An Agent Architecture for an Integrated Forest Ecosystem Management Decision Support System

A wide variety ofsofiware tools are available to s~ipport decision making in the management of ,forest eco.systems. These tools include rfatabase.~, growth and yield models, wildlife models, .silvicultz~ral expert .systems, ,financial models, geogi.aphica1 information systems, and visualization tools. Typically, each of these tools has its own complex inferface and data ,format. To use these tools in combination, a manager must learn each inter;/ace and manually convert data ,from one ,format to another. NED-2 is a robust, intelligent, goal-driven decision support system that integrates tools in each qf these categories. NED-2 uses a blackboard architecture and a set of semi-a14tonomous agents to manage these tools,for the user. Each ugent has the procedural knowlecige needed to operate a class qf c,'eci.sions szipport tools zcsed in,forest ecosystem management. The simulation ugent can set up input,for growth aiirl yield models and knows how to interpret the output,fiom these models. Meta-knowledge bases provide the sirnulation ugent with inforn~ation ahout the data,forma/s and control codes used by difirent growth simu1ator.s. The GIS agent can merge information with a shape,file and knows how to invoke a geographical infirmation system to display the information. The visualization agent can generate inpput ,for stand and landscape. visualization tools. The hlackboai-d system itself includes a powerfi:l agent that integrates a database and a set of Prolog clauses into a single blackboard. The inter:/ace agent provides access to all /he tools in the system through a single user interface. Other agents allow rievelopment of alternative treatment plans; provide analysis of timber, wildlife, water, ecology , and visual goals; and generate a wide variety of reports usejul in,forest management. The agent architecturce is designed to ,facilitate integration o f new third-part,v decision support tools as they become avuiluhle.