Search

Remotely sensed data are a key input to GIS-based spatial decision support systems for land cover and land use application areas. One of the major sources of error in the input of processed remotely sensed data to GIS is in the process of classification. Particularly important is the degradation of the data from the interval to nominal level of measurement. This is less significant in cultural landscapes where boundaries predominate, but it becomes an important source of error in natural, and disturbed natural, environments where gradients exist. Use of the Gi* local statistic as an alternative approach to processing remotely sensed data proved very successful, replicating the level of discrimination achieved by conventional classification and field labelling in a much shorter time, whilst avoiding the errors associated with conversion of the data from the interval to nominal level of measurement.

This paper further articulates the role of ubiquitous spatial technologies (e.g. Google Earth) as tools for analyzing, visualizing, and developing policy responses to predicted climate change impacts. Specifically, the efficiency and effectiveness of using the tools in the production of visualizations for the local level is studied. A brief background to climate change response reveals limited data and visualizations at the local level: ubiquitous spatial technologies can potentially fill the void. Case study data including temperature, rainfall and land suitability information from southwest Victoria (Australia) are used to test the hypothesis. The research team produced thirty short visualizations using minimal time, resources and a moderate skill base. The effectiveness of the visualizations was tested on a diverse group of stakeholders. It was found that the visuals provided contextual information and understandings of overarching climate change trends, however, integration with other datasets and higher levels of detail are required if the platform is to be used as a stand alone policy development tool. Moreover, the need to further develop design guidelines to guard against, or at least inform users about visual sensationalism is required.

A novel approach to the combination of a case based reasoning system and an artificial neural network is presented in which the neural network is integrated within the case based reasoning cycle so that its generalizing ability may be harnessed to provide improved case adaptation performance. The ensuing hybrid system has been applied to the task of oceanographic forecasting in a real-time environment and has produced very promising results. After presenting classifications of hybrid artificial intelligence problem-solving methods, the particular combination of case based reasoning and neural networks, as a problem-solving strategy, is discussed in greater depth. The hybrid artificial intelligence forecasting model is then explained and the experimental results obtained from trials at sea are outlined.

An investigation is described into the application of artificial intelligence to forecasting in the domain of oceanography. A hybrid approach to forecasting the thermal structure of the water ahead of a moving vessel is presented which combines the ability of a case-based reasoning system for identifying previously encountered similar situations and the generalizing ability of an artificial neural network to guide the adaptation stage of the case-based reasoning mechanism. The system has been successfully tested in real time in the Atlantic Ocean; the results obtained are presented and compared with those derived from other forecasting methods.

An instance-based problem solving model is presented in which the aim is to forecast, in real time, the physical parameter values of a complex and dynamic environment: the ocean. The situations in which the rules that determine a system are unknown, the prediction of the parameter values that determine the characteristic behaviour of the system can be a problematic task. In such a situation it has been found that an instance-based reasoning system can provide a more effective means of performing such predictions than other connectionist or symbolic techniques. The instance-based reasoning system incorporates a radial basis function artificial neural network for the instance adaptation. The results obtained from experiments, in which the system operated in real time in the oceanographic environment, are presented.