Results of applying the georadiolocation method to estimate snow volume in small avalanche assemblages on Sakhalin Island

. In this paper, the authors present the results of snow volume estimation in small avalanche accumulations on Sakhalin Island using GPR. Sakhalin Island using GPR. As a result of using the GPR method, data on snow cover height, snow volume and density on the slope were obtained. The values obtained using GPR correlate well with data from pits and observations from the Nevelsk weather station, which is located at a close altitude. GPR was modified to complete the study. We used metal markers in the control points to clarify the height of layers when linking them to radarograms. GeoScan32 software was used to determine the dielectric permittivity of the snow cover. The application of the georadiolocation method allows to safely and accurately estimate the snow volume in avalanche accumulation.


Introduction
One of the most important characteristics for the assessment of avalanche hazards is the height of the avalanche snow cover.In the normative document regulating the rules for avalanche hazard assessment in engineering surveys, one of the parameters required for determination is the average multiyear value of the maximum ten-day snow cover height [1].
The Sakhalin region is characterised by a high degree of avalanche damage, more than 70% of the territory of Sakhalin is prone to avalanches and from the point of view of avalanche threat to the population and economy of Sakhalin Island.Sakhalin is among the most dangerous territories in Russia [2].Avalanche activity here occurs at altitudes from 0 to 1600 m, from sea coasts to mountain peaks.Avalanche-prone slopes exceeding 20-200 m are the most dangerous for the population and economy of the region.The high degree of manifestation of avalanche processes is caused by the large depth of relief dissection (200-1000 m), large slope gradients (35-45°), large amount of solid precipitation (registered precipitation maximums: for winter in the mountains -1200 mm; for snowfall -350 mm; daily maximum -127 mm; hourly maximum -43 mm in (Chamginsky pass, 03.01.1991), duration of winter season (5-8 months) [3,4].In the region, 64 settlements and more than 500 km of roads and railways are located in the avalanche-prone zone, so the forecast of avalanche danger, in particular the estimation of possible avalanche volumes, is one of the conditions of population safety and sustainable development of the region.
Measurement of snow cover height can be made using various methods.The most commonly used methods are snow surveying, observations on installed snow measuring rails, and stereo surveying.In recent decades, laser scanning and radar have been increasingly used [5][6][7][8].
Snow surveying in avalanche areas is not safe, and it is not possible to install snow measuring rods at avalanche hotspots in the Sakhalin region, given the high degree of area affected by avalanche processes.The alternatives are laser survey, stereo survey, and georadilocation method.
A safe, non-contact remote method of measuring the snow cover height in the avalanche centre, which does not require large material and time costs, is the method of georadiolocation.In addition, there are currently theoretical developments that make it possible to consider the radiolocation method for studying snow cover properties [9][10][11].The relative simplicity and speed of obtaining such information would help to cover a large area of avalanche-prone territories, study the regularities of snow cover distribution, as well as the peculiarities of spatial occurrence of snow layers in avalanche formation, their structure, density, and water content.
In order to develop a georadiolocation method for estimating snow volume in small avalanche piles on Sakhalin Island, it is necessary to carry out numerous tests of the georadiolocation method.To develop a GPR method to estimate snow volume at small avalanche sites on Sakhalin Island, it is necessary to perform numerous GPR application tests under natural conditions of snow cover in different types of landscapes.Thus, the aim of this work is to apply GPR in avalanche hotspots on Sakhalin Island to determine the volume of snow cover in different landscape types.Sakhalin Island to determine the volume of snow accumulation.This paper reflects the results of the first GPR tests to achieve the set objectives.

