Bruce Carroll, Vaughan Landrum, and Lisa Pious

Timber Harvest Scheduling With Adjacency Constraints: Using ArcInfo To Make FORPLAN Realistic

ABSTRACT

FORPLAN is used by the Forest Service and private industry to schedule 
the management of forestland over time.  However, FORPLAN is non-
spatial and cannot model Oregon, Washington and California regulations 
that constrain timber harvesting based on whether adjacent stands have 
been recently harvested.  Therefore, FORPLAN may produce solutions 
that cannot be implemented in the field.  To resolve this problem, Boise 
Cascade Corporation retained Pacific Meridian Resources to link 
ArcInfo with FORPLAN to produce an application (the Spatial 
Feasibility Test, or SFT) that tests the feasibility of FORPLAN solutions 
against adjacency constraints.  SFT applies a FORPLAN solution to a 
coverage of timber stands (up to 100,000 polygons have been tested).  The 
user specifies a maximum adjoining size for any set of silvicultural 
treatments (e.g. maximum of 120 acres for clearcuts) in any period.  SFT 
can also apply a regulation in which stands that are adjoining for less than 
a given percent (e.g. ten percent in Washington) of their perimeter are not 
considered adjacent.  SFT produces reports and coverages that itemize 
how well SFT was able to apply the FORPLAN solution to the ground.  
Boise Cascade will use SFT in an iterative manner:  Run FORPLAN; run 
SFT to apply adjacency constraints; re-formulate FORPLAN to more 
closely match SFT outputs; re-run FORPLAN and SFT; and continue until 
SFT is able to fully allocate the FORPLAN solution.  The resulting 
FORPLAN solution will then be implementable under adjacency 
constraints.



FORPLAN

FORPLAN is a matrix-generator for linear programming that is 
specifically designed for management of large areas of forestland.  
FORPLAN accepts tabular information on the landbase (i.e. acres of each 
stand type), potential silvicultural and management prescriptions for each 
stand type, revenue and cost streams associated with each prescription, 
and any number of constraints on the landbase (e.g. harvest levels must be 
non-declining over time).  FORPLAN then submits the matrix to a linear 
program that attempts to maximize an objective (e.g. maximize lumber 
production) subject to the given constraints.  The result is a FORPLAN 
allocation of the prescriptions that must be applied to a given acreage of 
each stand type in each time period to produce the maximum objective.

The US Forest Service originally developed FORPLAN to help each 
National Forest develop Management Plans as required by law.  Boise 
Cascade Corporation uses a customized version of FORPLAN on its 
corporate timberlands to predict the potential benefits of management.



ADJACENCY CONSTRAINTS

The primary drawback to FORPLAN is that it is non-spatial.  There is no 
guarantee that a FORPLAN solution will be implementable, since 
FORPLAN cannot analyze site-specific cases such as seasonal 
inaccessiblity or logging practicalities.  When adjacency constraints are in 
effect, as they are in Washington, Oregon and California, FORPLAN 
alone may vastly overestimate the potential benefits of management.  A 
typical regulation is:

Oregon:  Maximum clearcut size of 120 acres, 4 year lag between 
adjoining harvests.

This implies the following adjacency constraint:

The combined area of adjoining polygons allocated to the clearcut 
prescription must not exceed 120 acres in any 4 year span.

Because of this adjacency constraint, there may be no way to distribute 
harvests across the landbase and meet the FORPLAN allocation while 
complying with the adjacency constraint.



SPATIAL FEASIBILITY TEST

To solve this problem, Boise Cascade retained Pacific Meridian Resources 
to produce SFT, the Spatial Feasibility Test.  SFT is an ArcInfo based 
application that attempts to "disaggregate" a FORPLAN solution by 
applying it to an actual coverage of stand polygons.  SFT can apply 
adjacency constraints to any silvicultural treatment code (such as clearcut) 
that the user specifies, with an associated maximum acreage (such as 120).

SFT is not an optimizer.  It progresses through the list of stand polygons, 
and attempts to allocate the current polygon according to the FORPLAN 
solution.  SFT checks the polygon's neighbors for adjacency violations for 
all time periods (e.g., there may be 20 5-year periods, for a simulation total 
of 100 years).  If none are found, the polygon is assigned the prescription 
specified by FORPLAN, and SFT moves to the next polygon.

This method is deterministic, based solely on the order in which the 
polygons are checked.  To counter this weakness, SFT allows the user to 
re-sort the polygon list on any database item, including a random number, 
and submit numerous runs to SFT for the same adjacency constraints.  
SFT automatically compares the set of runs, and ranks them according to a 
user-specified criteria.

The user can then apply the result of a chosen run to the original stand 
coverage, and produce maps and analyses using standard ArcInfo and 
ArcView.

Figures 1 through 3 demonstrate SFT output.  Figure 1 shows the effect of 
adjacency constraints on harvest scheduling.  Shaded areas are scheduled 
for harvest at some time in the 20 period run.  Unshaded areas are never 
harvested.  The unconstrained solution meets the FORPLAN allocation for 
the area.  The constrained solution falls far short of the FORPLAN goal.  
Figures 2 and 3 show the harvest pattern chosen by SFT for the first 4 
periods. 

ITERATIVE NATURE OF ANALYSIS
Boise Cascade is using SFT with FORPLAN in an iterative cycle.  
FORPLAN is run first, then numerous SFT runs are performed, using 
applicable adjacency constraints.  If SFT is able to allocate the full 
FORPLAN solution, no further cycling is needed, since SFT has proven 
that the FORPLAN solution is attainable given current adjacency 
constraints.

If SFT cannot allocate the full FORPLAN run, then FORPLAN will be 
rerun with tighter constraints.  For example, if the best SFT run was able 
to allocate only 75% of the FORPLAN solution, then FORPLAN may be 
rerun with an objective that is 75% of the previous FORPLAN run.  SFT is 
then run on the FORPLAN output, and the iterative cycle continues.  
When SFT is able to allocate the entire FORPLAN solution, then Boise 
Cascade is assured that the FORPLAN goals are attainable.  

SFT is not intended to be an intelligent harvest scheduler, as it does not 
create a logical harvesting pattern across the landscape based on 
transportation networks or working circles.  SFT's main purpose is to test 
the feasibility of Boise Cascade Corporation's FORPLAN runs under 
adjacency constraints.  With SFT, Boise Cascade Corporation will use 
FORPLAN with increased confidence that FORPLAN's results are 
realistic and implementable.  SFT is currently being enhanced to include 
such features as multiple adjacency constraints operating simultaneously, 
and the ability to allocate FORPLAN runs on a yearly basis.

Vaughan Landrum, Vice President
Pacific Meridian Resources
5915-B, Hollis St.
Emeryville, CA 94608
Tel: (510)654-6980
Fax: (510)654-5774
E-mail: pmr@crl.com

Lisa Pious, Senior Programmer
Pacific Meridian Resources
5915-B, Hollis St.
Emeryville, CA 94608
Tel: (510)654-6980
Fax: (510)654-5774
E-mail: pmr@crl.com

Presenting Author:
Vaughan Landrum