Detecting Landscape Changes in High Latitude Environments Using Landsat Trend Analysis: 2. Classification

摘要: Mapping landscape dynamics is necessary to assess cumulative impacts due climate change and development in Arctic regions. Landscape changes produce a range of temporal reflectance trajectories that can be obtained from remote sensing image time-series. these assumes their are unique characterized by magnitude shape. A companion paper this issue describes trajectory visualization method for assessing disturbances. This focusses on generating map using time-series calibrated Landsat Tasseled Cap indices 1985 2011. reference database covering the Mackenzie Delta region was created number ancillary datasets delineate polygons describing 21 natural human-induced Two approaches were tested classify generate maps. The first involved profile matching based shape distance, while second quantified with regression coefficients input decision tree classifier. Results indicate classification robust linear trend performed best. final assessed bootstrapping cross-validation, producing an overall accuracy 82.8% at level classes 87.3% when collapsed eight underlying processes.