*16th March 2012*
- Title: Coping with uncertainty in spatial environmental modelling
(by Gerard Heuvelink, Wageningen University, The Netherlands)
- Abstract: Uncertainty is all around us, in everyday life as well as in science and research. We may aim for a reduction of uncertainty by collecting more and better data and by improving our understanding and modelling of environmental processes, but it is an illusion to think that uncertainty can be completely eliminated. Instead, we must learn to cope with it and develop and use tools for handling uncertainty. Geoinformation suffers from errors and uncertainties in the position and attribute values of spatial objects, be it cities, roads, water bodies or the soil. Environmental models are only approximations of reality and will always depart from it to some degree. It is important that users are aware that spatial data and models are not error-free and that users are able to determine whether the accuracy of model outcomes is sufficient for the intended purpose. If it is not, then the results of a spatial analysis may lead to poor or even wrong decisions, with costly and irreversible consequences. This presentation starts by defining what uncertainty is and how it differs from error and fuzziness. Next it addresses how uncertainty can be characterised by probability distributions and how this applies to spatial data. Examples of positional and attribute uncertainty are given and used to illustrate various methods to visualise and communicate uncertainty. It is explained how the propagation of uncertainty through spatial environmental models can be analysed and the relative contribution of uncertainty sources quantified. As yet we still do not have an ‘uncertainty-aware GIS’, but important initiatives are underway to assist users with functionalities to handle uncertainties in spatial environmental modelling. One of these is the UncertWeb project, which develops an uncertainty-enabled model web that keeps track of uncertainty propagation in model chains.
*17th March 2012*
- Title: The Case for an Integrated Approach to Usability: Examples of Spatial Tasks (by Sven Bertel, Bauhaus-Universität Weimar, Germany)
- Abstract: Creating and evaluating human-machine interaction that humans can use well often necessitates an integrated stance that takes into account factors across perceptual, cognitive and computational levels. This is particularly true for complex tasks, for tasks with mixed initiative or when control and responsibilities are shared asymmetrically between human and machine partners. This talk will examine selected factors across different levels that matter for the usability of interactive systems. It will address chief methodological and disciplinary challenges that arise from the need of an integrated approach as well as novel perspectives that such an approach offers. A number of spatial tasks and tasks that involve spatial information will be presented as cases in point, along with matching approaches.