URN:NBN:fi:tsv-oa46363
DOI: 10.11143/46363
GIS and field data based modelling of snow water equivalent in
shrub tundra
YURY DVORNIKOV, ARTEM KHOMUTOV, DAMIR MULLANUROV, KSENIA ERMOKHINA,
ANATOLY GUBARKOV AND MARINA LEIBMAN
Yury Dvornikov, Artem Khomutov, Damir Mullanurov, Ksenia Ermokhina, Ana-
toly Gubarkov & Marina Leibman (2015). GIS and field data based modeling of
snow water equivalent in shrub tundra. Fennia 193: 1, 53–65. ISSN 1798-5617.
An approach for snow water equivalent (SWE) modelling in tundra environments
has been developed for the test area on the Yamal peninsula. Detailed mapping
of snow cover is very important for tundra areas under continuous permafrost
conditions, because the snow cover affects the active layer thickness (ALT) and
the ground temperature, acting as a heat-insulating agent. The information con-
cerning snow cover with specific regime of accumulation can support studies of
ground temperature distribution and other permafrost related aspects. Special
attention has been given to the presence of shrubs and microtopography, specifi-
cally ravines in a modelling approach. The methodology is based on statistical
analysis of snow survey data and on GIS (Geographical Information System)
analysis of a range of parameters: topography, wind, and shrub vegetation. The
topography significantly controls snow cover redistribution. This influence can be
expressed as increase of snow depth on concave and decrease on convex sur-
faces. Specifically, snow depth was related to curvature in the study area with a
correlation of R=0.83. An index is used to distinguish windward and leeward
slopes in order to explain wind redistribution of snow. It is calculated from aspect
data retrieved from a digital elevation model (obtained by field survey). It can be
shown that shrub vegetation can serve as a ‘trap’ for wind-blown snow but is not
a limiting factor for maximum snow depth, since the snow depth can be higher
or lower than shrub height dependent on other factors.
Keywords: snow water equivalent, GIS, modelling, topography, snow survey
Yury Dvornikov, Artem Khomutov, Damir Mullanurov & Ksenia Ermokhina, Earth
Cryosphere Institute, Russian Academy of Sciences, Siberian Branch, Malygin
street 86, Tyumen 625000, Russia, E-mails: ydvornikow@gmail.com, akhomu-
tov@gmail.com, damir.swat@mail.ru, diankina@gmail.com
Marina Leibman, Earth Cryosphere Institute, Russian Academy of Sciences, Si-
berian Branch, Malygin street 86, Tyumen 625000, Russia & Tyumen State Oil
and Gas University, Volodarsky street 38, Tyumen 625000, Russia, E-mail:
moleibman@mail.ru
Anatoly Gubarkov, Tyumen State Oil and Gas University, Volodarsky street 38,
Tyumen 625000, Russia, E-mail: agubarkov@gmail.com
Introduction
The systematic observations of snow cover in Rus-
sia have been initiated by A.Voeykov, who noted
that snow was an important component of the en-
vironment (Voeykov 1889). In permafrost regions,
snow cover plays an important role because it di-
rectly influences the thermal regime of frozen
ground, as it is a natural thermal insulator (Dos-
tovalov & Kudryavtsev 1967). Snow water equiva-
lent (SWE) can be used as a proxy for the integral
characteristics of snow depth and density, which
define the insulating properties of snow. SWE al-
lows the calculation of the land surface snow stor-
54 FENNIA 193: 1 (2015)Yury Dvornikov et al.
age and is also used for the assessment of the wa-
ter regime of rivers and lakes and the activation of
various cryogenic processes (Kuz’min 1960; Gray
& Male 1981; Marsh et al. 1995). In Arctic tundra
strong winds also play a significant role in snow
drift and can be responsible for the increase in
snow density (Kuz’min 1957).
Zhitkov (1913), based on his field observations
in Yamal, Western Siberia, Russia, already noted
that on the flat surfaces of tundra, unrelated to
hilltops, snow accumulates up to 20–30 cm
depth. However, it is documented in Trofimov
(1975) that snow with a depth up to 3–4 meter
can accumulate in depressions, while in most ar-
eas snow depth is usually less than 15 cm. This
small depth is caused by low winter precipitation
in the Arctic (Kuz’min 1960). However, there are
recent observations, which show the increase of
maximal snow storage in Russia, including the
northern part of Western Siberia (Krenke et al.
2000; Kitaev & Kislov 2008).
According to the “Map of snow depth” (Richter
1948), mean values of snow depth do not exceed
30–50 cm for the Yamal peninsula. These data are
consistent with those published by Kopanev (1978),
who noted that the average perennial snow depth
for Yamal does not exceed 30 cm in the second half
of March. More recent studies indicate that the
maximal snow storage for Yamal is about 150 mm
SWE (Kotlyakov 2004) or 50 cm of snow with the
average density of 0.3 g/cm3 (Trofimov 1975).
