Case-Based Support for Forestry Decisions: How to See the Wood from the Trees

In forestry, it is important to be able to accurately determine the volume of timber in a harvesting site and the products that could potentially be produced from that timber. We describe new terrestrial scanning technology that can produce a greater volume of higher quality data about individual trees. We show, however, that scanner data still often produces an incomplete profile of the individual trees. We describe Cabar, a case-based reasoning system that can interpolate missing sections in the scanner data and extrapolate to the upper reaches of the tree. Central to Cabar’s operation is a new asymmetric distance function, which we define in the paper. We report some preliminary experimental results that compare Cabar with a traditional approach used in Ireland. The results indicate that Cabar has the potential to better predict the market value of the products.