CHAPTER 5 SNOW LOADS - C5-1-

CHAPTER 5 SNOW LOADS

5.1 Procedure for Determining Snow Load on Roof C5-2

5.2 Snow Load on the Ground C5-4

5.2.1 Characteristic snow load on the ground C5-4
5.2.2 Snow depth on the ground C5-8
5.2.3 Equivalent unit weight for ground snow C5-15
5.2.4 Environmental coefficient C5-16

5.3 Ground Snow Load with Accumulation for n Days C5-18

5.3.1 Characteristic ground snow load with accumulation for n days C5-18
5.3.2 Ground snow depth accumulation for n days C5-19
5.3.3 Equivalent unit weight for ground snow with roof snow control C5-20

5.4 Snow Load on the Roof C5-22

5.4.1 Design snow load on the roof C5-22
5.4.2 Shape coefficient C5-22

5.5 Snow Load on the Roof with Control C5-25

5.5.1 Design snow load on the roof with control C5-25
5.5.2 Controlled snow load C5-25

5.6 Partial Snow Load on the Roof C5-27

5.7 Other Snow Loads C5-27

References C5-28
- C5-2 - Commentary on Recommendations for Loads on Buildings

CHAPTER 5 SNOW LOADS

5.1 Procedure for Determining Snow Load on Roof

Because the Japanese islands are surrounded by sea and scattered over a wide range of latitudes,
snow fall mechanisms and snow depth distributions vary from region to region. Hokkaido and
northwestern Honshu are covered with heavy snow during winter. These areas are subjected to the
snowfall mechanism shown in Fig. 5.1.1. However, areas on the Pacific Ocean side sometimes
receive heavy snow at the end of winter due to approaching low pressures in the temperate zone.

Figure 5.1.1 A typical snowfall mechanism in Japan

In spite of the regional climate differences and the two different snowfall mechanisms, the present
recommendation uses unified concepts as much as possible in determining snow loads on roofs, on the
basis of recent studies. A flowchart of the procedure for determining snow loads on roofs is shown in
Fig. 5.1.2. The main portion of the procedure is subdivided into three parts: setting up the design
concept for dealing with roof snow loads, considering snow conditions at the construction site, and
considering snow conditions on the roof itself.
First of all, the designer should set up the design concept for the roof snow load, whether or not an
active control system is to be introduced to reduce the snow load accumulating on the roof. An
appropriate system will reduce the design snow load on the roof. Because there have been many
accidents in which buildings with control apparatuses have failed, the reliability of the control
apparatuses needs to be critically assessed, also taking human error into account. Recent
developments in control system technologies have been remarkable in Japan. This new
recommendation is expected to encourage their further development.
Snow loads on roofs vary as a function of the characteristic snow load on the ground, climate
(including temperature and wind speed during the winter), roof shape, roofing material, and from one
winter to another. Most available snow load data comprises snow depth on the ground. This
recommendation assumes that the snow load on the roof is proportional to that on the ground.
As the second step, snow depth on the ground at the construction site is estimated by a statistical
treatment and by an interpolation technique. The basic load value in this recommendation is defined
as one for the snow depth on the ground with a return period of 100 years. If a building is designed
for another return period, a conversion factor is introduced to modify the value. The snow load on the
 CHAPTER 5 SNOW LOADS - C5-3-

ground is calculated from the snow depth multiplied by an equivalent snow density. The other ways
to estimate ground snow weight using observed precipitation and temperature are also introduced.

Figure 5.1.2 Flowchart of procedure for determining snow load on roof

- C5-4 - Commentary on Recommendations for Loads on Buildings

For the third step, the snow load on the roof is estimated from the snow load on the ground
multiplied by a ground-to-roof conversion factor, that is, a shape coefficient. Because the shape
coefficient varies according roof shape, temperature, snow type (dry or wet), wind direction and wind
speed, it is impossible to determine precisely. Therefore, it is recommended in principle that the shape
coefficient for the roof is estimated by an experimental study with models. However, tables in this
recommendation give values for several typical shapes. For a large roof, even if the shape is simple,
an experimental study is also recommended because the shape coefficient is usually different from
those used for small roofs.

5.2 Snow Load on the Ground

5.2.1 Characteristic snow load on the ground
(1) Estimation of ground snow weights based on the data of snow depth on the ground and equivalent
unit weight for ground snow
Because direct observation of snow loads on roofs are limited, characteristic snow loads on the
ground are used as characteristic values. For ordinary buildings without control systems, the annual
maximum load for the whole season is used as the characteristic snow load S0, which is given by
Eq.(5.2.1) as the product of factors:
S0 = kenv d0 p0 (5.2.1)
where
S0 (kN/m2) : characteristic snow load on the ground used for design when snow load on the roof
is not controlled
kenv : environmental coefficient, as defined in 5.2.4
d0 (m) : characteristic snow depth on the ground when the snow load on the roof is not
controlled, as defined in 5.2.2
3
p0 (kN/m ) : equivalent unit weight for ground snow as defined in 5.2.3

