall at the grid of 50 cm intervals within the marized in table 2. Data collected from the experiment
219 Page 4 of 16 Sådhanå (2024)49:219

Figure 1. Schematic view of the experiment setup.

were analyzed through variance analysis and nonlinear Efficiency (NSE), Index of Agreement (IOA), and Root
regression analysis. A two-sample t-test analysis was used Mean Square Error (RMSE). NSE is commonly used in
to compare the erosion rates (qs ) to parameter runoff (q), hydrology-related modeling whose model performance is
slope (S0 ), and rainfall intensity (i). The least significant determined based on the relative magnitude of the residual
difference (LSD) method in the probability of 0.05 was variance compared to the measured data variance. The NSE
used to distinguish the statistical difference between dif- value is considered good if it is more than 0.50 and very
ferent treatments. good if it is more than 0.75, however, its ideal value is 1.
The performance of the proposed volcanic ash erosion The NSE value is calculated using the following equation
model was evaluated by using the Nash-Sutcliffe model [44]:
Sådhanå (2024)49:219 Page 5 of 16 219

Figure 2. The coefficients b, c, and d of the erosion model.

Table 1. Control factors and their levels. precision. The ideal RMSE value is close to 0, and the
formula is expressed below:
Layer d10 d30 d50 d60 d90 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Pn 2
Tephra 0.023 0.085 0.112 0.124 0.200 i¼1 ðOi  Pi Þ
RMSE ¼ ð5Þ
Subsoil 0.030 0.108 0.220 0.300 1.180 N
Oi is the observed value at time i, Pi is predicted value at
Pn time i, O is the average of observed value, and N is the
ðOi  Pi Þ2 number of data.
NSE ¼ 1  Pi¼1
n  2 ð3Þ
To demonstrate the significance of the erosion model in
i¼1 Oi  O
this study compared to other erosion models, we also cal-
IOA is a statistical measure of the correlation of pre- culated the tephra erosion rate from our data experiment
dicted and observed values, which ideally has a value of 1. using the Universal Soil-Loss Equation (USLE) model for
If the value is more than 0.50, it means that the model barely sandy soil [46], Zhang et al sandy soil erosion model
performance is good. Its performance is considered great if on a steep slope [47], Ali et al process-based soil erosion
the value is more than 0.75. The IOA value is calculated model [48], and Aksoy et al bare soil with rill erosion
using the following formula [45]: model [49]. Those four models were chosen because of the
Pn similarity of sand soil properties and the variety of
ðOi  Pi Þ2 parameter ranges used in those experiments. The compar-
IOA ¼ 1  Pn  i¼1    ð4Þ
 Pi  O  þ  O i  O  2 ison of the tephra erosion model experiment and its
i¼1
parameter ranges to other erosion models is presented in
Meanwhile, RMSE shows the difference between the table 3.
predicted value and the observed value to perform model
 219

Table 2. Data measured from the experiment.

Page 6 of 16

S0 = 14.05% S0 = 26.79% S0 = 46.63%

Measured Measured Measured

q x 10-5 i qs 9 10-3 q 9 10-5 i qs 9 10-3 q 9 10-5 i qs 9 10-3
Tva (cm) (m3.s-1.m-1) (mm.min-1) (kg.s-1.m-1) (m3.s-1.m-1) (mm.min-1) (kg.s-1.m-1) (m3.s-1.m-1) (mm.min-1) (kg.s-1.m-1)
1.0 2.57 1.85 0.80 2.14 1.85 11.80 2.85 1.60 27.56
1.0 3.10 1.75 4.05 2.73 1.85 21.78 2.36 1.60 20.92
1.0 4.66 1.85 15.52 2.66 1.85 8.12 0.84 1.20 4.18
1.0 6.28 1.85 26.46 2.88 1.75 12.08 0.59 1.30 1.27
1.0 6.67 1.65 28.02 2.31 1.75 14.02 0.58 1.45 0.42
1.0 1.10 1.18 5.90
1.0 0.57 1.10 3.60
2.5 2.63 1.55 2.99 1.70 0.68 2.21 4.08 1.48 59.50
2.5 2.91 1.55 9.16 2.11 0.68 22.78 5.56 1.45 92.65
2.5 5.57 1.55 23.86 2.22 0.68 19.88 4.87 1.45 82.45
2.5 5.22 1.75 25.77 2.03 1.50 6.06
2.5 1.67 1.80 16.18 3.36 1.45 20.47
2.5 3.48 1.45 20.66
2.5 1.34 1.45 9.50
2.5 1.97 1.50 27.18
5.0 3.43 1.68 1.88 2.26 1.53 27.40 5.73 1.68 88.04
5.0 2.63 1.74 3.12 6.16 1.60 40.66 5.46 1.70 79.41
5.0 2.22 1.83 6.55 5.35 1.60 27.79 6.94 1.70 93.83
5.0 5.17 1.65 18.99 3.71 1.60 29.39 2.31 0.78 27.49
5.0 6.05 1.73 10.31 3.27 1.45 25.65 2.44 0.80 35.73
5.0 6.59 1.80 25.88 4.02 1.55 23.52 2.13 0.75 37.05
5.0 2.31 1.75 25.77 0.91 0.68 21.76
5.0 3.54 1.80 16.18 0.69 0.65 12.41
5.0 1.86 1.10 7.53 0.77 0.65 25.57
5.0 2.82 0.69 13.16 0.74 0.65 21.58
5.0 2.81 1.00 17.24
Sådhanå (2024)49:219
Table 3. The comparison of the tephra erosion model experiment and parameter ranges to other erosion models.
Sådhanå (2024)49:219

