2 0 0 1 2001 Postgraduate Symposium Submissions I I “ Surviving Your Thesis through Humour ”

This paper demonstrates the framework and methodology how weed population dynamics can be predicted using rule-base fuzzy logic as applied to GIS spatial image. Parthenium weed (parthenium hysterophorus L.) infestation in the Central Queensland region poses a serious threat to the environment and economic viability of the infected areas. Government agencies have taken steps to control and manage existing infestation and to curb future spread of this noxious weed. One of the tools used in these strategies is the prediction of parthenium weed population. Conventional weed forecasting methods utilises discrete values in exponential models and linear algorithms extensively. Attempts at predicting weed dispersal relied heavily on accuracy of the original charts or images to yield reasonable results. Using these methods, results of weed population forecasting are only as reliable as the data originally provided. This paper demonstrates that by using GIS spatial image categorised into themes, a fuzzy logic based forecasting methodology can be performed. Fuzzy logic is best suited to this type of problem as its requisite is in handling approximate data. The methodology would allow us to: 1) predict weed population reasonably well based on approximate data, 2) take into consideration additional parameters without re-writing the algorithm, 3) refine largescale forecasts to suit localised situations, 4) allow users to determine and inspect individual infestation factors, and 5) adapt the methodology to other weed or plant species. This paper also