(2) Estimation of ground snow weights based on the data of daily precipitation and daily mean air
temperature
In order to estimate the ground snow weights besides the conventional method using the equivalent
snow density in Chapter 5.2.3, this section presents a new method based on the meteorological data
such as daily precipitation and daily mean air temperature. Three kinds of equations due to Joh &
Sakurai, Kamimura & Uemura and Takahashi etc., are proposed as follows;
1) Joh and Sakurais equation1)
The authors have proposed a practical method to estimate ground snow weights during snow cover
period as defined below;

n "1
Pn = ! Pi (Ti < 2 C) (5.2.2)
i =1
 CHAPTER 5 SNOW LOADS - C5-5-

where
Pi (N/m2) : daily snowfall weight converted from the daily precipitation (mm)
Pn (N/m2) : ground snow weight of the n-th day: accumulated value of Pi from the first day
(i=1) of a period of continuous snow cover to the (n-1) day

Data of Pi should be accumulated when daily mean air temperature Ti () is less than the air
temperature limit , +2, which is used to judge that the data of Pi means the value of snowfall or not.
Here, the daily melting snow weight is neglected for practical application.
Figures 5.2.1(a)~(d) show the examples of the comparison of the estimated value due to Equation
(5.2.2) with the observed value of snow weights at Sapporo, Kamabuchi, Nagaoka and Tokamachi
located in heavy snow regions, respectively. In these four cases, the continuous snow cover periods
are almost three months at Sapporo and five months at other three observatories. It is obvious from
these examples that the proposed equation can approximately trace the increasing process of ground
snow weights from the starting day to the day of peak value.

2) Kamimura and Uemuras equation2)

This equation is expressed as follows;

Sm = Sm"1 + Cb # Pm " Ca # Tm (5.2.3)

"0 (T < 0C)

$
! Ca = #Ca 1 (0C & T and Cb Pm > 0)
$C
% a2 (0C & T and Cb Pm = 0)
#1 (T ' Tl )
%
Cb = $(Tu " T ) /(Tu " Tl ) (Tl < T ' Tu )
! %0
& (Tu < T )
where
Sm (N/m2) : estimated ground snow weight on m-th day after the starting day of calculation
!
Pm (N/m2) : snowfall weight on m-th day which is converted from the daily precipitation (mm)
Tm (C) : daily mean air temperature on m-th day
Ca (N/m2C) : snowmelt factor
Cb : reference coefficient to judge whether Pm means snow or rain
Tu (C) : upper limit temperature
Tl (C) : lower limit temperature

CbPm and CaTm in Equation (5.2.3) means the increasing snow weight and the melting snow

! weight on m-th day, respectively. Figures 5.2.2(a) and (b) show the examples at Nagaoka and
Tokamachi using Equation (5.2.3), in which the estimated values are compared with the observed
values for ground snow weights. It is clearly that this Equation can approximately trace the ground
- C5-6 - Commentary on Recommendations for Loads on Buildings

snow weights over the continuous snow cover period.

The calculation is done to use different and suitable value of snowmelt coefficient Ca at each
observatory, which can be obtained from the data of snow weights etc. Therefore, further
investigation is needed how to search the suitable snowmelt coefficient Ca at other observatory, which
has no data of snow weights.

Fig.5.2.1 Estimation of ground snow weight with Equation (5.2.2)1)

Fig.5.2.2 Estimation of ground snow weight with Equation (5.2.3)2)

 CHAPTER 5 SNOW LOADS - C5-7-

3) Takahashis equation3)
It is assumed that the each days snowfall formed one snow layer on the ground snow. After a
snowfall, the snow consolidates and metamorphoses in proportion to the number of days and air
temperature. Finally, it melts and it is assumed to percolate into the ground. In Equation (5.2.4),
numerator means the estimated snow weight of m-th snow layer and the denominator means the
estimated snow density of m-th snow layer.

m P" m Pc
m dn = (5.2.4)
# min n+ m k

where
m : number of days from the start of continuous snow cover
!
n : number of days from the snowfall of the layer
md n (m) : estimated depth of m-th snow layer
mP (kN/m2) : snowfall weight on m-th day which is converted from the daily precipitation (mm)
Pc (kN/m2) : melting snow weight on m-th day
%'# P = C (T " T ) (T > T )
m c 0 0
Pc = & m
'(0 (T $ T0 )
C (N/m2C) : snowmelt factor
T (C) : daily mean air temperature
!
T0 (C) : reference temperature for snowmelt fixed as -2 C
mk : correction number for the initial condition
2 2
m k = ( m "0 ) ("min )
m 0 (kN/m2) : unit snow weight of a snowfall on the m-th day
% " min (T $ 0C)
! '
m " 0 = & " min + mT ( " max # " min ) 3 (0C < T $ 3C)
'"
( max (3C < T )
min (kN/m2) : minimum unit snow weight of a snowfall
max (kN/m2) : maximum unit snow weight of a snowfall fixed as 4.9kN/m3
!
This equation is one-day step repeating calculation using minimum snow unit snow weight min and
snowmelt factor C as unknown factors. The minimum value of the estimated error between the
estimated daily snow depth and observed daily snow depth leads to the optimum solution for the
transition process of snow depth.
Figures 5.2.3(a)~(f) show the examples at Sapporo, Shinjo, Takada, Toyama and Fukui using
Equation (5.2.4), in which the estimated values are compared with the observed values for ground
snow weights. It is clear that this equation can approximately trace the ground snow weights over the
- C5-8 - Commentary on Recommendations for Loads on Buildings