Kılınç and
Description This study Richardson (1973) Zhang et al (2011) Ali et al. (2012) Aksoy et al. (2013)
Experiment scale Laboratory with 3 m 9 0.75 flume Laboratory with 6.5 m 9 1.36 m flume
Laboratory with Laboratory with 5 Laboratory with 3
4.57 m 9 1.52 m m 9 0,4 m flume m 9 0.5 m flume
flume
Soil source Woro tributary, Merapi Volcano, Indonesia Istambul, Türkiye
Foothills Campus, Yongding riverbed, Wageningen, the
Colorado State Beijing, China Netherlands
University, USA
Soil type Volcanic ash, fine sand Non-volcanic, fine to medium sand
Non-volcanic, Non-volcanic, Non-volcanic,
sandy soil medium to very medium to very
coarse sands coarse sands
Grain size Uniform sand, d50 = 0.10 mm Uniform sand, d50 = 0.15; 0.45 mm
90% of sand (0.06- Uniform sand, d50 Uniform sand, d50
2.00 mm); 10% = 0.10; 0.22; = 0.233; 0.536;
of clay and silt 0.41; 0.69; 1.16 0.719; 1.022 mm
(\0.06 mm) mm
Slope 14.1%; 26.8%; 46.6% 5%; 10%; 15%; 20%
5.7%; 10%; 15%; 8.7%; 17.4%; 5.2%; 8.7%;
20% 30%; 40% 25.9%; 34.2 %; 13.2%; 17.6 %
42.3%
Rainfall 0.65-1.85 mm.min-1 0.75; 1.08; 1.42; 1.75 mm.min-1
0.53; 0.95; 1.55; - -
1.95 mm.min-1
Runoff 0.04-10.95 n.a.
Page 7 of 16

cm3.s-1.cm-1 0.26-1.26 6.6-52.6 0.7-20.7

cm3.s-1.cm-1 cm3.s-1.cm-1 cm3.s-1.cm-1
219
219 Page 8 of 16 Sådhanå (2024)49:219

3. Result and discussion the erosion rate function of the slope covered by uncon-
solidated volcanic ash can be expressed as the following
3.1 Developing volcanic ash erosion empirical function:
model  
sc
qs ¼ f i; q; m; q; S0 ; X r ; ; g ð14Þ
The erosion rate equation model on slopes has been s0
developed by many scholars for years. One sheet erosion
model that has been popularly used is the Universal Soil- where qs is erosion rate of volcanic ash (kg s-1 m-1); q is
Loss Equation (USLE) model, which applies several prac- flow discharge (m3 s-1 m-1); i is rainfall intensity (m s-1);
Xr is slope length (m); q is the mass density of water (kg m-
tical coefficients to represent soil types ( Kb), vegetation 3
); m is the kinematic viscosity of water (m2 s-1); S0 is slope
b and land preparation ( P)
cover or crops ( C), b factors as
gradient; sc is critical shear stress (Pa), and s0 is applied
follows [50]: shear stress (Pa).
By implementing Buckingham p-theorem theory, an
qt ¼ 1; 7  108  S0 1;66 q2;035 Kb Cb Pb ð6Þ
erosion rate formula can be constructed by transforming
For sandy soils without vegetation, the USLE model in those variables in Eq. (14) into non-dimensional parame-
Eq. (6) can be simplified as follows [46]: ters. Therefore, the non-dimension erosion rate parameter
can be defined into four dimensionless parameters as seen
qt ¼ 2; 55  107  S0 1;66 q2;035 ð7Þ below:
The developed erosion models recently have increasingly  
qs iX r q sc
varied with diverse approaches. Similar to the USLE model ¼ f S0 ; ; ; ð15Þ
qv m m s0
form, Wang et al (2015) adjusted the exponents of erosion
parameters to apply on Loessial soil from dryland farming Based on Eq. (15), the erosion rate of volcanic ash by an
at the Loess Plateau, Yangling, China and developed this overland flow can be written as follows:
formula:
q c iX d  sc
e
r
qt ¼ 67; 68q1;2 S0 0;98 ð8Þ qs ¼ aqvS0 b 1 ð16Þ
m m s0
Using the linear regression method, Lei et al (2001) The dimensional analysis has transformed the erosion
formulated the erosion rate model as a function of slope and rate dimensionless parameter into the function of the slope
overland flow below: geometry (S0), rainfall dimensionless parameter (iXm r ), flow
qt ¼ 0; 3109 þ 0; 01718S0 þ 0; 12703Q ð9Þ dimensionless parameter (qm ), and the volcanic ash proper-
ties dimensionless parameter (ss0c Þ. This formula represents
Zhang et al (2011) and Ali et al (2012) developed an better the parameters involved in the erosion process than
erosion model by considering, not only slope and overland the earlier empiric models, which commonly define erosion
parameters but also the diameter of soil particles: rate partly as a function of parameters S0 and q, in which
0;345 the value of exponent b ranges between 0.98 to 2.89, while
Zhang etal equation : qt ¼ 2382; 32q1;269 S1;637
0 d50
the value of c is between 0.899 to 2.628. In this study, the
ð10Þ values of constants a, b, c, d, and e were obtained from the
0;5
laboratory experiment using a rainfall simulator by mea-
Ali et al equation : qt ¼ 0; 17  106 Q1;16 S2;89
0 d50 ð11Þ suring the erosion rate of volcanic ash on the demplot and
Meanwhile, Aksoy et al (2013) and Feng et al (2020) adjusting it to Eq. (16).
added the rainfall parameter to their model:
0;194
Aksoy et al equation : qs ¼ 0; 036q0;899 S1;146
0 r 0;073 d50 3.2 Determining the erosion model constants
ð12Þ from experimental data
From the volcanic ash erosion experiment, 64 data sets
Feng et al equation: qs ¼ 332775.i.Q.Sf ð13Þ
were obtained. Initially, the experimental data were pro-
The erosion of volcanic ash on a slope by overland flow cessed through outlier and significance checking, the data
involves a high variety of variables. Based on previous selection process is given in Supplementary Material. Four
studies on the erosion mechanism, the rate of soil particles data were identified as outliers and excluded from further
transported by the overland flow on uncovered land is analysis, thus only 60 data sets were used in the generation
defined as a function of parameters related to slope of the model. The t-test method assuming equal variance
geometry, rainfall intensity, runoff, and physical properties was used to find out the significance of parameters S0, i, and
of soil and fluid [47–52]. In line with this, it is assumed that q, as independent variables, against qs , as a dependent
Sådhanå (2024)49:219 Page 9 of 16 219

