The estimation of wind exposure for
windthrow hazard rating:
comparison between Strongblow,
MC2, Topex and a wind tunnel study
J-C. RUEL1, D. PIN1, L. SPACEK2, K. COOPER3 AND R. BENOIT 4
1
Groupc dc Rcchcrchcs en Am6nagement Durable et Integre des Forfits, Departement des Sciences du Bois
et de la Forfit, Ufiiversite Laval Sainte-Foy, Quebec, G1K 7P4, Canada
2
Departement des Sciences de la Terre, Universite du Quebec a Montreal, B.P. 8888, succ. Centre-Ville,
Montreal, Quebec, H3C 3P8, Canada
3
National Research Council Canada, Aerodynamics Laboratory, Montreal Road, Ottawa, Ontario, K1A
0R6, Canada
4
Recherche en Prevision Numeriquc, Environment Canada, 2121 Trans-Canada Highway, Dorval, Quebec,
H9P 1J3, Canada
Summary
This paper compares five different methods of wind exposure assessment in an attempt to identify a method that could serve in windthrow hazard assessment. The candidate methods that
were selected are: a wind tunnel simulation, two numerical models based on air flow behaviour
(MC2 and Strongblow) and two empirical methods (Topex and Topex-to-distance). Best correlations with the wind tunnel results were observed with Strongblow and Topex-to-distance. For
MC2, significant correlations were obtained at high windspeeds but not at lower windspeeds.
Ranking of windiness by topographic position was identical for the wind tunnel, Strongblow and
Topex-to-distance. Wind direction estimation by MC2 showed little sensitivity to topography,
especially at higher windspeeds. This fact can be attributed both to the version of MC2 used and
the conditions in which the simulations were performed. The estimation of windspeed by direction obtained with Strongblow also appeared not to be very reliable. The satisfactory performance of Topex-to-distance and the ease with which this method can be calculated would make
it appropriate for inclusion in a windthrow hazard rating system.
In forest ecosystems, the formation of canopy
gaps by single or multiple treefalls is one of the
O Iraorutt of Chartered Foratcn, 1997
most important types of local disturbance
(Schaetzl et al. 1989). The uprooting of trees
by wind is a complex process that greatly
Forestry, Vol. 70, No. 3, 1997
254
FORESTRY
influences stand and soil dynamics (Stephens,
1956; Schaetzl et al., 1989; Jonsson and Esseen,
1990). In extensively managed forests, relying on
large clearcuts as the main harvesting method,
windthrow is of particular concern at the buffer
zones along waterbodies and main roads. However, in North America, there is currently a
trend towards a more intensive use of partial
cutting (Franklin, 1990; Seymour and Hunter,
1992) which can bring an increase in vulnerability to uprooting. Moreover, there is a strong
pressure for reducing clearcut size and leaving
mature forest reserves and buffer strips for
water, wildlife and recreation (Seymour and
Hunter, 1992; Coates and Steventon, 1994) and
these practices could lead to increased treefall
(Laurance and Yensen, 1991; Esseen, 1994).
With better control of forest fires, windthrow is
also likely to play a more important role in the
rejuvenation of mature or overmature stands
(Schaetzl et al., 1989).
Many strategies can be adopted to cope with
the risk of windthrow (Quine, 1995). One can
consider it as a fatality against which nothing
can be done. This seems justified for very severe
storms that destroy almost everything (De
Champs, 1987). However, more often and
despite a certain randomness, wind damage can
be partially predicted and reduced by proper silvicultural actions. Such a response requires a
sound comprehension of the mechanisms
involved. Windthrow hazard depends upon the
interaction of numerous factors pertaining to
climate, topography, soil and stand characteristics.
Determination of regional wind climates is
generally considered a necessary first step for
windthrow hazard rating over broad areas
(Flesch and Wilson, 1993). On a more local
basis, however, topography, together with soil
and stand variables, exert a predominant influence.
Topographic features largely determine the
relative windiness at a local level and a definite
relation has been observed between the amount
of damage and certain features of the landscape
(Busby, 1965; Ruel and Pin, 1993; Busing and
Pauley, 1994). Although many general statements on the effect of topography on wind have
been made (Ruth and Yodcr, 1953; Buck, 1964;
Mitchell, 1995; Ruel, 1995), the wind climate
varies rapidly in complex terrain so that detailed
quantification of topographic influences is rarely
attempted (Busby, 1965; Miller et al., 1987).
Windiness would best be described by using an
intensive network of anemometers. However,
given the vast areas and the difficult access typical of boreal North American forests, foresters
must rely on an alternative method. Some methods of wind exposure assessment are available
and good reviews are presented by Miller et al
(1987) and Gardiner and Quine (1994). Each has
some advantages and some drawbacks, some
being easy to use, others, more complicated
(Hannah et al., 1991; Gardiner and Quine,
1994).
