Aggregation procedures in forest management planning using cluster analysis

Typical linear programming models used in forest planning can be very large. It is often of interest to analyze more compact, less detailed versions. One form of reducing the size of the problem is through an aggregation process. One way in which this has been done is through a column aggregation process, where sets of similar columns are replaced by one representative. A second alternative is to aggregate the original data, in which case the stands and management alternatives are grouped before building a model. Typical approaches for the aggregation processes have been analytical. We present an alternative approach for aggregation based on cluster analysis. Computational results for both types of aggregation show that using cluster analysis can be advantageous.