Table 4. The t-test result of laboratory experiment data.

So q (m3/s/m) i (m/s) qs (kg/s/m)

Mean 0.304 3:13  105 2:37  105 2:35  102
2 10
Variance 1:58  10 3:13  10 4:42  1011 5:06  104
Observations 60 60 60 60
Pooled Variance 8:17  103 2:53  104 2:53  104
Hypothesized Mean Difference 0 0 0
df (degree of freedom)* 118 118 118
t Stat 17.01 -8.09 -8.09
P(T\=t) one-tail 7:71  1034 2:99  1013 2:95  1013
t Critical one-tail 1.66 1.66 1.66
*degree of freedom (df) parameter So, q, and i (n1=60) regarding parameter qs (n2=60, df=n1?n2-2)

variable. Table 4 shows the results of the significance test q  0:474  1:139
1:572
1:371 iX r sc
resulting in the three parameters significantly affecting the qs ¼ 10:353qvS0 1
erosion rate of volcanic ash indicated by the P values for S0, m m s0
i; and q less than 0.05. Sequentially, the P values for S0, i; ð18Þ
and q are 7:71  1034 ; 2:99  1013 and 2:95  1013 . For 5 cm volcanic ash thickness:
From the P value, it can be found that the slope param-
eter is relatively more significant to erosion rate than q  0:306  1:139
1:769
1:191 iX r sc
rainfall intensity and runoff. It indicates that slope variation qs ¼ 10:353qvS0 1
m m s0
will greatly influence erosion rate. The slope gradient has a
stronger impact on sediment transport capacity than runoff ð19Þ
since the tangential component of the gravity force The three formula shows that the coefficients b, c, and d
increases with the slope gradient (48, 53). Meanwhile, the are changing by the variation of volcanic ash thickness. To
rainfall and runoff parameters have nearly the same P simplify the application of the formula for different vol-
values, possibly indicating the correlation of both parame- canic ash thickness, the coefficient values b, c, and d are
ters, as runoff is excess rainfall. plotted in a graph regarding the thickness of volcanic ash
To obtain the constants of the volcanic ash erosion rate (T va ) as shown in figure 2. It shows that the variation in the
model, optimization was carried out using the Excel Solver values of the three coefficients has a strong correlation with
with the GRG non-linear regression analysis method. In the the thickness of volcanic ash, where the relationship fol-
beginning, optimization of the model gave unsatisfactory lows the power function as the following equation:
results due to the high variability of parameters. Therefore,
to overcome this challenge, the optimization was conducted b ¼ 1:288T va 0:202 ð20Þ
by classifying the data based on the volcanic ash thickness.
The analysis results show that the coefficients a and e tend c ¼ 1:482T va 0:123 ð21Þ
to have constant values for any volcanic ash thickness,
namely 10.353 and 1.139. Meanwhile, the value of the d ¼ 0:589T va 0:366 ð22Þ
exponent b rises with the increase of volcanic ash thickness
where T va is volcanic ash thickness (cm). With Eq. (20),
values ranging from 1.280 to 1.769. Whilst, the constant c
(21), and (22), the value of each coefficient for any varia-
ranges from 1.191–1.458 and tends to decrease with
tion of volcanic ash thickness can be calculated.
increasing volcanic ash thickness, and so is the value of the
We identified that the coefficients a and e are correlated
constant d which is negative for all models with a range
with the physical properties of water (q, v) and volcanic ash
from -0.560 to -0.306. The complete formulas for volcanic
(ss0c ) which are the components of moving material in the
ash erosion rate are presented below:
For 1 cm volcanic ash thickness: erosion process, where water acts as the transport and
volcanic ash as the carried load. As the properties of both
q  0:560  1:139 materials were constant throughout the experiment, so as
1:459 iX r sc
qs ¼ 10:353qvS0 1:280 1 the values of coefficients a and e. Therefore, the values of
m m s0
both are not influenced by the increase in layer thickness.
ð17Þ Nevertheless, further studies on more geographically dis-
For 2.5 cm volcanic ash thickness: persed volcanoes are needed to ensure the consistency of
results for the case of different volcanic materials and
219 Page 10 of 16 Sådhanå (2024)49:219

Figure 3. The relationship between the measured runoff and the ratio of rainfall-runoff.