Description of the research area
The experiment on the application of the GPR method to determine the volume of snow accumulation in avalanche accumulation was conducted on the south-western coast of Sakhalin Island in the settlement area of Chekhov village (Figure 1).
Sakhalin is located in the Far East of Eurasia, washed mainly by the waters of the Sea of Okhotsk and from the west by the waters of the Sea of Japan (Tatar Strait), where the warm Kuroshio Current runs from south to north.The territory of Sakhalin is characterised by a wide variety of natural landscapes.The southern and central parts of the island are mainly occupied by mountain ranges stretching from north to south (the West Sakhalin and East Sakhalin Mountains) and separated by two relatively large river valleys -the Susunai and Tym-Poronai lowlands.The northern part of Sakhalin is a plain with a highly indented surface and residual denudation uplands in the central part with a height of 300-600 m (Vagis and Dagi ridges).
The climate of the study area is characterised by active cyclonic activity, which causes prolonged cloudy weather with abundant precipitation in the form of rain and snow.On Sakhalin, precipitation reaches its maximum values at the Makarov and Dolinsk meteorological stations in the south-eastern part of the island and at meteorological stations on the south-western coast (Nevelsk and Kholmsk) (Figure 1).Based on precipitation observations up to 1989, the highest average annual amount for the cold period is 348 mm in Dolinsk (south-central Sakhalin) and 736 mm for the warm period in Makarov (south-east coast of Sakhalin) [12].Sakhalin is one of the regions with high avalanche danger.The avalanche danger period lasts from four months in the coastal part of South Sakhalin to eight months in the mountainous part of the island.The widespread occurrence of avalanches on Sakhalin is due to a combination of geomorphological and meteorological factors.The climatic features of avalanche formation on Sakhalin are the large amount of solid precipitation (the sum of solid precipitation on Sakhalin averages from 150-300 mm on the coasts and in the valleys to 500-1200 mm in the mountains [12]) and the long duration of the winter season (the formation of stable snow cover on Sakhalin is observed from October in the mountainous part of Middle Sakhalin to the end of November in the coastal part of the southern part of the island [12]).
The geomorphological features of avalanche formation include small relative and absolute heights of avalanche accumulations.Avalanche formation on Sakhalin is possible already on slopes exceeding only 5 m, and the most frequent avalanche damage on Sakhalin is achieved by avalanche formation on slopes up to 200 m in relative height.The area affected by avalanche processes in settlements is high, for example, in the town of Kholmsk with a population of 25,000 people, located 45 kilometres south of the work area (Chekhov village) there are over 2,000 avalanche sites.Also, the motor road from Yasnoye to Zagorny village has more than 2 thousand avalanche sites.Yasnoye -Zagorny village road, which passed through the Chamginsky Pass (East Sakhalin Mountains), was one of the most avalancheprone roads in the country.
The village of Chekhov lies on the coast of the Tatar Strait and in the valleys of the Rudanovsky and Chekhovka Rivers, which originate on the western slopes of the Kamyshovy Range of the West Sakhalin Mountains.The avalanche site where GPR testing was conducted is located on the slopes of an intrusive massif (Figure 1).The avalanche site is covered with herbaceous vegetation dominated by Kuril bamboo.The surface of the avalanche pit is flat.The avalanche pit exceeds 120 m, the area is 1.1 ha, the slope in the avalanche breakaway zone reaches 39°.Formation of avalanches in the last decade occurs there once every 2-3 years (Table 1).On 13 January 2014, an avalanche came down in the neighbouring avalanche-shed, destroying a cinder-block transformer box and stopping in the school yard, damaging windows and a door on the ground floor of the building.

