Print this Page

Topical Session 3: Predictive Modelling of Resources and Hazards: Reliability and Uncertainty

Conveners: A. G. Fabbri (University of Milano-Bicocca, Italy), E. J. Carranza (State University of Campinas, Brazil)

With the ever increasing availability of personal computer facilities and digital imagery it has become attractive to construct spatial databases. They were mostly focused on mineral and energy exploration but eventually they were directed to the assessment and prediction of natural hazards and consequent risks. Spatial prediction modelling of future discoveries or of hazardous occurrences, however, has proven to be a complex challenge requiring mathematical models, assumptions, scenarios and suitable databases. Furthermore, many applications developed to date show that a great variety of processing strategies employed provide results that are difficult to visualize, interpret and compare. This session aims at debating the reliability, uncertainty and comparability of the results of predictive modelling, the prediction patterns. Hopefully a generalized framework and procedural strategy can be identified and databases shared for exhaustive experimentations. Some examples of issues requiring attentions are: (1) Data mining of hidden supporting evidence; (2) Establishing the “footprints” of a study area; (3) Training areas versus study areas; (4) Effectiveness of prediction patterns: number of classes and their uncertainty; (5) Model complexity versus study area complexity; (6) Continuous field versus categorical spatial evidence; (7) Relevance of spatial resolution in modelling; (8) Standardization of parameters and strategies of processing; (9) Interpretation of prediction patterns in physico-chemical terms or land use requirements; (10) Adjustment of spatial relationships by expert knowledge; (11) Comparison of models and prediction patterns; (12) Advanced software for spatial prediction modelling.

Permanent link to this article: http://www.iamg2018.org/index.php/topical-session-3-predictive-modelling-of-resources-and-hazards-reliability-and-uncertainty/