Observing lake- and river-ice decay with SAR: advantages and limitations of the unsupervised k-means classification approach

摘要: Large parts of the Arctic are covered by water bodies. Ice covers on lakes and rivers prohibit exchange heat vapor between body atmosphere. With melt onset, ecosystem is subjected to changes, making it important monitor ice decay. As ground-based monitoring these vast uninhabited areas expensive thus restricted a few locations, remote-sensing techniques need be applied. Here we evaluate performance unsupervised k-means classification for dividing fractions river channels from spaceborne radar data in comparison threshold-based methods. The analysis based six TerraSAR-X three RADARSAT-2 images, obtained during spring 2011 over central Lena Delta northern Siberia. found similar fixed-threshold approach. does not prior statistical backscatter analyses account configuration conditions, easier use than threshold method. In addition, that application low-pass filter closing results strongly improves overall results.