continuous snow cover period.

As a result, considering the difficulty to decide the suitable value about the regional characteristics
of snowmelt coefficient in advance, Equations (5.2.2) and (5.2.4) are recommended at present.

Fig.5.2.3 Estimation of ground snow weight with Equation (5.2.4)3)

5.2.2 Snow depth on the ground

For a roof snow load without a control system, the basic snow depth on the ground d0 is defined as
the annual maximum value for the whole season with a return period of 100 years. Statistical
properties of a snow depth at 423 meteorological observatory points in Japan were studied to estimate
a value with a long return period4). Five types of probability distribution functions were applied with
two plotting techniques: Hazen and Thomas. Statistical properties in Japan were subdivided into three
groups as shown in Fig. 5.2.4. Finally, it was concluded that statistical properties of the annual
maximum snow depth could not be explained by any single distribution or plotting technique. As a
result, a new method was proposed in which a linear regression analysis was applied to the upper 1/3
of all the data plotted on Gumbel probability paper.
 CHAPTER 5 SNOW LOADS - C5-9-

Figure 5.2.4 Type classification on Gumbel probability paper4)

In this recommendation, statistical values with a return period of 100 years were estimated by the
proposed method for the annual maximum snow depth for the whole season and for the annual
maximum increasing snow depth. Because the Gumbel distribution is given by Eq.(5.2.5), the snow
depth on the ground with a return period of r years (10<r<200) can be estimated by Eq.(5.2.6).

FX(x) = exp{-exp[-a(x-b)]} - ! <x< ! (5.2.5)

where
a: scale parameter
b: position parameter

d0 = b + 1/a ln(r) (5.2.6)

It is usually impossible to obtain meteorological data over a long period at the construction site.
Therefore, this recommendation gives an empirical equation for estimating the snow depth at a
location without any observatories. According to previous studies5), the dominant topographic factors
influencing snow depth on the ground are altitude and sea ratio, which is defined as the ratio of sea
area to total area around the site (see Fig. 5.2.5). In this recommendation, the annual maximum value
of snow depth on the ground is estimated for a given point by:

d = ( altitude) + ( sea ratio) + (5.2.7)

- C5-10 - Commentary on Recommendations for Loads on Buildings

where coefficients for altitude and for sea ratio and constant are given in tables in Appendix of
the original Japanese version. (Appendix 5.2, Table A5.2.4, A5.2.5)

Nishikawa-cho (Yamagata Prefecture)

Fig.5.2.5 Definition of sea ratio5) Fig. 5.2.6 Relation between snow depth
and altitude of observatories5)

Fig.5.2.7 Effectiveness of snow depth estimation based on two parameters5)

Snow depth is almost proportional to the altitude where the site is inland and in a narrow limited
area (Fig. 5.2.6). Although the altitude alone is not sufficient to interpolate in general (Fig. 5.2.7 (a)),
values estimated by Eq.(5.2.7) almost coincide with the observed data, as shown in Fig. 5.2.7 (b). Fig.
5.2.8 (a) ~ (h) shows the distribution of the annual maximum snow weight in Japan. The maps were
created with Kriging estimation which equation is expressed as follows.6)

1 2
" d ( ds ) =
2 N ( ds )
$ [ S ( s ) # S ( s )]
0 i 0 j (5.2.8)
N(ds )

where, N(ds) is the number of observatories with distance of ds, S0(si) is the observed value of
observatory si.
!
 CHAPTER 5 SNOW LOADS - C5-11-

Fig. 5.2.8 (a) Annual maximum snow weight (Return period: 100 years, Hokkaido)

Fig. 5.2.8 (b) Annual maximum snow weight (Return period: 100 years, Tohoku)
- C5-12 - Commentary on Recommendations for Loads on Buildings

Fig. 5.2.8 (c) Annual maximum snow weight (Return period: 100 years, Kanto)

Fig. 5.2.8 (d) Annual maximum snow weight (Return period: 100 years, Chubu)
 CHAPTER 5 SNOW LOADS - C5-13-

Fig. 5.2.8 (e) Annual maximum snow weight (Return period: 100 years, Kinki)

