Assessment of geo-disaster in the snowmelt season considering climate changes

In this study, we reproduced avalanches and debris flow phenomena using the debris flow analysis tool, iRIC Morpho2DH, and laboratory-test apparatus. In the laboratory tests, we reproduced aggregates of snow, water, and soil that form during avalanches and debris flows using a pot mill turntable. Consequently, the possibility of forming aggregates of snow, water, and soil during debris flow phenomena was analyzed. Evaluation of avalanche and debris flow phenomena during the snowmelt season was conducted by considering the characteristics of these aggregates. Based on the results, the validity of the analysis method for the Nozuka disaster in Japan was examined. Moreover, the risks associated with future avalanches and debris flow phenomena under changing climatic conditions were studied using the Policy Decision making for Future climate change (d4PDF) database.


Introduction
The climate of the Hokkaido region in Japan is characterized by cold temperatures and heavy snowfall compared to that of other regions.The region has been significantly affected by climate changes owing to its strong dependence on snowfall for its water resources [1].
Concerns exist regarding the increased risk of avalanches and debris flows during the snowmelt season as temperature increases and precipitation intensifies in the future.On March 9, 2018, an avalanche occurred near the Nozuka tunnel on Route 236 in Hokkaido due to increasing temperature and unprecedented heavy rainfall in March, causing damage to road facilities and disrupting traffic (figure 1 and figure 2).In addition to the avalanche, it was confirmed that soil sediments, such as debris flow, accumulated over a wide area, referred to as the Nozuka disaster [2].Considering that the accumulated rainfall during that time was 293 mm, it is necessary to investigate the occurrences of avalanches and debris flows.
In this study, we attempted to reproduce avalanche and debris flow phenomena using a debris flow analysis tool, iRIC Morpho2DH, and laboratory-test apparatus.In the laboratory tests, we simulated aggregates of snow, water, and soil that typically form during avalanche and debris flow phenomena using a pot mill turntable.We further explored the possibility of formation of these aggregates during snowmelt season and assessed the avalanche and debris flow phenomena by considering the characteristics of these aggregates.
Based on the results, we examined the validity of the analysis method for the Notsuka disaster and analyzed the potential risks associated with future avalanches and debris flow phenomena under changing climatic conditions using Policy Decision making for Future climate change (d4PDF) database [3].

Overview of disasters during the snowmelt season
The Nozuka mountain pass, where the disaster occurred, is located on the western part of the Hidaka mountains in Hokkaido Japan, and has a terrain prone to avalanches.On March 9, 2018, when the disaster occurred, warm and humid air inflow resulting from a weather front and a low air pressure led to increase in temperatures.Moreover, the accumulated rainfall just before the disaster was 293 mm, with the maximum hourly rainfall recorded at 37 mm (figure 3).It was also confirmed that the snow depth rapidly decreased by 22 cm.Avalanches are caused by heavy rainfall and rapid snow melting.The scale of the avalanche was approximately 60,000 m 3 in total, with approximately 7,000 m 3 deposited on the road (2 m high, 60 m wide, and 60 m long) [2,4].We presumed that a sediment flow (debris flow) coincided with the avalanche event.

Analysis method for avalanches and debris flow
The iRIC used in this study is a numerical simulation tool designed for hydraulic engineering purposes.We conducted analysis using the Morpho2DH solver [4], which is primarily based on one of the debris flow and mudflow models.In this analysis model, tractive force, drag coefficient, and flow bed level can be obtained.
The law of mass conservation for a mixture of water and sediment is expressed by the following equation: where t is time; h is flow depth of debris flow or mud flow; u and v are velocity components in x and y directions, respectively; c* is sediment concentration of soils in stationary sedimentary layer; and E is    erosion rate of flow bed.The relationship between each velocity component and the bed gradient in the flow direction is expressed as equation (2) and equation (3).
where θ, θx,, and θy are the bed gradients in the flow, x, and y directions, respectively; and θe is the equilibrium bed gradient in the flow direction with respect to the vertical average sediment concentration ｃ ̅̅̅ of the water and sediment mixture.
As shown in figure 4, the following relationship can be derived by considering the water and sediment mixture as a two-layer flow, with a laminar region near the bed and turbulent region above it, and assuming that the concentration ｃ ̅̅̅ is constant.
where φs is the internal friction angle (shear resistance angle) of the sediment, and hs/h is the laminar layer thickness ratio.Additionally, the mass conservation law of sediment in the flow is expressed by the following equation: Consequently, the following relationship is obtained from the law of conservation of momentum: where g is gravity, zb is the riverbed level, and P is pressure.Using the hydrostatic pressure approximation, where ρm is the density of sediment, ρ is the density of water, and σ is the density of soil particles.Furthermore, τbx and τby are tractive force components in x and y directions, respectively, and are expressed as follows: fb is the drag coefficient, which is expressed by equation (11) in turbulent region and equation ( 12) in laminar region.(Turbulent region) (Laminar region) where Cmu and  are coefficients related to mudflow resistance [4]; these values are set by the user based on the flow of the mudflow (in this study, fb≈72).kf = 0.16, kd = 0.0828, e is the coefficient of restitution of the particles, and d is the mean grain size of the sediment.The bed level is expressed by the following equation: Based on the above equations, avalanche and sediment flow phenomena were simulated using an analysis tool.Kawamura and Kido studied the variations in flow characteristics by changing snow and soil properties and clarified the effects of various factors on avalanche and debris flow phenomena [5].Herein, we presented the risks of avalanches and sediment flow (snow mudflow) phenomena in the future climate.