Previous studies on snow depth and SWE mod-
elling at the local scale have been carried out for
regions with different environmental conditions:
alpine (Purves et al. 1998; Winstral et al. 2002;
Geddes et al. 2005; Clow et al. 2012), and arctic
tundra (König & Sturm 1998; Essery & Pomeroy
2004). Practically the same parameters (inde-
pendent variables) have been used in these stud-
ies for modelling, such as topography, wind, veg-
etation, differing only in the importance assigned
to each variable.
The main purpose of this study is to model the
snow accumulation processes for tundra landscape
(elevation range 56 meters) where woody vegeta-
tion is absent. Three main parameters are consid-
ered: topography, wind, vegetation. The degree of
each control’s weight in the process of snow redis-
tribution is discussed as well. Additionally the use
of GIS enables the spatial and landscape consider-
ation of all the parameters. Analysis is character-
ized by a high level of detail corresponding to mi-
cro-level research (Gray & Male 1981).
Study area
The “Vaskiny Dachi” research station is located in
Central Yamal (70º20’N, 68º51’E) to the south-
east of the Bovanenkovo gas field (Leibman &
Kizyakov 2007). The topography is represented by
stepped plain, dissected by ravines, lake basins,
small rivers and complicated by a complex of
cryogenic processes, mainly cryogenic landslid-
ing (Leibman et al. 2012). The field observations
of snow cover in this area in the beginning of the
20th century have shown that the strong dissec-
tion has some influence on the distribution of
snow depth: the deflation from hilltops and accu-
mulation in depressions, such as the ravines and
foot of the slopes (Zhitkov 1913), under the influ-
ence of strong winds is a very significant compo-
nent of the study area (Trofimov 1975). According
to the Marre-Sale weather station records (www.
rp5.ru), prevailing wind direction for the cold sea-
son of 2012–2013 was from south-east. During
our snow survey in late March 2013 the same pre-
vailing ESE wind direction was observed with a
maximum wind speed of 14 m/sec, and an aver-
age wind speed of 7.4 m/sec. The total sum of pre-
cipitation during the cold season up to the end of
the snow survey amounted to 118 mm SWE.
The study area is located in the bioclimatic
subzone D (CAVM Team 2003), which is charac-
terized by about 9 °C mean July temperature.
Zonal vegetation on gently sloping upland sur-
faces consists of sedges, prostrate and erect dwarf
(<40 cm tall) shrubs and mosses (http://www.
a r c t i c a t l a s . o r g / m a p s / t h e m e s / c p /
cpbzUnit?queryID=D).
Shrub willows (Salix glauca, S. lanata) and
dwarf-birch (Betula nana) are widely distributed
here (Rebristaya & Khitun 1998). Plant commu-
nities with dense shrub layer mostly occupy val-
ley bottoms and gentle hill slopes. Willows
grow up to 2 m high in some places, so the
structure of plant communities may affect snow
distribution significantly.
Modelling was carried out for the local area
represented by a 1.65 km long and 250 m wide
transect (Fig. 1). The transect crosses the main
geomorphologic units of the study area (“Vaskiny
Dachi” research station) and is characterized by
different surface properties and vegetation com-
plexes. The transect includes a site for active layer
depth monitoring (CALM) with an extent of
100x100 m (Fig. 1, CALM site) (Brown et al.
2000, http://www.gwu.edu/~calm/).
FENNIA 193: 1 (2015) 55GIS and field data based modeling of snow water equivalent
Methodology
Snow survey
The field snow survey in the study area was under-
taken between 16th–31st March 2013. Snow depth
was measured using a metal ruler with 1 mm inter-
val and snow density with the snow sampler VS-43.
This device is mostly used in Russian meteorologi-
cal stations and designed for snow density meas-
urements during snow surveys (Slabovich 1985). It
consists of a metal cylinder and scales. At one end
of the cylinder is a ring with cutting elements while
the other end is closed by a lid.
Snow depth was measured at 233 points, in-
cluding 121 points on the CALM site, and density
was measured at 55 points.
Fig. 1. Study area with snow survey locations (March 2013), total 7 km2 (transect, monitoring sites, specific areas). DEM with
25 meter resolution pixels based on a 1:25,000 topographic map as background.
56 FENNIA 193: 1 (2015)Yury Dvornikov et al.
Digital elevation model
The application of a GIS allows the spatial analyses
of snow depth (Evans et al. 1989; Purves et al.
1998). Since topography significantly controls the
redistribution of snow in Arctic landscapes, a de-
tailed DEM is of high importance for SWE model-
ling (Litaor et al. 2008).
