Model-based growth and yield
comparisons between even-aged and
uneven-aged forests – evaluation of
current state-of-art in Finland
Jari Hynynen, Risto Ojansuu & Kalle Eerikäinen
CCF Workshop, 21.–22.8. 2013, Uppsala
”New” management alternatives
Due to the liberalization of forest management diversity of forest
management practices is likely to increase
– Short rotation management: ”new” tree species and new intensive
management regimes (pulpwood and energy wood production)
– Longer rotations, extensive forestry
– Intensive thinnings from above and selective cuttings
– Uneven-aged forest management
Focus on assessing the impacts of management alternatives on
–
–
–
–
Sustainability from the viewpoint of wood production and profitability
Biodiversity, landscape, carbon sequestration, nature tourism, …
At local, regional and national scales
Short term and long term impacts
Assessment of impacts set high demands for growth models
The most challenging task is to assess the impacts of CCF
Role of models in decision support
Model-based calculations are prevailing methods to apply
research-based information in decision support
Models are simplifications and predictions include
uncertainty
Measured data are required in model development:
models are never completely data independent
Applicability of models outside their “comfort zone”
depends on their extrapolation capability
An example on extrapolation capability of a model
Number of
seedlings
measurements
Model fitted to measurements
?-
applicability of a
model
”correct respones"
Sound theorethical background
provides better prerequisite for
model extrapolation
Interpolation
Extrapolation
Number of seed trees
Uneven-aged forest as a modelling challenge
Regeneration and early growth
– How many trees will be born and how they are distributed
– How many of them will survive
– Growth rate of seedlings
=> Ingrowth
Development of established trees
– Growth response to increased growing space (decreased competition)
– Damage risks
Which population a model represents?
– Subjective sampling (designed experiments and sample plots)
– Small number of experiments, need for intensive measurements
– Different models are based on different modelling data sets => Can we
compare the output of these models?
How to take into account large between-stand variation?
– e.g. large variation in ingrowth between stands
Comparison between even- and uneven-aged stands
Simulation of transition phase from even-aged to uneven-aged
An example of economically optimal management
of uneven-aged stand in equilibrium stage
(Tahvonen 2011)
Fig. 8. Stand structure in 5 cm size classes, temperature sum 1300 degree-days. (a) Interest rate 0.01; (b)
interest rate 0.02.
Treatment : Selective cutting with removal of 60 – 80 % of stand basal area
An example of economically optimal management of
uneven-aged stand in equilibrium stage
Tahvonen (2011)
Assumptions behind the applied
growth model (Pukkala et al.
2009):
• All trees are assumed to locate so
that they have adequate growing
space until harvesting
• Growth rate of trees is affected by
Fig. 8. Stand structure in 5 cm size classes,
temperature sum 1300 degree-days. (a) Interest
rate 0.01; (b) interest rate 0.02. Tahvonen (2011)
Stand development after cutting?
•
•
•
The impact of uneven spacing of trees?
Adaptation of trees to new competitive status?
Risks of damages?
– tree diameter
– stand density after selective
cutting
– relative tree size
= treatment history does not
affect tree growth
Some properties of existing models applied
to predict dynamics of uneven-aged forests
Regeneration and early growth
– Inadequate description of ingrowth
– Relevant properties, which are missing
• Amount of regeneration incl. large variation between stands
• Growth rate of seedlings & its variation
• Spatial distribution and the impacts of clustering
Response to selective cutting is inadequately described
– adaptation of trees to changed growing conditions
Damage risks are ignored
– Wind damages
– Root rot
Preliminary study based on measurement data:
Performance of models developed for even-aged stands in the
prediction of stand dynamics of uneven-aged stands
9
Evaluation data
Repeatedly measured growth and yield experiments
located in
– 6 even-aged
– 15 uneven-aged
stands of Norway spruce in Southern Finland
stands have been measured with 5-year interval
10-year growth period
– tree diameter growth
– tree height growth
10
Data from uneven-aged stands
Exp.
VES01
VES02
VES05
VES07
VES13
VES14
EVO02
EVO03
EVO04
LAP01
LAP05
LAP07
LAP13
VEP02
VEP04
Mean
Site type Dominant
age, yrs.