The testing of scaled topographic models in
boundary-layer wind tunnels can assist in the
assessment of wind exposure over extensive
areas of complex terrain (Miller et al., 1987).
Reasonable correlations between field and wind
tunnel data have been observed in gently rolling
terrain (Miller et al., 1987). However, this
approach requires sophisticated installations
that may not be available or affordable.
The assessment of topographic shelter can
also be estimated by simple empirical measures.
Thus, the Topex (Topographic exposure)
method was developed to provide consistency
and objectivity in assessing topographic shelter
(Miller et al., 1987). It has been introduced into
a windthrow hazard rating system for Great
Britain (Miller, 1985). This method uses angles
of elevation to the skyline as input. More
recently, modifications of the method have been
proposed to better differentiate between flat terrain and hilltops (Hannah et al., 1995).
Although Topex measurements do not accommodate local variations in wind behaviour due
to complex topographic effects, they have been
found to relate reasonably well to measures of
wind exposure by tatter flag studies or annual
mean windspeed (Miller et al., 1987). The pattern of Topex sector measurements has been
used to calculate the effect of funnelling and
aspect (Quine and White, 1994), adding to the
usefulness of the method in complex terrain.
Over the last decade, the development of
numerical airflow models created an interest for
this method as a windiness estimation technique
(Gardiner and Quine, 1994; Hannah et al., 1995;
Inglis et al., 1995; Wollenweber and Wollenwe-
ESTIMATION OF WIND EXPOSURE FOR WINDTHROW HAZARD RATING
ber, 1995). Present computing facilities make
such modelling a more feasible technique (Zack
and Minnich, 1991) but the reliability of this
approach is not yet well defined (Hannah et al.,
1995).
Other methods are also available. These
include the use of tatter flags and the close
observation of the trees themselves (Miller et al.,
1987; Robertson, 1987). However, comparative
studies still remain necessary to estimate the relative value of candidate methods for windiness
assessment.
This paper compares the reliability of different methods for windiness assessment. These
include: wind-tunnel trials, numerical models
and Topex. Comparisons will be made in the
context of a field monitoring of windthrow in
riparian buffer strips over a 5-year period in the
Laurentian Hills, in Eastern Canada.
Methods
255
Canada (Figure 1). It is mostly located in the balsam fir-paper birch ecoclimatic domain but
extends somewhat into the balsam fir-black
spruce domain (Thibault and Hotte, 1985). Balsam fir (Abies balsamea (L.) Mill.) is the predominant species, accompanied by paper birch
(Betula papyrifera Marsh.), black spruce (Picea
mariana (Mill.) B.S.P.) and white spruce (Picea
glauca (Moench.) Voss.). The climate is rather
cold (mean annual temperature = 0.3°C) and the
growing season is short (144 days). Prevailing
wind direction is from the north-west and mean
annual windspeed is 2.0 ms" 1 . The main surficial
deposits are glacial tills of varying depths overlying acidic bedrock. Soils are typically podzolic.
Two rivers flow through the study area, one
running NNW-SSE, and another NW-SE. A
third suspended valley runs SW—NE. Valley side
slopes generally lie between 10 and 40 per cent
but slopes up to 60 per cent are also found in
the study area. Altitude ranges from 670 m up
to 1150 m.
Study site
Wind-tunnel study
The study site is located in the Laurentian Hills,
approximately 80 km north of Quebec city,
A circular, l:5000-scale terrain model representing an area of 10.7 km diameter was
Figure 1. Location of the study area (indicated by star).
256
FORESTRY
constructed. This model was centred around the
NNW-SSE valley where most of the field windthrow survey was conducted. A decision was
made not to include any vertical exaggeration in
order to maintain proportionality in vertical and
horizontal flow. Layers of cardboard were cut
following the successive contours and assembled
together. Plaster was added to obtain a smooth
profile. Forty-one holes were drilled in the
model for installation of the sensors. Location
of the sensors corresponds to the field windthrow monitoring plots (sensors 6 to 20) to
which were added some additional sensors to
cover a wider range of topography (Figure 2).
The model was mounted on a turntable. The
approach wind profile was provided by an array
of 64 boundary-layer spires (Irwin, 1979;
Cooper et al., 1996). An omnidirectional, pressure-based sensor (Irwin, 1980) was used to
measure wind speed at a scale height of 10 m
above ground. The sensor permits the rapid
measurement at many locations with the use of
a Scanivalve. Mean flow directions were measured separately at each sensor location for
every test angle using black wool strands and
photographs taken from above.
Pressure measurements were made at an
absolute test speed of 45 m s~' over a 300° wind
direction range, in steps of 10°. No data are presented for winds coming from the East since the
model ended too abruptly in this direction. This
extremely high wind speed, if considered at the
model scale, was judged necessary to obtain
proper readings with the equipment used. The
>!
^m
</?:l\ote- *
Figure 2. Distriburion of the surface-wind sensors. Contour equidistance: 50 m.