Methodology
Previously, during the two previous winter seasons of 2020-2021 and 2021-2022, work was carried out on the application of the GPR method to study the snow cover, experience was accumulated in determining the snow-ground boundaries, indication of snow layers, application of metal markers in control points for accurate delineation of the media [10], which ultimately allows the authors to use the GPR method as a non-destructive method of studying the snow cover in vast areas and hard-to-reach areas.
For the first experiment on the application of the GPR method to estimate the snow volume on the slope, the most suitable variant of avalanche accumulation was chosen, which has transport accessibility, safe approaches to the summit, and the absence of woody and shrubby vegetation in the avalanche accumulation; GPR is a set of equipment, the main elements of which are: antenna unit, control unit, PC, power supply (power pack), motion sensor (if necessary).Selection of the operating mode, settings, data storage are made by the control unit.Depending on the design and application, the antenna unit (antennas) can be moved on wheels, sledges, mounted on a car using a special mount or fixed point to point [13,14].
When assessing the volume of snow in avalanches, the presence of a person in the danger zone is inadmissible, so the traditional use of GPR in conjunction with a person is not suitable for the experiment.The GPR developed by the authors is an autonomous device based on the GPR "OKO-2" with the antenna unit AB-1700 (1700 MHz).The equipment is installed in a plastic sled on two levels.On the first level there is an antenna unit in the zone of operation of which there are no other equipment and metal fasteners.On the second level, in a separate plastic shockproof box with protection degree IP 54 is the control unit, PC and power supply (power pack) (Figure 2).Due to the fact that field work takes place in rather extreme conditions, the equipment must meet the increased requirements to the conditions of use.The GPR "OKO-2" and the antenna unit were manufactured as field equipment, and the user selects the laptop independently for the tasks, so for the snow cover research the industrial laptop PC iROBO-7000-N420 with the installed software supplied with the GPR was used.
The winter season of 2023 started with the preparation of equipment and determination of field work locations, finalisation of the GPR and testing of its operability in the field work in a specially designated place on the territory of (Special Research Bureau for Automation of Marine Research of the Far Eastern Branch of the Russian Academy of Sciences).The experiment on application of GPR for estimation of snow volume in avalanche accumulation was conducted on 13 February 2023.At the selected avalanche site in the village of Chekhov the descent routes from three points indicated in the figure (Figure 5) were determined.In the upper part of the avalanche site the distance between the tracks is 10 metres, due to the trough funnel-shaped shape of the avalanche site the tracks taper down to one point, the length of each track was 45 metres.
At the beginning of the trace, a metal marker 5 metres from the edge of the cornice was installed in the snow layer at a given depth, with the help of which the heights of the layers were subsequently specified when linking them to the radarogram.After switching on the GPR, the operation modes were set up, the radarogram was probed and checked for the presence of reflected signals from the markers; these markers were also used as control points when conducting studies in avalanche formation.Then the GPR was lowered down the slope on a cable from three points in the avalanche zone, observing the same speed, thus carrying out sounding.All expedition members were in a safe place while working with the GPR on the slope.The length of the tether was 50 m (Figure 3).The record of the trace profile was stored in the PC memory.At the beginning of a new trace, a metal marker was again installed in the snow layer at a given depth, the snow cover height was recorded, and the sounding was repeated.Thus, three surveys were carried out along three traces.Comparison of the measured snow cover height with the actual one was made with the help of the reference pits.Since it is not safe to make a pothole directly in the avalanche pit itself, a place was chosen in the side part of the avalanche pit, where there were no cornices and blow-ups, and slopes did not exceed 15° (Figure 1).Also, the snow cover height on the southwestern coast was estimated using data from the nearest meteorological stations Ilyinsky, Kholmsk and Nevelsk (Table 2).The data from the meteorological stations were obtained from the website rp5.ru.

Results and discussion
Radarograms were obtained as a result of snow measuring works using GPR (Figure 4).On the third trace on the ledge the snow overhangs heavily and therefore the radarogram did not show the depth of sounding in the area of the ledge.On the first trace on the contrary, there is less snow on the edge of the slope than on the top of the avalanche pit and in the avalanche pit, which is why the radarogram shows a decrease.Despite the fact that the radarogram distinguishes snow layers with different densities, this paper estimates the snow volume in avalanche formation without indication of snow layers.To estimate the snow volume, it is necessary to link the layers on the radarogram to the snow cover thickness using the measured heights of the snow thickness at the top of the avalanche and metal markers.Then using the "Amplitude" function in GeoScan32 software, the data with snow thickness power and distance travelled are saved in a separate file.Since the displacement sensor was not used in the study, the sounding was carried out continuously, the GPR was descended uniformly.As a result, due to different descent rates, the number of points on the first profile was greater than on the third profile, but despite this, the snow height was estimated on the basis of high measurement detail: the measurement step was 7 cm on the first profile and 11 cm on the third profile (Table 2).
Using the results obtained, it is possible to find the total dielectric permittivity for the entire snow strata.After linking the layer on the radarogram to the layer in the study area, the values of the velocity and dielectric constant are read by the GeoScan32 programme automatically after setting the layer power.This method of dielectric permittivity determination is called velocity estimation or fitting method.It is performed when there is a time section and a radarogram with the arrival times of reflected waves, and the lithological composition of rocks and layer thicknesses are known from drilling data (applicable to geology).By analogy with geological studies for snow avalanches, drilling is replaced by stratigraphic columns.Having the distance, i.e., the power of the layer, and the time of arrival of the reflected wave, it is possible to calculate the propagation velocity of the electromagnetic wave and, accordingly, the dielectric permittivity for a given layer [15].Thus, the total dielectric permittivity for the entire snow column in the avalanche assembly was as follows ε= 1.9.
According to the formula of Robin G. [16] for dry snow we find the density ( . ): here  . ′ is dielectric constant of dry snow;  . defines density of dry snow, expressed in kg/m 3 .The density of total snow cover was 0.44 g/cm 3 .
The reference pits allow comparison of GPR-derived snow cover height and density values with those obtained by traditional validated methods.The height of the pits is measured using a feeler gauge and the density by weighing a cylinder of known volume filled with snow.
The height of the reference pits was 56 cm (Figure 5), which corresponds to the GPR measured average snow cover height (Table 2).To more correctly compare the GPR snow cover height data, a route snow survey should have been conducted at a nearby safe site, but no such sites were available.
The average snow density in the borehole was 0.44 g/cm 3 , which fully corresponds to the density obtained using GPR and the formula of density by Robin G. [16].where Sav is avalanche catchment area, hs.aver.defines average snow height in avalanche accumulation, obtained as the arithmetic mean value of the snow layer height measured by ground penetrating radar, Vs. is snow volume in the avalanche accumulation.
In this study, the snow volume in the entire avalanche accumulation was estimated.Thus, the avalanche accumulation area is 1.1 ha or 11000 m 2 , average GPR snow cover height -0.5 m, respectively, snow volume on the slope -5500 m 3 .Avalanches of this volume are regularly formed on Sakhalin.Knowing the density, volume and mass of snow, as well as the shape of the avalanche site profile and the stratigraphy of the snow column, it is possible to calculate the dynamic characteristics of a possible avalanche from this avalanche site.
Depending on the task, measurements can be made for the entire avalanche site, for the avalanche breakaway zone, or to determine the power of swells and cornices.
Since the average snow cover height value for the nearest weather station is often used in solving applied tasks, the values of this parameter for the nearest weather stations located on the south-western coast were analysed to compare the snow cover height values obtained by GPR (Figure 1, Table 3).* location of hydrometeorological stations is shown in Figure 1.Table 3 demonstrates that the value of snow height at the nearest HMS Kholmsk (45 km to the south of the study area) is more than 2 times less than the snow cover height obtained by the authors in Chekhov village.This is probably due to the different height of the site.It is known that the higher the terrain, the more precipitation falls there, the lower the air temperature, respectively, the greater the snow cover height.HMS Nevelsk (100 km south of Chekhov village), and located at an altitude of 165.7 m abs, shows a value of snow cover height comparable to the values obtained by the authors.Certainly, at this stage of work it is impossible to recommend using HMS Nevelsk to estimate the snow cover height in Chekhov village, however, based on the first results obtained, we can speak about the prospects of this research direction.