In summer 2011, a topographic survey was car-
ried out using a tachymeter TopCon GTS-235 (ac-
curacy of angle measurements 5') to produce a de-
tailed DEM with 5x5 m cell size. Based on this raster
dataset, terrain derivatives were calculated. Slope
aspects within 0–360 degree range were grouped
according to the main directions: N-NE-E-SE-S-SW-
W-NW, as well as a curvature ranging from -4 to +4.
Such a range of curvature corresponds to the dis-
sected topography of the site (Marchand & Killingt-
veit 2001). Curvature shows the convexity and con-
cavity degree of topographic patterns, i.e. of one
raster cell relative to other eight surrounding raster
cells (Zeverbergen & Thorne 1987). Negative curva-
ture values correspond to concave and positive to
convex patterns. The significance of this index in the
studies of snow cover has been noted by many re-
searchers (e.g. Freindlin & Shnyparkov 1985; Gold-
ing 1974; Woo et al. 1983; Sexstone & Fassnacht
2014), from which it has been observed that greater
snow accumulation occurs on concave slopes rather
than on convex hilltops.
A correlation analysis was conducted to reveal
the relationship, if any, of snow depth with the
above topographic parameters. For each point with
measured snow depth, and coverage by the de-
tailed DEM, aspect and curvature values were ex-
tracted according to their spatial referencing.
Wind-induced redistribution of snow
It is well documented that snow is blown away from
windward slopes and accumulated on leeward
slopes (Evans et al. 1989; Winstral et al. 2002; Litaor
et al. 2008). The prevailing wind direction is also
considered in our model. It is selected as south-
eastern according to the Marre-Sale weather station
records. To consider the influence of the wind, an
empirical correction W, which depends on the as-
pect value of each raster cell, is introduced (Eq. 1):
W = 0.5×(cosA-sinA)×K (1)
where W is a correction for the initial value of sim-
ulated snow depth, A the slope aspect, and K a
coefficient which is calculated empirically and de-
pends on the amount of snow precipitation as well
as on wind speed.
Vegetation-induced snow redistribution
Shrub vegetation on the key site is an important
factor for the redistribution of snow masses, being
a ‘trap’ for snow drift as well as a factor decreasing
the wind activity (Benson & Sturm 1993) for both
windward and leeward slopes (Essery & Pomeroy
2004). However, it is also documented that these
trapping processes do not always take place as
snow might be partially blown away from shrub
vegetation complexes (Pomeroy & Gray 1995).
During field work in summer 2011, shrubs (Salix
glauca, S. lanata, Betula nana) were documented in
terms of depth and shrub projective coverage for
each waypoint. For vegetation analysis in the GIS,
shrub contours on the transect were manually digi-
tized using a very high spatial resolution GeoEye-1
image (ID 2009081507005801603031603318)
acquired on 15th August 2009 (NGA license, Uni-
versity Alaska Fairbanks, NASA LCLUC Yamal).
Based on field descriptions from 2011 (unpub-
lished data), shrub height values were assigned to
each polygon.
Snow water equivalent model
A methodology was developed to build a map of
SWE, including the consideration of parameters:
topography, wind, and vegetation (independent
variables). The general scheme of GIS-based mod-
elling is shown in Figure 2. The DEM was convert-
ed to a point vector data model with 5 m spacing.
Curvature and aspect values for each point have
been added to the attribute table. A shrub height
measure was assigned based on the field descrip-
tions to shrub vector polygons retrieved from satel-
lite imagery interpretation. The initial value of
snow depth was calculated based on the estab-
lished dependence between the field measured
snow depth and extracted curvature values. To ob-
tain final values of calculated snow depth, correc-
tions based on the wind and vegetation data have
been made.
The field measurements of snow density were
used in transition from snow depth values to
SWE. If the density was measured layer-by-layer,
the average density for the entire section was
used in the model. Comparison of two parame-
ters (snow depth and SWE) showed a linear de-
FENNIA 193: 1 (2015) 57GIS and field data based modeling of snow water equivalent
pendence (R2 = 0.99), which has been used in
the calculation of SWE from snow depth. After
all, the raster surface was interpolated.
Results and discussion
Snow survey
The measured snow depth varies, depending on
the type of terrain, from 0 to 315 cm. The average
snow depth ranges on sub-horizontal surfaces be-
tween 15 and 30 cm. In depressions, snow depth
exceeds 1 m and can reach several meters. Den-
sity also varies depending on the geomorphology:
from 0.17 g/cm3 on the flat hill tops to 0.67 g/cm3
on concave portions of the slopes. The total results
of the snow survey are summarized in Table 1.
The data presented in Table 1 show that snow
depth is a very irregular parameter for our site
mainly due to relief control. In general, the results
of snow depth and snow density measurements
match that of previous research (Richter 1948;
Trofimov 1975; Kopanev 1978; Kotlyakov 2004).
