AN EXPERT-SYSTEM TO SUPPORT SITE PREPARATION DECISIONS RELATED TO REFORESTATION

Site preparation is commonly employed in forestry in order to improve reforestation success. Application of the proper treatment can affect the future ecology of the site significantly. Site preparation decisio ns are naturally complex as they relate to processes that involve inte raction of a multitude of ecological factors. There are three major ty pes of treatment - prescribed burning, mechanical, and chemical. The p referred treatment depends on the site, and for a given treatment, det ails of application have to be carefully chosen. A treatment may have several ecological outcomes. Of these, a crucial one is the effect of the treatment on the seedling productivity. It is important to be able to predict this effect in order to choose the most appropriate treatm ent. However, no analytic models to predict the effects of a treatment on a given situation exist. Moreover, the outcomes are measured in te rms of multiple variables which interact in a complex and not fully kn own way to affect seedling growth on the site. In practice, treatment decisions are usually based on experience, expertise, manuals, and gui des that are frequently heuristic in nature. The knowledge required to predict the outcomes of a site preparation treatment spans diverse do mains such as the science of soils, plant ecology and forestry. This s ituation motivated the development of an expert system to predict the ecological effects of site preparation treatments. Presently, the pres cribed burning and mechanical components have been deployed in one of the forest regions of British Columbia. The chemical treatment compone nt is being prototyped. This paper describes the nature of the problem and the process of developing and deploying the expert system. Issues discussed include development of the conceptual knowledge model, reco nciliation of knowledge obtained from multiple sources of expertise, k nowledge base validation, user interface principles, system evaluation , and the introduction of the system into the field. Conclusions are d rawn as to the process of expert systems development for problems of t his nature.