Conclusion
As a result of using the georadiolocation method to estimate snow volume in small avalanche sites on Sakhalin Island.Snow cover height, snow volume and snow density on the slope were obtained.
The snow cover height data obtained by GPR coincides with the measured snowpack height from the pits and is 0.5 and 0.56 metres, respectively.The snow cover height value also correlates with the snow cover height values from HMS Nevelsk, also located on the south-west coast 100 km south of the work area at an elevation of 165.7 m.
For the experiment, the GPR was upgraded to an autonomous device based on GPR OKO-2 with AB 1700 antenna unit.The whole set of equipment was installed in plastic sledge- sleds.Metal markers were used to check the GPR operation and to specify the heights of layers when linking them to the radarogram.
Application of GeoScan32 software product and selection method in determining the boundaries and power of layers on radarograms allows us to obtain the values of dielectric permittivity ε, which was 1.9.The formula-calculated value of the density of the entire snow column coincided completely with the density measured by the traditional method using a cylinder and scales in the pits and amounted to 0.44 g/cm 3 .
The volume of snow in the avalanche accumulation calculated on the basis of the avalanche accumulation area and average snow cover height according to GPR was 5500 m 3 .It can be assumed that the maximum volume of ground avalanche from this avalanche site will be about 5000 m 3 .Avalanches of this volume are formed annually on Sakhalin.
Thus, the use of georadiolocation as an alternative method for avalanche hazard forecasting allows to conduct studies promptly and without human presence in the danger zone.Also, the use of georadiolocation as one of the methods of snow cover research will allow us to approach the solution of other scientific and practical problems of glaciology.

Fig. 1 .
Fig. 1.Avalanche pit where GPR testing was carried out (the figures indicate: 1 -boundary of the avalanche pit, 2 -boundary of the work area, 3 -profile where snow height was measured by GPR, 4location of the pits, 5 -meteorological stations, 6 -location of the work).

E3SFig. 3 .
Fig. 3. GPR descent traces and the place of stratigraphic works, where 1 -points of descent traces start, 2 -place of stratigraphic works.

E3SFig. 5 .
Fig. 5. Pit photo and stratigraphic column of the snow column (experiment site, 13.02.2023).The following formula was used to calculate the snow volume in avalanche accumulation: Savhs.aver.=Vs,

Table 1 .
Parameters of the avalanche pit and avalanches formed in it.

Table 2 .
Snow cover height in avalanche formation obtained by GPR

Table 3 .
Snow height values based on data from meteorological stations (rp5.ru) and snow measurement works (Chekhov village)