Jc. Pierrat et al., MODELING OF THE STRUCTURE OF A STAND AND SIMULATION OF LOCAL AND SELECTIVE THINNING STRATEGIES, Canadian journal of forest research, 27(6), 1997, pp. 840-848
A method to predict the diameter distribution of a stand after a local
and selective thinning is presented, knowing the initial distribution
and taking Into account the spatial structure of the stand. It is sup
posed that trees can be ranked according to some criterion (e.g., diam
eter) and that the selection of trees to be thinned is based on their
local rank (i.e., rank in a plot). The stand is considered to be divid
ed into plots containing a constant number of trees and a plot is desc
ribed by the vector of numbers of trees In each class, For a randomly
sampled plot, the probability law of this vector is modeled by a mixtu
re of multinomial distributions, which reflects the spatial structure
at the plot scale. The parameters of the mixture are estimated by maxi
mum Likelihood with the EM algorithm, using a sample of plots. Then, f
or each component of the mixture, the probability laws of the order st
atistics are computed, from which the expected number of thinned trees
by class and by component is deduced Summing over the components give
s the expected number of thinned trees by class for the whole stand. T
he model is applied and validated on a real oak stand, In conclusion,
other possible applications are described, in particular to local and
selective thinning strategy in mixed stands.