HTTPS://JOURNALS.
UMS.
AC.
ID/INDEX.
PHP/FG/
ISSN: 0852-0682 | E-ISSN: 2460-3945
Research article Volcanic Pumice Rafts at Sea: Buoyancy and Infiltration with Micro-particles Sawada Sho1,*.
Gomez Christopher1.
Koi Takashi2 Kobe University.
Laboratory of Sediment Hazards and Disaster Risks (SABO) at the Faculty of Oceanology.
Higashinadaku Fukaeminamimachi 5-1-1.
Kobe City.
Japan.
2 Center for Natural Hazards Research.
Hokkaido University Kita 9 Nishi 9 Kita-Ku.
Sapporo.
Hokkaido 060-8589.
Japan Correspondence: 232w315w@stu.
kobe-u.
Citation:
Abstract Sho.
Christopher.
Takashi.
Pumice, characterised by its high vesicularity, often forms pumice rafts as it floats on water.
Although the vast majority of research has focused on understanding the behavior of pumice rafts, studies on the water infiltration properties of pumice remain scarce.
Moreover, the influence of underwater particles, such as microplastics in the ocean, on water infiltration through the pores of pumiceous materials is still unclear.
Therefore, this study investigates the water infiltration properties of pumice mate-rials and their behavior in different aqueous environments through laboratory experiments using pumiceous rocks (En-a, originating from Mount Eniw.
from the 2018 Hokkaido Eastern Iburi earthquake site.
Experiments were conducted in both still and slurry water conditions to examine the effects of particle size and sediment concentration on water infiltration rates.
Results showed that density variations follow a two-phase pattern: a rapid initial increase ( 0.
43 g/cmA and 0.
3931 g/cmA in the first 30s for small and large pumice, respectivel.
followed by a plateau phase ( 0.
022 g/cmA and 0.
0197 g/cmA in the next 60.
Statistical analysis revealed significant differences in infiltration rates based on particle size in still water conditions, with smaller pumices showing more heterogeneous infiltration pathways.
While sediment presence in slurry conditions did not significantly affect overall infiltration rates, it led to more stable data dispersion, particularly in larger samples.
The study also pro-poses a statistical framework for modeling pumice behavior, incorporating parameters such as vesicularity, particle concentration, and temperature.
These findings suggest that current pumice raft models need refinement to account for size-dependent infiltration behaviors and the influence of underwater particles, with implications for understanding the transport and environmental impact of pumice rafts in marine .
Volcanic Pumice Rafts at Sea:
Buoyancy and Infiltration with Micro-particles.
Forum Geografi.
, 53-63.
Article history:
Received: 2 December 2024 Revised: 23 January 2025 Accepted: 27 January 2025 Published: 15 March 2025 Keywords: Pumice.
Pumice Rafts.
Ocean Wastes.
Marine Ecosystems.
Introduction Conceptual Background This chapter outlines the background adequately.
Research problem should be elaborated with short review of literature and previous related publications.
Authors are required to include the citation of the previous published article from reputable journals.
Describe the purposes of the research to complete this part.
Pumice, characterized by its high vesicularity, remains buoyant on water and often forms extensive Aupumice raftsAy, resulting from massive amount of pumice discharged by volcanic eruptions within the ring of fire and from submarine volcanoes.
While the distribution within the ring of fire can lead to the pumice generation on the ocean, the investigations of underwater volcanic events remain scarce.
(Figure 1.
Table .
While they can influence alter ecosystems including coral reefs and mangroves (Ohno et al.
, 2.
, pumice rafts can, in turn, facilitate the spread of marine organisms, such as algae and bacteria, and particles in the In particular, pumice contemporarily reduce the concentration of micro plastics (MP.
on the ocean surface by capturing them through its pores.
However, the MPs entrapped in its porous structures can be reintroduced into the ocean environment due to its fragments, in the long run.
(Pradit et al.
, 2.
The MPs retained within pumice may potentially serve as geological markers of the Anthropocene in the future, and these complexities are likely to be preserved in stratigraphic layers, characterizing the Anthropocene epoch (Majcen et al.
, 2024.
Woodroffe & Stancheva.