ESTIMATION OF WIND EXPOSURE FOR WINDTHROW HAZARD RATING
signals were sampled for 10 s at a rate of 30 Hz
to estimate the average pressure. A reference
pressure was obtained from a sensor near the
border of the model and for a wind direction
where the local, gentle topography, was thought
to have minimal effects. Ratios of measured
pressure over reference speed were calculated
for each sensor and each run. Those overspeed
ratios will be used in the rest of the paper.
Unless otherwise stated, the values presented
correspond to the mean of all runs for a given
point.
Topex
For this empirical method, a clinometer and a
compass are generally used to measure the skyline elevation angle for the eight major compass
directions (Wilson, 1984). The sum of the eight
angle values constitutes the Topex value for the
study site. Since the presence of nearby stands
made impossible the determination of skyline
angle in some directions, Topex values were
derived from topographic maps by locating the
topographic features most likely to limit the
sight in a direction. In the British hazard rating
system, elevation is used in conjunction with the
Topex to estimate site windiness. In this study,
Topex alone was used since no correlation
between wind speed and elevation was available.
Some improvements could be brought to the
traditional Topex (Quine and White, 1994;
Hannah et al., 1995). Placing a limit on the distance within which the skyline is sought and
permitting negative values for the compass
directions where no higher ground is present
have been suggested. This method, referred to as
Topex-to-distance (Hannah et al., 1995), was
also assessed, using a limiting distance of 3 km.
Strongblow
The numerical model Strongblow was designed
for microcomputers and is sold by the Building
Research Establishment of the United Kingdom
(Cook et al., 1985). Its purpose is to estimate
wind speed at a specific site for the assessment
of wind loads on buildings. The input data set
includes altitude, latitude, terrain roughness and
topography, and a basic hourly-mean wind-
157
speed. In this study, a base windspeed of 14
m s"1 was used. It requires the measurement of
topographic parameters on topographic maps in
each direction.
Although initially designed for use in Britain,
this model can also be used for temperate overseas sites. However, some modifications are necessary regarding mainly the effect of altitude
and the directional factors. Thus, following the
recommendations by Cook et al. (1985), an
equal weight was applied to wind from each
direction and both the site and the terrain base
altitude were entered as zero.
For each of 12 directions, wind data are
derived for up to 10 heights above ground. For
general comparisons in this study, the mean
windspeed at 10 m height, averaged over all
directions, is used as an estimation of wind
exposure. For comparison with the other methods not considering land use, emphasis is put on
a trial using a uniform aerodynamic roughness
representative of the meteorological standard
(category 2 of Cook et al., 1985; zo = 0.03 m).
This value was selected to reflect the fact that
most of the area near the valleys has been
clearcut, leaving some patches and strips of
uncut timber. This model was run on a IBM PC
486 computer.
MC2
The model MC2 (Mesoscale Compressible
Community model) is a new model based on the
fully elastic non-hydrostatic equations of Tanguay et al. (1990). The model dynamics have
been developed jointly by scientists at the Universite du Quebec at Montreal and at the
Atmospheric Environment Service of Environment Canada. The model uses the complete set
of Euler equations. The model includes features
such as: 3-dimensional semi-Lagrangian advection scheme, complete set of physical processes,
automatic distribution of vertical levels and selfnesting capability for simulation at high resolution (Bergeron et al., 1994; Desgagne et al.,
1994). Trials with this model have shown it to
be accurate, efficient and versatile, allowing
excellent simulations over a wide spectrum of
scales (Benoit et al., 1994). Version 2.8 of the
model was run on a SUN SPARC10 workstation. However, due to limitations imposed by
258
FORESTRY
the hardware available, full use of the model
was not possible and only a simplified version of
the physics could be used, resulting essentially in
a frictionless flow near the surface.
Three wind events were simulated to obtain
windspeeds for the sensors location. These
events correspond to:
• a storm with relatively strong winds (windspeeds above 30 ms~');
• a condition of low windspeeds (less than 5
ms" 1 ); and
• a condition of intermediate windspeeds
(between 10 and 15 ms" 1 ), similar to those
used with Strongblow.
Data from the three wind events were
obtained through the meteorological offices and
run through a larger scale synoptic model.
Results were input into the MC2 model and
simulation was made in three successively finer
resolution scales: from 50 km to 5 km, and
finally to 500 m. At each step, data from the
larger scale was used as input for the model.
Results for each event consist of six simulations,
calculated every 15 min, centred around the
peak windspeed. Results of windspeed are averaged for each event separately.
Analysis of results
Correlation coefficients were computed relating
data obtained by each method. Since the data
obtained through MC2 were derived from a specific storm with a given direction, correlations
were also estimated for the corresponding wind
tunnel trial. A potential advantage of numerical
models would be to provide estimates, not only
of windspeed, but also of wind direction. To
verify the reliability of wind direction estimated
by MC2, the range of predicted directions was
examined and compared with the directions
observed in the appropriate wind tunnel trial.