" #
varied temperature regions as the viscosity of fluid will be q c iX c iX d iX c  sc
e
b r r r
changed by it. qs ¼ aqvS0 1
m m m m s0
Meanwhile, the value of coefficient b, which is the
exponent of the parameter S0 , increases with the thickness
of the volcanic ash layer. However, since the value of S0  c  cþd  e
b q iX r sc
must be less than 0, the increase of b as a matter of fact will qs ¼ aqvS0 1 ð23Þ
iX r m s0
reduce the amount of erosion rate. It is found that when the 
volcanic ash is wet its characteristics are changed and The ratio of iXqr represents the rainfall-runoff rela-
becomes sticky as clay. It is suspected that this is due to tionship on the slope covered by volcanic ash and the
 
flocculation between volcanic ash particles which are exponents of the parameter iXm r in Eq. (23) are approxi-
generally dominated by very fine material. In a thin vol- mately 0.90. To ensure the rainfall-runoff relationship, the

canic ash layer, the bonding between particles is weaker q
measured runoff and the ratio of rainfall-runoff iX r are
than the thicker one. But then again, even though the
influence of slope on erosion decreases, the potency of the plotted as shown in figure 3. The ratio of rainfall-runoff for
volcanic ash layer to move through the slope failure slope covered by 1 cm, 2.5 cm, and 5 cm volcanic ash
mechanism will increase, which depends on the infiltration respectively were 0.08-0.38 (av. 0.31), 0.18-0.77 (av. 0.55),
and slope parameters as well. and 0.21-0.82 (av. 0.51).
Interestingly, coefficient c and d have opposite values The increase in volcanic ash thickness resulted in a
which decrease as the volcanic ash layer thickness increase. higher ratio of rainfall-runoff. This finding is in line with
It indicates that the influence of parameter runoff and several previous studies. Through an experiment at Saku-
rainfall on volcanic ash erosion rate declines as its layer rajima Volcano slope, Teramoto et al (2006) found that
becomes thicker. It is known that runoff occurs whenever surface runoff increased about twice after an area covered
rainfall intensity exceeds the infiltration capacity of the soil, with 5 cm of volcanic ash. Additionally, the ratio of surface
therefore we assume that there is a correlation between runoff and the total amount of rainfall after volcanic ash
   spreading was greater than 0.15 with an average of 0.22.
parameter qm and iXm r which is related to the rainfall-
runoff relationship. Considering their correlation, Eq. (16) Another research at Mount Unzen slope also found that the
can be modified into the following equations: ratio of total overland flow and rainfall increased after and
before volcanic ash spreading around 0.156 and 0.269 [54].
Sådhanå (2024)49:219 Page 11 of 16 219

Figure 4. Scatter plot of the observed versus the predicted erosion rate with: (a) 1.00 cm; (b) 2.50 cm; and (c) 5.00 cm volcanic ash
layer.

Whilst, a 20 cm volcanic ash deposit at St. Helens slope erosion value of the model is highly related to the measured
caused the ratio of rainfall-runoff greater than 0.36 with an erosion value.
average of 0.58 [55]. Although many studies have revealed Figure 4 shows that Eq. (17) gives an overpredicting
the increase of rainfall-runoff ratio on a slope covered by result which is approximately ?0.0017 kg/s/m greater than
volcanic ash, further experiments on a wider range of measured data, while Eq. (18) and Eq. (19) provide
volcanic material properties, rainfall intensity, and slopes underpredicting which is lower than measured data, around
should be conducted to determine the appropriate runoff -0.0085 kg/s/m and -0.0068 kg/s/m. The differences in
coefficient for land covered by volcanic ash. model predictions regarding erosion rate measurement are
due to the influence of the subsoil layer, which is more
porous, on the infiltration process. The influence of the
3.3 Model performance testing subsoil infiltration process can be seen from its decreasing
effect with the increasing volcanic ash thickness. It is
Scatter plots of the observed data versus the predicted identified that there is a more significant slowdown in the
erosion rate from the formulas are presented in figure 4 to infiltration rate in thicker volcanic ash layers which causes
demonstrate the relative position of the regression model to
an increase in surface flow so that the erosion rate measured
the ideal line (1:1), the closer the model to the ideal line,
in conditions with a thicker layer of volcanic ash becomes
the more accurate the model is. The correlation value R2 of
higher [56]. The infiltration rate in the volcanic ash layer is
Eq. (17), (18), and (19) in sequences are 0.81, 0.92, and
low influenced by the formation of a crust on its surface due
0.93 as seen in figure 4. The R2 shows that the predicted to the impact force and compaction of raindrops falling on
219 Page 12 of 16 Sådhanå (2024)49:219

Table 5. The NSE, IOA, and RMSE values.