Yin et al.
, 2.
During volcanic eruptions, meanwhile, volcanic ash is also ejected alongside pumice.
Vol-canic ash can be produced in the process of pumice rafts, undergoing collision, friction, and chemical dissolution, influenced by ocean currents and wind, due to the exposure to saline seawater containing trace elements (Jutzeler et al.
, 2.
, after which they disperse widely as fine particles in the ocean via thermal convection and density currents, akin to MPs.
(Murch et , 2019, 2.
Copyright: A 2025 by the authors.
Submitted for possible open access publication under the terms and conditions of the Creative Commons Attribution (CC BY) license .
ttps://creativecommons.
org/licenses/by/4.
0/).
Sho et al.
Volcanic eruptions can affect both coastal habitats and human environments.
For instance, a total of 33 cases of volcanic ash-induced engine damage caused by volcanic ash encounters were reported (Christmann et al.
, 2.
Most of these incidents occurred in the Ring of Fire, including Southeast Asia.
New Zealand.
North America, and Europe .
specially around Icelan.
(Christmann et al.
, 2.
, and the primary causes were .
ash particles melting within the engine and Page 53 Forum Geografi, 39.
, 2025.
DOI: 10.
23917/forgeo.
recrystallizing on turbine blades .
clogging of air intake filters by ash, limiting the engineAos air supply, and .
high-temperature ash leading to physical abrasion or corrosion of turbine components (Casadevall, 1993.
Prata and Rose, 2015.
Christmann et al.
, 2.
Research has focused on establishing volcanic ash concentration standards and improving ash cloud prediction models to mitigate these risks .
Casadevall, 1993.
Whitham et al.
, 2012.
Prata and Rose, 2015.
Christmann et al.
, 2.
Also.
Skid resistance on road asphalt and runway concrete is notably reduced by volcanic ash, especially when the ash is rhyolitic or when conditions are wet.
Given that the rafts can affect both coastal habitats and maritime navigation, on the other hand, the prompt necessity for timely tracking and prediction methods has been pointed out (Jutzeler et al.
, 2.
Laboratory flow-tank experiments indicate that pumice movement is constrained by channel width, with wider pathways facilitating smoother flow, while narrow areas cause the accumulation (Nagayama & Izaki, 2.
On the other hand, pumice behaviour in aquatic environments has unique cooling properties.
Slowly cooling underwater pumice retains the buoyancy of its internal bubbles (Von Lichtan et al.
, 2.
Furthermore, the proportion of floating pumice varies depending on the eruption: the 2019 Havre event, for instance, produced over 90% floating pumice, whereas 30 to 70% for the eruption of Volcano-F in 2019, suggesting that eruption dynamics and vent location may play critical roles in the distribution (Yeo et al.
, 2.
Figure 1.
Representative Pumice Rafting Events.
Red Triangles: the Location of Volcanoes (The Source Of Pumice Rafts with a Larger Triangl.
Red Lines: The General Trajectory Paths of Pumice Rafts (After Bryan et al.
, 2.
Note: the Trajectory Data Source for c is Missing.
Table 1.
Representative Pumice Rafting Events Over the Last 200 Years.
Dates Eruptions Region Location References Tambora Indonesia c Oppenheimer .
Krakatau Indonesia c Jokiel and Cox .
Myojinsho Japan c Fiscke et al.
South Sandwich Islands British overseas territory c Sutherland .
Risso et al.
Surtsey Iceland c Gudmundsson and Ingolfsson .
(Unknow.
Tonga cu Bryan .
Tonga c Bryan et al.
Tristan de Cunha British overseas territory c Reagan et al.
Home Reef Tonga c Vaughan et al.
Havre Seamount Kermadec Arc c Jutzeler et al.
Tonga c Redick .
Yeo et al.
Fukutoku-okanoba Japan c Maeno et al.
Yoshida et al.
Sho et al.
Page 54 Forum Geografi, 39.
, 2025.
DOI: 10.
23917/forgeo.
Water Infiltration Properties of Pumice Materials Pumice absorption, with two different infiltration phases, can be simplified as FourierAos Law (Equation .
, mathematically similar to diffusion laws (Manville et al.