Comparisons were also made between windspeeds estimated for each direction by Strongblow and the wind tunnel.
Site windiness has often been related to topographic position (Buck, 1964). Hence, an analysis of variance was performed to identify if
differences in wind exposure could be related to
topographic position. Mean comparisons were
made by Fischer's protected l.s.d.
Results
Descriptive statistics for each method are presented in Table 1. Since the methods do not use
a common unit, direct comparisons of means
are meaningless. Nevertheless, it can be seen
that permitting negative angles and limiting the
distance in the estimation of Topex resulted in
a diminution of the mean and a marked increase
in the coefficient of variation. From this table, it
can be seen that large differences in the coefficients of variation occur suggesting different
sensitivities to topography.
Very good correlations with the wind tunnel
study were found with the Topex, Topex-to-distance and Strongblow (Table 2). The modifications brought to the standard Topex calculation
were successful in improving the correlation.
Very significant correlations, although somewhat poorer than the previous methods, were
obtained with MC2 at the highest windspeed.
Table 1: Descriptive statistics of wind exposure estimation obtained with the different methods tested
Method
Wind tunnel
Topex
Topex-to-distance
Strongblow
MC2 (strong winds)
MC2 (intermediate windspeed
MC2 (low windspeed)
Unit
Mean
Standard
deviation
overspeed ratio
degrees
degrees
ms- 1
ms-'
ms- 1
m s~'
1.34
47.88
18.65
28.98
33.7
13.07
4.41
0.33
33.82
71.11
3.60
2.28
0.94
0.66
Coefficient
of variation
(%)
24.9
70.6
381.1
12.4
6.8
7.2
15.0
ESTIMATION OF WIND EXPOSURE FOR WINDTHROW HAZARD RATING
259
Table 2: Correlation coefficients (r) between wind exposure estimated from the wind tunnel and the other
methods of estimation (« = 41)
Topexto-
Wind tunnel
Topex
Topex-to-distance
Strongblow
MC2 (strong winds)
MC2 (intermediate
windspeed
Topex
distance
Strongblow
-0.831"*
-0.916"*
0.951***
0.906***
-0.737***
-0.871***
MC2
(strong
winds)
0.712*"
-0.683***
-0.686***
0.564***
MC2
(intermediate
windspeed)
-0.207
0.227
0.230
-0.074
-0.359***
MC2
(low
windspeed)
0.113
-0.148
-0.061
0.030
0.373*
0.354*
* significant at P = 0.05; " significant at P = 0.01 and " * significant at P = 0.001.
However, correlation was not significant at the
other windspeeds. Generally, coefficients of correlation were positive, indicating that, as windspeed measured in the wind tunnel increases, so
does the estimation of windspeed from a given
method. The negative correlation obtained by
the Topex and the Topex-to-distance methods
is a desired feature since, as the topographic
shelter measured by the vertical angle increases,
windspeed is expected to be reduced.
Correlations between Topex and Topex to
distance are very good. Topex was highly correlated with Strongblow but modifications
through the Topex-to-distance approach considerably improved the correlation. Significant correlations between MC2 and other methods were
only observed with the strongest winds, the
highest being with the wind tunnel data. Correlations between the different MC2 runs were
significant but weak.
Since this general comparison gives an equal
weight to all directions sampled, it was felt necTable 3: Correlation coefficients (r) between wind
exposure estimation by MC2 and by the wind tunnel
trial corresponding to the same general wind direction (« = 41)
Method
MC2 (strong winds)
MC2 (intermediate windspeed)
MC2 (low windspeed)
*•* significant at P = 0.001.
r
0.817***
0.051
0.237
Table 4: Wind direction statistics from each of the
three MC2 trials at the maximum windspeed (n =
961)
Mean
direction
(degrees)
Range
degrees)
(min.-max.)
MC2 (strong winds)
219.1
12.8
(211.7-224.5)
MC2 (intermediate
windspeed)
132.9
28.5
(115.5-144)
MC2 (low
windspeed)
198.1
59.5
(157.4-216.9)
Method
essary to go into more detail to better evaluate
the results of MC2 which applies to a storm
with a specific wind direction. Comparison
made with the appropriate wind direction from
the wind tunnel increases the correlation at the
higher windspeed but not significantly at other
windspeeds (Table 3). At the highest windspeed,
wind direction over the study area varies very
little reflecting no important funnelling from the
valleys (Table 4 and Figure 3). This does not
conform to the wind tunnel trial for the same
mean direction where a definite funnelling effect
was observed (Figure 4). However, as windspeed decreases, the range in wind direction is
augmented, rising from 12.8° to 59.5°. Comparisons of windspeed by direction in the main
260
FORESTRY
-w
Figure 3. Wind direction estimated in the first MC2 trial (strong winds).
NNW—SSE valley were also performed for
Strongblow (Figure 5). They show a rather poor
ability to predict windspeed by direction. In
fact, the direction corresponding to the most
sheltered aspect in the wind tunnel is predicted
by Strongblow to have somewhat higher windspeeds than the other directions.