An anomalously high maximum value of snow
density measured in the field on the hilltop sur-
face (0.67 g/cm3) points to a strong wind influence
that has led to snow cover compression.
Topographic controls
The relation of measured snow depth to curvature ap-
pears to be very strong (Fig. 3), and hence the topog-
raphy has been selected as a main factor for calcula-
tion of the initial snow depth value for the transect.
The curvature has been used before as a second-
ary independent variable in snow modelling (Gold-
Fig. 2. Workflow for the
calculation of Snow Wa-
ter Equivalent (SWE) us-
ing GIS and remote sens-
ing (RS) data.
Snow depth, cm Snow density, g/cm3
Number of measurements 233 55
Minimum value 0 0.17
Maximum value 315 0.67
Average value 29 0.33
Standard deviation 38 0.09
Table 1. The results of the
snow survey in March
2013.
58 FENNIA 193: 1 (2015)Yury Dvornikov et al.
ing 1974; Woo et al. 1983; Winstral et al. 2002,
Clow et al. 2012). A study of the dependence be-
tween snow depth and curvature for the area of
Norway showed a very low correlation (Marchand
& Killingtveit 2001). The authors analyzed a linear
model of the dependence and used a low resolution
100x100 meters DEM, which – as they mentioned
– can adversely affect the degree of correlation. In
contrast, studies of snow cover in Khibiny moun-
tains (Freindlin & Shnyparkov 1985; Kontsevaya et
al. 1989) show the high correlation between these
two parameters and the authors used a linear re-
gression model for snow depth calculation. Kasurak
et al. (2009) and Sexstone and Fassnacht (2014)
also confirm that curvature is well correlated with
snowpack properties even for alpine terrain.
In this paper, this variable is used as a main pa-
rameter. It appears to be obvious that strong inverse
correlation (-0.83) between this parameter and
snow depth was revealed only because the detail of
the used DEM (5x5 m, which corresponds to 1:1000
mapping scale) allows an accurate description of
topography and therefore a more detailed derivative
surface. The DEM with lower resolution used for the
same area results in the substantially lower correla-
tion between these parameters (Fig. 4). Curvature is
very sensitive to the source DEM resolution because
the depth values of the eight surrounding cells are
taken into account in the calculation procedure for
each raster cell. It was also observed in the field that
the influence of microtopography is high too.
The measured snow depth varied up to 10–20 cm
in an area of less than 4–10 m2. However, the acqui-
sition of a DEM with such a detail is a very time-
consuming process. Processed stereo-pairs of air-
borne and satellite images or LIDAR data may serve
as an additional source along with a field topo-
graphic survey.
Modelling the wind-induced redistribution of
snow
Equation (1) allows to index leeward slopes pos-
itively and windward slopes negatively relative
to the prevailing wind direction. In Figure 5,
Fig. 3. The relation of field measured snow depth (cm) to surface curvature.
Fig. 4. The relation of field measured snow depth (cm) to surface curvature derived from a 25 meters resolution DEM based
on 1:25,000 topographic map.
FENNIA 193: 1 (2015) 59GIS and field data based modeling of snow water equivalent
correction values with a different aspect with re-
spect to the prevailing wind direction are sche-
matically shown. The ranges of correction (1)
were established according to the mean meas-
ured snow depth analysis on the slopes of differ-
ent aspect (Fig. 6).
Inventories of wind impact on snow redistri-
bution using GIS have been carried out by other
researchers (Purves et al. 1998; Winstral et al.
2002; Clow et al. 2012) with a similar scheme:
1) the identification of a prevailing wind direc-
tion; and 2) indexing a raster surface according
to the main established pattern; blowing of snow
away from windward slopes and accumulation
on leeward slopes.
Following our approach for a more accurate
analysis of snow redistribution by wind, the
field measurements were subdivided into two
groups: 1) collected within the sites with posi-
tive curvature values (convex sites), and 2) col-
lected within sites with negative curvature val-
ues (concave sites).
Fig. 5. Correction values to the modelled snow depth according vari-
ous slope aspects in relation to prevailing wind direction (see Eq. 1).
The data shown in Figure 6 demonstrate that,
on average, accumulation is up to 20–30 cm
more on the leeward slopes than on windward
slopes from which snow is blown away. In order
to be consistent with these data, the coefficient
K for the correction of W was assumed to be 25.
This value allows for the expansion of the value
of corrections up to ± 20 cm. This wind correc-
tion was entered in the initial calculated snow
depth value according to the graph in the at-
tributive table with aspect values. So far, the de-
pendence between snow depth and curvature –
wind group is not combined into one index be-
cause we are yet unable to account for the pre-
vailing wind direction variations from year to
year. It’s the main reason why the initial value
and wind correction are used separately and
this combination remains to be solved. Such
combinations have been successfully applied
using binary regression trees (Elder et al. 1995;
Erxleben et al. 2002; Winstral et al. 2002;
Molotch et al. 2005).
