This presentation investigates the role of forest structure in limiting the achievement of harvest and/or revenue goals of forest management. Results show that some spatial arrangements of stands suffer much greater reductions in goal achievement than others. Some means of ameliorating the reductions is also given.

1. Forest Structure & Spatial Restrictions:
Interactions & How They Affect
Harvest Goal Achievement
Karl R. Walters
Forest Planning Manager
1
© 2004 Forest Technology Group

2. Spatially explicit harvest scheduling
Stands are not spatially configured the way we would prefer them:
Stands are smaller than harvest blocks
Harvest blocks must exceed minimum size to be economical
Stands are larger than harvest blocks
Harvest blocks must not exceed some regulatory maximum opening size
Forest structure is too dispersed
Costs are minimized by harvesting large tracts in close proximity
Forest structure is too concentrated
Diversification of activities across a landscape is needed
Social and political limitations
Societal demands lead to legal remedies or self-regulation
Spatial restriction rules on opening size, adjacency, green-up
2

3. Stanley
Based on a hierarchical approach to planning
Solve strategic harvest schedule first
Allocate subset of harvest schedule tactically
Iterate as needed
Area-restriction model approach
Block configuration is not an input but part of solution
Heuristic solution methodology
3

4. Stanley input parameters
Opening size limitations on harvest blocks
Minimum feasible block size (economic considerations)=MINPARM
Maximum allowable block size (BMP’s, SFI, state law)=MAXPARM
Adjacency restrictions on harvest blocks
Minimum buffer distance (proximity to contemporary blocks)=PROX
Greenup (period to elapse before cutting adjacent block)=GREENUP
Adjustments to strategic plan for spatial feasibility
Deviations from strategic timing choices=TIMEDEV
% Variation relative to optimal flow profile=FLOWPARM
4

5. Stanley Input Parameters
FlowParm
Blue bars = strategic objective
Yellow bars = achieved volume
Maximum % variation is lowest
achieved % minus highest achieved %
5

6. Study Objectives
To quantify how forest structures interact with spatial restrictions
Why is there such variety in results in the literature?
Is there a reproducible metric that will predict how much of an impact
spatial restrictions will have in a given forest?
To gain a better understanding of the process so as to improve
implementation and/or affect policy
How much difference would it make to increase minimum block size
from 5ha to 10 ha? What is the impact of increasing the green-up
interval by 1 year?
Today, we will examine how GREENUP, PROX, TIMEDEV and
MINPARM together affect harvest level achievement
6

7. Experimental Design
Forest structure
32,000 ha, uniform site quality, single species
Uniform age-class distribution (1-40), 800 ha each
Volume objective, non-declining flow constraint
Minimum harvest age = 19
Planning horizon
Strategic = 80 yrs, tactical = 15 yrs
Varied spatial distribution of stands
Square or hexagon grids
20 ha cells (1600 cells in a 40x40 grid of squares & hexagons)
5 ha cells (6400 cells in a 80x80 grid of squares)
7

8. HX40R – 20 ha, random
8

9. SQ40R – 20ha, random
9

10. SQ80R – 5ha, random
10

11. SQ40S – 20 ha, systematic-random
11

12. SQ40C – 20ha, clustered
12

13. SQ40CS – 20 ha, systematic-cluster
13

14. Experimental Design
Stanley blocking and scheduling software, 15 year planning horizon
PROX – 3 levels of proximity (adjacents, 2nd ring, 3rd ring)
GREENUP – 3 levels of greenup interval (3, 4 & 5 years)
MINPARM – 2 levels (minimum 20 and 40 ha blocks)
MAXPARM – 3 levels (maximum 120, 180 and 240 ha blocks)
TIMEDEV – 3 levels (±5, ±10, ±15 yr deviations from timing)
FLOWPARM – 2 levels (5% & 10% variation about flow)
Factorial experiment
3888 obs. (6 factors x 54 levels x 2 replicates x 6 forests)
14

15. Ranking of Solutions
Highest average harvest volume AvgVol MinVol MaxVol BlockAvg
HX40R SQ80R 0.684180 0.658666 0.711808 56.71373
SQ40CS 0.539730 0.511570 0.575562 117.53030
Lowest periodic harvest volume SQ40C 0.692941 0.672847 0.718571 97.61975
SQ40S 0.550097 0.521164 0.575552 99.33364
SQ40CS HX40R 0.703730 0.676697 0.730613 55.82006
Highest periodic harvest volume SQ40R 0.656411 0.629567 0.683752 64.03164
HX40R AvgVol MinVol MaxVol BlockAvg
SQ80R 3 3 3 5
Largest average blocks SQ40CS 6 6 5 1
SQ40CS SQ40C 2 2 2 3
SQ40S 5 5 6 2
Smallest average blocks HX40R 1 1 1 6
HX40R SQ40R 4 4 4 4
15

16. ANOVA – 20 ha cell configurations
FACTOR F Value Pr(F)
Null hypothesis: no differences in PROX 2261.404 0
GREENUP 1410.437 0
average harvest due to spatial TIMEDEV 540.237 0
MINPARM 241.224 0
configuration MPFD 240.700 0
GREENUP:PROX 185.640 0
Rejected, alpha = 1% FLOWPARM 135.775 0
MAXPARM 117.744 0
All parameters significant at 1% MINPARM:MPFD 112.221 0
PROX:MPFD 71.509 0
Significant interaction MINPARM:TIMEDEV 52.115 0
GREENUP:MPFD 40.881 0
PROX:MAXPARM 27.184 0.0000002
GREENUP & PROX PROX:FLOWPARM 26.202 0.0000003
PROX:TIMEDEV 17.775 0.0000256
MPFD & MINPARM MINPARM:TIMEDEV:MPFD 17.355 0.0000318
GREENUP:PROX:TIMEDEV 14.720 0.0001272
MINPARM & TIMEDEV FLOWPARM:MPFD 11.714 0.0006285
GREENUP:FLOWPARM 9.747 0.0018124
MINPARM:FLOWPARM 8.721 0.0031699
FLOWPARM:TIMEDEV 8.521 0.0035367
PROX:MINPARM 8.259 0.0040816
MAXPARM:MPFD 4.919 0.0266416
PROX:MAXPARM:TIMEDEV 4.333 0.0374717
GREENUP:MINPARM:MPFD 4.123 0.0423895
16