The effect of topographic position on windiness estimation was very highly significant for
each of the methods tested. With the wind tunnel approach, it was observed that high hills
were the most exposed, followed by low hills
and saddles, and lastly by the three valleys
(Table 5). Groupings obtained through the wind
tunnel, Strongblow and Topex-to-distance were
identical and those from the Topex, quite simi-
lar. Groupings from MC2 differed in some
aspects, saddles rating higher in windiness than
with the other methods.
Discussion
The wind estimation methods showed many
similarities in explaining the pattern of wind
exposure over the topography. However, important differences were found between some methods and in some cases for the particular wind
event that was modelled.
The wind tunnel study gave very good correlations with the other methods used in this
study. In fact, best correlations were generally
ESTIMATION OF WIND EXPOSURE FOR WINDTHROW HAZARD RATING
found with this method, except for the correlation between Topex and Topex-to-distance
which are two very closely related approaches.
Ranking of topographic features by overspeed
ratio leads to the formation of three groups corresponding to increased windspeeds with
increasing elevation. Such increases have been
reported in many other cases (Buck, 1964;
Miller, 1985; Robertson, 1986; Finnigan and
Brunet, 1995).
Even though wind tunnel studies can involve
some artifacts, reasonable correlations have
been found between field and model data
(Miller et al., 1987). Moreover, significant correlations were found between the amount of
windthrow after 5 years in our field study and
261
the wind tunnel estimation for winds blowing at
right angles to the buffer strips. Windspeed
from the west explained 39 per cent of the variation in windthrow (Ruel etal., in preparation).
Thus, wind tunnel simulations seem a reliable
way of obtaining a wind exposure estimation
over a limited area. They seem able to provide
a quantitative estimation of windspeed and a
proper ranking of sites, similar to other methods. Such a quantitative estimation would be
needed to assess the risk of wind damage over
an area (Gardiner and Quine, 1994; Quine,
1994) while a relative ranking of sites could
meet the requirements of the extensive forestry
practised in Eastern Canada. However, this
approach has some drawbacks. Hence, for
Figure 4. Wind direction observed in the wind tunnel trial corresponding to the general wind direction of the
first MC2 trial (strong winds).
FORESTRY
262
Strong blow
1.4
30
1.2
1.0-20
0.8-
• 11111 !
Ill 11IIII!
11II1111!
0.6
0.4
02
Bwind tum*l • strongMow
0
ME
SE
sw
w
Figure 5. Comparison between overspeed ratios from
the wind tunnel and windspeed estimated by Strongblow for each main compass direction.
larger areas, it will become increasingly difficult
to represent adequately the topography without
any vertical scale exaggeration or having to
resort to a larger wind tunnel. Such equipment
is often not available to the forest manager.
Numerical airflow models are based on
sounder physical principles than geographic
empirical predictors and do not require calibration for a particular site even though they
require an initializing windspeed (Gardiner and
Quine, 1994). However, these models often
experience some difficulty in dealing with even
relatively gentle slopes (>18 per cent) and tend
to become conservative in such areas (Hannah
etal., 1991; Gardiner and Quine, 1994).
The numerical model MC2 was relatively reliable in estimating windspeeds when they were
around 30 m s~", showing relatively good corre-
lations with the other methods. However, at
lower windspeeds, estimations became less reliable, even when compared with Strongblow at a
similar windspeed (intermediate windspeed).
This does not necessarily imply that this model
cannot be used for wind exposure estimation
since storm winds are the main concern in windthrow hazard evaluation. Ranking of sites at the
highest windspeed differed from other methods,
saddles being qualified as more exposed. This
could partly be explained by the fact that, for
the particular wind event monitored, many saddles offered little shelter and rather tended to
funnel the wind. This phenomenon is reflected
in increased correlations when this storm was
compared with the wind tunnel trial corresponding to the same general wind direction.
Comparison with the appropriate wind tunnel
trial did not succeed in improving the correlations at the two other windspeeds. At the highest windspeed studied, wind direction estimated
by this model showed little sensitivity to topography.
The modest performance of MC2 in this
study can be attributed to two factors. First,
only a simplified version of the model could be
used with the hardware available. This could
result in overestimating windspeeds and underestimating the effect of topography. Second,
some limitations appear inherent to the version
2.8 of the model. Since these simulations, version 3.0 of the model has been issued. Recent
trials on a severe storm that occurred in November 1994 show better agreement with wind
direction and speed monitored at different
weather stations; there are some indications that
Table 5: Wind exposure ranking by topographic position for the main methods
Wind tunnel
(overspeed ratio)
High hill
Low hill
Saddle
SW-NE valley
NNW-SSE valley
NW-SE valley
1.93A
1~52B
1.43B
1.13c
1.12c
1.10c
Topex
0.5A
18.5AB
27.2B
56.8c
69.4c
78.2c
MC2
(strong winds)
(m s-1)
Topex-todistance
(degrees)
Strongblow
(m s-1)
-96.0A
-50.0B
-17.0B
35.2A
30.8B
29.6B
36.0AB
34.5BC
49.6c
68.2c
75.8 C
26.6c
27.1c
26.1c
33.8CD
36.2A
32.7 D
30E
Note: Two values associated with the same letter are not declared significantly different by Fischer's protected l.s.d. at P •
0.05.