Future climate data used and their characteristics
First, the weather characteristics under future climate scenarios were analyzed using the data from d4PDF database.The RCP8.5 scenario is the worst scenario with no mitigation measures taken to address global warming, and the prediction scenario is SST4, which corresponds to the highest sea surface temperature among the SST1.5 to SST4 scenarios.We focused on the most significant changes compared to present climatic conditions.The target periods included the present (September 1950 to August 2011) and future (September 2050 to August 2111) climates.An overview of the scenario is provided in table 1.
The analysis zone (red frame) is shown in figure 1.The temperature changes from March to May (snowmelt season in Japan) for both present and future climates of the region obtained through the analysis, which closely approximate the averages for each 20-year period, are shown in figure 5.The figure shows an increase in the temperature during the snowmelt period from March to May.Additionally, the temperatures became relatively high, indicating that the snowmelt season in the future climate is expected to commence approximately a month earlier than that in present climate.
The total and daily precipitations for three months during the same period are shown in figure 6.In the climate of 1992, the total precipitation for three months was 372 mm; however, in the future climate of 2079, it is projected to increase to 531 mm.For example, the maximum daily precipitation in the present climate is approximately 60 mm, whereas that in the future climate is approximately 70 mm, signifying an increase.Additionally, the number of days with precipitation of 30 mm or more increases  from 3 to 6 times, indicating an increase in the daily precipitation during the snowmelt season in the future climate.In Hokkaido, the incidence rate of landslide disasters during the snowmelt season surpasses that during other seasons.Therefore, it is important to implement countermeasures to mitigate such disasters during the snowmelt season [5].

Weather factors
Temperature, Precipitation, Snowmelt

Aggregates generated during avalanche flow and their evaluation
In this case study, it was observed that an avalanche occurred due to rapid snow melting induced by heavy rain and increasing temperatures, changing the dry snow to wet snow.Considering the mechanism of its occurrence, this phenomenon is similar to that of a snow mudflow.Aggregates (mixture) consisting of tephra, water, and snow are formed within snow mudflow, and the formation of these aggregates has a significant influence on the flow characteristics of snow mudflows.
We reproduced the aggregates formed during the avalanche process and determined that these aggregates affect the avalanches and sediment flow [5].The testing method employed to reproduce the aggregates is outlined as follows: (1) The mass ratio of soil and water was set at 47% and 38%, respectively; (2) each sample was maintained below 0 o C until immediately before the experiment; (3) the volume of water, snow, and soil is adjusted to 20% of the volume of the container.These conditions were set based on the scenarios under which the aggregates were most likely to be formed in preliminary experiments.Soil, snow, and water were sequentially added into a pot mill with a diameter of 100 mm and capacity of 500 mL.The pot mill was then rotated to replicate the flow phenomena caused by an avalanche.(4) The rotation time and speed of the pot mill were determined based on the estimated arrival time of the avalanche of 240 s and the flow distance of approximately 1 km in real field scenarios [5].
First, the existence rate (%); aggregate mass per unit length (maggr./m);and ratios of soil mass (msoil/m), snow mass (msnow/m), and water mass (mwater/m), which are crucial for the analysis, were examined by varying the mean grain size of soil particles (D50 in mm), where the total mass, m = maggre.+msoil+msnow+mwater,with maggre.representing the mass of aggregates, msoil representing the mass of soil particles, msnow representing the mass of snow, and mwater representing the mass of water.Previous studies have investigated the influence of fine particles in volcanic soil (referred to as Komaoka soil) on aggregate formation.In this study, soils obtained from the Nozuka disaster region were used to investigate the possibility of aggregate formation under in-situ conditions (natural water content, wn, is 2.5 and fine content, Fc, is 13.9 %).The particle size distribution and index properties of the Nozuka soil are shown in figure 7 and figure 8.The angle of internal friction () and cohesion (c) are determined through direct shear tests conducted under freeze-thaw and the no-freeze-thaw conditions, following the testing method outlined by the Japanese Geotechnical Society (JGS 0561).
A sample of aggregates simulated in this study is shown in figure 9 (the diameter is approximately 73 mm).The relationship between mean grain size of soils D50 (mm) and the existence rate (%) of aggregates, soil particles, and snow obtained from the above experiments is shown in figure 10.The figure shows that despite variations in soil grain size, each existence rate was approximately 16%, 38%, and 6% for aggregates, soil particle, and snow, respectively.Additionally, no aggregates were formed when the particle size exceeded 3 mm in this study.Based on the results, an analysis using the average abundance of aggregates, soil particles, and snow was conducted.

