Operational planning is normally based on stands as the primary unit of tre
atment. In recent years, interest has been directed towards the use of smal
ler area units such that the formation of treatment units becomes part of t
he operational planning. Here, the method of genetic algorithms was used in
order to delineate harvesting units in a forest described as a grid of 20
m x 20 m pixels. Remote sensing data and inventory plots were employed to d
erive pixel estimates of forest variables. Appropriate parameter settings f
or the algorithm were investigated for two cases that consisted of 26 x 25
pixels. Based on the experience from these tests, the algorithm was employe
d in forming treatment units in an area encompassing 100 x 100 grid cells,
or 400 ha. Due to the size of the problem, a two-stage procedure was sugges
ted. The solutions of the genetic algorithm were compared with a heuristic,
a cell removal algorithm, that was proposed in earlier studies of treatmen
t unit formation. The genetic algorithm performed better than the heuristic
for all the cases that were tested. (C) 2000 Elsevier Science B.V. All rig
hts reserved.