ESTIMATION OF WIND EXPOSURE FOR WINDTHROW HAZARD RATING
the simulations are more realistic even in less
homogeneous regions (S. Pellerin, Environment
Canada, personal communication).
Numerical models, such as MC2, represent
wind behaviour in detail for a specific storm but
do not provide an estimate of windiness over an
adequate time frame for a general site windiness
estimation. To do so, many runs would have to
be made and this would be computationally
expensive. In the present study, one complete
run including four different scales required 50
hours of computing time on a SUN SPARC10
workstation. Such models however remain very
informative when studying damage for a specific
storm. Recently, a very significantly advance has
been made with MC2 to reduce by a factor of
30 to 100 the cost of obtaining wind data over
high resolution orography by using a statistical
dynamical downscaling technique. This should
enable an application to much wider time
frames.
Strongblow correlated better than MC2 with
the other methods and required less qualified
personnel and reduced computing power. However, data are entered on a point by point basis
which means that derivation of the basic input
would be tedious for a large area. Even though
the representation of topography included in
this model is approximate, it was one of the
methods most closely related to the wind tunnel
for estimating windspeed and classifying topography. The inclusion of topographic features in
Strongblow is implemented by adjusting a
'topography factor'. However, a restriction on
flow separation limits the validity of the equations to shallow topography, which is equivalent to a limiting slope of about 30 per cent.
With steeper topography, all upwind slopes
behave approximately as if they were 30 per
cent (Cook et al., 1985). Since such slopes are
frequently exceeded in the topography typical of
the Laurentian Hills, this could be a serious
handicap of the method even though it did not
appear to invalidate the results in the present
study.
When comparing Strongblow with the other
methods, no roughness variations were included
in order to represent only the effect of topography as it was reproduced in those methods.
However, the possibility of incorporating land
use characteristics represents a potential advan-
263
tage of Strongblow. Thus, it would be possible
to account for clearcut size in the direction of
damaging winds and Steinblums et al. (1984)
have found this variable to be important in
explaining blowdown importance in riparian
strips.
Standard Topex determination has been in
use for some time and has proved to be in good
agreement with other measures of wind exposure assessment (Quine and White, 1993). In the
present study, standard Topex correlated reasonably well with the wind tunnel and Strongblow but results were improved with the
Topex-to-distance. Ranking of topographic features for wind exposure with Topex led to
results quite similar to the wind tunnel and
Strongblow whereas Topex-to-distance gave
identical rankings. Hannah et al. (1995) found
much better correlations between Topex-to-distance and windspeed measured on site, than
with Topex and windspeed. The performance of
the Topex-to-distance supports the statement
made by Quine and White (1994) regarding the
potential for improving the standard Topex by
placing a limit on the distance within which the
skyline angle is sought and by permitting negative angles. Under the standard Topex, both a
flat terrain and a hilltop would lead to a value
of 0. Under the modified version, however, the
flat terrain would remain with a value of 0 while
the hilltop would have a negative value. This
explains why the Topex-to-distance gave a
lower mean value and a higher coefficient of
variation. Even though both elevation and the
funnelling factor that have been proposed for
use in conjunction with the Topex-to-distance
(Quine and White, 1993) were not used here, the
results show that the Topex-to-distancc remains
useful in assessing windiness in a situation
highly conducive to such funnelling.
The Topex system alone does not consider
prevailing wind direction and this could represent a limitation of this approach. However,
even though this problem was not addressed in
this study, previous attempts to adjust Topex
scores for known directional differences have
been unsuccessful (Miller etal., 1987; Hannah et
al., 1995). To obtain a mean windspeed estimation for a particular site, calibration would be
needed. However, for comparative ranking purposes, such calibration is not needed. With the
264
FORESTRY
growing popularity of geographic information
systems, it becomes possible to present the terrain in digital form and to automatically derive
the Topex-to-distance or other similar value
(Wright and Quine, 1993; Quine and Bell, 1994).
This could represent a great advantage for
application over wide areas. To demonstrate
this application, a simple algorithm was developed. This algorithm searches for the maximum
difference in elevation in a given direction over
a finite distance. Then, the angle to this point is
calculated. This algorithm gave very good estimates of manually derived Topex-to-distance (r
= 0.972; P < 0.001) that correlates very well
with wind tunnel measurements (r = -0.930; P
< 0.001).