Volcanic Ash Thickness Proposed Erosion Model NSE IOA RMSE

q1:459 iX r 0:560  1:139
1.00 cm 0.77 0.99 0.005
qs ¼ 10:353qvS0 1:280 m m 1  ss0c
2.50 cm q1:371 iX 0:474  1:139 0.92 0.99 0.008
qs ¼ 10:353qvS0 1:572 m m
r
1  ss0c
5.00 cm q1:191 iX r 0:306  1:139 0.93 0.99 0.009
qs ¼ 10:353qvS0 1:769 m m 1  ss0c

the surface. Micro craters and peaks form on the surface of the proposed model, the erosion rate model developed by
volcanic ash deposits after the initiation of rain which is Kılınç and Richardson (1973), Zhang et al (2011), Ali et al
accompanied by the accumulation of very fine material (2012), and Aksoy et al (2016) was tested against other
(micro mass) in the form of one or several layers, with a erosion models. The data used for this analysis was
thickness of several tens to hundreds of micrometers [57]. experimental data in this study. A comparison of the
Such morphological changes lead to the formation of measured qs and predicted qs values from the volcanic ash
structural crusting in micro craters, which consists of a thin erosion rate model against other models is presented in
layer of clay-silt-sized dense volcanic ash material, coated figure 5.
by loose coarse-grained material. It is also stated that the Figure 5 shows the correlation value (R2) of the four
porosity of the surface crust decreases, where the value of models, which tends to be lower than the proposed models
the decrease in porosity depends on the size of the ash in this study. The values were less than 0.90, with the
particles. highest correlation value obtained by the model of Zhang
However, the results of the performance analysis of the et al (2011) of 0.85 and the lowest correlation value
volcanic ash erosion model on the slopes show satisfactory produced by Ali et al (2012) of 0.26. This indicates that
performance based on the NSE, IOA, and RMSE values as the correlation level of the erosion rate model developed
shown in table 5. The highest NSE value of 0.93 was in this study was more suitable for volcanic ash material
obtained for the erosion model on a slope covered with than other models. Although the other models were
5.00 cm of volcanic ash, while the lowest was 0.77 for a compiled based on erosion experiments of sandy soil
slope erosion model covered with 1.00 cm of volcanic ash. materials, their sources were not from volcanic material.
The NSE value for slopes covered with volcanic ash was The material used in the experiment of previous models
1.00 cm lower than the other two models because the was formed through a long process of rock weathering
residual variance of the model was lower than the variance and had been transported away from its source. This
of the measured data. Besides that, the IOA values of the process caused the shape of the sand to become spherical
three models showed great performance, above 0.95 for all and finer. Meanwhile, volcanic material used in this study
models. Additionally, the RMSE value was between 0.005 was mined from the river close to its source, so its shape
to 0.009. has not been changed and smoothed by the transportation
In the erosion model of slopes covered with 1 cm of process. Moreover, the shape of sand particles produced
volcanic ash, a slightly lower NSI value was found to be by an explosive eruption, such as in Mt. Merapi, is more
related to the condition of the volcanic ash layer being fragmented with sharp edges, so its porosity is lower than
thinner so that infiltration can easier penetrate the subsoil spherical sand particles.
which tends to be more permeable, as mentioned previ- In addition, the other erosion models tend to produce
ously. Additionally, as some parts of the volcanic ash layer values below the measured qs. Zhang et al model gave the
are transported, the raindrop directly falls on the subsoil so predicted qs around 0.74 of the measured qs, the Kılınç and
that it no longer represents the characteristics of slopes Richardson model gave the predicted qs about 0.19 of the
covered with volcanic ash. This resulted in the measured measured qs, even the Ali et al (2012) and Aksoy et al
erosion rate on the slope conditions being lower than the (2016) resulted in the predicted qs less than 0.05 of the
predicted sediment yield. measured qs. Of the four models, Zhang et al model had the
closest correlation and predictive values due to the char-
acteristics of the material used similar to the volcanic ash
3.4 Comparison of proposed model to other material, both in terms of density and gradation. However,
erosion models the lower predicted qs value is possible because they used
uniform and spherical-shaped sand material, which cer-
Erosion rate models on slopes have been developed by tainly affects the structure of the soil layer so that it tends to
many scholars. However, the erosion rate models for vol- be more porous. Moreover, the organic content of soil
canic ash are still very limited. To see the significance of resulting from the weathering process also contributed to
Sådhanå (2024)49:219 Page 13 of 16 219

Figure 5. Comparison of the measured qs and predicted qs values from the proposed model against other erosion models.