, 1.
yaaO = OeyaO O yaA yuIyae yuIya Where ycE O is the heat flux through the cross-sectional area ya, ycN is the temperature difference across the distance yuycu, and ya O is the thermal conductivity.
In the initial phase, water rapidly infiltrates the surface and outer vesicles.
Over time, however, the absorption rate decreases as water penetrates deeper, ultimately reaching saturation.
This delayed absorption can be explained by the differences in water absorption rates between the outer layer and the internal structure (Whitham and Sparks, 1986.
Manville et al.
, 1998, 2.
The time required for pumice to sink .
is highly dependent on pumice size and vesicle characteristics.
Manville et al.
have revealed that the tsink is proportional to the square of the pumiceAos radius, smaller clasts sinking more quickly due to faster water infiltration, while larger pumices remain afloat longer (Manville et al.
, 1.
Resin-based analysis has also shown that interconnected vesicles facilitate water infiltration and air escape, leading to sinking (Whitham and Sparks, 1.
Moreover, temperature also influences high-temperature pumice generates steam, rapidly expelling air from vesicles and causing immediate sinking, with a threshold of temperature varying depending on the density, suggesting that even low-density pumice can sink when sufficiently heated (Whitham and Sparks, 1.
While the vast majority of research has worked on revealing the behaviour of pumice rafts, the investigations into the water infiltration of pumice are very limited.
Also, the effect of underwater particles .
uch as microplastics in the ocea.
on the water infiltration of pumiceous materials has yet to be known.
Therefore, the authors aim to clarify the role of the material density of underwater pumice and particles in the vesicles on the water infiltration and the behaviour of pumice rafts.
The investigation of the water infiltration itself and the effect of particles on the water infiltration can contribute to clarify the sinking process (Whitham and Sparks, 1.
, leading to the modelling of the behaviour of pumice rafts accounting for the infiltration variation.
Research Methods Materials The present research relies on a set of laboratory experiments using the pumiceous rocks of the slope that collapsed during the 2018 Hokkaido Eastern Iburi earthquake in northern Japan.
total, 343 blocks of pumice were used for each repeat of experiments (Figure .
Figure 2.
Distribution of All Samples .
and the Samples from Each Region .
Methods The method of the experiments is divided into two different settings: a first one with still water, the second one was tested in slurry water with sediment, which is expected that the sediment should deter the water infiltration into a pumice (Table .
Sho et al.
Page 55 Forum Geografi, 39.
, 2025.
DOI: 10.
23917/forgeo.
Table 2.
The Samples Used in This Laboratory Experiment, and Details of the Experiment About the Period of Time.
Water Speed Type.
Scale Used in This Experiment, and the Pumice Conditions.
Period of time Sediment concentration And the diameter Scale 30 seconds y3 sets for each 0, 400 .
<125AAm 1g HL-200i First group of experiments: the pumices were first sunk in a water bath at air temperature (AustillAy wate.
using a tweezer to limit the contact between any manipulation instrument and the sample The sediment mass-change due to water infiltration was then measured using a 0.
1 g HL200i scale, with a measurement every 30 seconds for 90-second periods of time.
For the second group of experiments, the same time-step as the first was used in AuslurryAy water with sediment particles .
he diameter is <125AA.
created from the Iburi samples.
Afterwards, the density variation over time in each condition was deduced according to the following procedures (Figure .
Figure 3.
Flow Chart for this Experiment to Gain the Density Variation.
Sho et al.
Page 56 Forum Geografi, 39.
, 2025.
DOI: 10.
23917/forgeo.
Measurement of Initial Density The initial density was determined by the wax coating method (Orosa et al.
, 2.
, and the density for each sample is calculated as Equation 2.
a, yeoyeU ) = yea yeayeI Oe yeayeC yyIyeo Oe yyIyec where ycoyca : mass of dry sample in air, ycoycc : mass of coated specimen in air, ycoyc : mass of water spilled by coated specimen and yuUyc , yuUycy : specific gravity of water and paraffin.
In our contribution, the author dipped the samples into the melted wax to entrapped moisture, and the volume was calculated measuring the mass of water spilled from a silinder filled with water by putting the specimen into it.