Conclusion
Each of the candidate methods was able to
estimate to some degree the level of wind exposure at a site. Topex, Topex-to-distance,
Strongblow and the wind tunnel results correlated very well with each other. Ranking of
topography for wind exposure was also quite
similar for these methods. For MC2, significant
correlations were obtained with other methods
at high windspeeds but not at lower windspeeds. In general, wind direction or the estimation of wind speed by direction was more
problematic, even though a more recent version
of the model MC2 should also be investigated.
Hence, performance of the Topex-to-distance,
combined with its simplicity and the potential
of this method to automation using digitized
contour maps would make it an appropriate
method for inclusion in a windthrow hazard
rating system where only a relative ranking of
sites is needed. However, for risk rating, where
estimations of windspeed are required, numerical models would prove helpful. They would
also be an interesting tool in detailed studies of
particular wind events.
Acknowledgements
The authors wish to acknowledge the organizations
that contributed financially or materially to the realization of this study, namely: the Quebec's Ministry
of Natural Resources (field windthrow monitoring),
the National Research Council of Canada (wind tunnel), the Atmospheric Environment Service of Environment Canada (model MC2 and professional
support) and Hydro-Quebec (funding).
References
Benoit, R., Pellerin, P., Mailhot, J. and Lee, V. 1994
Modelling of late spring intense orographic precipitation over the Canadian portion of the Columbia
river basin. 6th. Conf. on Mesoscale Processes.
American Meteorological Soc., 18—22 July, Portland.
Bergon, G., Laprise, R. and Caya, D. 1994 The
Mesoscale Compressible Community (MC2) Model
Scientific Formalism. Documentation manual available at RPN, 2121 TransCanada Road, 5th floor,
Dorval, Quebec, H9P 1J3, Canada.
Buck, C.C. 1964 Winds Over Wildlands—a guide for
forest management. USDA For. Serv. Agr. Handbook No 272, 33pp.
Busby, J.A. 1965 Studies on the stability of conifer
stands. Scott. For. 19, 86-102.
Busing, R.T. and Pauley, E.F. 1994 Mortality trends
in a southern Appalachian red spruce population.
For. Ecol. Manage. 66, 41—45.
Coates, D. and Steventon, D. 1994 Principles of Patch
Retention Harvesting. British Columbia Forest
Serv., Prince Rupert Forest Region, Extension Note
02, 5pp.
Cook, N.J., Smith, B.W. and Huband, M.V. 1985
BRE Program Strongblow. Supplement 2 to: The
designer's guide to wind loading of building structures. Building Research Establishment, United
Kingdom.
Cooper, K.R., Ruel J-C. and Pin, D. 1996 Wind Tunnel Measurements of the Surface Winds on a Model
of the Montmorency Forest to Investigate the Effect
of Topography on the Windthrow of Trees.
National Research Council of Canada, report
LTR-A-4, Ottawa, April 1996.
De Champs, J. 1987 Mesures sylvicoles preventives.
Rev. For. Fr. 39, 313-322.
Desgagne, M., Benoit, R. and Chartier, Y. 1994 The
Mesoscale Compressible Community (MC2) Model
User's Manual. Documentation manual available at
RPN, 2121 TransCanada Road, 5th floor, Dorval,
Quebec H9P 1J3, Canada.
Esseen, P.A. 1994 Tree mortality patterns after experimental fragmentation of an old-growth conifer
forest. Biol. Conserv. 68, 19-28.
Finnigan, J.J. and Brunet, Y. 1995 Turbulent airflow
in forests on flat and hilly terrain. In Wind and
Trees. M.P. Courts and J. Grace (eds). Cambridge
University Press, Cambridge, 113-129.
ESTIMATION OF WIND EXPOSURE FOR WINDTHROW HAZARD RATING
Flesch, T.K. and Wilson, J.D. 1993 Extreme Value
Analysis of Wind Gusts in Alberta. Forestry
Canada-Alberta Land and Forest Services, 46pp.
Franklin, J.F. 1990 The Contribution of Old Growth
to the New Forestry. Univ. of Alberta, Faculty of
Agriculture and Forestry, Forest Industry Lecture
No 24, 4 April 1990.
Gardiner, B. and Quine, C.P. 1994 Wind damage to
forests. Biomtmetics 2, 139-147.
Hannah, P.J., Guo, X. and Quine, C.P. 1991 Reconstruction of time series in complex terrain. Wind
Energy Conversion 1991. Proc. of the 13th British
Wind Energy Ass. Conf, Swansea 10-12 April
1991, 281-288.
Hannah, P., Palutikof, J.F. and Quine, C.P. 1995 Predicting windspeeds for forest areas in complex terrain. In Wind and Trees. M.P. Courts and J. Grace
(eds). Cambridge University Press, Cambridge,
113-129.
Inglis, D.W.F., Choularton, T.W., Stromberg, I.M.F.,
Gardiner, B.A. and Hill, M. 1995 Testing of a linear airflow model for flow over complex terrain
and subject to stable, structured stratification. In
Wind and Trees. M.P. Courts and J. Grace
(eds). Cambridge University Press, Cambridge,
113-129.