Fig. 6. The distribution of average
measured snow depth (cm) for
slopes with different aspects:
field dataset for convex sites (cur-
vature > 0) (a); field dataset for
concave sites (curvature < 0) (b).
60 FENNIA 193: 1 (2015)Yury Dvornikov et al.
Modelling the vegetation-induced
redistribution of snow
Shrub willow distribution highly depends on re-
lief. The tallest shrubs (1–1.5 m) occupy concavi-
ties: the valleys of small rivers and slopes of ero-
sion features (Ukraintseva 1997) with the negative
values of curvature. Research carried out in the
tundra of Alaska also showed an increase of snow
density in relation to canopy height, branch diam-
eter (McFadden et al. 2001), as well as shrub den-
sity (Sturm et al. 2001). According to the data,
shrub communities accumulate 27% more snow
than hummocky tundra.
The relationship between snow depth, shrub
presence and shrub height was similar to the one
with the wind: the measurements were compared
depending on surface convexity/concavity. Figure
7 shows the correlation between shrub height and
snow depth versus the convexity/concavity index.
The data were graded according to this index to
study the impact of shrubs on snow specifically at
convex sites, as at concave areas this relationship
is overridden completely by relief as shown in Fig.
7. In the first case, strong dependence of snow
depth on shrub height is shown, in the second
case this dependence is not observed. This is due
to the biologically caused impossibility of shrub
growth over 1.5 m in sites where snow depth ex-
ceeds this value (Leibman 2004).
Figure 8 shows sorted values from the dataset
for convex slopes. This confirms that the shrub
vegetation can serve as a ‘trap’ for snow (Essery &
Pomeroy 2004). Following from this, sites with low
shrubs (less than 15 cm) control the snow depth,
but for higher shrubs (more than 15 cm) only a part
of blown snow is trapped and so they do not con-
trol the snow depth. It could be seen through the
behavior of curves on Figure 8: at lower shrub
height they correlate, but above 15 cm the “shrub
height” curve starts to be markedly higher than the
“snow depth” curve, which generally is going up
simultaneously. This threshold possibly depends
on the total amount of fallen snow.
According to the regression equation (Fig. 7), the
limitation value for calculated snow depth with the
presence of shrubs with certain height is (Eq. 2):
(2)
where X is snow depth, and Y is shrub height. The
value for each modelling point, which are defined
by the shrub vector polygons, was compared to
the initial SWE value and corrected for the wind
redistribution impact. When the limitation value
was higher, then the calculated value was assumed
equal to the limitation value. In this case, the shrub
vegetation was accepted as a limit for snow accu-
mulation according to the field data.
Fig. 7. The relation between
snow depth and shrub height.
The dataset is subdivided into
two groups: points located on
convex places (curvature > 0),
and points located on concave
places (curvature < 0).
� � �� � 10.5�1�1.0562
FENNIA 193: 1 (2015) 61GIS and field data based modeling of snow water equivalent
Results of this study provide the evidence that
shrub height can affect the snow depth, but does
not fully control this parameter. The relationship is
better understood when the topography factor is
excluded. The correlation between these parame-
ters is stronger in this case (e.g. Fig. 7).
The assumption that the vegetation acts as a sec-
ondary factor contradicts data for other regions,
where shrubs – and not topography – have been
established as a main control for snow depth (e.g.
McFadden et al. 2001). For the area, which is
densely dissected by erosion processes, topogra-
phy plays a more important role for the redistribu-
tion of snow. Our analysis shows a linear relation-
ship of the ‘trap’ effect of shrubs with snow depth.
Creation of SWE map
The SWE map is shown in Figure 9. The resolution
of the final raster model is equal to the source
DEM with 5x5 m.
Modelling and mapping of snow cover in this
paper is done at the local scale. Therefore, results
cannot well describe the distribution of snow
Fig. 8. The dependence between snow depth and shrub height (samples located on convex places).
within different areas, such as neighboring flood-
plains. Similar conclusions were suggested by
other researchers, who dealt with small test sites
(Litaor et al. 2008). Environmental features of the
area specify the number of independent variables
describing the distribution of snow in tundra
more accurately (Evans et al. 1989; McFadden et
al. 2001; Sturm et al. 2001; Essery & Pomeroy
2004). Even though independent variables, which
control SWE, do not correlate linearly with snow
distribution (Elder et al. 1995; Molotch et al.
2005), linear regression models have already
been used (Golding 1974). Generally, it is obvi-
ous that all the factors influencing the redistribu-
tion of snow in the site are interdependent and
complex.