Analysis method and conditions
Kawamura and Kido performed a back analysis to obtain the sediment area, sediment volume, and flow distance, which were compared for three types of phenomena: avalanche, debris flow, and snow mudflow [5].The results revealed that snow mudflow causes the most severe damage (table 2).As avalanche and debris flows are significantly influenced by topographical factors, our initial focus was on water flow (water flow analysis).Subsequently, an analysis was performed assuming the occurrence of snow mudflow.

Water flow analysis
First, we estimated the starting points where avalanches and debris flow are most likely to occur, based on the characteristics of water flow.The parameters and setting conditions used for the analysis are presented in table 3. The mean grain size of snow was set to 2 mm based on the study by Kamiisi et al [6].
In this analysis, a prescribed rainfall was applied to the area shown in figure 1.The amount of rainfall (the maximum daily precipitation) in future climate is assumed to be constant throughout the duration of the simulation.

Snow mudflow analysis
In the snow mudflow analysis, the sediment flow of soil samples collected at the Nozuka mountain pass and snow was assessed.

Results and discussion
We conducted water flow analysis based on the rainfall in the future climate.A cross-sectional view of typical slopes and the slope angle around Nozuka tunnel is shown in figure 11.The average slope angle was approximately 26° along the red line and approximately 16° along the yellow line.The results of water flow analysis are shown in figure 12, illustrating that water initially flows from two directions, which eventually converges and flows down near the tunnel entrance.The avalanche occurrence point estimated through back analysis based on the real amount of the avalanche-debris sediment is depicted by the red circle on the left side of the figure.This corresponds with the route shown on the left side of the analysis results, which consider the water flow characteristics.Additionally, if avalanches and debris flow occur simultaneously on two routes, as demonstrated in this analysis, the potential for damage may increase.The estimation results of the amount of the avalanche-debris sediment generated from the two main routes based on snow mudflow analysis are shown in figure 13.The left figure shows the avalanchedebris flow situation at 50 s after the start of analysis, whereas the right figure shows the deposition situation at 330 s after the start of the analysis.The slope failure area, depth, and maximum erosion depth calculated in this study are presented in table 5.The results show that the amount of sediment just before the convergence in the red line route was approximately 67,000 m 3 , and that in the yellow line path was approximately 24,000 m 3 , for a total of 95,000 m 3 .This represents an increase of approximately 1.6 times compared to the 60,000 m 3 estimated initially.These findings suggest that with increased precipitation due to the effects of climate change, snow and mudflows from multiple directions may merge, potentially causing extensive damage.

Conclusions
Through a series of analyses, the following conclusions were drawn: (1) In the future climate, the temperatures are projected to increase during the snowmelt season, accompanied by an increase in the number of days with heavy rainfall.This raises concerns about a potential increase in geo-disasters during the snowmelt season.
(2) Simulations of avalanche and debris flows (snow mudflow) were conducted by considering the physical characteristics of aggregates that are formed in the flow process, and disaster risks associated with future climate scenarios were estimated.
(3) The iRIC Morpho2DH, designed for debris flow analysis, proved capable of effectively simulating the snow and mudflow phenomena by appropriately configuring the parameters.

Figure 2 .
Figure 2. Area where disaster occurred (provided by Hokkaido Development Bureau, Japan).

Figure 3 .
Figure 3. Weather information at the time of the disaster (Nozuka Pass TM, Hokkaido).

Figure 5 .
Figure 5. Changes in daily average temperature during snowmelt season in present and future climatic conditions.

Figure 6 .
Figure 6.Changes in daily precipitation during snowmelt season in present and future climatic conditions.

Figure 7 .
Figure 7. Soil particle distribution of the Nozuka soil.

Figure 9 .
Figure 9. Sample of the aggregates of soil, snow, and water simulated in this study.

Figure 8 .
Figure 8. Index properties of the Nozuka soil.

Figure 11 .
Figure 11.Cross-sectional views and slope angle of the slope around the Nozuka Tunnel.

Figure 12 .
Figure 12. Results of water flow analysis.Figure 13. Results of snow mudflow analysis in future climate.

Figure 13 .
Figure 12. Results of water flow analysis.Figure 13. Results of snow mudflow analysis in future climate.

Table 3 .
Parameters and analytical conditions used in this study.