Irwin, H.P.A.H. 1979 Design and Use of Spires for
Natural Wind Simulation. National Research
Council, Canada. LTR-LA-233.
Irwin, H.P.A.H. 1980 A Simple Omnidirectional Sensor for Wind Tunnel Studies of Pedestrian Level
Winds. National Research Council, Canada. LTRLA-242.
Jonsson, B.C. and Esseen, P.A. 1990 Trecfall disturbance maintains high bryophyte diversity in a
boreal spruce forest. /. Ecol. 78, 924-936.
Laurance, W.F. and Yensen, E. 1991 Predicting the
impacts of edge effects in fragmented habitats.
Bio/. Conserv. 55, 77-92.
Miller, K.F. 1985 Windthrow hazard classification.
Forestry Commission Leaflet No 85. Forestry Commission, Edinburgh, 14pp.
Miller, K.F., Quine, C.P. and Hunt, J. 1987 The
assessment of wind exposure for forestry in upland
Britain. Forestry 60, 179-192.
Mitchell, S.J. 1995 The windthrow triangle: a relative
windthrow hazard assessment procedure for forest
managers. For. Chron. 71, 116 '150.
Quine, C.P. 1994 An improved understanding of
windthrow—moving from hazard towards risk.
Forestry Authority Research Information Note No.
257. Forestry Commission, Edinburgh, 6pp.
Quine, C.P. 1995 Assessing the risk of wind damage
to forests: practice and pitfalls. In Wind and Trees
(ed. M.P. Courts and J. Grace). Cambridge University Press, 379-403.
265
Quine, C.P. and Bell, P. 1994 Digital terrain models
in windthrow research. For. Br. Timber May,
28-32.
Quine, C.P. and White, I.M.S. 1993 Revised windiness scores for the windthrow hazard classification:
the revised scoring method. Forestry Commission
Research Information Note No 230. Forestry Commission, Edinburgh.
Quine, C.P. and White, I.M.S. 1994 Using the relationship between rate of tatter and topographic
variables to predict site windiness in Upland
Britain. Forestry 67, 245-256.
Robertson, A. 1986 Estimating mean windflow in
hilly terrain from tamarack {Lanx laricina (Du
Roi) K. Koch) deformation. Int. J. Biometeorol. 30,
333-349.
Robertson, A. 1987 The use of trees to study wind.
Arboric. ]. 11, 127-143.
Ruel, J-C. 1995 Understanding windthrow: silvicultural implications. For. Chron. 71, 434—445.
Ruel, J-C. and Pin, D. 1993 Windthrow along electrical distribution lines in a rural setting. J. Arboric.
19, 271-277.
Ruth, R.H. and Yoder, R.A. 1953 Reducing wind
damage in the forests of the Oregon Coast Range.
USDA For. Serv. Pac. North. For. and Range Exp.
Srn. 30pp.
Schaetzl, R.F., Burns, S.F., Johnson, D.L. and Small,
T.W. 1989 Tree uprooting: review of impacts on
forest ecology. Vegetatto 79, 165-176.
Seymour, R.S. and Hunter, M.L. 1992 New Forestry
in Eastern Spruce-Fir Forests: Principles and Applications to Maine. Maine Agric. Exp. Stn., Misc.
Publ. 716, 36pp.
Steinblums, I.J., Froehlich, H.A. and Lyons, J.K. 1984
Designing stable buffer strips for stream protection. ;. For. 82, 49-52.
Stephens, E.P. 1956 The uprooting of trees: a forest
process. Soil Sci. Soc. Am. Proc. 20, 113—116.
Tanguay, M., Robert, A. and Laprise, R. 1990 A
semi-implicit semi-lagrangian fully compressible
regional forecast model. Mon. Weath. Rev. 118,
1381-1395.
Thibault, M. and Hotte, D. 1985 Les Regions
Ecologiques du Quebec Miridional. Deuxieme
approximation. Color map at the scale 1/1 250
000. Ministere de I'Energie et des Ressources,
Service de la Recherche et Service de la Cartographic.
Wilson, J.D. 1984 Determining a TOPEX score.
Scott. For. 38, 251-256.
Wollenweber, G.C. and Wollenweber, F.G. 1995 Forest wind damage risk assessment for environmental
impact studies. In Wind and Trees. M.P. Courts
and J. Grace (eds). Cambridge University Press,
Cambridge, 404-^»23.
266
FORESTRY
Wright, J.A. and Quine, C.P. 1993 The use of a
geographical information system to investigate
storm damage to trees at Wykeham Forest, North
Yorkshire. Scott. For. 47, 166-174.
Zack, J.A. and Minnich, R.A. 1991 Integration of
geographic information systems with a diagnostic
wind field model for fire management. For. Sci. 37,
560-573.
Received 18 April 1996