Validation of results
The reliability of the model for snow depth distri-
bution was estimated through comparison with
the model based exclusively on field measure-
ments. For this purpose, the results of snow depth
measurements from the entire transect including
Number
of points
Snow depth
(min), cm
Snow depth
(max), cm
Snow depth
(mean), cm R
Volume of snow
(CALM site), m3
Field 0 315 29.12 2683
Model 190 0.94 276 33.06 0.82 2793
Table 2. The comparison of field and calculated snow depths for the transect.
62 FENNIA 193: 1 (2015)Yury Dvornikov et al.
Fig. 9. Snow water equivalent map for central part of the transect.
Fig. 10. Snow depth on the CALM site (a – interpolation of field data, b – calculated values).
CALM site (see Fig. 1) were used, as well as a
detailed DEM based on the ground survey. Field
data from the CALM site deliberately were not
used for the statistical analysis of the snow depth
controls. Snow depth was modelled for each
point on the transect. The analysis showed that
there is a quite high correlation factor between
the two value arrays (R=0.82). Field and mod-
elled datasets for CALM site are shown in Figure
10: the volume of snow was calculated for both
the in situ and the modelled version (Table 2).
This shows that the values of modeled snow
depth and accordingly the values of SWE reflect
the real snow cover distribution well.
Conclusions
Our observations of the snow cover in the study
area showed that the topography of the Central
Yamal tundra caused the extremely uneven distribu-
tion of snow. They also confirmed the importance of
FENNIA 193: 1 (2015) 63GIS and field data based modeling of snow water equivalent
wind drifting from the hilltops and accumulation of
snow in the concavities, as observed by other re-
searchers.
Snow redistribution is due to strong winds and
the lack of significant natural barriers. The data
analysis confirmed a pattern specified by many au-
thors previously: blowing away of snow from the
windward slopes and accumulation on the leeward
slopes in accordance with the prevailing wind di-
rection. To account for this pattern in the general
model proposed here we introduce a correction
based on the slope aspect. This correction allows for
an indexing of the slopes with respect to the prevail-
ing wind direction and facilitates automation.
The impact of shrubs is estimated as ‘traps’ for the
snow or limiting control, depending on the combi-
nation of shrub height and surface shape. The rela-
tionship between shrub height and snow depth is
well expressed when concave landforms do not
prejudge the accumulation of snow exceeding the
height of shrubs. As high shrubs in the key area are
linked to depressions, the limited impact of shrub
communities on the depth of the snow is observed.
However, shrubs retain part of the snow cover, pro-
viding a limiting effect which is used in the model.
Based on field data and identified relationships
between various controls and snow cover depth, a
technique for GIS-based modelling of the snow wa-
ter equivalent was developed, which is applicable
to a particular type of landscape. These studies can
be applicable in the detailed permafrost mapping
and modeling of the number of parameters, such as
active layer or ground temperature on the local
scale, as well as for the hydrological studies of the
tundra environments with the link to the various
cryogenic processes’ activation.
ACKNOWLEDGEMENTS
This research was conducted within the framework of
the Program of Fundamental Research Department of
Earth Sciences No. 12 “The processes in the atmosphere
and cryosphere as factors of environment changes”, the
RFBR grant 13-05-91001-ANF_a, Presidential grant for
scientific schools No. 5582.2012.5 and 3929.2014.5,
as well as International projects CALM and TSP.
REFERENCES
Benson CS & Sturm M 1993. Structure and wind
transport of seasonal snow on the Arctic slope of
Alaska. Annals of Glaciology 18, 261−267.
Brown J, Hinkel KM & Nelson FE 2000. The Circum-
polar active layer monitoring (CALM) program:
Research designs and initial results. Polar Geogra-
phy 24: 3, 165−258.
CAVM Team 2003. Circumpolar Arctic vegetation
map. Conservation of Arctic flora and fauna map
(CAFF) map no. 1. U.S. Fish and Wildlife Service,
Anchorage, AK.
Clow DW, Nanus L, Verdin KL & Schmidt J 2012.
Evaluation of SNODAS snow depth and snow wa-
ter equivalent estimates for the Colorado Rocky
Mountains, USA. Hydrological Processes 26: 17,
2583–2591. http://dx.doi.org/10.1002/hyp.9385.
Dostovalov BN & Kudryavtsev VA 1967. General per-
mafrost. Moscow State University press, Moscow
(in Russian).
Elder K, Michaelsen J & Dozier J 1995. Small basin
modelling of snow water equivalence using bina-
ry regression tree methods. In Tonnessen KA, Wil-
liams MW & Tranter M (eds). Biogeochemistry of
seasonally snow-covered catchments, 129–139.
IAHS Press, Wallingford.
Erxleben J, Elder K & Davis RE 2002. Comparison of
spatial interpolation methods for estimating snow
distribution in the Colorado Rocky Mountains.