Fig. 5.2.8 (f) Annual maximum snow weight (Return period: 100 years, Chugoku)
- C5-14 - Commentary on Recommendations for Loads on Buildings

Fig. 5.2.8 (g) Annual maximum snow weight (Return period: 100 years, Shikoku)

Fig. 5.2.8 (h) Annual maximum snow weight (Return period: 100 years, Kyushu)
 CHAPTER 5 SNOW LOADS - C5-15-

5.2.3 Equivalent unit weight for ground snow

Usually snow cover of the annual maximum depth comprises several layers, each of which has
accumulated one after another and with varying properties from snowflakes to granular snow, as
shown in Fig. 5.2.10. Fig. 5.2.9 shows typical examples of the relation between snow depth and the
snow mass that is defined as snow weight per unit area. Until the annual maximum snow depth is
recorded, the mass increases in proportion to snow depth. After that, the mass still increases and then
the maximum value is recorded, even though the depth decreases. Although there is usually a time lag
between the dates of the annual maximum depth and of the annual maximum mass, in most cases the
meteorological observation data give only the snow depth. Therefore, we need to use the equivalent
unit weight for ground snow, which is the ratio of the annual maximum snow mass to the annual
maximum snow depth, in order to evaluate the annual maximum snow mass from the annual
maximum snow depth.

Fig. 5.2.9 Relation between snow depth and Fig. 5.2.10 Layers of different types of snow
7)
snow mass (section of accumulated snow)8)

Statistical properties of the annual maximum snow depth and the annual maximum snow mass were
studied for data measured at 12 meteorological observatories which were located in heavy snow
regions in Japan. Fig. 5.2.11 shows the relation between snow depth d (m) and equivalent unit weight
for ground snow p0 (kN/m3) with return periods of 100 years and 10 years.9) The relation is described
by:

p 0 = 0.72 d 0 d ref + 2.32 (5.2.8)

- C5-16 - Commentary on Recommendations for Loads on Buildings

Fig.5.2.11 Relation between annual maximum snow depth and equivalent unit snow weight
at 12 observatories in Japan9)

Fig. 5.2.12 shows the relation between snow depth d and equivalent unit weight p for data calculated
by the method of snow weight evaluation using daily precipitations and daily average air temperatures
(mentioned in 5.1), which observed at 126 meteorological observatories not only in heavy snow
regions but also in little snow regions. Equation 5.2.8 shows overestimation for the snow depth less
than about one meter. Therefore it is possible to use smaller equivalent unit weight for the snow depth
less than about one meter against the equivalent unit weight obtained by the equation.

Fig.5.2.12 Relation between annual maximum snow depth and equivalent unit snow weight
with return period of 100 years calculated at 126 observatories in Japan10)

5.2.4 Environmental coefficient

Snow loads on roofs vary according to various factors due to the environmental conditions. Table
5.2.1 shows environmental factors influencing the depth of snow on roofs.
 CHAPTER 5 SNOW LOADS - C5-17-

Table 5.2.1 Environmental factors influencing depth of snow on roof

Macrofactors distance from the nearest seaside, altitude, topographical
inclination, topographical curvature, etc.
Mesofactors trees, neighboring buildings, local topography, etc.
Microfactors Thermal condition on the roof surface, roofing materials,
orientation of the roof, etc.

Macrofactors are mostly included in the estimated depth of snow on the ground. Microfactors are
too variable to consider in determining the design snow load. In this recommendation, only
mesofactors are considered by the environmental coefficient (kenv).
Table 5.2.2 shows an example of how the environmental condition influences the ground-to-roof
conversion factor, that is, shape coefficient. Fig. shows the location of three buildings with flat
roofs. Building A is usually heated and the snow on the roof is slowly melted. In years of ordinary
winter climates, the snow depth on the roof of building B is higher than those of the other two
buildings. This might be because building B is located to the leeward of building A. In heavy snow
years, the snow depth on the roof of building B is almost the same as that in ordinary years, though the
other two buildings are more heavily loaded than in ordinary years.

Table 5.2.2 Ratio of snow depths on the ground and on roofs7)

Ratio Roof of Bldg. A Roof of Bldg. B Roof of Bldg. C

Year / Ground / Ground / Ground
Light Snow Year
0.81 1.03 0.91
(1978 1979)
Heavy Snow Years
0.81 1.03 0.96
('79-'80, '80-'81)
Ordinary Snow Years
0.56 1.02 0.81
(77-78, 81-82 & 82-83)

Here, mean values are shown for heavy snow years and ordinary snow years.
The environmental coefficient (kenv) is generally defined as unity. When the snow depth on the
ground is estimated to increase locally because of the environmental condition, kenv should be
correspondingly larger than the unity.