17. Response of AvgVol to Greenup:Prox
17

18. Response conditioned by MPFD
MPFD MPFD
MPFD MPFD M PFD
18

19. SQ40C
Clustered age classes 11
Sensitivity to proximity
0.617
distance increases with greenup
9
interval
PROX
0.6
7 48
0.6
78
Relatively insensitive to 0.7
09
proximity at greenup = 3
5
0.7
39
3.0 3.5 4.0 4.5
Relatively insensitive to timing GREENUP
choice deviations for smaller
minimum blocks 0.67
8
Deviations initially help by
13
allowing more area to be 11
TIMEDEV
harvested 9
Blocks become more 0.678
7
heterogeneous in composition 5
20 25 30 35
MINPARM
19

20. SQ40CS
Clustered, systematic age classes 11 0.458
Sensitivity to proximity distance
increases with greenup interval
9
0.
39
7
PROX
Sensitivity to proximity distance is 0.45
7 8
higher at short greenup intervals
0.51
8
0.57
9
5
than SQ40C 0.63
9
0.700
3.0 3.5 4.0 4.5
More sensitive to timing choice GREENUP
deviations for smaller minimum
blocks than SQ40C 0.4
58
Deviations initially help by
0.51
13 8
allowing more area to be harvested 11
TIMEDEV
Blocks become more 9
0.518
heterogeneous in composition 7
5
20 25 30 35
MINPARM
20

21. SQ40S
Systematic, random age classes 11
0.456
Sensitivity to proximity distance
increases with greenup interval
9
PROX
0.45
6
Sensitivity to proximity distance is 7
0.519
higher at short greenup intervals
5 0.58
1
Relatively insensitive to timing
0.7
07
0.6
44
3.0 3.5 4.0 4.5
choice deviations for smaller GREENUP
minimum blocks
Deviations initially help by 0.4
56
0.51
allowing more area to be harvested
9
13
0.58
1
Blocks become more
11
TIMEDEV
heterogeneous in composition; 9
0.519
more pronounced than in 7
clustered age classes
56
0.4
5
20 25 30 35
MINPARM
21

22. SQ40R
Random age classes 11
Much lower sensitivity to 9
proximity at all greenup 0.6
2
PROX
9
intervals 7
Relatively more sensitive to 5
0.699
timing choice deviations than 3.0 3.5 4.0 4.5
GREENUP
clustered age classes
Blocks become more
heterogeneous in composition;
0.6
29
13
more pronounced than in 11
TIMEDEV
clustered age classes 9
0.629
99
0.6
7
9
0.55
5
20 25 30 35
MINPARM
22

23. HX40R
Random age classes
8
Insensitive to proximity at
0.6
7
5
9
short greenup intervals 6
PROX
More sensitive to proximity at
0.6
87
5
higher greenups than SQ40R 4 0.71
5
3
More sensitive to timing choice 3.0 3.5 4.0 4.5
GREENUP
deviations than clustered age
classes and SQ40R
13
0.68
7
11
TIMEDEV
9
0.687
0.659
5
71
7
0.
5
20 25 30 35
MINPARM
23

24. SQ80R
Random age classes
Relatively insensitive to 0.6
17 30
proximity at short greenup 13
PROX
intervals 9
Less sensitive to proximity at
0.
higher greenup intervals than
5
70
0
3.0 3.5 4.0 4.5
GREENUP
SQ40R
Decreasing sensitivity to timing
choice deviations at higher 13
minimum block sizes 11
TIMEDEV
9
0.630
7
0.700
5
20 25 30 35
MINPARM
24

25. Summary
Identical forests from a strategic standpoint
Yielded very different outcomes under spatial restrictions
Forest structure was significant determinant
Although contrived, forests have analogs in the real world
SQ40C – similar to disturbance dominated natural forests
SQ40CS – similar to plantation management of the US southeast
SQ40R – similar to forests of the Northeast (small, heterogeneous
stands)
25

26. Summary
Forest fragmentation
Highly fragmented forests are less sensitive to greenup interval
Neighboring stands are not likely to be harvested in same period anyway
In non-fragmented forests, sensitivity to proximity distance
increases with greenup interval
Neighboring stands are of similar age, and therefore likely candidates for
harvesting in same period; proximity distance determines how much of
this area is made unavailable during greenup
Allocation units
Substands (SQ80R) yielded better solutions than stands (SQ40R)
Smaller allocation units present more alternative block configurations
26

27. Summary
Block size
Minimum block size can be much more limiting on volume
achievement than maximum block size
Maximum block size is limiting only in non-fragmented forests
Mean block size is much smaller than maximum allowed
Mean approaches maximum only in clustered forests (SQ40CS, SQ40C)
Timing choice deviations
Mitigate shortfalls by allowing for the creation of larger harvest
blocks (sensitive to minimum block size)
Significant deviations from original timing can nullify gains arising
from increased harvest area
27

28. Forest Technology Group
Knowledge for Growth
28
© 2004 Forest Technology Group