Hydrological Processes 16: 18, 3627–3649.
http://dx.doi.org/10.1002/hyp.1239.
Essery RLH & Pomeroy JW 2004. Vegetation and topo-
graphic control of wind−blown snow distributions
in distributed and aggregated simulations for an
Arctic tundra basin. Journal of Hydrometeorology
5: 5, 734−744. http://dx.doi.org/10.1175/1525-
7541(2004)005%3C0735:VATCOW%3E2.0.CO;2.
Evans BM, Walker DA, Benson CS, Nordstrand EA &
Peterson GW 1989. Spatial interrelationships be-
tween terrain, snow distribution and vegetation
patterns at an arctic foothills site in Alaska. Hol-
arctic Ecology 12: 3, 270–278.
Freindlin V & Shnyparkov A 1985. Metodology of the
snow storage calculations for the small alpine
catchments. In Proceedings of glaciological inves-
tigations 53, 96–99. (in Russian).
Geddes CIA, Brown DG & Fagre DB 2005. Topogra-
phy and vegetation as predictors of snow water
equivalent across the alpine treeline ecotone at Lee
Ridge, Glacier National Park, Montana, USA. Arc-
tic Antarctic and Alpine Research 37: 2, 197−205.
h t t p : / / d x . d o i . o r g / 1 0 . 1 6 5 7 / 1 5 2 3 -
0430(2005)037%5B0197:TAVAPO%5D2.0.CO;2.
Golding DL 1974. The correlation of snowpack with
topography and snowmelt runoff on Marmot
Creek Basin, Alberta. Atmosphere 12: 1, 31–38.
Gray DM & Male DH 1981. Handbook of snow: Prin-
ciples, processes, management & use. Ontario
Pergamon Press Canada, Willowdale.
Kasurak A, Kelly R & Brenning A 2009. Mixed-effects
regression for snow distribution modeling. In Pro-
ceedings of the 66th eastern snow conference,
177–195. Niagara-on-the-lake, ON.
Kontsevaya VV, Sokolov VM & Freindlin VS 1989. In-
vestigations of snow accumulation on the differ-
64 FENNIA 193: 1 (2015)Yury Dvornikov et al.
ent topographic patterns in Khibiny mountains. In
Proceedings of the third Soviet Union conference
on avalanches, 232–240. Gidrometeoizdat, Len-
ingrad (in Russian).
Kopanev ID 1978. Snow cover on the territory of
USSR. Gidrometeoizdat, Leningrad (in Russian).
Kitaev LM & Kislov AV 2008. Regional differences of
snow accumulation – contemporary and future
changes (on the example of northern Europe and
northern part of West Siberia). Earth Cryosphere
(Kriosfera Zemli) XII: 2, 98–103. (in Russian).
Kotlyakov VM 2004. Snow cover and glaciers of the
Earth. Nauka, Moscow (In Russian).
Krenke AN, Kitaev LM & Razuvaev VN 2000. Snow
in the CIS and its regions under the global warm-
ing. Earth Cryosphere (Kriosfera Zemli) IV: 4, 97–
106. (in Russian).
Kuz’min PP 1957. Physical properties of snow cover.
Gidrometeoizdat, Leningrad (in Russian).
Kuz’min PP 1960. The formation of snow cover and
methods of snow storage calculation. Gidromete-
oizdat, Leningrad (in Russian).
König M & Sturm M 1998. Mapping snow distribu-
tion in the Alaskan Arctic using aerial photogra-
phy and topographic relationships. Water Re-
sources Research 34: 12, 3471–3483.
http://dx.doi.org/10.1029/98WR02514.
Leibman MO 2004. Mechanisms of formation of cry-
ogenic translational landslides and conditions for
their indication by high willow shrubs on Central
Yamal. Transactions of the Moscow Center of Rus-
sian geographical society. Biogeography 12, 89–
94. RASHN, Moscow (in Russian).
Leibman MO & Kizyakov AI 2007. Cryogenic land-
slides of the Yamal and Yugorsky Peninsulas. Earth
Cryosphere Institute SB RAS, Moscow (In Russian).
Leibman MO, Gubarkov AA & Khomutov AV 2012.
Research station Vaskiny Dachi. Excursion guide-
book. In Proceedings of the 10th international
conference on permafrost. Tyumen Pechatnik,
Salekhard.
Litaor MI, Williams M & Seastedt TR 2008. Topo-
graphic controls on snow distribution, soil mois-
ture, and species diversity of herbaceous alpine
vegetation, Niwot Ridge, Colorado. Journal of
Geophysical Research: Biogeosciences 113: G2.
http://dx.doi.org/10.1029/2007JG000419.