Figure 5.2.13 Block plan of the buildings7)

- C5-18 - Commentary on Recommendations for Loads on Buildings

5.3 Ground Snow Load with Accumulation for n Days

5.3.1 Characteristic ground snow load with accumulation for n days

(1) Estimation of ground snow weights based on the data of snow accumulation for n days on the
ground and equivalent unit weight for ground snow
A heavy short-term snowfall produces a critical condition in buildings whose snow loads are usually
controlled by reliable systems. In such cases, the annual maximum snow load for the whole season
leads to an overestimation and increasing snow depth for certain days needs to be statistically
estimated. In this recommendation, the increasing snow depth for n days is considered. The duration
n is determined by sufficient consideration. For buildings with reliable control systems, the
characteristic snow load on the ground Sn is given by Eq.(5.3.1) as the product of several factors:

Sn = kenv dn pn (5.3.1)

where
Sn: characteristic snow load (kN/m2) on the ground used for design when snow load on
the roof is controlled,
kenv: environmental coefficient, it is the same as defined in 5.2.4,
dn: characteristic snow depth (m) on the ground when the snow load on the roof is
controlled, as defined in 5.3.2,
pn: equivalent unit weight for ground snow (kN/m3), as defined in 5.3.3.

(2) Estimation of the snow weights on the ground based on the data of daily precipitation and daily
mean air temperature
In section 5.2.1(2), it is described that the new proposed method is practical and useful for
estimating ground snow weights. In order to estimate the characteristic snow weights on the ground,
the above-mentioned method can be applied easily. Fig. 5.3.1 shows the examples at Sapporo,
Kamabuchi, Nagaoka and Tokamachi using Equation (5.2.1), in which the estimated values are
compared with the observed values for snow weights increasing intensity for 7 days. As seen in the
Figures, the estimated values approximately agree with the observed values.
 CHAPTER 5 SNOW LOADS - C5-19-

Figure 5.3.1 Example of observed and estimated snow weight increasing intensity for 7 days due to
Eq. (5.2.2)10)

5.3.2 Ground snow depth accumulation for n days

The basic snow depth accumulation for n days on the ground dn is defined as the annual maximum
value of snow accumulation for n days with a return period of 100 years for snow load on a roof with a
reliable control system. The annual maximum increasing intensity of snow depth is defined as shown
in Fig. 5.3.2. Because of the limited number of meteorological observatories, the basic snow depth at
the construction site is estimated from topographical data.
To clarify the difference between the annual maximum snow depth for the whole season (AMD) and
the annual maximum increasing snow depth for 3 days (AMI-3) and for 7 days (AMI-7), two
antithetical examples plotted on Gumbel probability paper are shown in Fig. 5.3.3. In Hokkaido,
values of AMD become larger at the end of winter in spite of smaller AMI values because of the low
temperature. In Gifu, located in the middle of Honshu, however, values of AMI are closer to values of
AMD because this area is rather warmer and snow depth increases rather rapidly. A different snow
accumulation mechanism can also be recognized in Fig. 5.3.4.
- C5-20 - Commentary on Recommendations for Loads on Buildings

Fig. 5.3.2 Annual maximum increasing snow Fig. 5.3.4 Ratio of AMI-7 to AMD11)
depth for n days 11)

Fig. 5.3.3 Comparison of AMD, AMI-3 and AMI-7 (Gumbel probability paper, Hazen plot)11)

5.3.3 Equivalent unit weight for ground snow with roof snow control
Because there are quite few data of snow weight observed daily, the equivalent unit weight for
ground snow accumulation for n days (pn) is discussed here with data calculated by the method of
snow weight evaluation using daily precipitations and daily average air temperatures at 126
meteorological observatories. Fig. 5.3.5 shows the relation between snow depth dn and equivalent unit
weight pn based on the annual maximum snow depths accumulating for 3 and 7 days with return
periods of 100 years. The values of pn shown in Fig. 5.3.5 are smaller than those of p0 for the snow
depth larger than 0.5 m in heavy snow regions shown in Fig. 5.2.8, because the snow density increases
according to accumulation days. However they are similar to the values of p0 for the annual maximum
snow depth less than 0.5 m in little snow regions because snow melts away within several days. These
tendencies exhibit also in Fig. 5.3.7, that is the equivalent unit weights dn and d are almost constant for
Tokyo or Sendai located in Pacific Ocean side being relatively warm in winter, but they are different
 CHAPTER 5 SNOW LOADS - C5-21-

for Fukui or Sapporo located in Japan Sea side being cold.

The equivalent unit weight for 1, 3 and 7 days are 1.37, 1.41 and 1.57, respectively, which were
obtained as the average values of 126 observation points, as shown in Fig. 5.3.6. However, it is
considered for estimation in safety side that the equivalent unit weight for n-days accumulation snow
with roof snow control pn is the same as that for the annual maximum snow without roof snow control
p.