MT
63
MT
63
MT
63
MT
63
MT
88
OMT
100
MT
70
MT
70
MT
70
OMT
94
OMT
49
OMT
76
MT
55
MT
75
MT
75
72
Da,
cm
10.5
9.9
10.5
12.2
10.0
17.7
14.2
9.3
12.1
6.3
14.6
11.4
11.3
17.4
10.3
11.8
Dg,
cm
22.0
19.3
23.0
21.9
21.3
26.0
26.3
22.8
21.8
21.8
26
22.9
23.8
29
23.4
23.4
Stem number
ha-1
1578
1563
1078
918
925
481
694
731
1473
2293
1076
1318
1062
474
1248
1127
11
Volume,
m3ha-1
165
129
130
130
98
167
164
91
191
117
283
205
164
153
149
156
Data from even-aged stands
Exp.
LH505
LH511
LH513
Nyn01
NYN03
NYN04
NYN05
Mean
Site type Dominant
age, yrs.
EMT
53
OMT
39
OMT
54
MT
40
OMT
38
OMT
38
OMT
30
41.7
Da,
cm
13.5
13.0
12.1
11.7
10.8
12.2
10.2
11.9
Dg,
cm
16.9
15.2
19.4
13.2
16.3
16.2
11.7
15.6
Stem number
ha-1
1680
2055
1860
1920
2530
2060
3540
2235
12
Volume,
m3ha-1
142
202
153
129
190
187
194
171
Diameter distribution of an even-aged
and an uneven-aged stand
N ha-1
300
250
200
150
100
50
0
4
8
12
16
Even-aged (LH511)
20
24
28
32
Uneven-aged (VES05)
36
40
44
dbh, cm
13
Simulation and analysis
The development of each stand was simulated for 10year growth period with MOTTI simulator
Data from the first measurement were used as initial
data of the simulation
Simulated growth of each tree was compared with
observed growth
– bias = observed growth - predicted growth
– relative bias =100*(observed - predicted)/predicted
14
Average relative bias:
Diameter growth prediction
Relative bias,
%
60
40
20
-43.5%
-10.9%
0
-20
-40
-60
Uneven-aged stands
Even-aged stands
15
Average bias of the diameter growth prediction
Bias, cm
0,8
0,6
0,4
0,2
0
-0,2
-0,4
-0,6
-0,8
0
5
10
15
20
25
30
35
40
45
Tree diameter, cm
Even-aged
Uneven-aged
16
Average relative bias:
Tree height growth prediction
Relative bias, %
150
100
14.7%
32.7%
50
0
-50
-100
-150
Uneven-aged stands
Even-aged stands
17
Average bias of height growth prediction
Bias, cm
40
30
20
10
0
-10
-20
-30
-40
0
5
10
Even-aged
15
20
Uneven-aged
25
30
Tree height, m
35
18
Analysis of growth predictions for uneven-aged stands
Diameter growth
– small trees: systematic overprediction with trend
• model flattens out the effect of within-stand competition on the growth of
small trees
• suppressed trees were poorly represented in the modeling data obtained
from managed, even-aged stands
– large trees: systematic, but smaller overprediction, no trend
– bias in large trees (partly) originates from biased site index prediction
Height growth
– small trees: only small bias, no trend
– large trees: seriously biased
• systematic overprediction
• increased bias with increasing tree height
Conclusion
– poor performance of models for even-aged stands when applied to
uneven-aged stands
• Models include driving variables not applicable in UEF: H100, Hdom, Ddom
• Modelling data: different size distribution, different spatial distribution, etc.
• Different response to treatments (thinnings)
19
Conclusions
Applicability of existing growth and yield models
– Comparisons on treatment responses in even-aged stands
– Approximate predictions of the treatment responses of
established trees in uneven-aged forests
Model-based approach – not yet reliable for predicting
– Natural regeneration in continuous cover forestry
– Long-term dynamics of unmanaged stands
– Damage risks
• In intensively managed forests
• In old and/or unmanaged stands
• In uneven-aged forests
Current models are not reliable enough to be applied in economical
calculations on profitability of uneven-aged forestry or comparisons
with even-aged forestry
– Useful for research purposes
– Not useful for decision support in practice