Methodology and design of a decision support system to predict tree growth response from forest fertilization

Decision support system (DSS) technologies are becoming very important supporting tools for helping people in the decision-making process. DSS have been used in forestry and are evolving very rapidly as foresters are demanding more system functionality to improve forestry management operations along with development of dynamic and user-friendly software to cope with the increased demand for information.; The Intermountain Forest Tree Nutrition Cooperative (IFTNC) at the University of Idaho initiated a DSS project including, as its primary functional part, a real-time expert system prototype for Central Washington that focuses on tree nutrition management. The project's main objective was to design and develop a system methodology that involves a microcomputer program to predict Douglas-fir growth response based on fertilization treatments of 200 or 400 lb of nitrogen per acre during a six-year period. The system methodology began with the definition of the problem and ended with a preliminary design and operation of an integrated microcomputer application prototype.; Nine experts on forest fertilization management issues were interviewed with the purpose to acquire heuristic knowledge they use to conduct fertilization operations and also to obtain their input on how a fertilization-supporting tool could strengthen the decision-making process. In order to reinforce the qualitative information collected from the interviews, quantitative data from different sources were also gathered i.e., data mining. For instance, the IFTNC database was a source for individual tree measurements and site characteristics. Geographic Information Systems (GIS) database related to soil parent materials and potassium level were generalized to include these parameters as input attributes in the data set. Global Positional System (GPS) was used to locate stands and input site-specific conditions.; Interpretation and analysis included the interpretation of qualitative and quantitative information to look for system component functioning and then an analysis of how different components would operate under an integrated environment. To facilitate system understanding, a theoretical fertilization control system input was designed. This framework made it easy to design and operate the logic for various system module components. Also, preliminary definitions of the modeling and system architecture were studied.; Modeling consisted of searching for an appropriate mathematical technique for system prediction. The system mathematical module uses a neural network approach where input/output data pairs on individual tree measurements and physical site characteristics are trained to predict Douglas-fir growth response based on 200 or 400 lb of nitrogen per acre during a six-year span. This component proved to be a quick and robust prediction technique for the prototype under development.; Finally system development primarily dealt with the design and operation of the prototype itself. This system uses distinct software packages within an integrated personal computer environment using Microsoft Visual Basic as the integration development language. Besides the mathematical module, the prototype includes a visualization module implemented with MapObjects that produces maps of geographic stand location and also serves for data interpretation and analysis.; The research project involved designing process that involves a feasible and rapid way of using different tools to deal with forest fertilization management operations. It uses different software components within an integrated computer development environment, resulting in a state-of-the-art fertilization prediction tool. The system supports management decisions, and helps people involved in forest fertilization design and administer fertilization prescriptions in the Intermountain Northwest.