causing the material to absorb more water than volcanic ash • the empirical model for volcanic ash erosion is a
material. The material used by Zhang et al (2011) contains function of four dimensionless parameters,

organic material about 21.8 grams.kg-1. f S0 ; iXm r ; qm ; ss0c , related to slope, rainfall intensity,
runoff, and shear stress parameters.
4. Conclusion • the exponent coefficient of each erosion model
parameter changes regarding the volcanic ash layer
Volcanic eruption is a potential multi-hazard that can cause thickness, except for the shear stress parameter which
broad implications and disaster to society and environment remains constant.
near the volcanic area. Lahars are the most frequent sec- • the runoff and rainfall intensity parameters in the
ondary disasters that occurred in the post-eruption period. proposed erosion model represent the rainfall-runoff
Thus, understanding the phenomena of lahar initiation on relationship on the slope covered by volcanic ash,
the slope is essential to mitigate this potential hazard as where the ratio of rainfall-runoff (iXqr ) was greater as
early as possible. Nevertheless, observing the initiation the volcanic ash thickness increased.
process of lahar in the field is challenging and dangerous. • the proposed model performance was proved satisfying
Therefore, conducting a laboratory experiment in a more and well related to the selected predictors, with the
controllable environment and with a wider range of NSE values of the models were 0.77, 0.92, and 0.93
parameters is the best option. respectively; their IOA values were above 0.95, and
Based on laboratory experimental data within an inten- their RMSE values range between 0.005 to 0.009.
sity range of rainfall applied on a slope covered by • the comparison of the proposed model against other
unconsolidated tephra, we found: erosion models gave better predictions since the other
219 Page 14 of 16 Sådhanå (2024)49:219

models produced lower predictions even though sim- References

ilar properties of sand material were used to develop
those empirical equations because of the specific [1] Selva J, Acocella V, Bisson M, Caliro S, Costa A and Della
forming process of volcanic material itself. Seta M et al. 2019 Multiple natural hazards at volcanic
islands: a review for the Ischia volcano (Italy). J Appl.
However, it is important to notice that these proposed Volcanol. 8(1): 5
models are all empirical and valid only for the range of [2] Malawani M N, Lavigne F, Gomez C, Mutaqin B W and
experimental data used in this study. Therefore, advanced Hadmoko D S 2021 Review of local and global impacts of
field experiments are needed to validate the proposed volcanic eruptions and disaster management practices: the
model of volcanic ash erosion rate to confirm that this Indonesian example. Geosciences. 11(3): 109
model can be applied on a watershed scale. As the proposed [3] Barsotti S, Di Rienzo D I, Thordarson T, Björnsson B B and
model is limited based on experiments for a specific vol- Karlsdóttir S 2018 Assessing impact to infrastructures due to
canic material in Mt. Merapi, additional experiments on a tephra fallout from Öræfajökull Volcano (Iceland) by using a
scenario-based approach and a numerical model front. Earth
wider range of volcanic material properties from other
Sci. 6: 196
volcanoes should be conducted as well to ensure the model [4] Bonadonna C, Biass S, Menoni S and Gregg C E 2021
works for other cases. Last but not least, understanding Assessment of risk associated with tephra-related hazards. In
nature’s ability to recover after an eruption is also an Forecasting and Planning for Volcanic Hazards, Risks, and
interesting aspect to study further. It is known that the Disasters, pp. 329–78. Elsevier
impact of the eruption drastically changed the landscape [5] USGS 1980 Ash and Tephra Fall Hazards at Mount St.
due to the loss of vegetation. However, volcanic ash which Helens. USGS. www.usgs.gov/volcanoes/mount-st.-helens
is rich in minerals is a nutrient that fertilizes the soil so that [6] Goode L R, Handley H K, Cronin S J and Abdurrachman M
gradually vegetation will grow back on the slopes. The 2019 Insights into eruption dynamics from the 2014
speed of vegetation recovery is a variable that plays an pyroclastic deposits of Kelut volcano, Java, Indonesia, and
important role in reducing the rate of erosion over time. In implications for future hazards. J. Volcanol. Geotherm. Res.
382: 6–23
the future, it is hoped that this volcanic ash erosion model
[7] Capra Pedol L and Caballero Garcı́a L 2021 Rainfall on
can be implemented to predict the sediment yield of tephra active volcanoes: morphological response and associated
on a slope through the erosion process during a certain processes. In Precipitation, pp. 327–47. Elsevier
period so that tangible countermeasures as part of volcanic [8] Vallance J W and Iverson R M 2015 Lahars and their
disaster mitigation can be conducted to prevent its potential deposits. In The Encyclopedia of Volcanoes, pp. 649–64.
hazard. Elsevier
[9] Mothes P A and Vallance J W 2015 Chapter 6 - Lahars at
cotopaxi and Tungurahua Volcanoes, Ecuador: Highlights
Acknowledgments from Stratigraphy and Observational Records and Related
Downstream Hazards. In Volcanic Hazards, Risks and
The authors gratefully acknowledge the Faculty of Civil Disasters, pp. 141–68. Elsevier
[10] Huff W D and Owen L A 2015 Volcanic landforms and
and Environmental Engineering, Bandung Institute of
hazards. In Reference Module in Earth Systems and Envi-
Technology, Indonesia for the financial support of this ronmental Sciences, p. B9780124095489095129. Elsevier
research. The authors also want to thank the chief and [11] Jones R, Thomas R E, Peakall J and Manville V 2017
workers of Sabo Experimental Station, the Directorate Rainfall-runoff properties of tephra: simulated effects of
General of Water Resources, the Ministry of Public Works, grain-size and antecedent rainfall. Geomorphology. 282:
and Public Housing of Indonesia, for permission to access 39–51
the laboratory facilities during this research. [12] Gran K B, Montgomery D R and Halbur J C 2011 Long-term
elevated post-eruption sedimentation at Mount Pinatubo
Author Contribution This paper is part of the doctoral research of Philippines. Geology. 39(4): 367–370
the first author F.T.Y., which is supervised by I.S., J.N., and U.B.S. [13] Major J J and Mark L E 2006 Peak flow responses to
The conceptualization, data curation, methodology, investigation, and landscape disturbances caused by the cataclysmic 1980
analysis were conducted by F.T.Y. with guidance and critical feed- eruption of Mount St. Helens Washington. Geol. Soc. Am.
back from I.S., J.N., and U.B.S. All the authors contributed to find- Bull. 118(7–8): 938–958
ings, discussion, and conclusions.
[14] Tagata S, Yamakoshi T, Doi Y, Kurihara J, Terada H and
Sakai N 2005 Post-Eruption Characteristics of Rainfall
Declarations
Runoff and Sediment Discharge at the Miyakejima Volcano,
Conflicts of interest The funders had no role in the design of the Japan
study; in the collection, analyses, or interpretation of data; in the [15] Sasahara K, Yamakoshi T and Nishimoto H 2005 The effect
writing of the manuscript, or in the decision to publish the results. of rainfall intensity on sediment transport in a scoria-rich
river on Miyakejima Island, Japan. IAHS-AISH Publication,
pp. 214–21
Sådhanå (2024)49:219 Page 15 of 16 219