Fig. 5.3.5 Relation between annual maximum snow depth accumulating for 3 and 7 days and
equivalent unit snow weight with return period 100 years calculated at 126 observatories in
Japan10)

Fig. 5.3.6 Frequency of equivalent unit snow weight for 1, 3 and 7 days accumulation10)
- C5-22 - Commentary on Recommendations for Loads on Buildings

Fig. 5.3.7 Relation between accumulating days and equivalent unit snow weight in some cities in
Japan10)

5.4 Snow Load on the Roof

5.4.1 Design snow load on the roof
Generally, design snow load on the roof can not be determined directly. In this recommendation, it
is estimated from ground snow load. Snow load on the roof is usually less than on the ground because
of wind, sunshine, and so on. In addition, roof shape and slope influences the ratio of roof snow to the
ground snow. Therefore, shape coefficients 0 and n are introduced, taking into account
meteorological conditions. Partial snow load such as snowdrift caused by projecting structures, snow
eaves, and sliding snow from upper roofs at the eaves or lower levels of multilevel roofs are
considered in 5.6.
The format equations of design snow load on roofs differ depending on whether or not snow on the
roof is controlled. If the roof snow is not controlled, the design snow load on the roof S is the
characteristic snow load on the ground S0 multiplied by 0:

S = 0 S0 (5.4.1)

5.4.2 Shape coefficient

According to recent works, snow load on roofs depends on wind speed and temperature.12)13) In cold
regions, snow on roofs is easily removed, as shown in Photo 5.4.1 and 5.4.2. However, in warm
regions, especially in calm wind conditions, snow on the roof is similar to that on the ground.14)
 CHAPTER 5 SNOW LOADS - C5-23-

Photo 5.4.1 Snow drift on a roof Photo 5.4.2 Snow accumulation on a roof

In this recommendation, shape coefficient 0 is defined as:

0 = b + d + s (5.4.1)

where
b: basic shape coefficient
d: shape coefficient for irregular distribution caused by snowdrift
s: shape coefficient for irregular distribution caused by sliding

This coefficient is only applied to normal shaped buildings. For special forms or large scale
buildings, shape coefficient should be determined from special research or experimentation.

(1) Basic shape coefficient

The relationship between the ratio of roof snow depth to ground snow depth (basic shape
coefficient) and wind speed is shown in Fig. 5.4.1. In the figure, the solid line shows the former
recommendation of AIJ (1986), the pointed dashed line shows the recommendation of the former
Soviet Union15), the dashed line shows the equation proposed by Mihashi et. al. (1988)16), and the
dotted line shows the equation proposed by Tomabechi et. al. (1991)17). Every study shows that the
basic shape coefficient decreases as wind speed increases.
Roof snow also depends on roof shape. According to wind tunnel tests16), the ratio of roof snow
depth to ground snow depth increases for roof slopes up to 25 and decreases for roof slopes over 25.
In these ways, basic shape coefficient is determined as shown in Fig. 5.1. For roof slopes larger than
50, the same value as 50 is used.
- C5-24 - Commentary on Recommendations for Loads on Buildings

Figure 5.4.1 Relationship between ratio of roof snow depth to ground snow depth and wind speed

(2) Shape coefficient for irregular distribution caused by snow drift

In the troughs of M-shaped roofs, multiple pitched roofs and multispan roofs, snow accumulates by
snow drift and it is always deeper in the troughs than at the ridge, as shown in Fig. 5.4.218). This
phenomenon is also observed in wind tunnel tests19). Therefore, shape coefficient for irregular
distribution caused by snow drift d is defined for the troughs and is zero at the ridges. For multilevel
roofs, snow accumulation on lower roof levels increases especially at the edge of high roofs. The
same thing occurs when there is a parapet or similar structure on the roof.

Figure5.4.2 Observation of snow accumulated on multispan pitched roof (Ottawa)18)

(3) Shape coefficient for irregular distribution caused by sliding

Sliding of snow on the roof depends on characteristics of the snow, temperature, shape of the roof,
roofing materials, and so on. Its mechanism is difficult to accurately define. When (2b+d)d is larger
than roof gap shown as Fig. 5.4.3 (b), this should be adjusted as shown Fig. 5.4.3 (c).
 CHAPTER 5 SNOW LOADS - C5-25-

Figure5.4.3 Treatment of s

5.5 Snow Load on the Roof with Control

5.5.1 Design snow load on the roof with control
In this recommendation, design snow load on the roof with control is defined as:

S = S n n S c (5.5.1)
where
Sc: controlled snow load, as defined in 5.5.2.

In this case, n might be different from 0. However, because Sn is determined from snow
accumulation for n days, it is difficult to clarify the difference by wind tunnel tests or other estimation
methods. That also depends on the method of snow control. Therefore, in this recommendation, n
takes the same value as 0. When the snow load is controlled, Sc is calculated as difference of the
initial snow load of snow remaining on the roof when another heavy snowfall is expected and the
snow load removed by a device whose performance is guaranteed even during a heavy snowfall. The
performance should be estimated by research or experiments and then the value Sc should be
determined.