Also, this set of experiments was conducted at the end of all the procedures lest pores on the samples should be sealed before the experiments completed.
Calculation of Adimentional Mass Variation to Deduce the Density Variation The adimentional mass variation is introduced as Equation 3.
yeaO .
eiyeO , yeoyeU ) = yea.
eiyeO , yeoyeU ) yea.
a, yeoyeU ) .
where ycoO is adimentional mass [-], varied with time and size.
This adimensionalization serves to deduce the density variation, by using the initial density retrieved from Equation 2, owing to the definition of density and its modification as follows:
cycn , ycuyc ) = yco.
cycn , ycuyc ) ycO.
cuyc ) yco.
, ycuyc ) yco.
cycn , ycuyc ) ycO.
cuyc ) yco.
, ycuyc ) = E0 ycoO .
cycn , ycuyc ) (OA Equation 2 and Equation .
Hence, the density variation along the time and the volume can be expressed as Equation yyO.
eiyeO , yeoyeU ) = yyOya yeaO .
eiyeO , yeoyeU ) .
In this contribution, the authors determined the density variation by using the E0 as the average of each initial density calculated by .
, and the ycoO as the medium of each group of the mass.
Determination of Statistical Significance Parametric/non-parametric methods in statistic are commonly used to compare the means/medians of two/more than three groups.
In this contribution, the data retrieved from each experiment was divided into two depending on the parameter: Large and Small in AoSizeAo based on the median of all the original masses .
), for each experiment: AostillAo and AoslurryAo, where the threshold of the size was determined as the median of the sample group.
After that, the statistical significance was determined by the Mann-Whitney U and F tests for the difference and variance, because normality could not be confirmed by the results of the p-value of the Shapiro-Wilk test and the Q-Q plot (Figure .
In this contribution, the significance level was set to be 5%, and hypothesis of the p-value <0.
05 was concluded as the effect is statistically significant rejecting the null hypothesis.
The same analysis was conducted even for the dataset of initial density in order to investigate the difference in terms of material density depending on the Sho et al.
Page 57 Forum Geografi, 39.
, 2025.
DOI: 10.
23917/forgeo.
Figure 4.
The results of the Shapiro-Wilk test and the Q-Q plot.
Results and Discussion Adimentional Mass Variation The result in Figure 5 showed that the median of adimentional mass in every group of the size and water condition increased during 30 and 90 seconds .
n still water, 1.
58 to 1.
61, 1.
53 to 1.
in small/large group of samples, respectively.
in slurry water, 1.
55 to 1.
58, 1.
56 to 1.
58 in small/large group of samples, respectivel.
In regards with the differences in size, there is a tendency that the medians are inclined to decrease as the size is larger, although the difference is less significant in slurry water (The difference is 0.
, 0.
for still/slurry water, respectivel.
Furthermore, the variance was higher in small group of samples compared to the large .
n still water, 0.
0203 and 0.
, 0.
0180 and 0.
in large/small group samples, respectively.
in slurry water, 0.
0109 and 0.
0194 at 30s, 0.
00810 and 0179 at 90s in large/small group samples, respectivel.
Also, when comparing the same sample Sho et al.
Page 58 Forum Geografi, 39.
, 2025.
DOI: 10.
23917/forgeo.
group in each water condition of still and slurry, the difference for the larger group was more significant than the smaller one .
n large group, 0.
0155 and 0.
, 0.
0125 and 0.
in still/slurry water, respectively.
in smaller samples of group, 0.
0178 and 0.
, 018 and 0.
in still/slurry water, respectively.
Figure 5.
Adimentional mass variation for small/large of samples in still .
and slurry .
On the other hand, the statistical test was conducted using Mann Whitney-U test, as normality was not confirmed for any classes, with the p-value of the Shapiro-Wilk test showing all < 0.
except the two classes of large samples in slurry at 30 and 90 seconds, the Q-Q plots for the two classes indicating the deviation from the diagonal line (Figure .
The results showed that the significant differences were detected mostly in still water for both the changes of time and size, whereas in slurry water, almost no significant differences during the time of 30s to 90s was detected except for large group (Table 3.
