Portugal

Geography

Field surveying during the austral summer of 2007/08 and the analysis of a QuickBird satellite image, resulted on the production of a detailed geomorphological map of Deception Island (South Shetlands, Maritime Antarctic - 1:10 000) and allowed its analysis and spatial modelling of the geomorphological phenomena. The research was conducted in the vicinity of the Argentinean and Spanish Antarctic stations and allowed to continue the previous work by our group, and to provide new geomorphological observations aiming to understand how climate change and volcano activity affect permafrost and the present-day geomorphological dynamics in an area of large environmental sensitivity. The detailed geomorphological mapping was conducted in the area between Cerro Caliente and Crater Lake, as well as in the vicinity of the ruins of the Chilean Refuge. The mapping focused on landforms and deposits reflecting periglacial processes, such as rockfalls, talus slopes, debris cones, gullies, alluvial fans, outwash plains, thermokarst hollows, hummocky terrains, lag surfaces and nivation hollows. Some of these can be used as indicators of permafrost degradation. Observations on the processes involved in the genesis of these phenomena were also conducted. The analysis of the spatial distribution and characteristics of hummocky terrains, lag surfaces and nivation hollows, was complemented by GIS spatial modelling, intending to identify relevant controlling geographical factors. Models of the susceptibility of occurrence of these phenomena were created using two statistical methods: logistical regression, as a multivariate method; and the informative value method as a bivariate method. Success and prediction rate curves were used for model validation. The Area Under the Curve was used to quantify the level of performance and prediction of the models and to allow the comparison between the two methods. Although advantages and disadvantages have been identified in both methods, generally the results were of high quality. Thereby they demonstrated the potential of the models to predict which independent variables have more or less influence in the occurrence of the geomorphological phenomena and also that the models and data were reliable. Thus, regarding the hummocky terrains, the classes of the independent variables with more relevance in their occurrence, were: concave or rectilinear/flat areas; slopes between 8º and 12º; south and southwest aspects; tuff cones and maar deposits; global summer radiation between 1.8 and 2.0 TJ/m2; high values of permeability and Wetness Index. For the lag-surfaces the following classes were highlighted: convex areas; slopes between 14º and 26º; east aspect; tuff cones and maar deposits and strombolian scoria and lavas; medium and low values of permeability. Finally, for the nivation hollows, the controlling classes were: concave areas; slopes between 16º and 20º; southwest, northwest and west aspects; strombolian scoria and lavas; global summer radiation between 1.3 and 1.4 TJ/m2; and high values of permeability and Wetness Index.