作者: Jan Hjort , Joonas Ujanen , Miia Parviainen , Jon Tolgensbakk , Bernd Etzelmüller
DOI: 10.1016/J.GEOMORPH.2013.08.002
关键词:
摘要: Extrapolation potential of statistically-based geomorphological distribution models (GDMs) has not been scrutinized. Here, the possibility to transfer solifluction within and between six study areas in subarctic Arctic environments was examined. A generalized linear model, additive maximum entropy boosted regression tree methods were used analyses. The transferability success GDMs assessed by area under curve a receiver operating characteristic plot. Based on results, slope angle, mean annual air temperature remote sensing based index vegetation abundance most important variables contributing occurrence at landscape scale. In model extrapolation, over half calibrated transferable from one another. topographical conditions had greater effect than climate extrapolation potential. More precisely, it more difficult extrapolate high-relief environment an with moderate topography. On contrary, transferred better environments. conclusion, (i) region specific environmental may significantly affect relative importance GDMs, (ii) certain limitations across areas, (iii) range calibration critical factor for (iv) machine learning-based performed marginally parametric extrapolation. Extensive knowledge about space is needed before can be reliably change explorations.