5.5.2 Controlled snow load

There are three methods of artificially removing snow from a roof:

1. Mechanical snow removal (including snow removal with man power)

2. Positive snow sliding
3. Snow melting

A combination of methods 2 and 3 is also used. In methods 1 and 2, snow is removed after it
accumulates, and it is unclear how much snow has accumulated when removal starts. In method 3,
there are two possibilities: all snow is melted and some snow remains. Therefore, design snow load
on the roof should be decided by the planner. Fig. 5.5.1 (a) shows the daily snow depth and Fig. 5.5.1
(b) shows the ground snow load on the ground simulated from snow depth after Eq. (5.2.4).
- C5-26 - Commentary on Recommendations for Loads on Buildings

Figure 5.5.1 Simulations of roof snow control (in case of Sapporo 1980-1981)

When snow is removed by manpower, the time during which snow load exceeds the design snow
load on the roof should be considered. If the time is long as in Fig. 5.5.1 (c), the designer cannot
consider the controlled snow load. Fig. 5.5.1 (d) shows the case where snow is removed mechanically
with short delay. In the worst case, we should consider the case where a heavy snowfall starts when
the existing snow load is just over the design load level. Fig. 5.5.1 (e) shows reliable snow removal
system, and only such a case we can take controlled snow load into the design. Fig. 5.5.1 (f) simulates
a snow melting system that is run continuously during the snow season. Its capacity and reliability
against heavy snowfalls should be considered. When the performance of the device is guaranteed
even during heavy snow falls, the controlled snow load Sc may be reduced from Sn n as expressed in
Eq. (5.5.1).
According to some research, a melting system that melts all the snow on a roof all the time
consumes a lot of energy, and a melting system that only works after some snow has accumulated is
inefficient because of snow caving between the snow and the roof (see Fig. 5.5.2)20)21). After this trial,
a new method was studied22)23). With this method, a thin ice layer is created and then melted, thus
 CHAPTER 5 SNOW LOADS - C5-27-

inducing snow sliding and consuming little energy.

In snow control, it is important to ensure a reliable energy supply in a heavy snow fall. In Hokkaido,
since the 1970s, roads (national route) closed by heavy snow have been reopened in four days at most.

5.6 Partial Snow Load on the Roof

Snow load on roofs partially increases because of snowdrift caused by projecting objects like
chimneys, ventilators and so on. This load might not be critical for the main frame, but it should be
considered in design of secondary members. The load should be considered as concentrated.
Therefore, to determine the point of accumulation, wind direction should be considered. Snow eaves
are sometimes up to 2 to 3 meters long. Therefore, it should be considered as a concentrated load with
a density of about 4kN/m3. Snow sliding from the upper roof should be resisted first. However, if it
cannot be resisted, the distance and impact of sliding should be considered. The impact force might be
twice the gravity force.

5.7 Other Snow Loads

Side pressure from snow drifts on the outside walls of buildings may be categorized as A or B, as
shown in the Fig. 5.7.1. The load for case B might be larger than for case A. Side pressure might be
assumed as follows, after Matsushita et. al.24):

Fl = 9.8 "10#3 (15 ~ 20) ds2 (5.7.1)

where
Fl : Side pressure [kN/m2].
!

When the building might be buried under snow, snow load caused by sedimentation should be
!
considered because eaves or braces at the outside might be broken.
When snow adheres to the building or snow covers the building, snow mushrooms at the tops of the
projecting objects sometimes grow to over 1 meter, and therefore cannot be neglected. When snow
blows into balconies or outside corridors, it accumulates in sheltered zones.
- C5-28 - Commentary on Recommendations for Loads on Buildings

Figure 5.7.1 Side pressure and sedimentation25) Figure5.7.2 Observation of

side pressure24)

References
1) Sakurai, S., Joh, O. and Shibata, T.: Estimation of ground snow weight based on daily
precipitation and daily mean air temperature, Snow Engineering: Recent Advances, pp.185-192,
Balkema, 1997.
2) Kamimura, S., Umemura, T.: Estimation of daily snow mass on the ground using air temperature
and precipitation data, Second International Conference on Snow Engineering, CRREL Special
Report 92-27 pp.157-167, 1992.
3) Takahashi, T., Kawamura, T., Kuramoto, K.: Estimation of ground snow load using snow layer
model, Journal of Structural and Construction Engineering (Transactions of AIJ) No.545,
pp.35-41, 2001.
4) Izumi, M., Mihashi, H. and Takahashi, T., Statistical properties of the annual maximum series and
a new approach to estimate the extreme values for long return periods, Proc. of 1st. Int. Conf. on
Snow Engineering, CRREL Special Report 89-6, pp.25 - 34, 1989.
5) Takahashi, T., Fukaya, M., Mihashi, H. and Izumi, M., Influence of altitude and sea area ratio for
geographic distribution of snow depth, Summaries of Technical Papers of Annual Meeting, Vol. B,
AIJ, pp.219 - 220, 1992.8 (in Japanese).
6) Takahashi, T., Shitara, T. and Ellingwood, B.R.: Ground snow load estimation with Kriging, Snow
Engineering V, pp.135-140, Balkema, 2004.
7) Takahashi, H. and Nakamura, T., Disaster Prevention of Snow and Ice, Hakuashobo, pp.213 - 218,
1986 (in Japanese).
8) Maeda, H.: Fundamental Study on Estimation of Snow Load, Transactions of AIJ, No.319,
pp.32-37, 1982 (in Japanese with English abstract).
 CHAPTER 5 SNOW LOADS - C5-29-