Marchand W-D & Killingtveit Å 2001. Analyses of the
relation between spatial snow distribution and
terrain characteristics. In Hardy J & Frankenstein S
(eds). Proceedings of the 58th Eastern snow con-
ference, 71–84. Ottawa, ON.
Marsh P, Quinton W & Pomeroy JW 1995. Hydro-
logical processes and runoff at the arctic treeline
in northwestern Canada. In Sand K & Killingtveit
Å (eds). Proceedings of 10th international north-
ern basins symposium and workshop, 368–397.
SINTEF, Rep. STF 22. Trondheim, Norway.
McFadden J, Liston GE, Sturm M, Pielke R & Chapin
F 2001. Interactions of shrubs and snow in arctic
tundra: Measurements and models. In Dolman AJ,
Hall AJ, Kavvas ML, Oki T & Pomeroy JW (eds).
Sixth scientific assembly of the International Asso-
ciation of Hydrological Sciences, 317−325.
IAHS−AISH Publication, Maastricht.
Molotch NP, Bales RC, Colee MT & Dozier J 2005.
Estimating the spatial distribution of snow water
equivalent in an alpine basin using binary regres-
sion tree models: The impact of digital elevation
data and independent variable selection. Hydro-
logical Processes 19: 7, 1459–1479.
http://dx.doi.org/10.1002/hyp.5586.
Pomeroy JW & Gray DM 1995. Snow accumulation,
relocation and management. National Hydrology
Research Institute Science Report 7. Environment
Canada, Saskatoon.
Purves RS, Barton JS, Mackaness WA & Sugden DE
1998. The development of a rule based spatial
model of wind transport and deposition of snow.
Annals of Glaciology 26, 196−202.
Rebristaya OV & Khitun OV 1998. Botanical−geo-
graphic specifics of the Central Yamal’s flora. Bo-
tanical Journal (Botanicheskiy Zhurnal) 83: 7,
37–52. (in Russian).
Richter GD 1948. Role of snow cover in the physical−
geographical processes. Academy of Sciences USSR
publishing, Moscow and Leningrad (in Russian).
Sexstone GA & Fassnacht SR 2014. What drives basin
scale spatial variability of snowpack properties in
northern Colorado? The Cryosphere 8: 2, 329–
344. http://dx.doi.org/10.5194/tc-8-329-2014.
Slabovich GI (ed) 1985. Instructions for hydromete-
orological stations and posts. Gidrometeoizdat,
Leningrad (in Russian).
Sturm MJ, Holmgren JP, McFadden GE, Liston FS,
Chapin F & Racine C 2001. Snow − shrub interac-
tions in arctic tundra: A hypothesis with climatic
implications. Journal of Climate 14: 3, 336–344.
h t t p : / / d x . d o i . o r g / 1 0 . 1 1 7 5 / 1 5 2 0 -
0442(2001)014%3C0336:SSIIAT%3E2.0.CO;2.
Trofimov VT (ed) 1975. Yamal Peninsula. Moscow
State University press, Moscow (in Russian).
Ukraintseva NG 1997. Willow tundra in Yamal as the
indicator of salinity of superficial sediments. In
The results of fundamental research of the Earth
cryosphere in Arctic and Subarctic, 173–182.
Nauka, Novosibirsk (in Russian).
Ukraintseva NG 1998. Peculiarities of distribution of
shrubby tundra on Yamal. Transactions of the Mos-
cow Center of Russian Geographical Society. Bio-
geography, 46–53. RASHN, Moscow (in Russian).
Voeykov AI 1889. Snow cover, its impact on the soil,
climate and weather and ways of research. In Pro-
ceedings of Russian Imperial Geographical Soci-
ety XVIII 1. St−Petersburg (in Russian).
Winstral A, Elder K & Davis RE 2002. Spatial snow
modeling of wind-redistributed snow using terrain-
based parameters. Journal of Hydrometeorology 3:
5, 524−538. http://dx.doi.org/10.1175/1525-
7541(2002)003%3C0524:SSMOWR%3E2.0.CO;2.
Woo M, Heron R, Marsh P & Steer P 1983. Compari-
son of weather station snowfall with winter snow
FENNIA 193: 1 (2015) 65GIS and field data based modeling of snow water equivalent
accumulation in high Arctic basins. Atmosphere-
Ocean 21: 3, 312−325.
http://dx.doi.org/10.1080/07055900.1983.9649171.
Zeverbergen LW & Thorne CR 1987. Quantitative
analysis of land surface topography. Earth Surface
Processes and Landforms 12: 1, 47–56.
http://dx.doi.org/10.1002/esp.3290120107.
Zhitkov BM 1913. Yamal Peninsula. In Proceed-
ings of Imperial Russian Geographical Society
XLIX. Stasyulevich MM typography, St−Peters-
burg (in Russian).