Table .
In addition to that, there were no significant differences detected in water condition.
Moreover, the results of F test showed that there were significant differences of the variances in size, except in slurry water at 30s.
Also, the significant difference of a variance in water condition was detected only in large samples of group in 90s.
Table 3.
The results of statistical significance under the significance level of 5% for each group of size, region and water condition for the third round of experiment.
CND refers to as the abbreviation of condition.
Also, *, **, *** signs were used for different significance levels .
<0.
05, p<0.
01, p<0.
001, respectivel.
Sho et al.
CND1
CND2
Group Still Size Still Size Small Large Still Still Time Small Large Slurry Slurry Slurry Slurry Size Size Time Small Large Large Large Small Small Water Condition Water Condition Water Condition Water Condition P-Value Type of Tests Mann-Whitney U Mann-Whitney U Mann-Whitney U Mann-Whitney U Mann-Whitney U Mann-Whitney U Mann-Whitney U Mann-Whitney U Mann-Whitney U Mann-Whitney U Mann-Whitney U Mann-Whitney U Page 59 Forum Geografi, 39.
, 2025.
DOI: 10.
23917/forgeo.
Table 4.
The results of statistical significance under the significance level of 5% for each group of size, region and water condition for the third round of experiment.
CND refers to as the abbreviation of condition.
Also, *, **, *** signs were used for different significance levels .
<0.
05, p<0.
01, p<0.
001, respectivel.
CND1
CND2
Group Still Size Still Size Slurry Slurry Size Size Large Large Water Condition Water Condition P-Value Type of Tests *** *** F test F test F test F test F test F test Initial Density Regularity among Samples and Density Variation The result of Figure 6 showed that the initial density of pumice samples: the initial density was determined as 0.
734, as the median of a whole group of samples, based on the result that no significant differences in size were detected .
-value was 0.
Figure 6.
The initial density of the samples.
Equation 5 enabled us to gain the density variation (Figure .
, showing that there are two phases when pumice infiltrates water.
The density sharply increased ( 0.
43, 0.
3931 g/cm3 in the early 30s for small/large pumic.
and finally reached the platau ( 0.
022, 0.
0197 g/cm3 in the next 60s for small/large pumic.
Figure 7.
Density variation of the median of each group of dataset against time in still .
and the slurry .
3 Discussion The increasing amount of density over time can be considered due to the water infiltration.
However, the behavior varies depending on the effects of the particle size and the water conditions.
Water infiltration rates decrease as particle size increases, with smaller pumices reaching their threshold rapidly between 30 and 90 seconds.
Moreover, greater dispersion in small group of Sho et al.
Page 60 Forum Geografi, 39.
, 2025.
DOI: 10.
23917/forgeo.
samples was revealed as the results of variance analysis, indicating heterogeneous bubble structures that lead to variability in Auinfiltration pathwaysAy.
Such variability may derive from blockages within the bubble networks, as mentioned in previous works .
Yu et al.
, 2023.
Zhang et al.
, which could lead larger pumices to absorb water with slower yet stable infiltration rates due to consistent bubble structures, while the pathways in smaller pumices are likely to be often blocked, leading to inconsistent infiltration behaviour.
Taking the effect of size difference into account, it can be expected that normally, water should be infiltrated into small pumices more rapidly, with the pathways partially deferred due to the blockages, which means that the bubble structures vary depending on the sample, particularly this tendency being likely to be huge in smaller pumices.
Therefore, pumice raft models need to be refined with the consideration of the diverse behaviors associated with different size of clasts and inner structures.
Specifically, for the larger clasts, they could float and sink over time, as it should need a certain time for the infiltration.
on the other hands for the smaller pumices, the clasts could have individual infiltration properties due to different Auinfiltration pathwaysAy.
This experiment detected no significant differences across water conditions.
Nonetheless, slurry condition demonstrated more stable data dispersion.
In particular, the F test confirmed the significance in water condition towards the large group of samples, along with the results of significant difference that almost no significance was detected for size and time in slurry condition, except for larger samples.