9) Joh, O., Sakurai, S., Equivalent snow density in heavy snowing area, Journal of Snow Engineering,
JSSE, Vol.9, No.2, pp.112 - 114, 1993.4 (in Japanese)
10) Sakurai, S., Joh, O. and Shibata, T.: Estimation of equivalent density of snow accumulation for
short period, Snow Engineering, Recent Advances and Developments, pp.143-148, Balkema, 2000.
11) Izumi, M., Mihashi, H. and Takahashi, T.: Statistical Properties and Regional Characteristics of
Annual Maximum Increasing Intensity of Snow Depth, Journal of Structural and Construction
Engineering (Transactions of AIJ), No.392, pp.68-77, 1988.10 (in Japanese with English abstract).
12) Tomabechi, T., Izumi, M. and Endo, A, A Fundamental study on the evaluation method of
roof-snowfall-distributions of buildings, Journal of Structural Engineering, AIJ, Vol.32B, pp.49 -
62, 1986.3 (in Japanese with English abstract).
13) Izumi, M., Mihashi, H., Sasaki, T., Takahashi, T., Matsumura, T., Fundamental study on
estimation of roof snow loads part 15, Summaries of Technical Papers of Annual Meeting, Vol. B,
AIJ, pp.1405 - 1406, 1987.10 (in Japanese).
14) Yamada, K. and Matsumoto Y., Observation of snow cover on real flat roofs in the Hokuriku
region, 1982 - 1989, Summaries of Technical Papers of Annual Meeting, Vol. B, AIJ, pp.117 - 118,
1990.10 (in Japanese).
15) ISO 4355, Bases for design of structures - Determination of snow loads on roofs, 1998.
16) Mihashi, H., Takahashi, T. and Izumi, M., Wind effects on snow loads, Proc. of 1st. Int. Conf. on
Snow Engineering, CRREL Special Report 89-6, pp.158 - 167, 1989.
17) Tomabechi, T., Influence of meteorological conditions for snow depth on roofs, Summaries of
Technical Papers of Annual Meeting, Vol. B, AIJ, pp.51 - 52, 1989.10 (in Japanese).
18) Taylor, D.A., Roof snow loads in Canada, Canadian Journal of Civil Engineering, Vol.7, No.1,
pp.1 - 8, 1980.
19) Endo, A. and Tomabechi, T., Wind channel experiment of the forming conditions of the snow
depth on various roof with model snow, Memoirs of the Hokkaido Institute of Technology, No.11,
pp.163 - 178, 1983 (in Japanese with English abstract).
20) Morino, K., Kobayashi, M., Takebayashi, Y., Kawai, M., Kawashima, M., A snow melting
experiment on a pneumatic structure (air supported dome) No.2 a snow melting experiment by a
wind box, Summaries of Technical Papers of Annual Meeting, Planning division, AIJ, pp.789 -
790, 1984.10 (in Japanese).
21) Nishi, Y., Nishikawa, K., Ishii, H., Analysis of snow melting process of the air supported
structure -2, Summaries of Technical Papers of Annual Meeting, Vol. D, AIJ, pp.895 - 896, 1986.8
(in Japanese).
22) Ohtsuka, K., Joh, O., Homma, Y., Miyagawa, Y., Masumo, T. and Okada, H., A Study on the
removal of snow from membrane structures, Proc. of Membrane Structures Association of Japan,
No.4, pp.55 - 68, 1990 (in Japanese with English abstract).
23) Tomabechi, T., Yamaguchi, H., Ito, T., Hoshino, M., Fundamental study on sliding of snow on
the roof of membrane structure, Journal of Structural and Construction Engineering, AIJ, No.426,
- C5-30 - Commentary on Recommendations for Loads on Buildings

pp.99 - 105, 1991 (in Japanese with English abstract).

24) Matsushita, S., Nakajima, H. and Izumi, M., A study on side pressure of snow, Transactions of the
AIJ, No.89 (Summaries of Technical Papers of Annual Meeting) p.82, 1963.9 (in Japanese).
25) Kubodera, I.: Side pressure of snow on buildings, Kenchiku Gijutsu, No.384, pp.203-204, 1983
(in Japanese).