user_name Erik Valinger
user_organisation Swedish University of Agricultural Sciences
Faculty of Forestry
Department of Silviculture
S-901 83 Umea
Sweden
date 5th June 1996
model_name Tree characteristic model
description Describes probability of snow and wind damage at sites using
tree characteristics
software Developed in SAS 6.10 but Excel 7.0 will do
operating_system PC Windows 95
isumm There are three input variables at this stage of model development.
The risk of damage is modelled using upper diameter and height/diameter
at breast height ratio.
Valid only for Scots pine.
ispecfile
P
D
A
H
D
U
h
S
C
p
p
M
E
N
p
l
a
g
e
i
p
d
t
l
r
r
a
a
o
r
o
m
e
i
a
p
q
a
o
o
o
x
s
r
o
t
a
g
m
e
n
s
p
p
D
t
t
b
g
h
e
r
d
u
b
d
i
c
h
e
t
t
d
a
r
a
o
c
e
i
g
e
m
o
o
r
a
e
r
o
  1    0    17   10.5  16.0   12  0.65625  35    11  7.20917.2091 16.0 155890070967005.612
iexpspecfile Plot number
Damaged=1, undamaged=0
Age in years
Height measures in metres
Diameter measures in centi-metres
Upper diameter in centi-metres
HDQ in m/cm
Standage, year
closure in %
Proportion of birch, as a percentage %
Proportion of deciduous trees, as a percentage %
Maxd, maximal diameter at the plot, cm
Eastern co-ordinate in Swedish grid
Northern co-ordinate in Swedish grid
Calculated probability of damage from snow and wind, %
ivar1 upper diameter
ifile1 /air/sweden/acscotp.xls
idescr1 Diameter at 3 or 5 metre, cm
isensitiv1 Sensitive
isensdescr1 Logistic procedure
ivar2 hdq
ifile2 /air/sweden/acsotp.xls
idescr2 Height to diameter at breast height ratio
isensitiv2 Not Very Sensitive
isensdescr2 Logistic procedure
csumm Upperdia -0.0855
HDQ -4.0454
Model acprob=1-(1/(1+exp(0.8582-(0.0855*ovdia)-(4.0454*hdq))))
(then multiply by 100 to get the result in percent)
cspecfile None
cexpspecfile none
osumm acprob
ospecfile Excel sheet air/sweden/acscotp.xls
oexpspecfile See above
ovar1 acprob
ofile1 air/sweden/acscotp.xls
odescr1 Site probability of snow and wind damage
stat_meth Logistic procedure for probability of damage from snow and wind at a
site using tree characteristics when dominating species is
Scots pine

Chi-square for covariates
6.524 with 2 DF (p=0.0383)

Pr > Chi-square
Intercept 0.6283
Upperdia 0.0304
HDQ 0.1133
mech_meth  
log_meth Logistic model
acprob=1-(1/(1+exp(0.8582-(0.0855*ovdia)-(4.0454*hdq))))
(then the result is multiplied by 100 to get it in percent)
scope Model developed from a sample of data from the county of
Vasterbotten in northern Sweden (286 sample plots). Thirteen
of the plots had one or more damaged sample trees, damage=1.
Model have been proven to predict damage in a neighbouring well, but fails to predict damage
in a county in southern Sweden, i.e. have to be further improved for other geographic areas.
other ?
copyright Swedish University of Agricultural Sciences
Faculty of Forestry
Department of Silviculture
contact Associate Professor Erik Valinger