One possibility for this is that the data dispersion can be considered due to the dispersion of the Auinner pathways to infiltrateAy, which can be deduced that it is more likely that sediment particles could be either/both excluded in the process of AustillAy water or/and intrude into the pores in large group of clasts, whereas small group of clasts, with their insufficient pore sizes, may not have allowed the particles intrusion.
Either ways, this suggests the influence of particles on the permeability, the current model may not consider this Aublack-boxedAy process, indicating that the model may not be able to be applied when considering the effect of increasing amount of MPs or volcanic ash in the ocean on the infiltration of pumices in a raft.
This study categorised size and sediment concentration .
ater conditio.
as Large/Small and Still/Slurry in a binary framework, with significance difference in density variation and its dispersion.
However, by rewriting Equation 5 into the form of the probability density function yce (Equation .
, the distribution of adimentional mass for each time, size and water condition shown in Figure 5 can be interpreted as Aucross-sectionAy of the probability density function yce2 for adimentional mass variation:
f = f1 (E0 .
1 , x2 .
A ) UI f2 (EO .
1 , y2 .
A ) .
where, yce1 , yce2 : the probability density functions for original density and adimentional density variation, yua0 , yua O : original density, adimentional density variation, ycuycn , ycyc .
cn, yc = 1, 2.
A ): the parameters to determine the original density, adimentional density, including the vesicularity and its coalescence, time, diameter of pumices, particle concentration and particle diameter and.
In addition, yua O can be expressed using Equation 5:
EO .
c, yc.
c, yc.
= ycoO .
c, yc.
Assuming that this study merely treats size and sediment concentration not as AucontinuousAy but as AybinaryAy, the eight distribution maps in Figure 4 can be interpreted as representing conditional probabilities of yce2 for adimentional density under the conditions Equation 7:
yc = 30, 90 yce2 = .
ua O | .
cc = yccyaycaycyciyce , yccycIycoycaycoyco )) yca = ycaycycycnycoyco , ycaycycoycycycyc where, yc: time .
, ycc: the diameter of pumices .
, and yccyaycaycyciyce , yccycIycoycaycoyco is the average particle sizes for the Large/Small groups of pumices used in this study.
Additionally, yca: the volumetric sediment concentration in the fluid, ycaycycycnycoyco , ycaycycoycycycyc corresponds to sediment concentrations under the Still/Slurry water.
The same interpretation can be applied even to Figure 6, which can be considered as the conditional probability derived from the experimental setup for this study.
This suggests that determining yce1 and yce2 through experimental sampling allows for the formulation of the water infiltration of pumice, ultimately providing a statistical model equipped with a wide range of interdisciplinary parameters.
In this contribution, the experiments and results are based on the samples from a limited particle size ranges (<5.
8 c.
and a specific research area (Ena pumice derived from the Eniwa Volcano in Atsuma.
Hokkaid.
However, the research including Sho et al.
Page 61 Forum Geografi, 39.
, 2025.
DOI: 10.
23917/forgeo.
this study, suggest that the water absorption property of pumice vary depending on its size and internal structure .
Mitchell et al.
, 2.
Therefore, further research using pumice with different sizes and internal structures that reflect various eruption processes may be necessary.
Also, the investigation was limited to fixed sediment concentration, c, and particle size, yuo .
ca = 400 , yuo < 125 .
while a significant difference in the data variability was observed between still and slurry conditions.
Future research is required to clarify the effect of different concentration and particle size on water infiltration of pumice Conclusion Thus, the results of the laboratory experiment implied that the density of pumiceous materials could vary depending on time and size .
164, 1.
1271 g/cm3 with 30 seconds.
186, 1.
1468 g/cm3 with 90 seconds, for small/large samples, respectivel.
Also, the behaviour has the potential to influence the behaviour of pumice rafts, illustrating the necessity of the current model improvement according to the density variation due to the water infiltration, taking it into an account that the behaviour and the data dispersion could vary with size and even whether or not they are immersed into the condition filled with particles .
ccording to the experimental results, p-value of F test was at a minimum of 0.
00851 be-tween small and large samples in slurr.
Additionally, the results of the distribution sug-gest the possibility of a cross-disciplinary statistical model for pumice behaviours.
Acknowledgements References