[16] Miyata S, Ahemad H, Fujita M, Tsujimoto H and Teratani on tephra-covered slopes with different fine particle content.
T 2016 Temporal change of infiltration characteristics of display
volcanic ash layer and its effect on rainfallrunoff [32] Williams G T, Jenkins S F, Lee D W J and Wee S J 2021
processess How rainfall influences tephra fall loading — an experi-
[17] Teramoto Y, Shimokawa E and Jitousono T 2006 Effects of mental approach. Bull Volcanol. 83(6): 42
volcanic ash on the runoff process in Sakurajima Volcano [33] Duhita A D P, Rahardjo A P and Hairani A 2021 Effect of
[18] Kisa H, Yamakoshi T and Ishizuka T 2014 Impact of short- slope on infiltration capacity and erosion of mount merapi
term temporal changes in volcanic ash fall on rainfall slope materials. J. Civ. Eng. Forum. 7(1): 71–84
threshold for debris flow occurrence in Sakurajima, Japan. [34] Hudson N 1963 Raindrop size distribution in high intensity
Int. J. Erosion Control Eng. 7(3): 75–84 storms. Rhod. J. Agric. Res. 1(1): 6–11
[19] Alexander J, Barclay J, Sušnik J, Loughlin S C, Herd R A [35] Horne M 2017 Design and Construction of a Rainfall
and Darnell A et al. 2010 Sediment-charged flash floods on Simulator for Large-Scale Testing of Erosion Control
Montserrat: the influence of synchronous tephra fall and Practices and Products. Auburn University, Auburn
varying extent of vegetation damage. J. Volcanol. Geotherm. [36] Sousa Júnior S F de, Mendes T A and Siqueira E Q de 2017
Res. 194(4): 127–138 Development and calibration of a rainfall simulator for
[20] Pierson T C, Major J J, Amigo Á and Moreno H 2013 Acute hydrological studies. RBRH. 22(0):
sedimentation response to rainfall following the explosive [37] Laws J O and Parsons D A 1943 The relation of raindrop-
phase of the 2008–2009 eruption of Chaitén volcano. Chile. size to intensity. Trans. AGU. 24(2): 452
Bull Volcanol. 75(5): 723 [38] van Dijk A I J M, Meesters A G C A and Bruijnzeel L A
[21] Major J J, Janda R J and Daag A S 1996 Watershed 2002 Exponential distribution theory and the interpretation
disturbance and lahars on the east side of Mount Pinatubo of splash detachment and transport experiments. Soil Sci.
during the mid-June 1991 eruptions. In: Fire and Mud: Soc. Am. J. 66(5): 1466–1474
Eruptions and Lahars of Mount Pinatubo (eds) Newhall C G [39] Yakubu M L, Yusop Z and Yusof F 2014 The modelled
and Punongbayan R S, University of Washington Press, raindrop size distribution of Skudai, Peninsular Malaysia,
Washington, pp 895–919 using exponential and lognormal distributions. Sci. World J.
[22] Cascini L, Cuomo S, Pastor M and Sorbino G 2010 2014: 1–7
Modeling of rainfall-induced shallow landslides of the [40] Kelkar V N 2021 Size distribution of raindrops -Part I.
flow-type. J. Geotech. Geoenviron. Eng. 136(1): 85–98 MAUSAM. 10(2): 125–136
[23] Iverson R M and LaHusen R G 1989 Dynamic pore-pressure [41] Kara T, Ekmekci E and Apan M 2008 Determining the
fluctuations in rapidly shearing granular materials. Science. uniformity coefficient and water distribution characteristics
246(4931): 796–799 of some sprinklers. Pak. J. Biol. Sci. 11(2): 214–219
[24] Volentik A C M, Connor C B, Connor L J and Bonadonna C [42] Maroufpoor E, Faryabi A, Ghamarnia H and Moshrefi G Y
2009 Aspects of volcanic hazard assessment for the Bataan 2010 Evaluation of uniformity coefficients for sprinkler
nuclear power plant, Luzon Peninsula, Philippines. In: irrigation systems under different field conditions in Kurdis-
Volcanic and tectonic hazard assessment for nuclear facil- tan Province (Northwest of Iran). Soil Water Res. 5(4):
ities (eds) Connor C B, Chapman N A and Connor L J, 139–145
Cambridge University Press, Cambridge, pp 229–256 [43] Standard Test Method for Determination of Rolled Erosion
[25] Zanchetta G, Sulpizio R, Pareschi M T, Leoni F M and Control Product (RECP) Performance in Protecting Hill-
Santacroce R 2004 Characteristics of May 5–6, 1998 slopes from Rainfall-Induced Erosion. 2019. ASTM-D6459-
volcaniclastic debris flows in the Sarno area (Campania, 19, American Society for Testing and Materials
southern Italy): relationships to structural damage and hazard [44] Nash J E and Sutcliffe J V 1970 River flow forecasting
zonation. J. Volcanolo. Geotherm. Res. 133(1–4): 377–393 through conceptual models part I — a discussion of
[26] Collins B D and Dunne T 1986 Erosion of tephra from the principles. J. Hydrol. 10(3): 282–290
1980 eruption of Mount St Helens. Geol. Soc. Am. Bull. [45] Willmott C J 1981 On the validation of models. Phys. Geogr.
5(97): 896–905 2(2): 184–194
[27] Cuomo S, Della Sala M and Novità A 2015 Physically based [46] Kilinc M and Richardson E V 1973 Mechanics of Soil
modelling of soil erosion induced by rainfall in small Erosion from Overland Flow Generated by Simulated
mountain basins. Geomorphology. 243: 106–115 Rainfall. Colorado State University, Fort Collins
[28] de Bélizal E, Lavigne F, Hadmoko D S, Degeai J-P, [47] Zhang G-H, Wang L-L, Tang K-M, Luo R-T and Zhang X C
Dipayana G A and Mutaqin B W et al. 2013 Rain-triggered 2011 Effects of sediment size on transport capacity of
lahars following the 2010 eruption of Merapi volcano, overland flow on steep slopes. Hydrol. Sci. J. 56(7):
Indonesia: a major risk. J. Volcanol.Geotherm. Res. 261: 1289–1299
330–347 [48] Ali M, Sterk G, Seeger M, Boersema M and Peters P 2012
[29] Hairani A and Rahardjo A P 2021 A Theoretical Model for Effect of hydraulic parameters on sediment transport capac-
Debris Flow Initiation by Considering Effect of Hydrody- ity in overland flow over erodible beds. Hydrol. Earth Syst.
namic Force: Sci. 16(2): 591–601
[30] Hasegawa S 1997 Evaluation of rainfall infiltration charac- [49] Aksoy H, Unal N E, Cokgor S, Gedikli A, Yoon J and Koca
teristics in a volcanic ash soil by time domain reflectometry K et al. 2013 Laboratory experiments of sediment transport
method. Hydrol. Earth Syst. Sci. 1(2): 303–312 from bare soil with a rill. Hydrol. Sci. J. 58(7): 1505–1518
[31] Hiraoka M, Imamori N, Shimizu T and Ishida K 2022 [50] Julien P Y 2010 Erosion and sedimentation. 2nd edn.
Laboratory rainfall simulation for surface runoff generation Cambridge University Press, Cambridge, p 371
219 Page 16 of 16 Sådhanå (2024)49:219

[51] Lei T W, Zhang Q, Zhao J and Tang Z 2001 Laboratory the 1980 eruption of Mount St. Helens, Washington, USA.
study of sediment transport capacity in the dynamic process Hydrol. Sci. J.. 34(3): 339–353
of rill erosion. Trans. ASAE. 44(6): [56] Nugroho J, Soekarno I, Yunita F T, Kuntoro A A 2022
[52] Feng Q, Linyao D, Jigen L, Bei S, Honghu L and Jiesheng H Kajian Laju Infiltrasi Pada Lereng Yang Tertutup Abu
et al. 2020 Equations for predicting interrill erosion on steep Vulkanik Berdasarkan Eksperimen Skala Laboratorium.
slopes in the three gorges reservoir China. J. Hydrol. JSDA. 18(1):41–54. https://doi.org/10.32679/jsda.v18i1.744
Hydromech. 68(1): 51–59 [57] Tarasenko I, Bielders C L, Guevara A and Delmelle P 2019
[53] Jing X, Chen Y, Pan C, Yin T, Wang W and Fan X 2019 Surface crusting of volcanic ash deposits under simulated
Erosion failure of a soil slope by heavy rain: laboratory rainfall. Bull Volcanol. 81(5): 30
investigation and modified GA model of soil slope failure.
IJERPH. 16(6): 1075
[54] Ogawa Y, Daimaru H and Shimizu A 2007 Experimental Springer Nature or its licensor (e.g. a society or other partner)
study of post-eruption overland flow and sediment load from holds exclusive rights to this article under a publishing agreement
slopes overlain by pyroclastic-flow deposits, Unzen volcano with the author(s) or other rightsholder(s); author self-archiving of
Japan. Geomorphologie. 13(3): 237–246 the accepted manuscript version of this article is solely governed
[55] Leavesley G H, Lusby G C and Lichty R W 1989 Infiltration by the terms of such publishing agreement and applicable law.
and erosion characteristics of selected tephra deposits from